external/ton
1
0
Fork 0
mirror of https://github.com/ton-blockchain/ton synced 2025-02-12 11:12:16 +00:00
Code Issues Releases Wiki Activity

[Tolk] Initial commit of TOLK Language: fork all sources from FunC

Browse source 
The Tolk Language will be positioned as "next-generation FunC".
It's literally a fork of a FunC compiler,
introducing familiar syntax similar to TypeScript,
but leaving all low-level optimizations untouched.

Note, that FunC sources are partially stored
in the parser/ folder (shared with TL/B).
In Tolk, nothing is shared.
Everything from parser/ is copied into tolk/ folder.
This commit is contained in:
tolk-vm 2024-10-31 10:51:07 +04:00
parent eed3153ace
commit 82648ebd6a
No known key found for this signature in database
GPG key ID: 7905DD7FE0324B12
43 changed files with 13674 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.github/script/amd64-20.04.Dockerfile vendored
View file

@ -17,4 +17,4 @@ ENV CC clang
ENV CXX clang++
ENV CCACHE_DISABLE 1
RUN cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DPORTABLE=1 -DTON_ARCH= -DCMAKE_CXX_FLAGS="-mavx2" ..
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func validator-engine validator-engine-console create-state generate-random-id create-hardfork dht-server lite-client
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func tolk validator-engine validator-engine-console create-state generate-random-id create-hardfork dht-server lite-client

2
.github/script/amd64-22.04.Dockerfile vendored
View file

@ -17,4 +17,4 @@ ENV CC clang
ENV CXX clang++
ENV CCACHE_DISABLE 1
RUN cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DPORTABLE=1 -DTON_ARCH= -DCMAKE_CXX_FLAGS="-mavx2" ..
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func validator-engine validator-engine-console create-state generate-random-id create-hardfork dht-server lite-client
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func tolk validator-engine validator-engine-console create-state generate-random-id create-hardfork dht-server lite-client

2
.github/script/arm64-20.04.Dockerfile vendored
View file

@ -17,4 +17,4 @@ ENV CC clang
ENV CXX clang++
ENV CCACHE_DISABLE 1
RUN cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DPORTABLE=1 -DTON_ARCH= ..
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func validator-engine validator-engine-console create-state generate-random-id dht-server lite-client
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func tolk validator-engine validator-engine-console create-state generate-random-id dht-server lite-client

2
.github/script/arm64-22.04.Dockerfile vendored
View file

@ -17,4 +17,4 @@ ENV CC clang
ENV CXX clang++
ENV CCACHE_DISABLE 1
RUN cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DPORTABLE=1 -DTON_ARCH= ..
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func validator-engine validator-engine-console create-state generate-random-id dht-server lite-client
RUN ninja storage-daemon storage-daemon-cli tonlibjson blockchain-explorer fift func tolk validator-engine validator-engine-console create-state generate-random-id dht-server lite-client

32
.github/workflows/create-release.yml vendored
View file

@ -167,6 +167,14 @@ jobs:
asset_name: func.exe
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Windows 2019 single artifact - tolk
uses: svenstaro/upload-release-action@v2
with:
repo_token: ${{ secrets.GITHUB_TOKEN }}
file: artifacts/ton-win-binaries/tolk.exe
asset_name: tolk.exe
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Windows 2019 single artifact - lite-client
uses: svenstaro/upload-release-action@v2
with:
@ -257,6 +265,14 @@ jobs:
asset_name: func-mac-x86-64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Mac x86-64 single artifact - tolk
uses: svenstaro/upload-release-action@v2
with:
repo_token: ${{ secrets.GITHUB_TOKEN }}
file: artifacts/ton-x86_64-macos-binaries/tolk
asset_name: tolk-mac-x86-64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Mac x86-64 single artifact - lite-client
uses: svenstaro/upload-release-action@v2
with:
@ -348,6 +364,14 @@ jobs:
asset_name: func-mac-arm64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Mac arm64 single artifact - tolk
uses: svenstaro/upload-release-action@v2
with:
repo_token: ${{ secrets.GITHUB_TOKEN }}
file: artifacts/ton-arm64-macos-binaries/tolk
asset_name: tolk-mac-arm64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Mac arm64 single artifact - lite-client
uses: svenstaro/upload-release-action@v2
with:
@ -438,6 +462,14 @@ jobs:
asset_name: func-linux-x86_64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Linux x86-64 single artifact - tolk
uses: svenstaro/upload-release-action@v2
with:
repo_token: ${{ secrets.GITHUB_TOKEN }}
file: artifacts/ton-x86_64-linux-binaries/tolk
asset_name: tolk-linux-x86_64
tag: ${{ steps.tag.outputs.TAG }}
- name: Upload Linux x86-64 single artifact - lite-client
uses: svenstaro/upload-release-action@v2
with:

1
.github/workflows/ton-arm64-macos.yml vendored
View file

@ -29,6 +29,7 @@ jobs:
artifacts/lite-client -V
artifacts/fift -V
artifacts/func -V
artifacts/tolk -v
- name: Upload artifacts
uses: actions/upload-artifact@master

1
.github/workflows/ton-x86-64-linux.yml vendored
View file

@ -33,6 +33,7 @@ jobs:
artifacts/lite-client -V
artifacts/fift -V
artifacts/func -V
artifacts/tolk -v
- name: Upload artifacts
uses: actions/upload-artifact@master

1
.github/workflows/ton-x86-64-macos.yml vendored
View file

@ -29,6 +29,7 @@ jobs:
artifacts/lite-client -V
artifacts/fift -V
artifacts/func -V
artifacts/tolk -v
- name: Upload artifacts
uses: actions/upload-artifact@master

2
.gitignore vendored
View file

@ -13,6 +13,8 @@ test/regression-tests.cache/
**/*build*/
.idea
.vscode
.DS_Store
dev/
zlib/
libsodium/
libmicrohttpd-0.9.77-w32-bin/

1
CMakeLists.txt
View file

@ -413,6 +413,7 @@ add_subdirectory(adnl)
add_subdirectory(crypto)
add_subdirectory(lite-client)
add_subdirectory(emulator)
add_subdirectory(tolk)
#BEGIN tonlib
add_subdirectory(tonlib)

6
assembly/native/build-macos-portable.sh
View file

@ -153,7 +153,7 @@ test $? -eq 0 || { echo "Can't configure ton"; exit 1; }
if [ "$with_tests" = true ]; then
ninja storage-daemon storage-daemon-cli blockchain-explorer \
tonlib tonlibjson tonlib-cli validator-engine func fift \
tonlib tonlibjson tonlib-cli validator-engine func tolk fift \
lite-client pow-miner validator-engine-console generate-random-id json2tlo dht-server \
http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork tlbc emulator \
test-ed25519 test-ed25519-crypto test-bigint test-vm test-fift test-cells test-smartcont \
@ -162,7 +162,7 @@ if [ "$with_tests" = true ]; then
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
else
ninja storage-daemon storage-daemon-cli blockchain-explorer \
tonlib tonlibjson tonlib-cli validator-engine func fift \
tonlib tonlibjson tonlib-cli validator-engine func tolk fift \
lite-client pow-miner validator-engine-console generate-random-id json2tlo dht-server \
http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork tlbc emulator
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
@ -173,6 +173,7 @@ strip -s storage/storage-daemon/storage-daemon-cli
strip -s blockchain-explorer/blockchain-explorer
strip -s crypto/fift
strip -s crypto/func
strip -s tolk/tolk
strip -s crypto/create-state
strip -s crypto/tlbc
strip -s validator-engine-console/validator-engine-console
@ -197,6 +198,7 @@ if [ "$with_artifacts" = true ]; then
cp build/blockchain-explorer/blockchain-explorer artifacts/
cp build/crypto/fift artifacts/
cp build/crypto/func artifacts/
cp build/tolk/tolk artifacts/
cp build/crypto/create-state artifacts/
cp build/crypto/tlbc artifacts/
cp build/validator-engine-console/validator-engine-console artifacts/

6
assembly/native/build-macos-shared.sh
View file

@ -81,7 +81,7 @@ test $? -eq 0 || { echo "Can't configure ton"; exit 1; }
if [ "$with_tests" = true ]; then
ninja storage-daemon storage-daemon-cli blockchain-explorer \
tonlib tonlibjson tonlib-cli validator-engine func fift \
tonlib tonlibjson tonlib-cli validator-engine func tolk fift \
lite-client pow-miner validator-engine-console generate-random-id json2tlo dht-server \
http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork tlbc emulator \
test-ed25519 test-ed25519-crypto test-bigint test-vm test-fift test-cells test-smartcont \
@ -90,7 +90,7 @@ if [ "$with_tests" = true ]; then
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
else
ninja storage-daemon storage-daemon-cli blockchain-explorer \
tonlib tonlibjson tonlib-cli validator-engine func fift \
tonlib tonlibjson tonlib-cli validator-engine func tolk fift \
lite-client pow-miner validator-engine-console generate-random-id json2tlo dht-server \
http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork tlbc emulator
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
@ -102,6 +102,7 @@ strip -s storage/storage-daemon/storage-daemon-cli
strip -s blockchain-explorer/blockchain-explorer
strip -s crypto/fift
strip -s crypto/func
strip -s tolk/tolk
strip -s crypto/create-state
strip -s crypto/tlbc
strip -s validator-engine-console/validator-engine-console
@ -126,6 +127,7 @@ if [ "$with_artifacts" = true ]; then
cp build/blockchain-explorer/blockchain-explorer artifacts/
cp build/crypto/fift artifacts/
cp build/crypto/func artifacts/
cp build/tolk/tolk artifacts/
cp build/crypto/create-state artifacts/
cp build/crypto/tlbc artifacts/
cp build/validator-engine-console/validator-engine-console artifacts/

7
assembly/native/build-ubuntu-portable.sh
View file

@ -144,7 +144,7 @@ cmake -GNinja .. \
test $? -eq 0 || { echo "Can't configure ton"; exit 1; }
if [ "$with_tests" = true ]; then
ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
ninja storage-daemon storage-daemon-cli fift func tolk tonlib tonlibjson tonlib-cli \
validator-engine lite-client pow-miner validator-engine-console blockchain-explorer \
generate-random-id json2tlo dht-server http-proxy rldp-http-proxy \
adnl-proxy create-state emulator test-ed25519 test-ed25519-crypto test-bigint \
@ -153,7 +153,7 @@ ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
test-fec test-tddb test-db test-validator-session-state test-emulator
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
else
ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
ninja storage-daemon storage-daemon-cli fift func tolk tonlib tonlibjson tonlib-cli \
validator-engine lite-client pow-miner validator-engine-console blockchain-explorer \
generate-random-id json2tlo dht-server http-proxy rldp-http-proxy \
adnl-proxy create-state emulator
@ -166,6 +166,7 @@ strip -s storage/storage-daemon/storage-daemon \
crypto/fift \
crypto/tlbc \
crypto/func \
tolk/tolk \
crypto/create-state \
validator-engine-console/validator-engine-console \
tonlib/tonlib-cli \
@ -195,7 +196,7 @@ if [ "$with_artifacts" = true ]; then
mkdir artifacts
mv build/tonlib/libtonlibjson.so.0.5 build/tonlib/libtonlibjson.so
cp build/storage/storage-daemon/storage-daemon build/storage/storage-daemon/storage-daemon-cli \
build/crypto/fift build/crypto/tlbc build/crypto/func build/crypto/create-state build/blockchain-explorer/blockchain-explorer \
build/crypto/fift build/crypto/tlbc build/crypto/func build/tolk/tolk build/crypto/create-state build/blockchain-explorer/blockchain-explorer \
build/validator-engine-console/validator-engine-console build/tonlib/tonlib-cli \
build/tonlib/libtonlibjson.so build/http/http-proxy build/rldp-http-proxy/rldp-http-proxy \
build/dht-server/dht-server build/lite-client/lite-client build/validator-engine/validator-engine \

7
assembly/native/build-ubuntu-shared.sh
View file

@ -52,7 +52,7 @@ cmake -GNinja -DTON_USE_JEMALLOC=ON .. \
test $? -eq 0 || { echo "Can't configure ton"; exit 1; }
if [ "$with_tests" = true ]; then
ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
ninja storage-daemon storage-daemon-cli fift func tolk tonlib tonlibjson tonlib-cli \
validator-engine lite-client pow-miner validator-engine-console blockchain-explorer \
generate-random-id json2tlo dht-server http-proxy rldp-http-proxy \
adnl-proxy create-state emulator test-ed25519 test-ed25519-crypto test-bigint \
@ -61,7 +61,7 @@ ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
test-fec test-tddb test-db test-validator-session-state test-emulator
test $? -eq 0 || { echo "Can't compile ton"; exit 1; }
else
ninja storage-daemon storage-daemon-cli fift func tonlib tonlibjson tonlib-cli \
ninja storage-daemon storage-daemon-cli fift func tolk tonlib tonlibjson tonlib-cli \
validator-engine lite-client pow-miner validator-engine-console blockchain-explorer \
generate-random-id json2tlo dht-server http-proxy rldp-http-proxy \
adnl-proxy create-state emulator
@ -74,6 +74,7 @@ strip -s storage/storage-daemon/storage-daemon \
crypto/fift \
crypto/tlbc \
crypto/func \
tolk/tolk \
crypto/create-state \
validator-engine-console/validator-engine-console \
tonlib/tonlib-cli \
@ -105,7 +106,7 @@ if [ "$with_artifacts" = true ]; then
mkdir artifacts
mv build/tonlib/libtonlibjson.so.0.5 build/tonlib/libtonlibjson.so
cp build/storage/storage-daemon/storage-daemon build/storage/storage-daemon/storage-daemon-cli \
build/crypto/fift build/crypto/tlbc build/crypto/func build/crypto/create-state build/blockchain-explorer/blockchain-explorer \
build/crypto/fift build/crypto/tlbc build/crypto/func build/tolk/tolk build/crypto/create-state build/blockchain-explorer/blockchain-explorer \
build/validator-engine-console/validator-engine-console build/tonlib/tonlib-cli \
build/tonlib/libtonlibjson.so build/http/http-proxy build/rldp-http-proxy/rldp-http-proxy \
build/dht-server/dht-server build/lite-client/lite-client build/validator-engine/validator-engine \

5
assembly/native/build-windows-2019.bat
View file

@ -155,7 +155,7 @@ IF %errorlevel% NEQ 0 (
)
IF "%1"=="-t" (
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tonlib tonlibjson ^
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tolk tonlib tonlibjson ^
tonlib-cli validator-engine lite-client pow-miner validator-engine-console generate-random-id ^
json2tlo dht-server http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork emulator ^
test-ed25519 test-ed25519-crypto test-bigint test-vm test-fift test-cells test-smartcont test-net ^
@ -166,7 +166,7 @@ IF %errorlevel% NEQ 0 (
exit /b %errorlevel%
)
) else (
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tonlib tonlibjson ^
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tolk tonlib tonlibjson ^
tonlib-cli validator-engine lite-client pow-miner validator-engine-console generate-random-id ^
json2tlo dht-server http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork emulator
IF %errorlevel% NEQ 0 (
@ -204,6 +204,7 @@ build\blockchain-explorer\blockchain-explorer.exe ^
build\crypto\fift.exe ^
build\crypto\tlbc.exe ^
build\crypto\func.exe ^
build\tolk\tolk.exe ^
build\crypto\create-state.exe ^
build\validator-engine-console\validator-engine-console.exe ^
build\tonlib\tonlib-cli.exe ^

5
assembly/native/build-windows.bat
View file

@ -156,7 +156,7 @@ IF %errorlevel% NEQ 0 (
)
IF "%1"=="-t" (
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tonlib tonlibjson ^
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tolk tonlib tonlibjson ^
tonlib-cli validator-engine lite-client pow-miner validator-engine-console generate-random-id ^
json2tlo dht-server http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork emulator ^
test-ed25519 test-ed25519-crypto test-bigint test-vm test-fift test-cells test-smartcont test-net ^
@ -167,7 +167,7 @@ IF %errorlevel% NEQ 0 (
exit /b %errorlevel%
)
) else (
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tonlib tonlibjson ^
ninja storage-daemon storage-daemon-cli blockchain-explorer fift func tolk tonlib tonlibjson ^
tonlib-cli validator-engine lite-client pow-miner validator-engine-console generate-random-id ^
json2tlo dht-server http-proxy rldp-http-proxy adnl-proxy create-state create-hardfork emulator
IF %errorlevel% NEQ 0 (
@ -205,6 +205,7 @@ build\blockchain-explorer\blockchain-explorer.exe ^
build\crypto\fift.exe ^
build\crypto\tlbc.exe ^
build\crypto\func.exe ^
build\tolk\tolk.exe ^
build\crypto\create-state.exe ^
build\validator-engine-console\validator-engine-console.exe ^
build\tonlib\tonlib-cli.exe ^

1
assembly/nix/build-linux-arm64-nix.sh
View file

@ -43,6 +43,7 @@ sudo strip -s storage-daemon \
fift \
tlbc \
func \
tolk \
create-state \
validator-engine-console \
tonlib-cli \

1
assembly/nix/build-linux-x86-64-nix.sh
View file

@ -43,6 +43,7 @@ sudo strip -s storage-daemon \
fift \
tlbc \
func \
tolk \
create-state \
validator-engine-console \
tonlib-cli \

1
assembly/nix/build-macos-nix.sh
View file

@ -43,6 +43,7 @@ sudo strip -xSX storage-daemon \
fift \
tlbc \
func \
tolk \
create-state \
validator-engine-console \
tonlib-cli \

937
crypto/smartcont/mathlib.tolk Normal file
View file

@ -0,0 +1,937 @@
{-
-
- Tolk fixed-point mathematical library
- (initially copied from mathlib.fc)
-
-}
{-
This file is part of TON Tolk Standard Library.
Tolk Standard Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
Tolk Standard Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
-}
{---------------- HIGH-LEVEL FUNCTION DECLARATIONS -----------------}
{-
Most functions declared here work either with integers or with fixed-point numbers of type `fixed248`.
`fixedNNN` informally denotes an alias for type `int` used to represent fixed-point numbers with scale 2^NNN.
Prefix `fixedNNN::` is prepended to the names of high-level functions that accept arguments and return values of type `fixedNNN`.
-}
{- function declarations have been commented out, otherwise they are not inlined by the current Tolk compiler
;; nearest integer to sqrt(a*b) for non-negative integers or fixed-point numbers a and b
int geom_mean(int a, int b) inline_ref;
;; integer square root
int sqrt(int a) inline;
;; fixed-point square root
;; fixed248 sqrt(fixed248 x)
int fixed248::sqrt(int x) inline;
int fixed248::sqr(int x) inline;
const int fixed248::One;
;; log(2) as fixed248
int fixed248::log2_const() inline;
;; Pi as fixed248
int fixed248::Pi_const() inline;
;; fixed248 exp(fixed248 x)
int fixed248::exp(int x) inline_ref;
;; fixed248 exp2(fixed248 x)
int fixed248::exp2(int x) inline_ref;
;; fixed248 log(fixed248 x)
int fixed248::log(int x) inline_ref;
;; fixed248 log2(fixed248 x)
int fixed248::log2(int x) inline;
;; fixed248 pow(fixed248 x, fixed248 y)
int fixed248::pow(int x, int y) inline_ref;
;; (fixed248, fixed248) sincos(fixed248 x);
(int, int) fixed248::sincos(int x) inline_ref;
;; fixed248 sin(fixed248 x);
int fixed248::sin(int x) inline;
;; fixed248 cos(fixed248 x);
int fixed248::cos(int x) inline;
;; fixed248 tan(fixed248 x);
int fixed248::tan(int x) inline_ref;
;; fixed248 cot(fixed248 x);
int fixed248::cot(int x) inline_ref;
;; fixed248 asin(fixed248 x);
int fixed248::asin(int x) inline;
;; fixed248 acos(fixed248 x);
int fixed248::acos(int x) inline;
;; fixed248 atan(fixed248 x);
int fixed248::atan(int x) inline_ref;
;; fixed248 acot(fixed248 x);
int fixed248::acot(int x) inline_ref;
;; random number uniformly distributed in [0..1)
;; fixed248 random();
int fixed248::random() impure inline;
;; random number with standard normal distribution (2100 gas on average)
;; fixed248 nrand();
int fixed248::nrand() impure inline;
;; generates a random number approximately distributed according to the standard normal distribution (1200 gas)
;; (fails chi-squared test, but it is shorter and faster than fixed248::nrand())
;; fixed248 nrand_fast();
int fixed248::nrand_fast() impure inline;
-} ;; end (declarations)
{-------------------- INTERMEDIATE FUNCTIONS -----------------------}
{-
Intermediate functions are used in the implementations of high-level `fixedNNN::...` functions
if necessary, they can be used to define additional high-level functions for other fixed-point types, such as fixed128, outside this library. They can be also used in a hypothetical floating-point Tolk library.
For these reasons, the declarations of these functions are collected here.
-}
{- function declarations have been commented out, otherwise they are not inlined by the current Tolk compiler
;; fixed258 tanh(fixed258 x, int steps);
int tanh_f258(int x, int n);
;; computes exp(x)-1 for |x| <= log(2)/2.
;; fixed257 expm1(fixed257 x);
int expm1_f257(int x);
;; computes (sin(x+xe),-cos(x+xe)) for |x| <= Pi/4, xe very small
;; this function is very accurate, error less than 0.7 ulp (consumes ~ 5500 gas)
;; (fixed256, fixed256) sincosn(fixed256 x, fixed259 xe)
(int, int) sincosn_f256(int x, int xe);
;; compute (sin(x),1-cos(x)) in fixed256 for |x| < 16*atan(1/16) = 0.9987
;; (fixed256, fixed257) sincosm1_f256(fixed256 x);
;; slightly less accurate than sincosn_f256() (error up to 3/2^256), but faster (~ 4k gas) and shorter
(int, int) sincosm1_f256(int x);
;; compute (p, q) such that p/q = tan(x) for |x|<2*atan(1/2)=1899/2048=0.927
;; (int, int) tan_aux(fixed256 x);
(int, int) tan_aux_f256(int x);
;; returns (y, s) such that log(x) = y/2^256 + s*log(2) for positive integer x
;; this function is very precise (error less than 0.6 ulp) and consumes < 7k gas
;; (fixed256, int) log_aux_f256(int x);
(int, int) log_aux_f256(int x);
;; returns (y, s) such that log2(x) = y/2^256 + s for positive integer x
;; this function is very precise (error less than 0.6 ulp) and consumes < 7k gas
;; (fixed256, int) log2_aux_f256(int x);
(int, int) log2_aux_f256(int x);
;; compute (q, z) such that atan(x)=q*atan(1/32)+z for -1 <= x < 1
;; this function is reasonably accurate (error < 7 ulp with ulp = 2^-261), but it consumes >7k gas
;; this is sufficient for most purposes
;; (int, fixed261) atan_aux(fixed256 x)
(int, int) atan_aux_f256(int x);
;; fixed255 atan(fixed255 x);
int atan_f255(int x);
;; for -1 <= x < 1 only
;; fixed256 atan_small(fixed256 x);
int atan_f256_small(int x);
;; fixed255 asin(fixed255 x);
int asin_f255(int x);
;; fixed254 acos(fixed255 x);
int acos_f255(int x);
;; generates normally distributed pseudo-random number
;; fixed252 nrand();
int nrand_f252(int x);
;; a faster and shorter variant of nrand_f252() that fails chi-squared test
;; (should suffice for most purposes)
;; fixed252 nrand_fast();
int nrand_fast_f252(int x);
-} ;; end (declarations)
{---------------- MISSING OPERATIONS AND BUILT-INS -----------------}
int sgn(int x) asm "SGN";
;; compute floor(log2(x))+1
int log2_floor_p1(int x) asm "UBITSIZE";
int mulrshiftr(int x, int y, int s) asm "MULRSHIFTR";
int mulrshiftr256(int x, int y) asm "256 MULRSHIFTR#";
(int, int) mulrshift256mod(int x, int y) asm "256 MULRSHIFT#MOD";
(int, int) mulrshiftr256mod(int x, int y) asm "256 MULRSHIFTR#MOD";
(int, int) mulrshiftr255mod(int x, int y) asm "255 MULRSHIFTR#MOD";
(int, int) mulrshiftr248mod(int x, int y) asm "248 MULRSHIFTR#MOD";
(int, int) mulrshiftr5mod(int x, int y) asm "5 MULRSHIFTR#MOD";
(int, int) mulrshiftr6mod(int x, int y) asm "6 MULRSHIFTR#MOD";
(int, int) mulrshiftr7mod(int x, int y) asm "7 MULRSHIFTR#MOD";
int lshift256divr(int x, int y) asm "256 LSHIFT#DIVR";
(int, int) lshift256divmodr(int x, int y) asm "256 LSHIFT#DIVMODR";
(int, int) lshift255divmodr(int x, int y) asm "255 LSHIFT#DIVMODR";
(int, int) lshift2divmodr(int x, int y) asm "2 LSHIFT#DIVMODR";
(int, int) lshift7divmodr(int x, int y) asm "7 LSHIFT#DIVMODR";
(int, int) lshiftdivmodr(int x, int y, int s) asm "LSHIFTDIVMODR";
(int, int) rshiftr256mod(int x) asm "256 RSHIFTR#MOD";
(int, int) rshiftr248mod(int x) asm "248 RSHIFTR#MOD";
(int, int) rshiftr4mod(int x) asm "4 RSHIFTR#MOD";
(int, int) rshift3mod(int x) asm "3 RSHIFT#MOD";
;; computes y - x (Tolk compiler does not try to use this by itself)
int sub_rev(int x, int y) asm "SUBR";
int nan() asm "PUSHNAN";
int is_nan(int x) asm "ISNAN";
{------------------------ SQUARE ROOTS ----------------------------}
;; computes sqrt(a*b) exactly rounded to the nearest integer
;; for all 0 <= a, b <= 2^256-1
;; may be used with b=1 or b=scale of fixed-point numbers
int geom_mean(int a, int b) inline_ref {
ifnot (min(a, b)) {
return 0;
}
int s = log2_floor_p1(a); ;; throws out of range error if a < 0 or b < 0
int t = log2_floor_p1(b);
;; NB: (a-b)/2+b == (a+b)/2, but without overflow for large a and b
int x = (s == t ? (a - b) / 2 + b : 1 << ((s + t) / 2));
do {
;; if always used with b=2^const, may be optimized to "const LSHIFTDIVC#"
;; it is important to use `muldivc` here, not `muldiv` or `muldivr`
int q = (muldivc(a, b, x) - x) / 2;
x += q;
} until (q == 0);
return x;
}
;; integer square root, computes round(sqrt(a)) for all a>=0.
;; note: `inline` is better than `inline_ref` for such simple functions
int sqrt(int a) inline {
return geom_mean(a, 1);
}
;; version for fixed248 = fixed-point numbers with scale 2^248
;; fixed248 sqrt(fixed248 x)
int fixed248::sqrt(int x) inline {
return geom_mean(x, 1 << 248);
}
;; fixed255 sqrt(fixed255 x)
int fixed255::sqrt(int x) inline {
return geom_mean(x, 1 << 255);
}
;; fixed248 sqr(fixed248 x);
int fixed248::sqr(int x) inline {
return muldivr(x, x, 1 << 248);
}
;; fixed255 sqr(fixed255 x);
int fixed255::sqr(int x) inline {
return muldivr(x, x, 1 << 255);
}
const int fixed248::One = (1 << 248);
const int fixed255::One = (1 << 255);
{-------------------- USEFUL CONSTANTS --------------------}
;; store huge constants in inline_ref functions for reuse
;; (y,z) where y=round(log(2)*2^256), z=round((log(2)*2^256-y)*2^128)
;; then log(2) = y/2^256 + z/2^384
(int, int) log2_xconst_f256() inline_ref {
return (80260960185991308862233904206310070533990667611589946606122867505419956976172, -32272921378999278490133606779486332143);
}
;; (y,z) where Pi = y/2^254 + z/2^382
(int, int) Pi_xconst_f254() inline_ref {
return (90942894222941581070058735694432465663348344332098107489693037779484723616546, 108051869516004014909778934258921521947);
}
;; atan(1/16) as fixed260
int Atan1_16_f260() inline_ref {
return 115641670674223639132965820642403718536242645001775371762318060545014644837101; ;; true value is ...101.0089...
}
;; atan(1/8) as fixed259
int Atan1_8_f259() inline_ref {
return 115194597005316551477397594802136977648153890007566736408151129975021336532841; ;; correction -0.1687...
}
;; atan(1/32) as fixed261
int Atan1_32_f261() inline_ref {
return 115754418570128574501879331591757054405465733718902755858991306434399246026247; ;; correction 0.395...
}
;; inline is better than inline_ref for such very small functions
int log2_const_f256() inline {
(int c, _) = log2_xconst_f256();
return c;
}
int fixed248::log2_const() inline {
return log2_const_f256() ~>> 8;
}
int Pi_const_f254() inline {
(int c, _) = Pi_xconst_f254();
return c;
}
int fixed248::Pi_const() inline {
return Pi_const_f254() ~>> 6;
}
{--------------- HYPERBOLIC TANGENT AND EXPONENT -------------------}
;; hyperbolic tangent of small x via n+2 terms of Lambert's continued fraction
;; n=17: good for |x| < log(2)/4 = 0.173
;; fixed258 tanh_f258(fixed258 x, int n)
int tanh_f258(int x, int n) inline_ref {
int x2 = muldivr(x, x, 1 << 255); ;; x^2 as fixed261
int c = int a = (2 * n + 5) << 250; ;; a=2n+5 as fixed250
int Two = (1 << 251); ;; 2. as fixed250
repeat (n) {
a = (c -= Two) + muldivr(x2, 1 << 239, a); ;; a := 2k+1+x^2/a as fixed250, k=n+1,n,...,2
}
a = (touch(3) << 254) + muldivr(x2, 1 << 243, a); ;; a := 3+x^2/a as fixed254
;; y = x/(1+a') = x - x*a'/(1+a') = x - x*x^2/(a+x^2) where a' = x^2/a
return x - (muldivr(x, x2, a + (x2 ~>> 7)) ~>> 7);
}
;; fixed257 expm1_f257(fixed257 x)
;; computes exp(x)-1 for small x via 19 terms of Lambert's continued fraction for tanh(x/2)
;; good for |x| < log(2)/2 = 0.347 (n=17); consumes ~3500 gas
int expm1_f257(int x) inline_ref {
;; (almost) compute tanh(x/2) first; x/2 as fixed258 = x as fixed257
int x2 = muldivr(x, x, 1 << 255); ;; x^2 as fixed261
int Two = (1 << 251); ;; 2. as fixed250
int c = int a = touch(39) << 250; ;; a=2n+5 as fixed250
repeat (17) {
a = (c -= Two) + muldivr(x2, 1 << 239, a); ;; a := 2k+1+x^2/a as fixed250, k=n+1,n,...,2
}
a = (touch(3) << 254) + muldivr(x2, 1 << 243, a); ;; a := 3+x^2/a as fixed254
;; now tanh(x/2) = x/(1+a') where a'=x^2/a ; apply exp(x)-1=2*tanh(x/2)/(1-tanh(x/2))
int t = (x ~>> 4) - a; ;; t:=x-a as fixed254
return x - muldivr(x2, t / 2, a + mulrshiftr256(x, t) ~/ 4) ~/ 4; ;; x - x^2 * (x-a) / (a + x*(x-a))
}
;; expm1_f257() may be used to implement specific fixed-point exponentials
;; example:
;; fixed248 exp(fixed248 x)
int fixed248::exp(int x) inline_ref {
var (l2c, l2d) = log2_xconst_f256();
;; divide x by log(2) and convert to fixed257
;; (int q, x) = muldivmodr(x, 256, l2c); ;; unfortunately, no such built-in
(int q, x) = lshiftdivmodr(x, l2c, 8);
x = 2 * x - muldivr(q, l2d, 1 << 127);
int y = expm1_f257(x);
;; result is (1 + y) * (2^q) --> ((1 << 257) + y) >> (9 - q)
return (y ~>> (9 - q)) - (-1 << (248 + q));
;; note that (y ~>> (9 - q)) + (1 << (248 + q)) leads to overflow when q=8
}
;; compute 2^x in fixed248
;; fixed248 exp2(fixed248 x)
int fixed248::exp2(int x) inline_ref {
;; (int q, x) = divmodr(x, 1 << 248); ;; no such built-in
(int q, x) = rshiftr248mod(x);
x = muldivr(x, log2_const_f256(), 1 << 247);
int y = expm1_f257(x);
return (y ~>> (9 - q)) - (-1 << (248 + q));
}
{--------------------- TRIGONOMETRIC FUNCTIONS -----------------------}
;; fixed260 tan(fixed260 x);
;; computes tan(x) for small |x|<atan(1/16) via 16 terms of Lambert's continued fraction
int tan_f260_inlined(int x) inline {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed264
int Two = (1 << 251); ;; 2. as fixed250
int c = int a = touch(33) << 250; ;; a=2n+5 as fixed250
repeat (14) {
a = (c -= Two) - muldivr(x2, 1 << 236, a); ;; a := 2k+1-x^2/a as fixed250, k=n+1,n,...,2
}
a = (touch(3) << 254) - muldivr(x2, 1 << 240, a); ;; a := 3-x^2/a as fixed254
;; y = x/(1-a') = x + x*a'/(1-a') = x + x*x^2/(a-x^2) where a' = x^2/a
return x + (muldivr(x / 2, x2, a - (x2 ~>> 10)) ~>> 9);
}
;; fixed260 tan(fixed260 x);
int tan_f260(int x) inline_ref {
return tan_f260_inlined(x);
}
;; fixed258 tan(fixed258 x);
;; computes tan(x) for small |x|<atan(1/4) via 20 terms of Lambert's continued fraction
int tan_f258_inlined(int x) inline {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed260
int Two = (1 << 251); ;; 2. as fixed250
int c = int a = touch(41) << 250; ;; a=2n+5 as fixed250
repeat (18) {
a = (c -= Two) - muldivr(x2, 1 << 240, a); ;; a := 2k+1-x^2/a as fixed250, k=n+1,n,...,2
}
a = (touch(3) << 254) - muldivr(x2, 1 << 244, a); ;; a := 3-x^2/a as fixed254
;; y = x/(1-a') = x + x*a'/(1-a') = x + x*x^2/(a-x^2) where a' = x^2/a
return x + (muldivr(x / 2, x2, a - (x2 ~>> 6)) ~>> 5);
}
;; fixed258 tan(fixed258 x);
int tan_f258(int x) inline_ref {
return tan_f258_inlined(x);
}
;; (fixed259, fixed263) sincosm1(fixed259 x)
;; computes (sin(x), 1-cos(x)) for small |x|<2*atan(1/16)
(int, int) sincosm1_f259_inlined(int x) inline {
int t = tan_f260_inlined(x); ;; t=tan(x/2) as fixed260
int tt = mulrshiftr256(t, t); ;; t^2 as fixed264
int y = tt ~/ 512 + (1 << 255); ;; 1+t^2 as fixed255
;; 2*t/(1+t^2) as fixed259 and 2*t^2/(1+t^2) as fixed263
;; return (muldivr(t, 1 << 255, y), muldivr(tt, 1 << 255, y));
return (t - muldivr(t / 2, tt, y) ~/ 256, tt - muldivr(tt / 2, tt, y) ~/ 256);
}
(int, int) sincosm1_f259(int x) inline_ref {
return sincosm1_f259_inlined(x);
}
;; computes (sin(x+xe),-cos(x+xe)) for |x| <= Pi/4, xe very small
;; this function is very accurate, error less than 0.7 ulp (consumes ~ 5500 gas)
;; (fixed256, fixed256) sincosn(fixed256 x, fixed259 xe)
(int, int) sincosn_f256(int x, int xe) inline_ref {
;; var (q, x1) = muldivmodr(x, 8, Atan1_8_f259()); ;; no muldivmodr() builtin
var (q, x1) = lshift2divmodr(abs(x), Atan1_8_f259()); ;; reduce mod theta where theta=2*atan(1/8)
var (si, co) = sincosm1_f259(x1 * 2 + xe);
var (a, b, c) = (-1, 0, 1);
repeat (q) { ;; (a+b*I) *= (8+I)^2 = 63+16*I
(a, b, c) = (63 * a - 16 * b, 16 * a + 63 * b, 65 * c);
}
;; now a/c = cos(q*theta), b/c = sin(q*theta) exactly(!)
;; compute (a+b*I)*(1-co+si*I)/c
;; (b, a) = (lshift256divr(b, c), lshift256divr(a, c));
(b, int br) = lshift256divmodr(b, c); br = muldivr(br, 128, c);
(a, int ar) = lshift256divmodr(a, c); ar = muldivr(ar, 128, c);
return (sgn(x) * (((mulrshiftr256(b, co) - br) ~/ 16 - mulrshiftr256(a, si)) ~/ 8 - b),
a - ((mulrshiftr256(a, co) - ar) ~/ 16 + mulrshiftr256(b, si)) ~/ 8);
}
;; compute (sin(x),1-cos(x)) in fixed256 for |x| < 16*atan(1/16) = 0.9987
;; (fixed256, fixed257) sincosm1_f256(fixed256 x);
;; slightly less accurate than sincosn_f256() (error up to 3/2^256), but faster (~ 4k gas) and shorter
(int, int) sincosm1_f256(int x) inline_ref {
var (si, co) = sincosm1_f259_inlined(x); ;; compute (sin,1-cos)(x/8) in (fixed259,fixed263)
int r = 7;
repeat (r / 2) {
;; 1-cos(2*x) = 2*sin(x)^2, sin(2*x) = 2*sin(x)*cos(x)
(co, si) = (mulrshiftr256(si, si), si - (mulrshiftr256(si, co) ~>> r));
r -= 2;
}
return (si, co);
}
;; compute (p, q) such that p/q = tan(x) for |x|<2*atan(1/2)=1899/2048=0.927
;; (int, int) tan_aux(fixed256 x);
(int, int) tan_aux_f256(int x) inline_ref {
int t = tan_f258_inlined(x); ;; t=tan(x/4) as fixed258
;; t:=2*t/(1-t^2)=2*(t-t^3/(t^2-1))
int tt = mulrshiftr256(t, t); ;; t^2 as fixed260
t = muldivr(t, tt, tt ~/ 16 + (-1 << 256)) ~/ 16 - t; ;; now t=-tan(x/2) as fixed259
return (t, mulrshiftr256(t, t) ~/ 4 + (-1 << 256)); ;; return (2*t, t^2-1) as fixed256
}
;; sincosm1_f256() and sincosn_f256() may be used to implement trigonometric functions for different fixed-point types
;; example:
;; (fixed248, fixed248) sincos(fixed248 x);
(int, int) fixed248::sincos(int x) inline_ref {
var (Pic, Pid) = Pi_xconst_f254();
;; (int q, x) = muldivmodr(x, 128, Pic); ;; no muldivmodr() builtin
(int q, x) = lshift7divmodr(x, Pic); ;; reduce mod Pi/2
x = 2 * x - muldivr(q, Pid, 1 << 127);
(int si, int co) = sincosm1_f256(x); ;; doesn't make sense to use more accurate sincosn_f256()
co = (1 << 248) - (co ~>> 9);
si ~>>= 8;
repeat (q & 3) {
(si, co) = (co, - si);
}
return (si, co);
}
;; fixed248 sin(fixed248 x);
;; inline is better than inline_ref for such simple functions
int fixed248::sin(int x) inline {
(int si, _) = fixed248::sincos(x);
return si;
}
;; fixed248 cos(fixed248 x);
int fixed248::cos(int x) inline {
(_, int co) = fixed248::sincos(x);
return co;
}
;; similarly, tan_aux_f256() may be used to implement tan() and cot() for specific fixed-point formats
;; fixed248 tan(fixed248 x);
;; not very accurate when |tan(x)| is very large (difficult to do better without floating-point numbers)
;; however, the relative accuracy is approximately 2^-247 in all cases, which is good enough for arguments given up to 2^-249
int fixed248::tan(int x) inline_ref {
var (Pic, Pid) = Pi_xconst_f254();
;; (int q, x) = muldivmodr(x, 128, Pic); ;; no muldivmodr() builtin
(int q, x) = lshift7divmodr(x, Pic); ;; reduce mod Pi/2
x = 2 * x - muldivr(q, Pid, 1 << 127);
var (a, b) = tan_aux_f256(x); ;; now a/b = tan(x')
if (q & 1) {
(a, b) = (b, - a);
}
return muldivr(a, 1 << 248, b); ;; either -b/a or a/b as fixed248
}
;; fixed248 cot(fixed248 x);
int fixed248::cot(int x) inline_ref {
var (Pic, Pid) = Pi_xconst_f254();
(int q, x) = lshift7divmodr(x, Pic); ;; reduce mod Pi/2
x = 2 * x - muldivr(q, Pid, 1 << 127);
var (b, a) = tan_aux_f256(x); ;; now b/a = tan(x')
if (q & 1) {
(a, b) = (b, - a);
}
return muldivr(a, 1 << 248, b); ;; either -b/a or a/b as fixed248
}
{----------------- INVERSE HYPERBOLIC TANGENT AND LOGARITHMS -----------------}
;; inverse hyperbolic tangent of small x, evaluated by means of n terms of the continued fraction
;; valid for |x| < 2^-2.5 ~ 0.18 if n=37 (slightly less accurate with n=36)
;; |x| < 1/8 if n=32; |x| < 2^-3.5 if n=28; |x| < 1/16 if n=25
;; |x| < 2^-4.5 if n=23; |x| < 1/32 if n=21; |x| < 1/64 if n=18
;; fixed258 atanh(fixed258 x);
int atanh_f258(int x, int n) inline_ref {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed260
int One = (1 << 254);
int a = One ~/ n + (1 << 255); ;; a := 2 + 1/n as fixed254
repeat (n - 1) {
;; a := 1 + (1 - x^2 / a)(1 + 1/n) as fixed254
int t = One - muldivr(x2, 1 << 248, a); ;; t := 1 - x^2 / a
a = muldivr(t, n, (int n1 = n - 1)) + One;
n = n1;
}
;; x / (1 - x^2 / a) = x / (1 - d) = x + x * d / (1 - d) for d = x^2 / a
;; int d = muldivr(x2, 1 << 255, a - (x2 ~>> 6)); ;; d/(1-d) = x^2/(a-x^2) as fixed261
;; return x + (mulrshiftr256(x, d) ~>> 5);
return x + muldivr(x, x2 / 2, a - x2 ~/ 64) ~/ 32;
}
;; number of terms n should be chosen as for atanh_f258()
;; fixed261 atanh(fixed261 x);
int atanh_f261_inlined(int x, int n) inline {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed266
int One = (1 << 254);
int a = One ~/ n + (1 << 255); ;; a := 2 + 1/n as fixed254
repeat (n - 1) {
;; a := 1 + (1 - x^2 / a)(1 + 1/n) as fixed254
int t = One - muldivr(x2, 1 << 242, a); ;; t := 1 - x^2 / a
a = muldivr(t, n, (int n1 = n - 1)) + One;
n = n1;
}
;; x / (1 - x^2 / a) = x / (1 - d) = x + x * d / (1 - d) for d = x^2 / a
;; int d = muldivr(x2, 1 << 255, a - (x2 ~>> 12)); ;; d/(1-d) = x^2/(a-x^2) as fixed267
;; return x + (mulrshiftr256(x, d) ~>> 11);
return x + muldivr(x, x2, a - x2 ~/ 4096) ~/ 4096;
}
;; fixed261 atanh(fixed261 x);
int atanh_f261(int x, int n) inline_ref {
return atanh_f261_inlined(x, n);
}
;; returns (y, s) such that log(x) = y/2^257 + s*log(2) for positive integer x
;; (fixed257, int) log_aux(int x)
(int, int) log_aux_f257(int x) inline_ref {
int s = log2_floor_p1(x);
x <<= 256 - s;
int t = touch(-1 << 256);
if ((x >> 249) <= 90) {
;; t~touch();
t >>= 1;
s -= 1;
}
x += t;
int 2x = 2 * x;
int y = lshift256divr(2x, (x >> 1) - t);
;; y = 2x - (mulrshiftr256(2x, y) ~>> 2); ;; this line could improve precision on very rare occasions
return (atanh_f258(y, 36), s);
}
;; computes 33^m for small m
int pow33(int m) inline {
int t = 1;
repeat (m) { t *= 33; }
return t;
}
;; computes 33^m for small 0<=m<=22
;; slightly faster than pow33()
int pow33b(int m) inline {
(int mh, int ml) = m /% 5;
int t = 1;
repeat (ml) { t *= 33; }
repeat (mh) { t *= 33 * 33 * 33 * 33 * 33; }
return t;
}
;; returns (s, q, y) such that log(x) = s*log(2) + q*log(33/32) + y/2^260 for positive integer x
;; (int, int, fixed260) log_auxx_f260(int x);
(int, int, int) log_auxx_f260(int x) inline_ref {
int s = log2_floor_p1(x) - 1;
x <<= 255 - s; ;; rescale to 1 <= x < 2 as fixed255
int t = touch(2873) << 244; ;; ~ (33/32)^11 ~ sqrt(2) as fixed255
int x1 = (x - t) >> 1;
int q = muldivr(x1, 65, x1 + t) + 11; ;; crude approximation to round(log(x)/log(33/32))
;; t = 1; repeat (q) { t *= 33; } ;; t:=33^q, 0<=q<=22
t = pow33b(q);
t <<= (51 - q) * 5; ;; t:=(33/32)^q as fixed255, nearest power of 33/32 to x
x -= t;
int y = lshift256divr(x << 4, (x >> 1) + t); ;; y = (x-t)/(x+t) as fixed261
y = atanh_f261(y, 18); ;; atanh((x-t)/(x+t)) as fixed261, or log(x/t) as fixed260
return (s, q, y);
}
;; returns (y, s) such that log(x) = y/2^256 + s*log(2) for positive integer x
;; this function is very precise (error less than 0.6 ulp) and consumes < 7k gas
;; (fixed256, int) log_aux_f256(int x);
(int, int) log_aux_f256(int x) inline_ref {
var (s, q, y) = log_auxx_f260(x);
var (yh, yl) = rshiftr4mod(y); ;; y ~/% 16 , but Tolk does not optimize this to RSHIFTR#MOD
;; int Log33_32 = 3563114646320977386603103333812068872452913448227778071188132859183498739150; ;; log(33/32) as fixed256
;; int Log33_32_l = -3769; ;; log(33/32) = Log33_32 / 2^256 + Log33_32_l / 2^269
yh += (yl * 512 + q * -3769) ~>> 13; ;; compensation, may be removed if slightly worse accuracy is acceptable
int Log33_32 = 3563114646320977386603103333812068872452913448227778071188132859183498739150; ;; log(33/32) as fixed256
return (yh + q * Log33_32, s);
}
;; returns (y, s) such that log2(x) = y/2^256 + s for positive integer x
;; this function is very precise (error less than 0.6 ulp) and consumes < 7k gas
;; (fixed256, int) log2_aux_f256(int x);
(int, int) log2_aux_f256(int x) inline_ref {
var (s, q, y) = log_auxx_f260(x);
y = lshift256divr(y, log2_const_f256()) ~>> 4; ;; y/log(2) as fixed256
int Log33_32 = 5140487830366106860412008603913034462883915832139695448455767612111363481357; ;; log_2(33/32) as fixed256
;; Log33_32/2^256 happens to be a very precise approximation to log_2(33/32), no compensation required
return (y + q * Log33_32, s);
}
;; functions log_aux_f256() and log2_aux_f256() may be used to implement specific fixed-point instances of log() and log2()
;; fixed248 log(fixed248 x)
int fixed248::log(int x) inline_ref {
var (y, s) = log_aux_f256(x);
return muldivr(s - 248, log2_const_f256(), 1 << 8) + (y ~>> 8);
;; return muldivr(s - 248, 80260960185991308862233904206310070533990667611589946606122867505419956976172, 1 << 8) + (y ~>> 8);
}
;; fixed248 log2(fixed248 x)
int fixed248::log2(int x) inline {
var (y, s) = log2_aux_f256(x);
return ((s - 248) << 248) + (y ~>> 8);
}
;; computes x^y as exp(y*log(x)), x >= 0
;; fixed248 pow(fixed248 x, fixed248 y);
int fixed248::pow(int x, int y) inline_ref {
ifnot (y) {
return 1 << 248; ;; x^0 = 1
}
if (x <= 0) {
int bad = (x | y) < 0;
return 0 >> bad; ;; 0^y = 0 if x=0 and y>=0; "out of range" exception otherwise
}
var (l, s) = log2_aux_f256(x);
s -= 248; ;; log_2(x) = s+l, l is fixed256, 0<=l<1
;; compute (s+l)*y = q+ll
var (q1, r1) = mulrshiftr248mod(s, y); ;; muldivmodr(s, y, 1 << 248)
var (q2, r2) = mulrshift256mod(l, y);
r2 >>= 247;
var (q3, r3) = rshiftr248mod(q2); ;; divmodr(q2, 1 << 248);
var (q, ll) = rshiftr248mod(r1 + r3);
ll = 512 * ll + r2;
q += q1 + q3;
;; now log_2(x^y) = y*log_2(x) = q + ll, ss integer, ll fixed257, -1/2<=ll<1/2
int sq = q + 248;
if (sq <= 0) {
return - (sq == 0); ;; underflow
}
int y = expm1_f257(mulrshiftr256(ll, log2_const_f256()));
return (y ~>> (9 - q)) - (-1 << sq);
}
{--------------------- INVERSE TRIGONOMETRIC FUNCTIONS -------------------}
;; number of terms n should be chosen as for atanh_f258()
;; fixed259 atan(fixed259 x);
int atan_f259(int x, int n) inline_ref {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed262
int One = (1 << 254);
int a = One ~/ n + (1 << 255); ;; a := 2 + 1/n as fixed254
repeat (n - 1) {
;; a := 1 + (1 + x^2 / a)(1 + 1/n) as fixed254
int t = One + muldivr(x2, 1 << 246, a); ;; t := 1 + x^2 / a
a = muldivr(t, n, (int n1 = n - 1)) + One;
n = n1;
}
;; x / (1 + x^2 / a) = x / (1 + d) = x - x * d / (1 + d) = x - x * x^2/(a+x^2) for d = x^2 / a
return x - muldivr(x, x2, a + x2 ~/ 256) ~/ 256;
}
;; number of terms n should be chosen as for atanh_f261()
;; fixed261 atan(fixed261 x);
int atan_f261_inlined(int x, int n) inline {
int x2 = mulrshiftr256(x, x); ;; x^2 as fixed266
int One = (1 << 254);
int a = One ~/ n + (1 << 255); ;; a := 2 + 1/n as fixed254
repeat (n - 1) {
;; a := 1 + (1 + x^2 / a)(1 + 1/n) as fixed254
int t = One + muldivr(x2, 1 << 242, a); ;; t := 1 + x^2 / a
a = muldivr(t, n, (int n1 = n - 1)) + One;
n = n1;
}
;; x / (1 + x^2 / a) = x / (1 + d) = x - x * d / (1 + d) = x - x * x^2/(a+x^2) for d = x^2 / a
return x - muldivr(x, x2, a + x2 ~/ 4096) ~/ 4096;
}
;; fixed261 atan(fixed261 x);
int atan_f261(int x, int n) inline_ref {
return atan_f261_inlined(x, n);
}
;; computes (q,a,b) such that q is approximately atan(x)/atan(1/32) and a+b*I=(1+I/32)^q as fixed255
;; then b/a=atan(q*atan(1/32)) exactly, and (a,b) is almost a unit vector pointing in the direction of (1,x)
;; must have |x|<1.1, x is fixed24
;; (int, fixed255, fixed255) atan_aux_prereduce(fixed24 x);
(int, int, int) atan_aux_prereduce(int x) inline_ref {
int xu = abs(x);
int tc = 7214596; ;; tan(13*theta) as fixed24 where theta=atan(1/32)
int t1 = muldivr(xu - tc, 1 << 88, xu * tc + (1 << 48)); ;; tan(x') as fixed64 where x'=atan(x)-13*theta
;; t1/(3+t1^2) * 3073/32 = x'/3 * 3072/32 = x' / (96/3072) = x' / theta
int q = muldivr(t1 * 3073, 1 << 59, t1 * t1 + (touch(3) << 128)) + 13; ;; approximately round(atan(x)/theta), 0<=q<=25
var (pa, pb) = (33226912, 5232641); ;; (32+I)^5
var (qh, ql) = q /% 5;
var (a, b) = (1 << (5 * (51 - q)), 0); ;; (1/32^q, 0) as fixed255
repeat (ql) { ;; a+b*I *= 32+I
(a, b) = (sub_rev(touch(b), 32 * a), a + 32 * b); ;; same as (32 * a - b, 32 * b + a), but more efficient
}
repeat (qh) { ;; a+b*I *= (32+I)^5 = pa + pb*I
(a, b) = (a * pa - b * pb, a * pb + b * pa);
}
int xs = sgn(x);
return (xs * q, a, xs * b);
}
;; compute (q, z) such that atan(x)=q*atan(1/32)+z for -1 <= x < 1
;; this function is reasonably accurate (error < 7 ulp with ulp = 2^-261), but it consumes >7k gas
;; this is sufficient for most purposes
;; (int, fixed261) atan_aux(fixed256 x)
(int, int) atan_aux_f256(int x) inline_ref {
var (q, a, b) = atan_aux_prereduce(x ~>> 232); ;; convert x to fixed24
;; now b/a = tan(q*atan(1/32)) exactly, where q is near atan(x)/atan(1/32); so b/a is near x
;; compute y = u/v = (a*x-b)/(a+b*x) as fixed261 ; then |y|<0.0167 = 1.07/64 and atan(x)=atan(y)+q*atan(1/32)
var (u, ul) = mulrshiftr256mod(a, x);
u = (ul ~>> 250) + ((u - b) << 6); ;; |u| < 1/32, convert fixed255 -> fixed261
int v = a + mulrshiftr256(b, x); ;; v is scalar product of (a,b) and (1,x), it is approximately in [1..sqrt(2)] as fixed255
int y = muldivr(u, 1 << 255, v); ;; y = u/v as fixed261
int z = atan_f261_inlined(y, 18); ;; z = atan(x)-q*atan(1/32)
return (q, z);
}
;; compute (q, z) such that atan(x)=q*atan(1/32)+z for -1 <= x < 1
;; this function is very accurate (error < 2 ulp), but it consumes >7k gas
;; in most cases, faster function atan_aux_f256() should be used
;; (int, fixed261) atan_auxx(fixed256 x)
(int, int) atan_auxx_f256(int x) inline_ref {
var (q, a, b) = atan_aux_prereduce(x ~>> 232); ;; convert x to fixed24
;; now b/a = tan(q*atan(1/32)) exactly, where q is near atan(x)/atan(1/32); so b/a is near x
;; compute y = (a*x-b)/(a+b*x) as fixed261 ; then |y|<0.0167 = 1.07/64 and atan(x)=atan(y)+q*atan(1/32)
;; use sort of double precision arithmetic for this
var (u, ul) = mulrshiftr256mod(a, x);
ul /= 2;
u -= b; ;; |u| < 1/32 as fixed255
var (v, vl) = mulrshiftr256mod(b, x);
vl /= 2;
v += a; ;; v is scalar product of (a,b) and (1,x), it is approximately in [1..sqrt(2)] as fixed255
;; y = (u + ul*eps) / (v + vl*eps) = u/v + (ul - vl * u/v)/v * eps where eps=1/2^255
var (y, r) = lshift255divmodr(u, v); ;; y = u/v as fixed255
int yl = muldivr(ul + r, 1 << 255, v) - muldivr(vl, y, v); ;; y/2^255 + yl/2^510 represent u/v
y = (yl ~>> 249) + (y << 6); ;; convert y to fixed261
int z = atan_f261_inlined(y, 18); ;; z = atan(x)-q*atan(1/32)
return (q, z);
}
;; consumes ~ 8k gas
;; fixed255 atan(fixed255 x);
int atan_f255(int x) inline_ref {
int s = (x ~>> 256);
touch(x);
if (s) {
x = lshift256divr(-1 << 255, x); ;; x:=-1/x as fixed256
} else {
x *= 2; ;; convert to fixed256
}
var (q, z) = atan_aux_f256(x);
;; now atan(x) = z + q*atan(1/32) + s*(Pi/2), z is fixed261
var (Pi_h, Pi_l) = Pi_xconst_f254(); ;; Pi/2 as fixed255 + fixed383
var (qh, ql) = mulrshiftr6mod (q, Atan1_32_f261());
return qh + s * Pi_h + (z + ql + muldivr(s, Pi_l, 1 << 122)) ~/ 64;
}
;; computes atan(x) for -1 <= x < 1 only
;; fixed256 atan_small(fixed256 x);
int atan_f256_small(int x) inline_ref {
var (q, z) = atan_aux_f256(x);
;; now atan(x) = z + q*atan(1/32), z is fixed261
var (qh, ql) = mulrshiftr5mod (q, Atan1_32_f261());
return qh + (z + ql) ~/ 32;
}
;; fixed255 asin(fixed255 x);
int asin_f255(int x) inline_ref {
int a = fixed255::One - fixed255::sqr(x); ;; a:=1-x^2
ifnot (a) {
return sgn(x) * Pi_const_f254(); ;; Pi/2 or -Pi/2
}
int y = fixed255::sqrt(a); ;; sqrt(1-x^2)
int t = - lshift256divr(x, (-1 << 255) - y); ;; t = x/(1+sqrt(1-x^2)) avoiding overflow
return atan_f256_small(t); ;; asin(x)=2*atan(t)
}
;; fixed254 acos(fixed255 x);
int acos_f255(int x) inline_ref {
int Pi = Pi_const_f254();
if (x == (-1 << 255)) {
return Pi; ;; acos(-1) = Pi
}
Pi /= 2;
int y = fixed255::sqrt(fixed255::One - fixed255::sqr(x)); ;; sqrt(1-x^2)
int t = lshift256divr(x, (-1 << 255) - y); ;; t = -x/(1+sqrt(1-x^2)) avoiding overflow
return Pi + atan_f256_small(t) ~/ 2; ;; acos(x)=Pi/2 + 2*atan(t)
}
;; consumes ~ 10k gas
;; fixed248 asin(fixed248 x)
int fixed248::asin(int x) inline {
return asin_f255(x << 7) ~>> 7;
}
;; consumes ~ 10k gas
;; fixed248 acos(fixed248 x)
int fixed248::acos(int x) inline {
return acos_f255(x << 7) ~>> 6;
}
;; consumes ~ 7500 gas
;; fixed248 atan(fixed248 x);
int fixed248::atan(int x) inline_ref {
int s = (x ~>> 249);
touch(x);
if (s) {
s = sgn(s);
x = lshift256divr(-1 << 248, x); ;; x:=-1/x as fixed256
} else {
x <<= 8; ;; convert to fixed256
}
var (q, z) = atan_aux_f256(x);
;; now atan(x) = z + q*atan(1/32) + s*(Pi/2), z is fixed261
return (z ~/ 64 + s * Pi_const_f254() + muldivr(q, Atan1_32_f261(), 64)) ~/ 128; ;; compute in fixed255, then convert
}
;; fixed248 acot(fixed248 x);
int fixed248::acot(int x) inline_ref {
int s = (x ~>> 249);
touch(x);
if (s) {
x = lshift256divr(-1 << 248, x); ;; x:=-1/x as fixed256
s = 0;
} else {
x <<= 8; ;; convert to fixed256
s = sgn(x);
}
var (q, z) = atan_aux_f256(x);
;; now acot(x) = - z - q*atan(1/32) + s*(Pi/2), z is fixed261
return (s * Pi_const_f254() - z ~/ 64 - muldivr(q, Atan1_32_f261(), 64)) ~/ 128; ;; compute in fixed255, then convert
}
{--------------------- PSEUDO-RANDOM NUMBERS -------------------}
;; random number with standard normal distribution N(0,1)
;; generated by Kinderman--Monahan ratio method modified by J.Leva
;; spends ~ 2k..3k gas on average
;; fixed252 nrand();
int nrand_f252() impure inline_ref {
var (x, s, t, A, B, r0) = (nan(), touch(29483) << 236, touch(-3167) << 239, 12845, 16693, 9043);
;; 4/sqrt(e*Pi) = 1.369 loop iterations on average
do {
var (u, v) = (random() / 16 + 1, muldivr(random() - (1 << 255), 7027, 1 << 16)); ;; fixed252; 7027=ceil(sqrt(8/e)*2^12)
int va = abs(v);
var (u1, v1) = (u - s, va - t); ;; (u - 29483/2^16, abs(v) + 3167/2^13) as fixed252
;; Q := u1^2 + v1 * (A*v1 - B*u1) as fixed252 where A=12845/2^16, B=16693/2^16
int Q = muldivr(u1, u1, 1 << 252) + muldivr(v1, muldivr(v1, A, 1 << 16) - muldivr(u1, B, 1 << 16), 1 << 252);
;; must have 9043 / 2^15 < Q < 9125 / 2^15, otherwise accept if smaller, reject if larger
int Qd = (Q >> 237) - r0;
if ((Qd < 9125 - 9043) & (va / u < 16)) {
x = muldivr(v, 1 << 252, u); ;; x:=v/u as fixed252; reject immediately if |v/u| >= 16
if (Qd >= 0) { ;; immediately accept if Qd < 0
;; rarely taken branch - 0.012 times per call on average
;; check condition v^2 < -4*u^2*log(u), or equivalent condition u < exp(-x^2/4) for x=v/u
int xx = mulrshiftr256(x, x) ~/ 4; ;; x^2/4 as fixed248
int ex = fixed248::exp(- xx) * 16; ;; exp(-x^2/4) as fixed252
if (u > ex) {
x = nan(); ;; condition false, reject
}
}
}
} until (~ is_nan(x));
return x;
}
;; generates a random number approximately distributed according to the standard normal distribution
;; much faster than nrand_f252(), should be suitable for most purposes when only several random numbers are needed
;; fixed252 nrand_fast();
int nrand_fast_f252() impure inline_ref {
int t = touch(-3) << 253; ;; -6. as fixed252
repeat (12) {
t += random() / 16; ;; add together 12 uniformly random numbers
}
return t;
}
;; random number uniformly distributed in [0..1)
;; fixed248 random();
int fixed248::random() impure inline {
return random() >> 8;
}
;; random number with standard normal distribution
;; fixed248 nrand();
int fixed248::nrand() impure inline {
return nrand_f252() ~>> 4;
}
;; generates a random number approximately distributed according to the standard normal distribution
;; fixed248 nrand_fast();
int fixed248::nrand_fast() impure inline {
return nrand_fast_f252() ~>> 4;
}

638
crypto/smartcont/stdlib.tolk Normal file
View file

@ -0,0 +1,638 @@
;; Standard library for Tolk
;; (initially copied from stdlib.fc)
;;
{-
This file is part of TON Tolk Standard Library.
Tolk Standard Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
Tolk Standard Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
-}
{-
# Tuple manipulation primitives
The names and the types are mostly self-explaining.
Note that currently values of atomic type `tuple` can't be cast to composite tuple type (e.g. `[int, cell]`)
and vise versa.
-}
{-
# Lisp-style lists
Lists can be represented as nested 2-elements tuples.
Empty list is conventionally represented as TVM `null` value (it can be obtained by calling [null()]).
For example, tuple `(1, (2, (3, null)))` represents list `[1, 2, 3]`. Elements of a list can be of different types.
-}
;;; Adds an element to the beginning of lisp-style list.
forall X -> tuple cons(X head, tuple tail) asm "CONS";
;;; Extracts the head and the tail of lisp-style list.
forall X -> (X, tuple) uncons(tuple list) asm "UNCONS";
;;; Extracts the tail and the head of lisp-style list.
forall X -> (tuple, X) list_next(tuple list) asm( -> 1 0) "UNCONS";
;;; Returns the head of lisp-style list.
forall X -> X car(tuple list) asm "CAR";
;;; Returns the tail of lisp-style list.
tuple cdr(tuple list) asm "CDR";
;;; Creates tuple with zero elements.
tuple empty_tuple() asm "NIL";
;;; Appends a value `x` to a `Tuple t = (x1, ..., xn)`, but only if the resulting `Tuple t' = (x1, ..., xn, x)`
;;; is of length at most 255. Otherwise throws a type check exception.
forall X -> tuple tpush(tuple t, X value) asm "TPUSH";
forall X -> (tuple, ()) ~tpush(tuple t, X value) asm "TPUSH";
;;; Creates a tuple of length one with given argument as element.
forall X -> [X] single(X x) asm "SINGLE";
;;; Unpacks a tuple of length one
forall X -> X unsingle([X] t) asm "UNSINGLE";
;;; Creates a tuple of length two with given arguments as elements.
forall X, Y -> [X, Y] pair(X x, Y y) asm "PAIR";
;;; Unpacks a tuple of length two
forall X, Y -> (X, Y) unpair([X, Y] t) asm "UNPAIR";
;;; Creates a tuple of length three with given arguments as elements.
forall X, Y, Z -> [X, Y, Z] triple(X x, Y y, Z z) asm "TRIPLE";
;;; Unpacks a tuple of length three
forall X, Y, Z -> (X, Y, Z) untriple([X, Y, Z] t) asm "UNTRIPLE";
;;; Creates a tuple of length four with given arguments as elements.
forall X, Y, Z, W -> [X, Y, Z, W] tuple4(X x, Y y, Z z, W w) asm "4 TUPLE";
;;; Unpacks a tuple of length four
forall X, Y, Z, W -> (X, Y, Z, W) untuple4([X, Y, Z, W] t) asm "4 UNTUPLE";
;;; Returns the first element of a tuple (with unknown element types).
forall X -> X first(tuple t) asm "FIRST";
;;; Returns the second element of a tuple (with unknown element types).
forall X -> X second(tuple t) asm "SECOND";
;;; Returns the third element of a tuple (with unknown element types).
forall X -> X third(tuple t) asm "THIRD";
;;; Returns the fourth element of a tuple (with unknown element types).
forall X -> X fourth(tuple t) asm "3 INDEX";
;;; Returns the first element of a pair tuple.
forall X, Y -> X pair_first([X, Y] p) asm "FIRST";
;;; Returns the second element of a pair tuple.
forall X, Y -> Y pair_second([X, Y] p) asm "SECOND";
;;; Returns the first element of a triple tuple.
forall X, Y, Z -> X triple_first([X, Y, Z] p) asm "FIRST";
;;; Returns the second element of a triple tuple.
forall X, Y, Z -> Y triple_second([X, Y, Z] p) asm "SECOND";
;;; Returns the third element of a triple tuple.
forall X, Y, Z -> Z triple_third([X, Y, Z] p) asm "THIRD";
;;; Push null element (casted to given type)
;;; By the TVM type `Null` Tolk represents absence of a value of some atomic type.
;;; So `null` can actually have any atomic type.
forall X -> X null() asm "PUSHNULL";
;;; Moves a variable [x] to the top of the stack
forall X -> (X, ()) ~impure_touch(X x) impure asm "NOP";
;;; Returns the current Unix time as an Integer
int now() asm "NOW";
;;; Returns the internal address of the current smart contract as a Slice with a `MsgAddressInt`.
;;; If necessary, it can be parsed further using primitives such as [parse_std_addr].
slice my_address() asm "MYADDR";
;;; Returns the balance of the smart contract as a tuple consisting of an int
;;; (balance in nanotoncoins) and a `cell`
;;; (a dictionary with 32-bit keys representing the balance of "extra currencies")
;;; at the start of Computation Phase.
;;; Note that RAW primitives such as [send_raw_message] do not update this field.
[int, cell] get_balance() asm "BALANCE";
;;; Returns the logical time of the current transaction.
int cur_lt() asm "LTIME";
;;; Returns the starting logical time of the current block.
int block_lt() asm "BLOCKLT";
;;; Computes the representation hash of a `cell` [c] and returns it as a 256-bit unsigned integer `x`.
;;; Useful for signing and checking signatures of arbitrary entities represented by a tree of cells.
int cell_hash(cell c) asm "HASHCU";
;;; Computes the hash of a `slice s` and returns it as a 256-bit unsigned integer `x`.
;;; The result is the same as if an ordinary cell containing only data and references from `s` had been created
;;; and its hash computed by [cell_hash].
int slice_hash(slice s) asm "HASHSU";
;;; Computes sha256 of the data bits of `slice` [s]. If the bit length of `s` is not divisible by eight,
;;; throws a cell underflow exception. The hash value is returned as a 256-bit unsigned integer `x`.
int string_hash(slice s) asm "SHA256U";
{-
# Signature checks
-}
;;; Checks the Ed25519-`signature` of a `hash` (a 256-bit unsigned integer, usually computed as the hash of some data)
;;; using [public_key] (also represented by a 256-bit unsigned integer).
;;; The signature must contain at least 512 data bits; only the first 512 bits are used.
;;; The result is `−1` if the signature is valid, `0` otherwise.
;;; Note that `CHKSIGNU` creates a 256-bit slice with the hash and calls `CHKSIGNS`.
;;; That is, if [hash] is computed as the hash of some data, these data are hashed twice,
;;; the second hashing occurring inside `CHKSIGNS`.
int check_signature(int hash, slice signature, int public_key) asm "CHKSIGNU";
;;; Checks whether [signature] is a valid Ed25519-signature of the data portion of `slice data` using `public_key`,
;;; similarly to [check_signature].
;;; If the bit length of [data] is not divisible by eight, throws a cell underflow exception.
;;; The verification of Ed25519 signatures is the standard one,
;;; with sha256 used to reduce [data] to the 256-bit number that is actually signed.
int check_data_signature(slice data, slice signature, int public_key) asm "CHKSIGNS";
{---
# Computation of boc size
The primitives below may be useful for computing storage fees of user-provided data.
-}
;;; Returns `(x, y, z, -1)` or `(null, null, null, 0)`.
;;; Recursively computes the count of distinct cells `x`, data bits `y`, and cell references `z`
;;; in the DAG rooted at `cell` [c], effectively returning the total storage used by this DAG taking into account
;;; the identification of equal cells.
;;; The values of `x`, `y`, and `z` are computed by a depth-first traversal of this DAG,
;;; with a hash table of visited cell hashes used to prevent visits of already-visited cells.
;;; The total count of visited cells `x` cannot exceed non-negative [max_cells];
;;; otherwise the computation is aborted before visiting the `(max_cells + 1)`-st cell and
;;; a zero flag is returned to indicate failure. If [c] is `null`, returns `x = y = z = 0`.
(int, int, int) compute_data_size(cell c, int max_cells) impure asm "CDATASIZE";
;;; Similar to [compute_data_size?], but accepting a `slice` [s] instead of a `cell`.
;;; The returned value of `x` does not take into account the cell that contains the `slice` [s] itself;
;;; however, the data bits and the cell references of [s] are accounted for in `y` and `z`.
(int, int, int) slice_compute_data_size(slice s, int max_cells) impure asm "SDATASIZE";
;;; A non-quiet version of [compute_data_size?] that throws a cell overflow exception (`8`) on failure.
(int, int, int, int) compute_data_size?(cell c, int max_cells) asm "CDATASIZEQ NULLSWAPIFNOT2 NULLSWAPIFNOT";
;;; A non-quiet version of [slice_compute_data_size?] that throws a cell overflow exception (8) on failure.
(int, int, int, int) slice_compute_data_size?(cell c, int max_cells) asm "SDATASIZEQ NULLSWAPIFNOT2 NULLSWAPIFNOT";
;;; Throws an exception with exit_code excno if cond is not 0 (commented since implemented in compilator)
;; () throw_if(int excno, int cond) impure asm "THROWARGIF";
{--
# Debug primitives
Only works for local TVM execution with debug level verbosity
-}
;;; Dumps the stack (at most the top 255 values) and shows the total stack depth.
() dump_stack() impure asm "DUMPSTK";
{-
# Persistent storage save and load
-}
;;; Returns the persistent contract storage cell. It can be parsed or modified with slice and builder primitives later.
cell get_data() asm "c4 PUSH";
;;; Sets `cell` [c] as persistent contract data. You can update persistent contract storage with this primitive.
() set_data(cell c) impure asm "c4 POP";
{-
# Continuation primitives
-}
;;; Usually `c3` has a continuation initialized by the whole code of the contract. It is used for function calls.
;;; The primitive returns the current value of `c3`.
cont get_c3() impure asm "c3 PUSH";
;;; Updates the current value of `c3`. Usually, it is used for updating smart contract code in run-time.
;;; Note that after execution of this primitive the current code
;;; (and the stack of recursive function calls) won't change,
;;; but any other function call will use a function from the new code.
() set_c3(cont c) impure asm "c3 POP";
;;; Transforms a `slice` [s] into a simple ordinary continuation `c`, with `c.code = s` and an empty stack and savelist.
cont bless(slice s) impure asm "BLESS";
{---
# Gas related primitives
-}
;;; Sets current gas limit `gl` to its maximal allowed value `gm`, and resets the gas credit `gc` to zero,
;;; decreasing the value of `gr` by `gc` in the process.
;;; In other words, the current smart contract agrees to buy some gas to finish the current transaction.
;;; This action is required to process external messages, which bring no value (hence no gas) with themselves.
;;;
;;; For more details check [accept_message effects](https://ton.org/docs/#/smart-contracts/accept).
() accept_message() impure asm "ACCEPT";
;;; Sets current gas limit `gl` to the minimum of limit and `gm`, and resets the gas credit `gc` to zero.
;;; If the gas consumed so far (including the present instruction) exceeds the resulting value of `gl`,
;;; an (unhandled) out of gas exception is thrown before setting new gas limits.
;;; Notice that [set_gas_limit] with an argument `limit ≥ 2^63 − 1` is equivalent to [accept_message].
() set_gas_limit(int limit) impure asm "SETGASLIMIT";
;;; Commits the current state of registers `c4` (“persistent data”) and `c5` (“actions”)
;;; so that the current execution is considered “successful” with the saved values even if an exception
;;; in Computation Phase is thrown later.
() commit() impure asm "COMMIT";
;;; Not implemented
;;() buy_gas(int gram) impure asm "BUYGAS";
;;; Computes the amount of gas that can be bought for `amount` nanoTONs,
;;; and sets `gl` accordingly in the same way as [set_gas_limit].
() buy_gas(int amount) impure asm "BUYGAS";
;;; Computes the minimum of two integers [x] and [y].
int min(int x, int y) asm "MIN";
;;; Computes the maximum of two integers [x] and [y].
int max(int x, int y) asm "MAX";
;;; Sorts two integers.
(int, int) minmax(int x, int y) asm "MINMAX";
;;; Computes the absolute value of an integer [x].
int abs(int x) asm "ABS";
{-
# Slice primitives
It is said that a primitive _loads_ some data,
if it returns the data and the remainder of the slice
(so it can also be used as modifying method).
It is said that a primitive _preloads_ some data, if it returns only the data
(it can be used as non-modifying method).
Unless otherwise stated, loading and preloading primitives read the data from a prefix of the slice.
-}
;;; Converts a `cell` [c] into a `slice`. Notice that [c] must be either an ordinary cell,
;;; or an exotic cell (see [TVM.pdf](https://ton-blockchain.github.io/docs/tvm.pdf), 3.1.2)
;;; which is automatically loaded to yield an ordinary cell `c'`, converted into a `slice` afterwards.
slice begin_parse(cell c) asm "CTOS";
;;; Checks if [s] is empty. If not, throws an exception.
() end_parse(slice s) impure asm "ENDS";
;;; Loads the first reference from the slice.
(slice, cell) load_ref(slice s) asm( -> 1 0) "LDREF";
;;; Preloads the first reference from the slice.
cell preload_ref(slice s) asm "PLDREF";
{- Functions below are commented because are implemented on compilator level for optimisation -}
;;; Loads a signed [len]-bit integer from a slice [s].
;; (slice, int) ~load_int(slice s, int len) asm(s len -> 1 0) "LDIX";
;;; Loads an unsigned [len]-bit integer from a slice [s].
;; (slice, int) ~load_uint(slice s, int len) asm( -> 1 0) "LDUX";
;;; Preloads a signed [len]-bit integer from a slice [s].
;; int preload_int(slice s, int len) asm "PLDIX";
;;; Preloads an unsigned [len]-bit integer from a slice [s].
;; int preload_uint(slice s, int len) asm "PLDUX";
;;; Loads the first `0 ≤ len ≤ 1023` bits from slice [s] into a separate `slice s''`.
;; (slice, slice) load_bits(slice s, int len) asm(s len -> 1 0) "LDSLICEX";
;;; Preloads the first `0 ≤ len ≤ 1023` bits from slice [s] into a separate `slice s''`.
;; slice preload_bits(slice s, int len) asm "PLDSLICEX";
;;; Loads serialized amount of TonCoins (any unsigned integer up to `2^120 - 1`).
(slice, int) load_grams(slice s) asm( -> 1 0) "LDGRAMS";
(slice, int) load_coins(slice s) asm( -> 1 0) "LDGRAMS";
;;; Returns all but the first `0 ≤ len ≤ 1023` bits of `slice` [s].
slice skip_bits(slice s, int len) asm "SDSKIPFIRST";
(slice, ()) ~skip_bits(slice s, int len) asm "SDSKIPFIRST";
;;; Returns the first `0 ≤ len ≤ 1023` bits of `slice` [s].
slice first_bits(slice s, int len) asm "SDCUTFIRST";
;;; Returns all but the last `0 ≤ len ≤ 1023` bits of `slice` [s].
slice skip_last_bits(slice s, int len) asm "SDSKIPLAST";
(slice, ()) ~skip_last_bits(slice s, int len) asm "SDSKIPLAST";
;;; Returns the last `0 ≤ len ≤ 1023` bits of `slice` [s].
slice slice_last(slice s, int len) asm "SDCUTLAST";
;;; Loads a dictionary `D` (HashMapE) from `slice` [s].
;;; (returns `null` if `nothing` constructor is used).
(slice, cell) load_dict(slice s) asm( -> 1 0) "LDDICT";
;;; Preloads a dictionary `D` from `slice` [s].
cell preload_dict(slice s) asm "PLDDICT";
;;; Loads a dictionary as [load_dict], but returns only the remainder of the slice.
slice skip_dict(slice s) asm "SKIPDICT";
;;; Loads (Maybe ^Cell) from `slice` [s].
;;; In other words loads 1 bit and if it is true
;;; loads first ref and return it with slice remainder
;;; otherwise returns `null` and slice remainder
(slice, cell) load_maybe_ref(slice s) asm( -> 1 0) "LDOPTREF";
;;; Preloads (Maybe ^Cell) from `slice` [s].
cell preload_maybe_ref(slice s) asm "PLDOPTREF";
;;; Returns the depth of `cell` [c].
;;; If [c] has no references, then return `0`;
;;; otherwise the returned value is one plus the maximum of depths of cells referred to from [c].
;;; If [c] is a `null` instead of a cell, returns zero.
int cell_depth(cell c) asm "CDEPTH";
{-
# Slice size primitives
-}
;;; Returns the number of references in `slice` [s].
int slice_refs(slice s) asm "SREFS";
;;; Returns the number of data bits in `slice` [s].
int slice_bits(slice s) asm "SBITS";
;;; Returns both the number of data bits and the number of references in `slice` [s].
(int, int) slice_bits_refs(slice s) asm "SBITREFS";
;;; Checks whether a `slice` [s] is empty (i.e., contains no bits of data and no cell references).
int slice_empty?(slice s) asm "SEMPTY";
;;; Checks whether `slice` [s] has no bits of data.
int slice_data_empty?(slice s) asm "SDEMPTY";
;;; Checks whether `slice` [s] has no references.
int slice_refs_empty?(slice s) asm "SREMPTY";
;;; Returns the depth of `slice` [s].
;;; If [s] has no references, then returns `0`;
;;; otherwise the returned value is one plus the maximum of depths of cells referred to from [s].
int slice_depth(slice s) asm "SDEPTH";
{-
# Builder size primitives
-}
;;; Returns the number of cell references already stored in `builder` [b]
int builder_refs(builder b) asm "BREFS";
;;; Returns the number of data bits already stored in `builder` [b].
int builder_bits(builder b) asm "BBITS";
;;; Returns the depth of `builder` [b].
;;; If no cell references are stored in [b], then returns 0;
;;; otherwise the returned value is one plus the maximum of depths of cells referred to from [b].
int builder_depth(builder b) asm "BDEPTH";
{-
# Builder primitives
It is said that a primitive _stores_ a value `x` into a builder `b`
if it returns a modified version of the builder `b'` with the value `x` stored at the end of it.
It can be used as non-modifying method.
All the primitives below first check whether there is enough space in the `builder`,
and only then check the range of the value being serialized.
-}
;;; Creates a new empty `builder`.
builder begin_cell() asm "NEWC";
;;; Converts a `builder` into an ordinary `cell`.
cell end_cell(builder b) asm "ENDC";
;;; Stores a reference to `cell` [c] into `builder` [b].
builder store_ref(builder b, cell c) asm(c b) "STREF";
;;; Stores an unsigned [len]-bit integer `x` into `b` for `0 ≤ len ≤ 256`.
;; builder store_uint(builder b, int x, int len) asm(x b len) "STUX";
;;; Stores a signed [len]-bit integer `x` into `b` for` 0 ≤ len ≤ 257`.
;; builder store_int(builder b, int x, int len) asm(x b len) "STIX";
;;; Stores `slice` [s] into `builder` [b]
builder store_slice(builder b, slice s) asm "STSLICER";
;;; Stores (serializes) an integer [x] in the range `0..2^120 − 1` into `builder` [b].
;;; The serialization of [x] consists of a 4-bit unsigned big-endian integer `l`,
;;; which is the smallest integer `l ≥ 0`, such that `x < 2^8l`,
;;; followed by an `8l`-bit unsigned big-endian representation of [x].
;;; If [x] does not belong to the supported range, a range check exception is thrown.
;;;
;;; Store amounts of TonCoins to the builder as VarUInteger 16
builder store_grams(builder b, int x) asm "STGRAMS";
builder store_coins(builder b, int x) asm "STGRAMS";
;;; Stores dictionary `D` represented by `cell` [c] or `null` into `builder` [b].
;;; In other words, stores a `1`-bit and a reference to [c] if [c] is not `null` and `0`-bit otherwise.
builder store_dict(builder b, cell c) asm(c b) "STDICT";
;;; Stores (Maybe ^Cell) to builder:
;;; if cell is null store 1 zero bit
;;; otherwise store 1 true bit and ref to cell
builder store_maybe_ref(builder b, cell c) asm(c b) "STOPTREF";
{-
# Address manipulation primitives
The address manipulation primitives listed below serialize and deserialize values according to the following TL-B scheme:
```TL-B
addr_none$00 = MsgAddressExt;
addr_extern$01 len:(## 8) external_address:(bits len)
= MsgAddressExt;
anycast_info$_ depth:(#<= 30) { depth >= 1 }
rewrite_pfx:(bits depth) = Anycast;
addr_std$10 anycast:(Maybe Anycast)
workchain_id:int8 address:bits256 = MsgAddressInt;
addr_var$11 anycast:(Maybe Anycast) addr_len:(## 9)
workchain_id:int32 address:(bits addr_len) = MsgAddressInt;
_ _:MsgAddressInt = MsgAddress;
_ _:MsgAddressExt = MsgAddress;
int_msg_info$0 ihr_disabled:Bool bounce:Bool bounced:Bool
src:MsgAddress dest:MsgAddressInt
value:CurrencyCollection ihr_fee:Grams fwd_fee:Grams
created_lt:uint64 created_at:uint32 = CommonMsgInfoRelaxed;
ext_out_msg_info$11 src:MsgAddress dest:MsgAddressExt
created_lt:uint64 created_at:uint32 = CommonMsgInfoRelaxed;
```
A deserialized `MsgAddress` is represented by a tuple `t` as follows:
- `addr_none` is represented by `t = (0)`,
i.e., a tuple containing exactly one integer equal to zero.
- `addr_extern` is represented by `t = (1, s)`,
where slice `s` contains the field `external_address`. In other words, `
t` is a pair (a tuple consisting of two entries), containing an integer equal to one and slice `s`.
- `addr_std` is represented by `t = (2, u, x, s)`,
where `u` is either a `null` (if `anycast` is absent) or a slice `s'` containing `rewrite_pfx` (if anycast is present).
Next, integer `x` is the `workchain_id`, and slice `s` contains the address.
- `addr_var` is represented by `t = (3, u, x, s)`,
where `u`, `x`, and `s` have the same meaning as for `addr_std`.
-}
;;; Loads from slice [s] the only prefix that is a valid `MsgAddress`,
;;; and returns both this prefix `s'` and the remainder `s''` of [s] as slices.
(slice, slice) load_msg_addr(slice s) asm( -> 1 0) "LDMSGADDR";
;;; Decomposes slice [s] containing a valid `MsgAddress` into a `tuple t` with separate fields of this `MsgAddress`.
;;; If [s] is not a valid `MsgAddress`, a cell deserialization exception is thrown.
tuple parse_addr(slice s) asm "PARSEMSGADDR";
;;; Parses slice [s] containing a valid `MsgAddressInt` (usually a `msg_addr_std`),
;;; applies rewriting from the anycast (if present) to the same-length prefix of the address,
;;; and returns both the workchain and the 256-bit address as integers.
;;; If the address is not 256-bit, or if [s] is not a valid serialization of `MsgAddressInt`,
;;; throws a cell deserialization exception.
(int, int) parse_std_addr(slice s) asm "REWRITESTDADDR";
;;; A variant of [parse_std_addr] that returns the (rewritten) address as a slice [s],
;;; even if it is not exactly 256 bit long (represented by a `msg_addr_var`).
(int, slice) parse_var_addr(slice s) asm "REWRITEVARADDR";
{-
# Dictionary primitives
-}
;;; Sets the value associated with [key_len]-bit key signed index in dictionary [dict] to [value] (cell),
;;; and returns the resulting dictionary.
cell idict_set_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTISETREF";
(cell, ()) ~idict_set_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTISETREF";
;;; Sets the value associated with [key_len]-bit key unsigned index in dictionary [dict] to [value] (cell),
;;; and returns the resulting dictionary.
cell udict_set_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTUSETREF";
(cell, ()) ~udict_set_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTUSETREF";
cell idict_get_ref(cell dict, int key_len, int index) asm(index dict key_len) "DICTIGETOPTREF";
(cell, int) idict_get_ref?(cell dict, int key_len, int index) asm(index dict key_len) "DICTIGETREF" "NULLSWAPIFNOT";
(cell, int) udict_get_ref?(cell dict, int key_len, int index) asm(index dict key_len) "DICTUGETREF" "NULLSWAPIFNOT";
(cell, cell) idict_set_get_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTISETGETOPTREF";
(cell, cell) udict_set_get_ref(cell dict, int key_len, int index, cell value) asm(value index dict key_len) "DICTUSETGETOPTREF";
(cell, int) idict_delete?(cell dict, int key_len, int index) asm(index dict key_len) "DICTIDEL";
(cell, int) udict_delete?(cell dict, int key_len, int index) asm(index dict key_len) "DICTUDEL";
(slice, int) idict_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTIGET" "NULLSWAPIFNOT";
(slice, int) udict_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTUGET" "NULLSWAPIFNOT";
(cell, slice, int) idict_delete_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTIDELGET" "NULLSWAPIFNOT";
(cell, slice, int) udict_delete_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTUDELGET" "NULLSWAPIFNOT";
(cell, (slice, int)) ~idict_delete_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTIDELGET" "NULLSWAPIFNOT";
(cell, (slice, int)) ~udict_delete_get?(cell dict, int key_len, int index) asm(index dict key_len) "DICTUDELGET" "NULLSWAPIFNOT";
cell udict_set(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTUSET";
(cell, ()) ~udict_set(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTUSET";
cell idict_set(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTISET";
(cell, ()) ~idict_set(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTISET";
cell dict_set(cell dict, int key_len, slice index, slice value) asm(value index dict key_len) "DICTSET";
(cell, ()) ~dict_set(cell dict, int key_len, slice index, slice value) asm(value index dict key_len) "DICTSET";
(cell, int) udict_add?(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTUADD";
(cell, int) udict_replace?(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTUREPLACE";
(cell, int) idict_add?(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTIADD";
(cell, int) idict_replace?(cell dict, int key_len, int index, slice value) asm(value index dict key_len) "DICTIREPLACE";
cell udict_set_builder(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTUSETB";
(cell, ()) ~udict_set_builder(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTUSETB";
cell idict_set_builder(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTISETB";
(cell, ()) ~idict_set_builder(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTISETB";
cell dict_set_builder(cell dict, int key_len, slice index, builder value) asm(value index dict key_len) "DICTSETB";
(cell, ()) ~dict_set_builder(cell dict, int key_len, slice index, builder value) asm(value index dict key_len) "DICTSETB";
(cell, int) udict_add_builder?(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTUADDB";
(cell, int) udict_replace_builder?(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTUREPLACEB";
(cell, int) idict_add_builder?(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTIADDB";
(cell, int) idict_replace_builder?(cell dict, int key_len, int index, builder value) asm(value index dict key_len) "DICTIREPLACEB";
(cell, int, slice, int) udict_delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTUREMMIN" "NULLSWAPIFNOT2";
(cell, (int, slice, int)) ~udict::delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTUREMMIN" "NULLSWAPIFNOT2";
(cell, int, slice, int) idict_delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTIREMMIN" "NULLSWAPIFNOT2";
(cell, (int, slice, int)) ~idict::delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTIREMMIN" "NULLSWAPIFNOT2";
(cell, slice, slice, int) dict_delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTREMMIN" "NULLSWAPIFNOT2";
(cell, (slice, slice, int)) ~dict::delete_get_min(cell dict, int key_len) asm(-> 0 2 1 3) "DICTREMMIN" "NULLSWAPIFNOT2";
(cell, int, slice, int) udict_delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTUREMMAX" "NULLSWAPIFNOT2";
(cell, (int, slice, int)) ~udict::delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTUREMMAX" "NULLSWAPIFNOT2";
(cell, int, slice, int) idict_delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTIREMMAX" "NULLSWAPIFNOT2";
(cell, (int, slice, int)) ~idict::delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTIREMMAX" "NULLSWAPIFNOT2";
(cell, slice, slice, int) dict_delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTREMMAX" "NULLSWAPIFNOT2";
(cell, (slice, slice, int)) ~dict::delete_get_max(cell dict, int key_len) asm(-> 0 2 1 3) "DICTREMMAX" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_min?(cell dict, int key_len) asm (-> 1 0 2) "DICTUMIN" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_max?(cell dict, int key_len) asm (-> 1 0 2) "DICTUMAX" "NULLSWAPIFNOT2";
(int, cell, int) udict_get_min_ref?(cell dict, int key_len) asm (-> 1 0 2) "DICTUMINREF" "NULLSWAPIFNOT2";
(int, cell, int) udict_get_max_ref?(cell dict, int key_len) asm (-> 1 0 2) "DICTUMAXREF" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_min?(cell dict, int key_len) asm (-> 1 0 2) "DICTIMIN" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_max?(cell dict, int key_len) asm (-> 1 0 2) "DICTIMAX" "NULLSWAPIFNOT2";
(int, cell, int) idict_get_min_ref?(cell dict, int key_len) asm (-> 1 0 2) "DICTIMINREF" "NULLSWAPIFNOT2";
(int, cell, int) idict_get_max_ref?(cell dict, int key_len) asm (-> 1 0 2) "DICTIMAXREF" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_next?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTUGETNEXT" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_nexteq?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTUGETNEXTEQ" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_prev?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTUGETPREV" "NULLSWAPIFNOT2";
(int, slice, int) udict_get_preveq?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTUGETPREVEQ" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_next?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTIGETNEXT" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_nexteq?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTIGETNEXTEQ" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_prev?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTIGETPREV" "NULLSWAPIFNOT2";
(int, slice, int) idict_get_preveq?(cell dict, int key_len, int pivot) asm(pivot dict key_len -> 1 0 2) "DICTIGETPREVEQ" "NULLSWAPIFNOT2";
;;; Creates an empty dictionary, which is actually a null value. Equivalent to PUSHNULL
cell new_dict() asm "NEWDICT";
;;; Checks whether a dictionary is empty. Equivalent to cell_null?.
int dict_empty?(cell c) asm "DICTEMPTY";
{- Prefix dictionary primitives -}
(slice, slice, slice, int) pfxdict_get?(cell dict, int key_len, slice key) asm(key dict key_len) "PFXDICTGETQ" "NULLSWAPIFNOT2";
(cell, int) pfxdict_set?(cell dict, int key_len, slice key, slice value) asm(value key dict key_len) "PFXDICTSET";
(cell, int) pfxdict_delete?(cell dict, int key_len, slice key) asm(key dict key_len) "PFXDICTDEL";
;;; Returns the value of the global configuration parameter with integer index `i` as a `cell` or `null` value.
cell config_param(int x) asm "CONFIGOPTPARAM";
;;; Checks whether c is a null. Note, that Tolk also has polymorphic null? built-in.
int cell_null?(cell c) asm "ISNULL";
;;; Creates an output action which would reserve exactly amount nanotoncoins (if mode = 0), at most amount nanotoncoins (if mode = 2), or all but amount nanotoncoins (if mode = 1 or mode = 3), from the remaining balance of the account. It is roughly equivalent to creating an outbound message carrying amount nanotoncoins (or b − amount nanotoncoins, where b is the remaining balance) to oneself, so that the subsequent output actions would not be able to spend more money than the remainder. Bit +2 in mode means that the external action does not fail if the specified amount cannot be reserved; instead, all remaining balance is reserved. Bit +8 in mode means `amount <- -amount` before performing any further actions. Bit +4 in mode means that amount is increased by the original balance of the current account (before the compute phase), including all extra currencies, before performing any other checks and actions. Currently, amount must be a non-negative integer, and mode must be in the range 0..15.
() raw_reserve(int amount, int mode) impure asm "RAWRESERVE";
;;; Similar to raw_reserve, but also accepts a dictionary extra_amount (represented by a cell or null) with extra currencies. In this way currencies other than TonCoin can be reserved.
() raw_reserve_extra(int amount, cell extra_amount, int mode) impure asm "RAWRESERVEX";
;;; Sends a raw message contained in msg, which should contain a correctly serialized object Message X, with the only exception that the source address is allowed to have dummy value addr_none (to be automatically replaced with the current smart contract address), and ihr_fee, fwd_fee, created_lt and created_at fields can have arbitrary values (to be rewritten with correct values during the action phase of the current transaction). Integer parameter mode contains the flags. Currently mode = 0 is used for ordinary messages; mode = 128 is used for messages that are to carry all the remaining balance of the current smart contract (instead of the value originally indicated in the message); mode = 64 is used for messages that carry all the remaining value of the inbound message in addition to the value initially indicated in the new message (if bit 0 is not set, the gas fees are deducted from this amount); mode' = mode + 1 means that the sender wants to pay transfer fees separately; mode' = mode + 2 means that any errors arising while processing this message during the action phase should be ignored. Finally, mode' = mode + 32 means that the current account must be destroyed if its resulting balance is zero. This flag is usually employed together with +128.
() send_raw_message(cell msg, int mode) impure asm "SENDRAWMSG";
;;; Creates an output action that would change this smart contract code to that given by cell new_code. Notice that this change will take effect only after the successful termination of the current run of the smart contract
() set_code(cell new_code) impure asm "SETCODE";
;;; Generates a new pseudo-random unsigned 256-bit integer x. The algorithm is as follows: if r is the old value of the random seed, considered as a 32-byte array (by constructing the big-endian representation of an unsigned 256-bit integer), then its sha512(r) is computed; the first 32 bytes of this hash are stored as the new value r' of the random seed, and the remaining 32 bytes are returned as the next random value x.
int random() impure asm "RANDU256";
;;; Generates a new pseudo-random integer z in the range 0..range−1 (or range..−1, if range < 0). More precisely, an unsigned random value x is generated as in random; then z := x * range / 2^256 is computed.
int rand(int range) impure asm "RAND";
;;; Returns the current random seed as an unsigned 256-bit Integer.
int get_seed() impure asm "RANDSEED";
;;; Sets the random seed to unsigned 256-bit seed.
() set_seed(int) impure asm "SETRAND";
;;; Mixes unsigned 256-bit integer x into the random seed r by setting the random seed to sha256 of the concatenation of two 32-byte strings: the first with the big-endian representation of the old seed r, and the second with the big-endian representation of x.
() randomize(int x) impure asm "ADDRAND";
;;; Equivalent to randomize(cur_lt());.
() randomize_lt() impure asm "LTIME" "ADDRAND";
;;; Checks whether the data parts of two slices coinside
int equal_slice_bits (slice a, slice b) asm "SDEQ";
;;; Concatenates two builders
builder store_builder(builder to, builder from) asm "STBR";

51
tolk/CMakeLists.txt Normal file
View file

@ -0,0 +1,51 @@
cmake_minimum_required(VERSION 3.5 FATAL_ERROR)
set(TOLK_SOURCE
srcread.cpp
lexer.cpp
symtable.cpp
keywords.cpp
unify-types.cpp
parse-tolk.cpp
abscode.cpp
gen-abscode.cpp
analyzer.cpp
asmops.cpp
builtins.cpp
stack-transform.cpp
optimize.cpp
codegen.cpp
tolk.cpp
)
add_executable(tolk tolk-main.cpp ${TOLK_SOURCE})
target_include_directories(tolk PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}>)
target_link_libraries(tolk PUBLIC git ton_crypto) # todo replace with ton_crypto_core in the future
if (WINGETOPT_FOUND)
target_link_libraries_system(tolk wingetopt)
endif ()
if (USE_EMSCRIPTEN)
add_executable(tolkfiftlib tolk-wasm.cpp ${TOLK_SOURCE})
target_include_directories(tolkfiftlib PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}>)
target_link_libraries(tolkfiftlib PUBLIC fift-lib git)
target_link_options(tolkfiftlib PRIVATE
-sEXPORTED_RUNTIME_METHODS=FS,ccall,cwrap,UTF8ToString,stringToUTF8,lengthBytesUTF8,addFunction,removeFunction,setValue
-sEXPORTED_FUNCTIONS=_tolk_compile,_version,_malloc,_free,_setThrew
-sEXPORT_NAME=CompilerModule
-sERROR_ON_UNDEFINED_SYMBOLS=0
-sFILESYSTEM=1 -lnodefs.js
-Oz
-sIGNORE_MISSING_MAIN=1
-sAUTO_NATIVE_LIBRARIES=0
-sMODULARIZE=1
-sTOTAL_MEMORY=33554432
-sALLOW_MEMORY_GROWTH=1
-sALLOW_TABLE_GROWTH=1
--embed-file ${CMAKE_CURRENT_SOURCE_DIR}/../crypto/fift/lib@/fiftlib
-fexceptions
)
target_compile_options(tolkfiftlib PRIVATE -fexceptions -fno-stack-protector)
endif ()
install(TARGETS tolk RUNTIME DESTINATION bin)

526
tolk/abscode.cpp Normal file
View file

@ -0,0 +1,526 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* ABSTRACT CODE
*
*/
TmpVar::TmpVar(var_idx_t _idx, int _cls, TypeExpr* _type, SymDef* sym, const SrcLocation* loc)
: v_type(_type), idx(_idx), cls(_cls), coord(0) {
if (sym) {
name = sym->sym_idx;
sym->value->idx = _idx;
}
if (loc) {
where = std::make_unique<SrcLocation>(*loc);
}
if (!_type) {
v_type = TypeExpr::new_hole();
}
if (cls == _Named) {
undefined = true;
}
}
void TmpVar::set_location(const SrcLocation& loc) {
if (where) {
*where = loc;
} else {
where = std::make_unique<SrcLocation>(loc);
}
}
void TmpVar::dump(std::ostream& os) const {
show(os);
os << " : " << v_type << " (width ";
v_type->show_width(os);
os << ")";
if (coord > 0) {
os << " = _" << (coord >> 8) << '.' << (coord & 255);
} else if (coord < 0) {
int n = (~coord >> 8), k = (~coord & 0xff);
if (k) {
os << " = (_" << n << ".._" << (n + k - 1) << ")";
} else {
os << " = ()";
}
}
os << std::endl;
}
void TmpVar::show(std::ostream& os, int omit_idx) const {
if (cls & _Named) {
os << symbols.get_name(name);
if (omit_idx && (omit_idx >= 2 || (cls & _UniqueName))) {
return;
}
}
os << '_' << idx;
}
std::ostream& operator<<(std::ostream& os, const TmpVar& var) {
var.show(os);
return os;
}
void VarDescr::show_value(std::ostream& os) const {
if (val & _Int) {
os << 'i';
}
if (val & _Const) {
os << 'c';
}
if (val & _Zero) {
os << '0';
}
if (val & _NonZero) {
os << '!';
}
if (val & _Pos) {
os << '>';
}
if (val & _Neg) {
os << '<';
}
if (val & _Bool) {
os << 'B';
}
if (val & _Bit) {
os << 'b';
}
if (val & _Even) {
os << 'E';
}
if (val & _Odd) {
os << 'O';
}
if (val & _Finite) {
os << 'f';
}
if (val & _Nan) {
os << 'N';
}
if (int_const.not_null()) {
os << '=' << int_const;
}
}
void VarDescr::show(std::ostream& os, const char* name) const {
if (flags & _Last) {
os << '*';
}
if (flags & _Unused) {
os << '?';
}
if (name) {
os << name;
}
os << '_' << idx;
show_value(os);
}
void VarDescr::set_const(long long value) {
return set_const(td::make_refint(value));
}
void VarDescr::set_const(td::RefInt256 value) {
int_const = std::move(value);
if (!int_const->signed_fits_bits(257)) {
int_const.write().invalidate();
}
val = _Const | _Int;
int s = sgn(int_const);
if (s < -1) {
val |= _Nan | _NonZero;
} else if (s < 0) {
val |= _NonZero | _Neg | _Finite;
if (*int_const == -1) {
val |= _Bool;
}
} else if (s > 0) {
val |= _NonZero | _Pos | _Finite;
} else if (!s) {
//if (*int_const == 1) {
// val |= _Bit;
//}
val |= _Zero | _Neg | _Pos | _Finite | _Bool | _Bit;
}
if (val & _Finite) {
val |= int_const->get_bit(0) ? _Odd : _Even;
}
}
void VarDescr::set_const(std::string value) {
str_const = value;
val = _Const;
}
void VarDescr::set_const_nan() {
set_const(td::make_refint());
}
void VarDescr::operator|=(const VarDescr& y) {
val &= y.val;
if (is_int_const() && y.is_int_const() && cmp(int_const, y.int_const) != 0) {
val &= ~_Const;
}
if (!(val & _Const)) {
int_const.clear();
}
}
void VarDescr::operator&=(const VarDescr& y) {
val |= y.val;
if (y.int_const.not_null() && int_const.is_null()) {
int_const = y.int_const;
}
}
void VarDescr::set_value(const VarDescr& y) {
val = y.val;
int_const = y.int_const;
}
void VarDescr::set_value(VarDescr&& y) {
val = y.val;
int_const = std::move(y.int_const);
}
void VarDescr::clear_value() {
val = 0;
int_const.clear();
}
void VarDescrList::show(std::ostream& os) const {
if (unreachable) {
os << "<unreachable> ";
}
os << "[";
for (const auto& v : list) {
os << ' ' << v;
}
os << " ]\n";
}
void Op::flags_set_clear(int set, int clear) {
flags = (flags | set) & ~clear;
for (auto& op : block0) {
op.flags_set_clear(set, clear);
}
for (auto& op : block1) {
op.flags_set_clear(set, clear);
}
}
void Op::split_vars(const std::vector<TmpVar>& vars) {
split_var_list(left, vars);
split_var_list(right, vars);
for (auto& op : block0) {
op.split_vars(vars);
}
for (auto& op : block1) {
op.split_vars(vars);
}
}
void Op::split_var_list(std::vector<var_idx_t>& var_list, const std::vector<TmpVar>& vars) {
int new_size = 0, changes = 0;
for (var_idx_t v : var_list) {
int c = vars.at(v).coord;
if (c < 0) {
++changes;
new_size += (~c & 0xff);
} else {
++new_size;
}
}
if (!changes) {
return;
}
std::vector<var_idx_t> new_var_list;
new_var_list.reserve(new_size);
for (var_idx_t v : var_list) {
int c = vars.at(v).coord;
if (c < 0) {
int n = (~c >> 8), k = (~c & 0xff);
while (k-- > 0) {
new_var_list.push_back(n++);
}
} else {
new_var_list.push_back(v);
}
}
var_list = std::move(new_var_list);
}
void Op::show(std::ostream& os, const std::vector<TmpVar>& vars, std::string pfx, int mode) const {
if (mode & 2) {
os << pfx << " [";
for (const auto& v : var_info.list) {
os << ' ';
if (v.flags & VarDescr::_Last) {
os << '*';
}
if (v.flags & VarDescr::_Unused) {
os << '?';
}
os << vars[v.idx];
if (mode & 4) {
os << ':';
v.show_value(os);
}
}
os << " ]\n";
}
std::string dis = disabled() ? "<disabled> " : "";
if (noreturn()) {
dis += "<noret> ";
}
if (!is_pure()) {
dis += "<impure> ";
}
switch (cl) {
case _Undef:
os << pfx << dis << "???\n";
break;
case _Nop:
os << pfx << dis << "NOP\n";
break;
case _Call:
os << pfx << dis << "CALL: ";
show_var_list(os, left, vars);
os << " := " << (fun_ref ? fun_ref->name() : "(null)") << " ";
if ((mode & 4) && args.size() == right.size()) {
show_var_list(os, args, vars);
} else {
show_var_list(os, right, vars);
}
os << std::endl;
break;
case _CallInd:
os << pfx << dis << "CALLIND: ";
show_var_list(os, left, vars);
os << " := EXEC ";
show_var_list(os, right, vars);
os << std::endl;
break;
case _Let:
os << pfx << dis << "LET ";
show_var_list(os, left, vars);
os << " := ";
show_var_list(os, right, vars);
os << std::endl;
break;
case _Tuple:
os << pfx << dis << "MKTUPLE ";
show_var_list(os, left, vars);
os << " := ";
show_var_list(os, right, vars);
os << std::endl;
break;
case _UnTuple:
os << pfx << dis << "UNTUPLE ";
show_var_list(os, left, vars);
os << " := ";
show_var_list(os, right, vars);
os << std::endl;
break;
case _IntConst:
os << pfx << dis << "CONST ";
show_var_list(os, left, vars);
os << " := " << int_const << std::endl;
break;
case _SliceConst:
os << pfx << dis << "SCONST ";
show_var_list(os, left, vars);
os << " := " << str_const << std::endl;
break;
case _Import:
os << pfx << dis << "IMPORT ";
show_var_list(os, left, vars);
os << std::endl;
break;
case _Return:
os << pfx << dis << "RETURN ";
show_var_list(os, left, vars);
os << std::endl;
break;
case _GlobVar:
os << pfx << dis << "GLOBVAR ";
show_var_list(os, left, vars);
os << " := " << (fun_ref ? fun_ref->name() : "(null)") << std::endl;
break;
case _SetGlob:
os << pfx << dis << "SETGLOB ";
os << (fun_ref ? fun_ref->name() : "(null)") << " := ";
show_var_list(os, right, vars);
os << std::endl;
break;
case _Repeat:
os << pfx << dis << "REPEAT ";
show_var_list(os, left, vars);
os << ' ';
show_block(os, block0.get(), vars, pfx, mode);
os << std::endl;
break;
case _If:
os << pfx << dis << "IF ";
show_var_list(os, left, vars);
os << ' ';
show_block(os, block0.get(), vars, pfx, mode);
os << " ELSE ";
show_block(os, block1.get(), vars, pfx, mode);
os << std::endl;
break;
case _While:
os << pfx << dis << "WHILE ";
show_var_list(os, left, vars);
os << ' ';
show_block(os, block0.get(), vars, pfx, mode);
os << " DO ";
show_block(os, block1.get(), vars, pfx, mode);
os << std::endl;
break;
case _Until:
os << pfx << dis << "UNTIL ";
show_var_list(os, left, vars);
os << ' ';
show_block(os, block0.get(), vars, pfx, mode);
os << std::endl;
break;
case _Again:
os << pfx << dis << "AGAIN ";
show_var_list(os, left, vars);
os << ' ';
show_block(os, block0.get(), vars, pfx, mode);
os << std::endl;
break;
default:
os << pfx << dis << "<???" << cl << "> ";
show_var_list(os, left, vars);
os << " -- ";
show_var_list(os, right, vars);
os << std::endl;
break;
}
}
void Op::show_var_list(std::ostream& os, const std::vector<var_idx_t>& idx_list,
const std::vector<TmpVar>& vars) const {
if (!idx_list.size()) {
os << "()";
} else if (idx_list.size() == 1) {
os << vars.at(idx_list[0]);
} else {
os << "(" << vars.at(idx_list[0]);
for (std::size_t i = 1; i < idx_list.size(); i++) {
os << "," << vars.at(idx_list[i]);
}
os << ")";
}
}
void Op::show_var_list(std::ostream& os, const std::vector<VarDescr>& list, const std::vector<TmpVar>& vars) const {
auto n = list.size();
if (!n) {
os << "()";
} else {
os << "( ";
for (std::size_t i = 0; i < list.size(); i++) {
if (i) {
os << ", ";
}
if (list[i].is_unused()) {
os << '?';
}
os << vars.at(list[i].idx) << ':';
list[i].show_value(os);
}
os << " )";
}
}
void Op::show_block(std::ostream& os, const Op* block, const std::vector<TmpVar>& vars, std::string pfx, int mode) {
os << "{" << std::endl;
std::string pfx2 = pfx + " ";
for (const Op& op : block) {
op.show(os, vars, pfx2, mode);
}
os << pfx << "}";
}
void CodeBlob::flags_set_clear(int set, int clear) {
for (auto& op : ops) {
op.flags_set_clear(set, clear);
}
}
std::ostream& operator<<(std::ostream& os, const CodeBlob& code) {
code.print(os);
return os;
}
// flags: +1 = show variable definition locations; +2 = show vars after each op; +4 = show var abstract value info after each op; +8 = show all variables at start
void CodeBlob::print(std::ostream& os, int flags) const {
os << "CODE BLOB: " << var_cnt << " variables, " << in_var_cnt << " input\n";
if ((flags & 8) != 0) {
for (const auto& var : vars) {
var.dump(os);
if (var.where && (flags & 1) != 0) {
var.where->show(os);
os << " defined here:\n";
var.where->show_context(os);
}
}
}
os << "------- BEGIN --------\n";
for (const auto& op : ops) {
op.show(os, vars, "", flags);
}
os << "-------- END ---------\n\n";
}
var_idx_t CodeBlob::create_var(int cls, TypeExpr* var_type, SymDef* sym, const SrcLocation* location) {
vars.emplace_back(var_cnt, cls, var_type, sym, location);
if (sym) {
sym->value->idx = var_cnt;
}
return var_cnt++;
}
bool CodeBlob::import_params(FormalArgList arg_list) {
if (var_cnt || in_var_cnt || op_cnt) {
return false;
}
std::vector<var_idx_t> list;
for (const auto& par : arg_list) {
TypeExpr* arg_type;
SymDef* arg_sym;
SrcLocation arg_loc;
std::tie(arg_type, arg_sym, arg_loc) = par;
list.push_back(create_var(arg_sym ? (TmpVar::_In | TmpVar::_Named) : TmpVar::_In, arg_type, arg_sym, &arg_loc));
}
emplace_back(loc, Op::_Import, list);
in_var_cnt = var_cnt;
return true;
}
} // namespace tolk

916
tolk/analyzer.cpp Normal file
View file

@ -0,0 +1,916 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* ANALYZE AND PREPROCESS ABSTRACT CODE
*
*/
void CodeBlob::simplify_var_types() {
for (TmpVar& var : vars) {
TypeExpr::remove_indirect(var.v_type);
var.v_type->recompute_width();
}
}
int CodeBlob::split_vars(bool strict) {
int n = var_cnt, changes = 0;
for (int j = 0; j < var_cnt; j++) {
TmpVar& var = vars[j];
if (strict && var.v_type->minw != var.v_type->maxw) {
throw ParseError{var.where.get(), "variable does not have fixed width, cannot manipulate it"};
}
std::vector<TypeExpr*> comp_types;
int k = var.v_type->extract_components(comp_types);
tolk_assert(k <= 254 && n <= 0x7fff00);
tolk_assert((unsigned)k == comp_types.size());
if (k != 1) {
var.coord = ~((n << 8) + k);
for (int i = 0; i < k; i++) {
auto v = create_var(vars[j].cls, comp_types[i], 0, vars[j].where.get());
tolk_assert(v == n + i);
tolk_assert(vars[v].idx == v);
vars[v].name = vars[j].name;
vars[v].coord = ((int)j << 8) + i + 1;
}
n += k;
++changes;
} else if (strict && var.v_type->minw != 1) {
throw ParseError{var.where.get(),
"cannot work with variable or variable component of width greater than one"};
}
}
if (!changes) {
return 0;
}
for (auto& op : ops) {
op.split_vars(vars);
}
return changes;
}
bool CodeBlob::compute_used_code_vars() {
VarDescrList empty_var_info;
return compute_used_code_vars(ops, empty_var_info, true);
}
bool CodeBlob::compute_used_code_vars(std::unique_ptr<Op>& ops_ptr, const VarDescrList& var_info, bool edit) const {
tolk_assert(ops_ptr);
if (!ops_ptr->next) {
tolk_assert(ops_ptr->cl == Op::_Nop);
return ops_ptr->set_var_info(var_info);
}
// here and below, bitwise | (not logical ||) are used to execute both left and right parts
return static_cast<int>(compute_used_code_vars(ops_ptr->next, var_info, edit)) |
static_cast<int>(ops_ptr->compute_used_vars(*this, edit));
}
bool operator==(const VarDescrList& x, const VarDescrList& y) {
if (x.size() != y.size()) {
return false;
}
for (std::size_t i = 0; i < x.size(); i++) {
if (x.list[i].idx != y.list[i].idx || x.list[i].flags != y.list[i].flags) {
return false;
}
}
return true;
}
bool same_values(const VarDescr& x, const VarDescr& y) {
if (x.val != y.val || x.int_const.is_null() != y.int_const.is_null()) {
return false;
}
if (x.int_const.not_null() && cmp(x.int_const, y.int_const) != 0) {
return false;
}
return true;
}
bool same_values(const VarDescrList& x, const VarDescrList& y) {
if (x.size() != y.size()) {
return false;
}
for (std::size_t i = 0; i < x.size(); i++) {
if (x.list[i].idx != y.list[i].idx || !same_values(x.list[i], y.list[i])) {
return false;
}
}
return true;
}
bool Op::set_var_info(const VarDescrList& new_var_info) {
if (var_info == new_var_info) {
return false;
}
var_info = new_var_info;
return true;
}
bool Op::set_var_info(VarDescrList&& new_var_info) {
if (var_info == new_var_info) {
return false;
}
var_info = std::move(new_var_info);
return true;
}
bool Op::set_var_info_except(const VarDescrList& new_var_info, const std::vector<var_idx_t>& var_list) {
if (!var_list.size()) {
return set_var_info(new_var_info);
}
VarDescrList tmp_info{new_var_info};
tmp_info -= var_list;
return set_var_info(tmp_info);
}
bool Op::set_var_info_except(VarDescrList&& new_var_info, const std::vector<var_idx_t>& var_list) {
if (var_list.size()) {
new_var_info -= var_list;
}
return set_var_info(std::move(new_var_info));
}
std::vector<var_idx_t> sort_unique_vars(const std::vector<var_idx_t>& var_list) {
std::vector<var_idx_t> vars{var_list}, unique_vars;
std::sort(vars.begin(), vars.end());
vars.erase(std::unique(vars.begin(), vars.end()), vars.end());
return vars;
}
VarDescr* VarDescrList::operator[](var_idx_t idx) {
auto it = std::lower_bound(list.begin(), list.end(), idx);
return it != list.end() && it->idx == idx ? &*it : nullptr;
}
const VarDescr* VarDescrList::operator[](var_idx_t idx) const {
auto it = std::lower_bound(list.begin(), list.end(), idx);
return it != list.end() && it->idx == idx ? &*it : nullptr;
}
std::size_t VarDescrList::count(const std::vector<var_idx_t> idx_list) const {
std::size_t res = 0;
for (var_idx_t idx : idx_list) {
if (operator[](idx)) {
++res;
}
}
return res;
}
std::size_t VarDescrList::count_used(const std::vector<var_idx_t> idx_list) const {
std::size_t res = 0;
for (var_idx_t idx : idx_list) {
auto v = operator[](idx);
if (v && !v->is_unused()) {
++res;
}
}
return res;
}
VarDescrList& VarDescrList::operator-=(var_idx_t idx) {
auto it = std::lower_bound(list.begin(), list.end(), idx);
if (it != list.end() && it->idx == idx) {
list.erase(it);
}
return *this;
}
VarDescrList& VarDescrList::operator-=(const std::vector<var_idx_t>& idx_list) {
for (var_idx_t idx : idx_list) {
*this -= idx;
}
return *this;
}
VarDescrList& VarDescrList::add_var(var_idx_t idx, bool unused) {
auto it = std::lower_bound(list.begin(), list.end(), idx);
if (it == list.end() || it->idx != idx) {
list.emplace(it, idx, VarDescr::_Last | (unused ? VarDescr::_Unused : 0));
} else if (it->is_unused() && !unused) {
it->clear_unused();
}
return *this;
}
VarDescrList& VarDescrList::add_vars(const std::vector<var_idx_t>& idx_list, bool unused) {
for (var_idx_t idx : idx_list) {
add_var(idx, unused);
}
return *this;
}
VarDescr& VarDescrList::add(var_idx_t idx) {
auto it = std::lower_bound(list.begin(), list.end(), idx);
if (it == list.end() || it->idx != idx) {
it = list.emplace(it, idx);
}
return *it;
}
VarDescr& VarDescrList::add_newval(var_idx_t idx) {
auto it = std::lower_bound(list.begin(), list.end(), idx);
if (it == list.end() || it->idx != idx) {
return *list.emplace(it, idx);
} else {
it->clear_value();
return *it;
}
}
VarDescrList& VarDescrList::clear_last() {
for (auto& var : list) {
if (var.flags & VarDescr::_Last) {
var.flags &= ~VarDescr::_Last;
}
}
return *this;
}
VarDescrList VarDescrList::operator+(const VarDescrList& y) const {
VarDescrList res;
auto it1 = list.cbegin();
auto it2 = y.list.cbegin();
while (it1 != list.cend() && it2 != y.list.cend()) {
if (it1->idx < it2->idx) {
res.list.push_back(*it1++);
} else if (it1->idx > it2->idx) {
res.list.push_back(*it2++);
} else {
res.list.push_back(*it1++);
res.list.back() += *it2++;
}
}
while (it1 != list.cend()) {
res.list.push_back(*it1++);
}
while (it2 != y.list.cend()) {
res.list.push_back(*it2++);
}
return res;
}
VarDescrList& VarDescrList::operator+=(const VarDescrList& y) {
return *this = *this + y;
}
VarDescrList VarDescrList::operator|(const VarDescrList& y) const {
if (y.unreachable) {
return *this;
}
if (unreachable) {
return y;
}
VarDescrList res;
auto it1 = list.cbegin();
auto it2 = y.list.cbegin();
while (it1 != list.cend() && it2 != y.list.cend()) {
if (it1->idx < it2->idx) {
it1++;
} else if (it1->idx > it2->idx) {
it2++;
} else {
res.list.push_back(*it1++);
res.list.back() |= *it2++;
}
}
return res;
}
VarDescrList& VarDescrList::operator|=(const VarDescrList& y) {
if (y.unreachable) {
return *this;
} else {
return *this = *this | y;
}
}
VarDescrList& VarDescrList::operator&=(const VarDescrList& values) {
for (const VarDescr& vd : values.list) {
VarDescr* item = operator[](vd.idx);
if (item) {
*item &= vd;
}
}
unreachable |= values.unreachable;
return *this;
}
VarDescrList& VarDescrList::import_values(const VarDescrList& values) {
if (values.unreachable) {
set_unreachable();
} else
for (auto& vd : list) {
auto new_vd = values[vd.idx];
if (new_vd) {
vd.set_value(*new_vd);
} else {
vd.clear_value();
}
}
return *this;
}
bool Op::std_compute_used_vars(bool disabled) {
// left = OP right
// var_info := (var_info - left) + right
VarDescrList new_var_info{next->var_info};
new_var_info -= left;
new_var_info.clear_last();
if (args.size() == right.size() && !disabled) {
for (const VarDescr& arg : args) {
new_var_info.add_var(arg.idx, arg.is_unused());
}
} else {
new_var_info.add_vars(right, disabled);
}
return set_var_info(std::move(new_var_info));
}
bool Op::compute_used_vars(const CodeBlob& code, bool edit) {
tolk_assert(next);
const VarDescrList& next_var_info = next->var_info;
if (cl == _Nop) {
return set_var_info_except(next_var_info, left);
}
switch (cl) {
case _IntConst:
case _SliceConst:
case _GlobVar:
case _Call:
case _CallInd:
case _Tuple:
case _UnTuple: {
// left = EXEC right;
if (!next_var_info.count_used(left) && is_pure()) {
// all variables in `left` are not needed
if (edit) {
disable();
}
return std_compute_used_vars(true);
}
return std_compute_used_vars();
}
case _SetGlob: {
// GLOB = right
if (right.empty() && edit) {
disable();
}
return std_compute_used_vars(right.empty());
}
case _Let: {
// left = right
std::size_t cnt = next_var_info.count_used(left);
tolk_assert(left.size() == right.size());
auto l_it = left.cbegin(), r_it = right.cbegin();
VarDescrList new_var_info{next_var_info};
new_var_info -= left;
new_var_info.clear_last();
std::vector<var_idx_t> new_left, new_right;
for (; l_it < left.cend(); ++l_it, ++r_it) {
if (std::find(l_it + 1, left.cend(), *l_it) == left.cend()) {
auto p = next_var_info[*l_it];
new_var_info.add_var(*r_it, edit && (!p || p->is_unused()));
new_left.push_back(*l_it);
new_right.push_back(*r_it);
}
}
if (new_left.size() < left.size()) {
left = std::move(new_left);
right = std::move(new_right);
}
if (!cnt && edit) {
// all variables in `left` are not needed
disable();
}
return set_var_info(std::move(new_var_info));
}
case _Return: {
// return left
if (var_info.count(left) == left.size()) {
return false;
}
std::vector<var_idx_t> unique_vars = sort_unique_vars(left);
var_info.list.clear();
for (var_idx_t i : unique_vars) {
var_info.list.emplace_back(i, VarDescr::_Last);
}
return true;
}
case _Import: {
// import left
std::vector<var_idx_t> unique_vars = sort_unique_vars(left);
var_info.list.clear();
for (var_idx_t i : unique_vars) {
var_info.list.emplace_back(i, next_var_info[i] ? 0 : VarDescr::_Last);
}
return true;
}
case _If: {
// if (left) then block0 else block1
// VarDescrList nx_var_info = next_var_info;
// nx_var_info.clear_last();
code.compute_used_code_vars(block0, next_var_info, edit);
VarDescrList merge_info;
if (block1) {
code.compute_used_code_vars(block1, next_var_info, edit);
merge_info = block0->var_info + block1->var_info;
} else {
merge_info = block0->var_info + next_var_info;
}
merge_info.clear_last();
merge_info += left;
return set_var_info(std::move(merge_info));
}
case _While: {
// while (block0 || left) block1;
// ... block0 left { block1 block0 left } next
VarDescrList new_var_info{next_var_info};
bool changes = false;
do {
VarDescrList after_cond{new_var_info};
after_cond += left;
code.compute_used_code_vars(block0, after_cond, changes);
code.compute_used_code_vars(block1, block0->var_info, changes);
std::size_t n = new_var_info.size();
new_var_info += block1->var_info;
new_var_info.clear_last();
if (changes) {
break;
}
changes = (new_var_info.size() == n);
} while (changes <= edit);
new_var_info += left;
code.compute_used_code_vars(block0, new_var_info, edit);
return set_var_info(block0->var_info);
}
case _Until: {
// until (block0 || left);
// .. { block0 left } block0 left next
VarDescrList after_cond_first{next_var_info};
after_cond_first += left;
code.compute_used_code_vars(block0, after_cond_first, false);
VarDescrList new_var_info{block0->var_info};
bool changes = false;
do {
VarDescrList after_cond{new_var_info};
after_cond += next_var_info;
after_cond += left;
code.compute_used_code_vars(block0, after_cond, changes);
std::size_t n = new_var_info.size();
new_var_info += block0->var_info;
new_var_info.clear_last();
if (changes) {
break;
}
changes = (new_var_info.size() == n);
} while (changes <= edit);
return set_var_info(std::move(new_var_info) + next_var_info);
}
case _Repeat: {
// repeat (left) block0
// left { block0 } next
VarDescrList new_var_info{next_var_info};
bool changes = false;
do {
code.compute_used_code_vars(block0, new_var_info, changes);
std::size_t n = new_var_info.size();
new_var_info += block0->var_info;
new_var_info.clear_last();
if (changes) {
break;
}
changes = (new_var_info.size() == n);
} while (changes <= edit);
tolk_assert(left.size() == 1);
bool last = new_var_info.count_used(left) == 0;
new_var_info += left;
if (last) {
new_var_info[left[0]]->flags |= VarDescr::_Last;
}
return set_var_info(std::move(new_var_info));
}
case _Again: {
// for(;;) block0
// { block0 }
VarDescrList new_var_info;
bool changes = false;
do {
code.compute_used_code_vars(block0, new_var_info, changes);
std::size_t n = new_var_info.size();
new_var_info += block0->var_info;
new_var_info.clear_last();
if (changes) {
break;
}
changes = (new_var_info.size() == n);
} while (changes <= edit);
return set_var_info(std::move(new_var_info));
}
case _TryCatch: {
code.compute_used_code_vars(block0, next_var_info, edit);
code.compute_used_code_vars(block1, next_var_info, edit);
VarDescrList merge_info = block0->var_info + block1->var_info + next_var_info;
merge_info -= left;
merge_info.clear_last();
return set_var_info(std::move(merge_info));
}
default:
std::cerr << "fatal: unknown operation <??" << cl << "> in compute_used_vars()\n";
throw ParseError{where, "unknown operation"};
}
}
bool prune_unreachable(std::unique_ptr<Op>& ops) {
if (!ops) {
return true;
}
Op& op = *ops;
if (op.cl == Op::_Nop) {
if (op.next) {
ops = std::move(op.next);
return prune_unreachable(ops);
}
return true;
}
bool reach;
switch (op.cl) {
case Op::_IntConst:
case Op::_SliceConst:
case Op::_GlobVar:
case Op::_SetGlob:
case Op::_Call:
case Op::_CallInd:
case Op::_Tuple:
case Op::_UnTuple:
case Op::_Import:
reach = true;
break;
case Op::_Let: {
reach = true;
break;
}
case Op::_Return:
reach = false;
break;
case Op::_If: {
// if left then block0 else block1; ...
VarDescr* c_var = op.var_info[op.left[0]];
if (c_var && c_var->always_true()) {
op.block0->last().next = std::move(op.next);
ops = std::move(op.block0);
return prune_unreachable(ops);
} else if (c_var && c_var->always_false()) {
op.block1->last().next = std::move(op.next);
ops = std::move(op.block1);
return prune_unreachable(ops);
} else {
reach = static_cast<int>(prune_unreachable(op.block0)) | static_cast<int>(prune_unreachable(op.block1));
}
break;
}
case Op::_While: {
// while (block0 || left) block1;
if (!prune_unreachable(op.block0)) {
// computation of block0 never returns
ops = std::move(op.block0);
return prune_unreachable(ops);
}
VarDescr* c_var = op.block0->last().var_info[op.left[0]];
if (c_var && c_var->always_false()) {
// block1 never executed
op.block0->last().next = std::move(op.next);
ops = std::move(op.block0);
return prune_unreachable(ops);
} else if (c_var && c_var->always_true()) {
if (!prune_unreachable(op.block1)) {
// block1 never returns
op.block0->last().next = std::move(op.block1);
ops = std::move(op.block0);
return false;
}
// infinite loop
op.cl = Op::_Again;
op.block0->last().next = std::move(op.block1);
op.left.clear();
reach = false;
} else {
if (!prune_unreachable(op.block1)) {
// block1 never returns, while equivalent to block0 ; if left then block1 else next
op.cl = Op::_If;
std::unique_ptr<Op> new_op = std::move(op.block0);
op.block0 = std::move(op.block1);
op.block1 = std::make_unique<Op>(op.next->where, Op::_Nop);
new_op->last().next = std::move(ops);
ops = std::move(new_op);
}
reach = true; // block1 may be never executed
}
break;
}
case Op::_Repeat: {
// repeat (left) block0
VarDescr* c_var = op.var_info[op.left[0]];
if (c_var && c_var->always_nonpos()) {
// loop never executed
ops = std::move(op.next);
return prune_unreachable(ops);
}
if (c_var && c_var->always_pos()) {
if (!prune_unreachable(op.block0)) {
// block0 executed at least once, and it never returns
// replace code with block0
ops = std::move(op.block0);
return false;
}
} else {
prune_unreachable(op.block0);
}
reach = true;
break;
}
case Op::_Until:
case Op::_Again: {
// do block0 until left; ...
if (!prune_unreachable(op.block0)) {
// block0 never returns, replace loop by block0
ops = std::move(op.block0);
return false;
}
reach = (op.cl != Op::_Again);
break;
}
case Op::_TryCatch: {
reach = static_cast<int>(prune_unreachable(op.block0)) | static_cast<int>(prune_unreachable(op.block1));
break;
}
default:
std::cerr << "fatal: unknown operation <??" << op.cl << ">\n";
throw ParseError{op.where, "unknown operation in prune_unreachable()"};
}
if (reach) {
return prune_unreachable(op.next);
} else {
while (op.next->next) {
op.next = std::move(op.next->next);
}
return false;
}
}
void CodeBlob::prune_unreachable_code() {
if (prune_unreachable(ops)) {
throw ParseError{loc, "control reaches end of function"};
}
}
void CodeBlob::fwd_analyze() {
VarDescrList values;
tolk_assert(ops && ops->cl == Op::_Import);
for (var_idx_t i : ops->left) {
values += i;
if (vars[i].v_type->is_int()) {
values[i]->val |= VarDescr::_Int;
}
}
ops->fwd_analyze(values);
}
void Op::prepare_args(VarDescrList values) {
if (args.size() != right.size()) {
args.clear();
for (var_idx_t i : right) {
args.emplace_back(i);
}
}
for (std::size_t i = 0; i < right.size(); i++) {
const VarDescr* val = values[right[i]];
if (val) {
args[i].set_value(*val);
// args[i].clear_unused();
} else {
args[i].clear_value();
}
args[i].clear_unused();
}
}
VarDescrList Op::fwd_analyze(VarDescrList values) {
var_info.import_values(values);
switch (cl) {
case _Nop:
case _Import:
break;
case _Return:
values.set_unreachable();
break;
case _IntConst: {
values.add_newval(left[0]).set_const(int_const);
break;
}
case _SliceConst: {
values.add_newval(left[0]).set_const(str_const);
break;
}
case _Call: {
prepare_args(values);
auto func = dynamic_cast<const SymValAsmFunc*>(fun_ref->value);
if (func) {
std::vector<VarDescr> res;
res.reserve(left.size());
for (var_idx_t i : left) {
res.emplace_back(i);
}
AsmOpList tmp;
func->compile(tmp, res, args, where); // abstract interpretation of res := f (args)
int j = 0;
for (var_idx_t i : left) {
values.add_newval(i).set_value(res[j++]);
}
} else {
for (var_idx_t i : left) {
values.add_newval(i);
}
}
break;
}
case _Tuple:
case _UnTuple:
case _GlobVar:
case _CallInd: {
for (var_idx_t i : left) {
values.add_newval(i);
}
break;
}
case _SetGlob:
break;
case _Let: {
std::vector<VarDescr> old_val;
tolk_assert(left.size() == right.size());
for (std::size_t i = 0; i < right.size(); i++) {
const VarDescr* ov = values[right[i]];
if (!ov && verbosity >= 5) {
std::cerr << "FATAL: error in assignment at right component #" << i << " (no value for _" << right[i] << ")"
<< std::endl;
for (auto x : left) {
std::cerr << '_' << x << " ";
}
std::cerr << "= ";
for (auto x : right) {
std::cerr << '_' << x << " ";
}
std::cerr << std::endl;
}
// tolk_assert(ov);
if (ov) {
old_val.push_back(*ov);
} else {
old_val.emplace_back();
}
}
for (std::size_t i = 0; i < left.size(); i++) {
values.add_newval(left[i]).set_value(std::move(old_val[i]));
}
break;
}
case _If: {
VarDescrList val1 = block0->fwd_analyze(values);
VarDescrList val2 = block1 ? block1->fwd_analyze(std::move(values)) : std::move(values);
values = val1 | val2;
break;
}
case _Repeat: {
bool atl1 = (values[left[0]] && values[left[0]]->always_pos());
VarDescrList next_values = block0->fwd_analyze(values);
while (true) {
VarDescrList new_values = values | next_values;
if (same_values(new_values, values)) {
break;
}
values = std::move(new_values);
next_values = block0->fwd_analyze(values);
}
if (atl1) {
values = std::move(next_values);
}
break;
}
case _While: {
auto values0 = values;
values = block0->fwd_analyze(values);
if (values[left[0]] && values[left[0]]->always_false()) {
// block1 never executed
block1->fwd_analyze(values);
break;
}
while (true) {
VarDescrList next_values = values | block0->fwd_analyze(values0 | block1->fwd_analyze(values));
if (same_values(next_values, values)) {
break;
}
values = std::move(next_values);
}
break;
}
case _Until:
case _Again: {
while (true) {
VarDescrList next_values = values | block0->fwd_analyze(values);
if (same_values(next_values, values)) {
break;
}
values = std::move(next_values);
}
values = block0->fwd_analyze(values);
break;
}
case _TryCatch: {
VarDescrList val1 = block0->fwd_analyze(values);
VarDescrList val2 = block1->fwd_analyze(std::move(values));
values = val1 | val2;
break;
}
default:
std::cerr << "fatal: unknown operation <??" << cl << ">\n";
throw ParseError{where, "unknown operation in fwd_analyze()"};
}
if (next) {
return next->fwd_analyze(std::move(values));
} else {
return values;
}
}
bool Op::set_noreturn(bool nr) {
if (nr) {
flags |= _NoReturn;
} else {
flags &= ~_NoReturn;
}
return nr;
}
bool Op::mark_noreturn() {
switch (cl) {
case _Nop:
if (!next) {
return set_noreturn(false);
}
// fallthrough
case _Import:
case _IntConst:
case _SliceConst:
case _Let:
case _Tuple:
case _UnTuple:
case _SetGlob:
case _GlobVar:
case _CallInd:
case _Call:
return set_noreturn(next->mark_noreturn());
case _Return:
return set_noreturn(true);
case _If:
case _TryCatch:
return set_noreturn((static_cast<int>(block0->mark_noreturn()) & static_cast<int>(block1 && block1->mark_noreturn())) | static_cast<int>(next->mark_noreturn()));
case _Again:
block0->mark_noreturn();
return set_noreturn(true);
case _Until:
return set_noreturn(static_cast<int>(block0->mark_noreturn()) | static_cast<int>(next->mark_noreturn()));
case _While:
block1->mark_noreturn();
return set_noreturn(static_cast<int>(block0->mark_noreturn()) | static_cast<int>(next->mark_noreturn()));
case _Repeat:
block0->mark_noreturn();
return set_noreturn(next->mark_noreturn());
default:
std::cerr << "fatal: unknown operation <??" << cl << ">\n";
throw ParseError{where, "unknown operation in mark_noreturn()"};
}
}
void CodeBlob::mark_noreturn() {
ops->mark_noreturn();
}
} // namespace tolk

372
tolk/asmops.cpp Normal file
View file

@ -0,0 +1,372 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
#include <iostream>
namespace tolk {
/*
*
* ASM-OP LIST FUNCTIONS
*
*/
int is_pos_pow2(td::RefInt256 x) {
if (sgn(x) > 0 && !sgn(x & (x - 1))) {
return x->bit_size(false) - 1;
} else {
return -1;
}
}
int is_neg_pow2(td::RefInt256 x) {
return sgn(x) < 0 ? is_pos_pow2(-x) : 0;
}
std::ostream& operator<<(std::ostream& os, AsmOp::SReg stack_reg) {
int i = stack_reg.idx;
if (i >= 0) {
if (i < 16) {
return os << 's' << i;
} else {
return os << i << " s()";
}
} else if (i >= -2) {
return os << "s(" << i << ')';
} else {
return os << i << " s()";
}
}
AsmOp AsmOp::Const(int arg, std::string push_op, td::RefInt256 origin) {
std::ostringstream os;
os << arg << ' ' << push_op;
return AsmOp::Const(os.str(), origin);
}
AsmOp AsmOp::make_stk2(int a, int b, const char* str, int delta) {
std::ostringstream os;
os << SReg(a) << ' ' << SReg(b) << ' ' << str;
int c = std::max(a, b) + 1;
return AsmOp::Custom(os.str(), c, c + delta);
}
AsmOp AsmOp::make_stk3(int a, int b, int c, const char* str, int delta) {
std::ostringstream os;
os << SReg(a) << ' ' << SReg(b) << ' ' << SReg(c) << ' ' << str;
int m = std::max(a, std::max(b, c)) + 1;
return AsmOp::Custom(os.str(), m, m + delta);
}
AsmOp AsmOp::BlkSwap(int a, int b) {
std::ostringstream os;
if (a == 1 && b == 1) {
return AsmOp::Xchg(0, 1);
} else if (a == 1) {
if (b == 2) {
os << "ROT";
} else {
os << b << " ROLL";
}
} else if (b == 1) {
if (a == 2) {
os << "-ROT";
} else {
os << a << " -ROLL";
}
} else {
os << a << " " << b << " BLKSWAP";
}
return AsmOp::Custom(os.str(), a + b, a + b);
}
AsmOp AsmOp::BlkPush(int a, int b) {
std::ostringstream os;
if (a == 1) {
return AsmOp::Push(b);
} else if (a == 2 && b == 1) {
os << "2DUP";
} else {
os << a << " " << b << " BLKPUSH";
}
return AsmOp::Custom(os.str(), b + 1, a + b + 1);
}
AsmOp AsmOp::BlkDrop(int a) {
std::ostringstream os;
if (a == 1) {
return AsmOp::Pop();
} else if (a == 2) {
os << "2DROP";
} else {
os << a << " BLKDROP";
}
return AsmOp::Custom(os.str(), a, 0);
}
AsmOp AsmOp::BlkDrop2(int a, int b) {
if (!b) {
return BlkDrop(a);
}
std::ostringstream os;
os << a << " " << b << " BLKDROP2";
return AsmOp::Custom(os.str(), a + b, b);
}
AsmOp AsmOp::BlkReverse(int a, int b) {
std::ostringstream os;
os << a << " " << b << " REVERSE";
return AsmOp::Custom(os.str(), a + b, a + b);
}
AsmOp AsmOp::Tuple(int a) {
switch (a) {
case 1:
return AsmOp::Custom("SINGLE", 1, 1);
case 2:
return AsmOp::Custom("PAIR", 2, 1);
case 3:
return AsmOp::Custom("TRIPLE", 3, 1);
}
std::ostringstream os;
os << a << " TUPLE";
return AsmOp::Custom(os.str(), a, 1);
}
AsmOp AsmOp::UnTuple(int a) {
switch (a) {
case 1:
return AsmOp::Custom("UNSINGLE", 1, 1);
case 2:
return AsmOp::Custom("UNPAIR", 1, 2);
case 3:
return AsmOp::Custom("UNTRIPLE", 1, 3);
}
std::ostringstream os;
os << a << " UNTUPLE";
return AsmOp::Custom(os.str(), 1, a);
}
AsmOp AsmOp::IntConst(td::RefInt256 x) {
if (x->signed_fits_bits(8)) {
return AsmOp::Const(dec_string(x) + " PUSHINT", x);
}
if (!x->is_valid()) {
return AsmOp::Const("PUSHNAN", x);
}
int k = is_pos_pow2(x);
if (k >= 0) {
return AsmOp::Const(k, "PUSHPOW2", x);
}
k = is_pos_pow2(x + 1);
if (k >= 0) {
return AsmOp::Const(k, "PUSHPOW2DEC", x);
}
k = is_pos_pow2(-x);
if (k >= 0) {
return AsmOp::Const(k, "PUSHNEGPOW2", x);
}
if (!x->mod_pow2_short(23)) {
return AsmOp::Const(dec_string(x) + " PUSHINTX", x);
}
return AsmOp::Const(dec_string(x) + " PUSHINT", x);
}
AsmOp AsmOp::BoolConst(bool f) {
return AsmOp::Const(f ? "TRUE" : "FALSE");
}
AsmOp AsmOp::Parse(std::string custom_op) {
if (custom_op == "NOP") {
return AsmOp::Nop();
} else if (custom_op == "SWAP") {
return AsmOp::Xchg(1);
} else if (custom_op == "DROP") {
return AsmOp::Pop(0);
} else if (custom_op == "NIP") {
return AsmOp::Pop(1);
} else if (custom_op == "DUP") {
return AsmOp::Push(0);
} else if (custom_op == "OVER") {
return AsmOp::Push(1);
} else {
return AsmOp::Custom(custom_op);
}
}
AsmOp AsmOp::Parse(std::string custom_op, int args, int retv) {
auto res = Parse(custom_op);
if (res.is_custom()) {
res.a = args;
res.b = retv;
}
return res;
}
void AsmOp::out(std::ostream& os) const {
if (!op.empty()) {
os << op;
return;
}
switch (t) {
case a_none:
break;
case a_xchg:
if (!a && !(b & -2)) {
os << (b ? "SWAP" : "NOP");
break;
}
os << SReg(a) << ' ' << SReg(b) << " XCHG";
break;
case a_push:
if (!(a & -2)) {
os << (a ? "OVER" : "DUP");
break;
}
os << SReg(a) << " PUSH";
break;
case a_pop:
if (!(a & -2)) {
os << (a ? "NIP" : "DROP");
break;
}
os << SReg(a) << " POP";
break;
default:
throw Fatal{"unknown assembler operation"};
}
}
void AsmOp::out_indent_nl(std::ostream& os, bool no_eol) const {
for (int i = 0; i < indent; i++) {
os << " ";
}
out(os);
if (!no_eol) {
os << std::endl;
}
}
std::string AsmOp::to_string() const {
if (!op.empty()) {
return op;
} else {
std::ostringstream os;
out(os);
return os.str();
}
}
bool AsmOpList::append(const std::vector<AsmOp>& ops) {
for (const auto& op : ops) {
if (!append(op)) {
return false;
}
}
return true;
}
const_idx_t AsmOpList::register_const(Const new_const) {
if (new_const.is_null()) {
return not_const;
}
unsigned idx;
for (idx = 0; idx < constants_.size(); idx++) {
if (!td::cmp(new_const, constants_[idx])) {
return idx;
}
}
constants_.push_back(std::move(new_const));
return (const_idx_t)idx;
}
Const AsmOpList::get_const(const_idx_t idx) {
if ((unsigned)idx < constants_.size()) {
return constants_[idx];
} else {
return {};
}
}
void AsmOpList::show_var(std::ostream& os, var_idx_t idx) const {
if (!var_names_ || (unsigned)idx >= var_names_->size()) {
os << '_' << idx;
} else {
var_names_->at(idx).show(os, 2);
}
}
void AsmOpList::show_var_ext(std::ostream& os, std::pair<var_idx_t, const_idx_t> idx_pair) const {
auto i = idx_pair.first;
auto j = idx_pair.second;
if (!var_names_ || (unsigned)i >= var_names_->size()) {
os << '_' << i;
} else {
var_names_->at(i).show(os, 2);
}
if ((unsigned)j < constants_.size() && constants_[j].not_null()) {
os << '=' << constants_[j];
}
}
void AsmOpList::out(std::ostream& os, int mode) const {
if (!(mode & 2)) {
for (const auto& op : list_) {
op.out_indent_nl(os);
}
} else {
std::size_t n = list_.size();
for (std::size_t i = 0; i < n; i++) {
const auto& op = list_[i];
if (!op.is_comment() && i + 1 < n && list_[i + 1].is_comment()) {
op.out_indent_nl(os, true);
os << '\t';
do {
i++;
} while (i + 1 < n && list_[i + 1].is_comment());
list_[i].out(os);
os << std::endl;
} else {
op.out_indent_nl(os, false);
}
}
}
}
bool apply_op(StackTransform& trans, const AsmOp& op) {
if (!trans.is_valid()) {
return false;
}
switch (op.t) {
case AsmOp::a_none:
return true;
case AsmOp::a_xchg:
return trans.apply_xchg(op.a, op.b, true);
case AsmOp::a_push:
return trans.apply_push(op.a);
case AsmOp::a_pop:
return trans.apply_pop(op.a);
case AsmOp::a_const:
return !op.a && op.b == 1 && trans.apply_push_newconst();
case AsmOp::a_custom:
return op.is_gconst() && trans.apply_push_newconst();
default:
return false;
}
}
} // namespace tolk

1231
tolk/builtins.cpp Normal file
View file

File diff suppressed because it is too large Load diff

910
tolk/codegen.cpp Normal file
View file

@ -0,0 +1,910 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* GENERATE TVM STACK CODE
*
*/
StackLayout Stack::vars() const {
StackLayout res;
res.reserve(s.size());
for (auto x : s) {
res.push_back(x.first);
}
return res;
}
int Stack::find(var_idx_t var, int from) const {
for (int i = from; i < depth(); i++) {
if (at(i).first == var) {
return i;
}
}
return -1;
}
// finds var in [from .. to)
int Stack::find(var_idx_t var, int from, int to) const {
for (int i = from; i < depth() && i < to; i++) {
if (at(i).first == var) {
return i;
}
}
return -1;
}
// finds var outside [from .. to)
int Stack::find_outside(var_idx_t var, int from, int to) const {
from = std::max(from, 0);
if (from >= to) {
return find(var);
} else {
int t = find(var, 0, from);
return t >= 0 ? t : find(var, to);
}
}
int Stack::find_const(const_idx_t cst, int from) const {
for (int i = from; i < depth(); i++) {
if (at(i).second == cst) {
return i;
}
}
return -1;
}
void Stack::forget_const() {
for (auto& vc : s) {
if (vc.second != not_const) {
vc.second = not_const;
}
}
}
void Stack::issue_pop(int i) {
validate(i);
if (output_enabled()) {
o << AsmOp::Pop(i);
}
at(i) = get(0);
s.pop_back();
modified();
}
void Stack::issue_push(int i) {
validate(i);
if (output_enabled()) {
o << AsmOp::Push(i);
}
s.push_back(get(i));
modified();
}
void Stack::issue_xchg(int i, int j) {
validate(i);
validate(j);
if (i != j && get(i) != get(j)) {
if (output_enabled()) {
o << AsmOp::Xchg(i, j);
}
std::swap(at(i), at(j));
modified();
}
}
int Stack::drop_vars_except(const VarDescrList& var_info, int excl_var) {
int dropped = 0, changes;
do {
changes = 0;
int n = depth();
for (int i = 0; i < n; i++) {
var_idx_t idx = at(i).first;
if (((!var_info[idx] || var_info[idx]->is_unused()) && idx != excl_var) || find(idx, 0, i - 1) >= 0) {
// unneeded
issue_pop(i);
changes = 1;
break;
}
}
dropped += changes;
} while (changes);
return dropped;
}
void Stack::show(int flags) {
std::ostringstream os;
for (auto i : s) {
os << ' ';
o.show_var_ext(os, i);
}
o << AsmOp::Comment(os.str());
mode |= _Shown;
}
void Stack::forget_var(var_idx_t idx) {
for (auto& x : s) {
if (x.first == idx) {
x = std::make_pair(_Garbage, not_const);
modified();
}
}
}
void Stack::push_new_var(var_idx_t idx) {
forget_var(idx);
s.emplace_back(idx, not_const);
modified();
}
void Stack::push_new_const(var_idx_t idx, const_idx_t cidx) {
forget_var(idx);
s.emplace_back(idx, cidx);
modified();
}
void Stack::assign_var(var_idx_t new_idx, var_idx_t old_idx) {
int i = find(old_idx);
tolk_assert(i >= 0 && "variable not found in stack");
if (new_idx != old_idx) {
at(i).first = new_idx;
modified();
}
}
void Stack::do_copy_var(var_idx_t new_idx, var_idx_t old_idx) {
int i = find(old_idx);
tolk_assert(i >= 0 && "variable not found in stack");
if (find(old_idx, i + 1) < 0) {
issue_push(i);
tolk_assert(at(0).first == old_idx);
}
assign_var(new_idx, old_idx);
}
void Stack::enforce_state(const StackLayout& req_stack) {
int k = (int)req_stack.size();
for (int i = 0; i < k; i++) {
var_idx_t x = req_stack[i];
if (i < depth() && s[i].first == x) {
continue;
}
while (depth() > 0 && std::find(req_stack.cbegin(), req_stack.cend(), get(0).first) == req_stack.cend()) {
// current TOS entry is unused in req_stack, drop it
issue_pop(0);
}
int j = find(x);
if (j >= depth() - i) {
issue_push(j);
j = 0;
}
issue_xchg(j, depth() - i - 1);
tolk_assert(s[i].first == x);
}
while (depth() > k) {
issue_pop(0);
}
tolk_assert(depth() == k);
for (int i = 0; i < k; i++) {
tolk_assert(s[i].first == req_stack[i]);
}
}
void Stack::merge_const(const Stack& req_stack) {
tolk_assert(s.size() == req_stack.s.size());
for (std::size_t i = 0; i < s.size(); i++) {
tolk_assert(s[i].first == req_stack.s[i].first);
if (s[i].second != req_stack.s[i].second) {
s[i].second = not_const;
}
}
}
void Stack::merge_state(const Stack& req_stack) {
enforce_state(req_stack.vars());
merge_const(req_stack);
}
void Stack::rearrange_top(const StackLayout& top, std::vector<bool> last) {
while (last.size() < top.size()) {
last.push_back(false);
}
int k = (int)top.size();
for (int i = 0; i < k; i++) {
for (int j = i + 1; j < k; j++) {
if (top[i] == top[j]) {
last[i] = false;
break;
}
}
}
int ss = 0;
for (int i = 0; i < k; i++) {
if (last[i]) {
++ss;
}
}
for (int i = 0; i < k; i++) {
var_idx_t x = top[i];
// find s(j) containing x with j not in [ss, ss+i)
int j = find_outside(x, ss, ss + i);
if (last[i]) {
// rearrange x to be at s(ss-1)
issue_xchg(--ss, j);
tolk_assert(get(ss).first == x);
} else {
// create a new copy of x
issue_push(j);
issue_xchg(0, ss);
tolk_assert(get(ss).first == x);
}
}
tolk_assert(!ss);
}
void Stack::rearrange_top(var_idx_t top, bool last) {
int i = find(top);
if (last) {
issue_xchg(0, i);
} else {
issue_push(i);
}
tolk_assert(get(0).first == top);
}
bool Op::generate_code_step(Stack& stack) {
stack.opt_show();
stack.drop_vars_except(var_info);
stack.opt_show();
bool inline_func = stack.mode & Stack::_InlineFunc;
switch (cl) {
case _Nop:
case _Import:
return true;
case _Return: {
stack.enforce_state(left);
if (stack.o.retalt_ && (stack.mode & Stack::_NeedRetAlt)) {
stack.o << "RETALT";
}
stack.opt_show();
return false;
}
case _IntConst: {
auto p = next->var_info[left[0]];
if (!p || p->is_unused()) {
return true;
}
auto cidx = stack.o.register_const(int_const);
int i = stack.find_const(cidx);
if (i < 0) {
stack.o << push_const(int_const);
stack.push_new_const(left[0], cidx);
} else {
tolk_assert(stack.at(i).second == cidx);
stack.do_copy_var(left[0], stack[i]);
}
return true;
}
case _SliceConst: {
auto p = next->var_info[left[0]];
if (!p || p->is_unused()) {
return true;
}
stack.o << AsmOp::Const("x{" + str_const + "} PUSHSLICE");
stack.push_new_var(left[0]);
return true;
}
case _GlobVar:
if (dynamic_cast<const SymValGlobVar*>(fun_ref->value)) {
bool used = false;
for (auto i : left) {
auto p = next->var_info[i];
if (p && !p->is_unused()) {
used = true;
}
}
if (!used || disabled()) {
return true;
}
std::string name = symbols.get_name(fun_ref->sym_idx);
stack.o << AsmOp::Custom(name + " GETGLOB", 0, 1);
if (left.size() != 1) {
tolk_assert(left.size() <= 15);
stack.o << AsmOp::UnTuple((int)left.size());
}
for (auto i : left) {
stack.push_new_var(i);
}
return true;
} else {
tolk_assert(left.size() == 1);
auto p = next->var_info[left[0]];
if (!p || p->is_unused() || disabled()) {
return true;
}
stack.o << "CONT:<{";
stack.o.indent();
auto func = dynamic_cast<SymValAsmFunc*>(fun_ref->value);
if (func) {
// TODO: create and compile a true lambda instead of this (so that arg_order and ret_order would work correctly)
std::vector<VarDescr> args0, res;
TypeExpr::remove_indirect(func->sym_type);
tolk_assert(func->get_type()->is_map());
auto wr = func->get_type()->args.at(0)->get_width();
auto wl = func->get_type()->args.at(1)->get_width();
tolk_assert(wl >= 0 && wr >= 0);
for (int i = 0; i < wl; i++) {
res.emplace_back(0);
}
for (int i = 0; i < wr; i++) {
args0.emplace_back(0);
}
func->compile(stack.o, res, args0, where); // compile res := f (args0)
} else {
std::string name = symbols.get_name(fun_ref->sym_idx);
stack.o << AsmOp::Custom(name + " CALLDICT", (int)right.size(), (int)left.size());
}
stack.o.undent();
stack.o << "}>";
stack.push_new_var(left.at(0));
return true;
}
case _Let: {
tolk_assert(left.size() == right.size());
int i = 0;
std::vector<bool> active;
active.reserve(left.size());
for (std::size_t k = 0; k < left.size(); k++) {
var_idx_t y = left[k]; // "y" = "x"
auto p = next->var_info[y];
active.push_back(p && !p->is_unused());
}
for (std::size_t k = 0; k < left.size(); k++) {
if (!active[k]) {
continue;
}
var_idx_t x = right[k]; // "y" = "x"
bool is_last = true;
for (std::size_t l = k + 1; l < right.size(); l++) {
if (right[l] == x && active[l]) {
is_last = false;
}
}
if (is_last) {
auto info = var_info[x];
is_last = (info && info->is_last());
}
if (is_last) {
stack.assign_var(--i, x);
} else {
stack.do_copy_var(--i, x);
}
}
i = 0;
for (std::size_t k = 0; k < left.size(); k++) {
if (active[k]) {
stack.assign_var(left[k], --i);
}
}
return true;
}
case _Tuple:
case _UnTuple: {
if (disabled()) {
return true;
}
std::vector<bool> last;
for (var_idx_t x : right) {
last.push_back(var_info[x] && var_info[x]->is_last());
}
stack.rearrange_top(right, std::move(last));
stack.opt_show();
int k = (int)stack.depth() - (int)right.size();
tolk_assert(k >= 0);
if (cl == _Tuple) {
stack.o << AsmOp::Tuple((int)right.size());
tolk_assert(left.size() == 1);
} else {
stack.o << AsmOp::UnTuple((int)left.size());
tolk_assert(right.size() == 1);
}
stack.s.resize(k);
for (int i = 0; i < (int)left.size(); i++) {
stack.push_new_var(left.at(i));
}
return true;
}
case _Call:
case _CallInd: {
if (disabled()) {
return true;
}
SymValFunc* func = (fun_ref ? dynamic_cast<SymValFunc*>(fun_ref->value) : nullptr);
auto arg_order = (func ? func->get_arg_order() : nullptr);
auto ret_order = (func ? func->get_ret_order() : nullptr);
tolk_assert(!arg_order || arg_order->size() == right.size());
tolk_assert(!ret_order || ret_order->size() == left.size());
std::vector<var_idx_t> right1;
if (args.size()) {
tolk_assert(args.size() == right.size());
for (int i = 0; i < (int)right.size(); i++) {
int j = arg_order ? arg_order->at(i) : i;
const VarDescr& arg = args.at(j);
if (!arg.is_unused()) {
tolk_assert(var_info[arg.idx] && !var_info[arg.idx]->is_unused());
right1.push_back(arg.idx);
}
}
} else if (arg_order) {
for (int i = 0; i < (int)right.size(); i++) {
right1.push_back(right.at(arg_order->at(i)));
}
} else {
right1 = right;
}
std::vector<bool> last;
for (var_idx_t x : right1) {
last.push_back(var_info[x] && var_info[x]->is_last());
}
stack.rearrange_top(right1, std::move(last));
stack.opt_show();
int k = (int)stack.depth() - (int)right1.size();
tolk_assert(k >= 0);
for (int i = 0; i < (int)right1.size(); i++) {
if (stack.s[k + i].first != right1[i]) {
std::cerr << stack.o;
}
tolk_assert(stack.s[k + i].first == right1[i]);
}
auto exec_callxargs = [&](int args, int ret) {
if (args <= 15 && ret <= 15) {
stack.o << exec_arg2_op("CALLXARGS", args, ret, args + 1, ret);
} else {
tolk_assert(args <= 254 && ret <= 254);
stack.o << AsmOp::Const(PSTRING() << args << " PUSHINT");
stack.o << AsmOp::Const(PSTRING() << ret << " PUSHINT");
stack.o << AsmOp::Custom("CALLXVARARGS", args + 3, ret);
}
};
if (cl == _CallInd) {
exec_callxargs((int)right.size() - 1, (int)left.size());
} else {
auto func = dynamic_cast<const SymValAsmFunc*>(fun_ref->value);
if (func) {
std::vector<VarDescr> res;
res.reserve(left.size());
for (var_idx_t i : left) {
res.emplace_back(i);
}
func->compile(stack.o, res, args, where); // compile res := f (args)
} else {
auto fv = dynamic_cast<const SymValCodeFunc*>(fun_ref->value);
std::string name = symbols.get_name(fun_ref->sym_idx);
bool is_inline = (fv && (fv->flags & 3));
if (is_inline) {
stack.o << AsmOp::Custom(name + " INLINECALLDICT", (int)right.size(), (int)left.size());
} else if (fv && fv->code && fv->code->require_callxargs) {
stack.o << AsmOp::Custom(name + (" PREPAREDICT"), 0, 2);
exec_callxargs((int)right.size() + 1, (int)left.size());
} else {
stack.o << AsmOp::Custom(name + " CALLDICT", (int)right.size(), (int)left.size());
}
}
}
stack.s.resize(k);
for (int i = 0; i < (int)left.size(); i++) {
int j = ret_order ? ret_order->at(i) : i;
stack.push_new_var(left.at(j));
}
return true;
}
case _SetGlob: {
tolk_assert(fun_ref && dynamic_cast<const SymValGlobVar*>(fun_ref->value));
std::vector<bool> last;
for (var_idx_t x : right) {
last.push_back(var_info[x] && var_info[x]->is_last());
}
stack.rearrange_top(right, std::move(last));
stack.opt_show();
int k = (int)stack.depth() - (int)right.size();
tolk_assert(k >= 0);
for (int i = 0; i < (int)right.size(); i++) {
if (stack.s[k + i].first != right[i]) {
std::cerr << stack.o;
}
tolk_assert(stack.s[k + i].first == right[i]);
}
if (right.size() > 1) {
stack.o << AsmOp::Tuple((int)right.size());
}
if (!right.empty()) {
std::string name = symbols.get_name(fun_ref->sym_idx);
stack.o << AsmOp::Custom(name + " SETGLOB", 1, 0);
}
stack.s.resize(k);
return true;
}
case _If: {
if (block0->is_empty() && block1->is_empty()) {
return true;
}
if (!next->noreturn() && (block0->noreturn() != block1->noreturn())) {
stack.o.retalt_ = true;
}
var_idx_t x = left[0];
stack.rearrange_top(x, var_info[x] && var_info[x]->is_last());
tolk_assert(stack[0] == x);
stack.opt_show();
stack.s.pop_back();
stack.modified();
if (inline_func && (block0->noreturn() || block1->noreturn())) {
bool is0 = block0->noreturn();
Op* block_noreturn = is0 ? block0.get() : block1.get();
Op* block_other = is0 ? block1.get() : block0.get();
stack.mode &= ~Stack::_InlineFunc;
stack.o << (is0 ? "IF:<{" : "IFNOT:<{");
stack.o.indent();
Stack stack_copy{stack};
block_noreturn->generate_code_all(stack_copy);
stack.o.undent();
stack.o << "}>ELSE<{";
stack.o.indent();
block_other->generate_code_all(stack);
if (!block_other->noreturn()) {
next->generate_code_all(stack);
}
stack.o.undent();
stack.o << "}>";
return false;
}
if (block1->is_empty() || block0->is_empty()) {
bool is0 = block1->is_empty();
Op* block = is0 ? block0.get() : block1.get();
// if (left) block0; ...
// if (!left) block1; ...
if (block->noreturn()) {
stack.o << (is0 ? "IFJMP:<{" : "IFNOTJMP:<{");
stack.o.indent();
Stack stack_copy{stack};
stack_copy.mode &= ~Stack::_InlineFunc;
stack_copy.mode |= next->noreturn() ? 0 : Stack::_NeedRetAlt;
block->generate_code_all(stack_copy);
stack.o.undent();
stack.o << "}>";
return true;
}
stack.o << (is0 ? "IF:<{" : "IFNOT:<{");
stack.o.indent();
Stack stack_copy{stack}, stack_target{stack};
stack_target.disable_output();
stack_target.drop_vars_except(next->var_info);
stack_copy.mode &= ~Stack::_InlineFunc;
block->generate_code_all(stack_copy);
stack_copy.drop_vars_except(var_info);
stack_copy.opt_show();
if ((is0 && stack_copy == stack) || (!is0 && stack_copy.vars() == stack.vars())) {
stack.o.undent();
stack.o << "}>";
if (!is0) {
stack.merge_const(stack_copy);
}
return true;
}
// stack_copy.drop_vars_except(next->var_info);
stack_copy.enforce_state(stack_target.vars());
stack_copy.opt_show();
if (stack_copy.vars() == stack.vars()) {
stack.o.undent();
stack.o << "}>";
stack.merge_const(stack_copy);
return true;
}
stack.o.undent();
stack.o << "}>ELSE<{";
stack.o.indent();
stack.merge_state(stack_copy);
stack.opt_show();
stack.o.undent();
stack.o << "}>";
return true;
}
if (block0->noreturn() || block1->noreturn()) {
bool is0 = block0->noreturn();
Op* block_noreturn = is0 ? block0.get() : block1.get();
Op* block_other = is0 ? block1.get() : block0.get();
stack.o << (is0 ? "IFJMP:<{" : "IFNOTJMP:<{");
stack.o.indent();
Stack stack_copy{stack};
stack_copy.mode &= ~Stack::_InlineFunc;
stack_copy.mode |= (block_other->noreturn() || next->noreturn()) ? 0 : Stack::_NeedRetAlt;
block_noreturn->generate_code_all(stack_copy);
stack.o.undent();
stack.o << "}>";
block_other->generate_code_all(stack);
return !block_other->noreturn();
}
stack.o << "IF:<{";
stack.o.indent();
Stack stack_copy{stack};
stack_copy.mode &= ~Stack::_InlineFunc;
block0->generate_code_all(stack_copy);
stack_copy.drop_vars_except(next->var_info);
stack_copy.opt_show();
stack.o.undent();
stack.o << "}>ELSE<{";
stack.o.indent();
stack.mode &= ~Stack::_InlineFunc;
block1->generate_code_all(stack);
stack.merge_state(stack_copy);
stack.opt_show();
stack.o.undent();
stack.o << "}>";
return true;
}
case _Repeat: {
var_idx_t x = left[0];
//stack.drop_vars_except(block0->var_info, x);
stack.rearrange_top(x, var_info[x] && var_info[x]->is_last());
tolk_assert(stack[0] == x);
stack.opt_show();
stack.s.pop_back();
stack.modified();
if (block0->noreturn()) {
stack.o.retalt_ = true;
}
if (true || !next->is_empty()) {
stack.o << "REPEAT:<{";
stack.o.indent();
stack.forget_const();
if (block0->noreturn()) {
Stack stack_copy{stack};
StackLayout layout1 = stack.vars();
stack_copy.mode &= ~Stack::_InlineFunc;
stack_copy.mode |= Stack::_NeedRetAlt;
block0->generate_code_all(stack_copy);
} else {
StackLayout layout1 = stack.vars();
stack.mode &= ~Stack::_InlineFunc;
stack.mode |= Stack::_NeedRetAlt;
block0->generate_code_all(stack);
stack.enforce_state(std::move(layout1));
stack.opt_show();
}
stack.o.undent();
stack.o << "}>";
return true;
} else {
stack.o << "REPEATEND";
stack.forget_const();
StackLayout layout1 = stack.vars();
block0->generate_code_all(stack);
stack.enforce_state(std::move(layout1));
stack.opt_show();
return false;
}
}
case _Again: {
stack.drop_vars_except(block0->var_info);
stack.opt_show();
if (block0->noreturn()) {
stack.o.retalt_ = true;
}
if (!next->is_empty() || inline_func) {
stack.o << "AGAIN:<{";
stack.o.indent();
stack.forget_const();
StackLayout layout1 = stack.vars();
stack.mode &= ~Stack::_InlineFunc;
stack.mode |= Stack::_NeedRetAlt;
block0->generate_code_all(stack);
stack.enforce_state(std::move(layout1));
stack.opt_show();
stack.o.undent();
stack.o << "}>";
return true;
} else {
stack.o << "AGAINEND";
stack.forget_const();
StackLayout layout1 = stack.vars();
block0->generate_code_all(stack);
stack.enforce_state(std::move(layout1));
stack.opt_show();
return false;
}
}
case _Until: {
// stack.drop_vars_except(block0->var_info);
// stack.opt_show();
if (block0->noreturn()) {
stack.o.retalt_ = true;
}
if (true || !next->is_empty()) {
stack.o << "UNTIL:<{";
stack.o.indent();
stack.forget_const();
auto layout1 = stack.vars();
stack.mode &= ~Stack::_InlineFunc;
stack.mode |= Stack::_NeedRetAlt;
block0->generate_code_all(stack);
layout1.push_back(left[0]);
stack.enforce_state(std::move(layout1));
stack.opt_show();
stack.o.undent();
stack.o << "}>";
stack.s.pop_back();
stack.modified();
return true;
} else {
stack.o << "UNTILEND";
stack.forget_const();
StackLayout layout1 = stack.vars();
block0->generate_code_all(stack);
layout1.push_back(left[0]);
stack.enforce_state(std::move(layout1));
stack.opt_show();
return false;
}
}
case _While: {
// while (block0 | left) block1; ...next
var_idx_t x = left[0];
stack.drop_vars_except(block0->var_info);
stack.opt_show();
StackLayout layout1 = stack.vars();
bool next_empty = false && next->is_empty();
if (block0->noreturn()) {
stack.o.retalt_ = true;
}
stack.o << "WHILE:<{";
stack.o.indent();
stack.forget_const();
stack.mode &= ~Stack::_InlineFunc;
stack.mode |= Stack::_NeedRetAlt;
block0->generate_code_all(stack);
stack.rearrange_top(x, !next->var_info[x] && !block1->var_info[x]);
stack.opt_show();
stack.s.pop_back();
stack.modified();
stack.o.undent();
Stack stack_copy{stack};
stack.o << (next_empty ? "}>DO:" : "}>DO<{");
if (!next_empty) {
stack.o.indent();
}
stack_copy.opt_show();
block1->generate_code_all(stack_copy);
stack_copy.enforce_state(std::move(layout1));
stack_copy.opt_show();
if (!next_empty) {
stack.o.undent();
stack.o << "}>";
return true;
} else {
return false;
}
}
case _TryCatch: {
if (block0->is_empty() && block1->is_empty()) {
return true;
}
if (block0->noreturn() || block1->noreturn()) {
stack.o.retalt_ = true;
}
Stack catch_stack{stack.o};
std::vector<var_idx_t> catch_vars;
std::vector<bool> catch_last;
for (const VarDescr& var : block1->var_info.list) {
if (stack.find(var.idx) >= 0) {
catch_vars.push_back(var.idx);
catch_last.push_back(!block0->var_info[var.idx]);
}
}
const size_t block_size = 255;
for (size_t begin = catch_vars.size(), end = begin; end > 0; end = begin) {
begin = end >= block_size ? end - block_size : 0;
for (size_t i = begin; i < end; ++i) {
catch_stack.push_new_var(catch_vars[i]);
}
}
catch_stack.push_new_var(left[0]);
catch_stack.push_new_var(left[1]);
stack.rearrange_top(catch_vars, catch_last);
stack.opt_show();
stack.o << "c4 PUSH";
stack.o << "c5 PUSH";
stack.o << "c7 PUSH";
stack.o << "<{";
stack.o.indent();
if (block1->noreturn()) {
catch_stack.mode |= Stack::_NeedRetAlt;
}
block1->generate_code_all(catch_stack);
catch_stack.drop_vars_except(next->var_info);
catch_stack.opt_show();
stack.o.undent();
stack.o << "}>CONT";
stack.o << "c7 SETCONT";
stack.o << "c5 SETCONT";
stack.o << "c4 SETCONT";
for (size_t begin = catch_vars.size(), end = begin; end > 0; end = begin) {
begin = end >= block_size ? end - block_size : 0;
stack.o << std::to_string(end - begin) + " PUSHINT";
stack.o << "-1 PUSHINT";
stack.o << "SETCONTVARARGS";
}
stack.s.erase(stack.s.end() - catch_vars.size(), stack.s.end());
stack.modified();
stack.o << "<{";
stack.o.indent();
if (block0->noreturn()) {
stack.mode |= Stack::_NeedRetAlt;
}
block0->generate_code_all(stack);
if (block0->noreturn()) {
stack.s = std::move(catch_stack.s);
} else if (!block1->noreturn()) {
stack.merge_state(catch_stack);
}
stack.opt_show();
stack.o.undent();
stack.o << "}>CONT";
stack.o << "c1 PUSH";
stack.o << "COMPOSALT";
stack.o << "SWAP";
stack.o << "TRY";
return true;
}
default:
std::cerr << "fatal: unknown operation <??" << cl << ">\n";
throw ParseError{where, "unknown operation in generate_code()"};
}
}
void Op::generate_code_all(Stack& stack) {
int saved_mode = stack.mode;
auto cont = generate_code_step(stack);
stack.mode = (stack.mode & ~Stack::_ModeSave) | (saved_mode & Stack::_ModeSave);
if (cont && next) {
next->generate_code_all(stack);
}
}
void CodeBlob::generate_code(AsmOpList& out, int mode) {
Stack stack{out, mode};
tolk_assert(ops && ops->cl == Op::_Import);
auto args = (int)ops->left.size();
for (var_idx_t x : ops->left) {
stack.push_new_var(x);
}
ops->generate_code_all(stack);
stack.apply_wrappers(require_callxargs && (mode & Stack::_InlineAny) ? args : -1);
if (!(mode & Stack::_DisableOpt)) {
optimize_code(out);
}
}
void CodeBlob::generate_code(std::ostream& os, int mode, int indent) {
AsmOpList out_list(indent, &vars);
generate_code(out_list, mode);
out_list.out(os, mode);
}
} // namespace tolk

449
tolk/gen-abscode.cpp Normal file
View file

@ -0,0 +1,449 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include <numeric>
#include "tolk.h"
using namespace std::literals::string_literals;
namespace tolk {
/*
*
* EXPRESSIONS
*
*/
Expr* Expr::copy() const {
auto res = new Expr{*this};
for (auto& arg : res->args) {
arg = arg->copy();
}
return res;
}
Expr::Expr(int c, sym_idx_t name_idx, std::initializer_list<Expr*> _arglist) : cls(c), args(std::move(_arglist)) {
sym = lookup_symbol(name_idx);
if (!sym) {
}
}
void Expr::chk_rvalue(const Lexem& lem) const {
if (!is_rvalue()) {
lem.error_at("rvalue expected before `", "`");
}
}
void Expr::chk_lvalue(const Lexem& lem) const {
if (!is_lvalue()) {
lem.error_at("lvalue expected before `", "`");
}
}
void Expr::chk_type(const Lexem& lem) const {
if (!is_type()) {
lem.error_at("type expression expected before `", "`");
}
}
bool Expr::deduce_type(const Lexem& lem) {
if (e_type) {
return true;
}
switch (cls) {
case _Apply: {
if (!sym) {
return false;
}
SymVal* sym_val = dynamic_cast<SymVal*>(sym->value);
if (!sym_val || !sym_val->get_type()) {
return false;
}
std::vector<TypeExpr*> arg_types;
for (const auto& arg : args) {
arg_types.push_back(arg->e_type);
}
TypeExpr* fun_type = TypeExpr::new_map(TypeExpr::new_tensor(arg_types), TypeExpr::new_hole());
try {
unify(fun_type, sym_val->sym_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "cannot apply function " << sym->name() << " : " << sym_val->get_type() << " to arguments of type "
<< fun_type->args[0] << ": " << ue;
lem.error(os.str());
}
e_type = fun_type->args[1];
TypeExpr::remove_indirect(e_type);
return true;
}
case _VarApply: {
tolk_assert(args.size() == 2);
TypeExpr* fun_type = TypeExpr::new_map(args[1]->e_type, TypeExpr::new_hole());
try {
unify(fun_type, args[0]->e_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "cannot apply expression of type " << args[0]->e_type << " to an expression of type " << args[1]->e_type
<< ": " << ue;
lem.error(os.str());
}
e_type = fun_type->args[1];
TypeExpr::remove_indirect(e_type);
return true;
}
case _Letop: {
tolk_assert(args.size() == 2);
try {
// std::cerr << "in assignment: " << args[0]->e_type << " from " << args[1]->e_type << std::endl;
unify(args[0]->e_type, args[1]->e_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "cannot assign an expression of type " << args[1]->e_type << " to a variable or pattern of type "
<< args[0]->e_type << ": " << ue;
lem.error(os.str());
}
e_type = args[0]->e_type;
TypeExpr::remove_indirect(e_type);
return true;
}
case _LetFirst: {
tolk_assert(args.size() == 2);
TypeExpr* rhs_type = TypeExpr::new_tensor({args[0]->e_type, TypeExpr::new_hole()});
try {
// std::cerr << "in implicit assignment of a modifying method: " << rhs_type << " and " << args[1]->e_type << std::endl;
unify(rhs_type, args[1]->e_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "cannot implicitly assign an expression of type " << args[1]->e_type
<< " to a variable or pattern of type " << rhs_type << " in modifying method `" << symbols.get_name(val)
<< "` : " << ue;
lem.error(os.str());
}
e_type = rhs_type->args[1];
TypeExpr::remove_indirect(e_type);
// std::cerr << "result type is " << e_type << std::endl;
return true;
}
case _CondExpr: {
tolk_assert(args.size() == 3);
auto flag_type = TypeExpr::new_atomic(_Int);
try {
unify(args[0]->e_type, flag_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "condition in a conditional expression has non-integer type " << args[0]->e_type << ": " << ue;
lem.error(os.str());
}
try {
unify(args[1]->e_type, args[2]->e_type);
} catch (UnifyError& ue) {
std::ostringstream os;
os << "the two variants in a conditional expression have different types " << args[1]->e_type << " and "
<< args[2]->e_type << " : " << ue;
lem.error(os.str());
}
e_type = args[1]->e_type;
TypeExpr::remove_indirect(e_type);
return true;
}
}
return false;
}
int Expr::define_new_vars(CodeBlob& code) {
switch (cls) {
case _Tensor:
case _MkTuple:
case _TypeApply: {
int res = 0;
for (const auto& x : args) {
res += x->define_new_vars(code);
}
return res;
}
case _Var:
if (val < 0) {
val = code.create_var(TmpVar::_Named, e_type, sym, &here);
return 1;
}
break;
case _Hole:
if (val < 0) {
val = code.create_var(TmpVar::_Tmp, e_type, nullptr, &here);
}
break;
}
return 0;
}
int Expr::predefine_vars() {
switch (cls) {
case _Tensor:
case _MkTuple:
case _TypeApply: {
int res = 0;
for (const auto& x : args) {
res += x->predefine_vars();
}
return res;
}
case _Var:
if (!sym) {
tolk_assert(val < 0 && here.defined());
if (prohibited_var_names.count(symbols.get_name(~val))) {
throw ParseError{
here, PSTRING() << "symbol `" << symbols.get_name(~val) << "` cannot be redefined as a variable"};
}
sym = define_symbol(~val, false, here);
// std::cerr << "predefining variable " << symbols.get_name(~val) << std::endl;
if (!sym) {
throw ParseError{here, std::string{"redefined variable `"} + symbols.get_name(~val) + "`"};
}
sym->value = new SymVal{SymVal::_Var, -1, e_type};
return 1;
}
break;
}
return 0;
}
var_idx_t Expr::new_tmp(CodeBlob& code) const {
return code.create_tmp_var(e_type, &here);
}
void add_set_globs(CodeBlob& code, std::vector<std::pair<SymDef*, var_idx_t>>& globs, const SrcLocation& here) {
for (const auto& p : globs) {
auto& op = code.emplace_back(here, Op::_SetGlob, std::vector<var_idx_t>{}, std::vector<var_idx_t>{ p.second }, p.first);
op.flags |= Op::_Impure;
}
}
std::vector<var_idx_t> Expr::pre_compile_let(CodeBlob& code, Expr* lhs, Expr* rhs, const SrcLocation& here) {
while (lhs->is_type_apply()) {
lhs = lhs->args.at(0);
}
while (rhs->is_type_apply()) {
rhs = rhs->args.at(0);
}
if (lhs->is_mktuple()) {
if (rhs->is_mktuple()) {
return pre_compile_let(code, lhs->args.at(0), rhs->args.at(0), here);
}
auto right = rhs->pre_compile(code);
TypeExpr::remove_indirect(rhs->e_type);
auto unpacked_type = rhs->e_type->args.at(0);
std::vector<var_idx_t> tmp{code.create_tmp_var(unpacked_type, &rhs->here)};
code.emplace_back(lhs->here, Op::_UnTuple, tmp, std::move(right));
auto tvar = new Expr{_Var};
tvar->set_val(tmp[0]);
tvar->set_location(rhs->here);
tvar->e_type = unpacked_type;
pre_compile_let(code, lhs->args.at(0), tvar, here);
return tmp;
}
auto right = rhs->pre_compile(code);
std::vector<std::pair<SymDef*, var_idx_t>> globs;
auto left = lhs->pre_compile(code, &globs);
for (var_idx_t v : left) {
code.on_var_modification(v, here);
}
code.emplace_back(here, Op::_Let, std::move(left), right);
add_set_globs(code, globs, here);
return right;
}
std::vector<var_idx_t> pre_compile_tensor(const std::vector<Expr *> args, CodeBlob &code,
std::vector<std::pair<SymDef*, var_idx_t>> *lval_globs,
std::vector<int> arg_order) {
if (arg_order.empty()) {
arg_order.resize(args.size());
std::iota(arg_order.begin(), arg_order.end(), 0);
}
tolk_assert(args.size() == arg_order.size());
std::vector<std::vector<var_idx_t>> res_lists(args.size());
struct ModifiedVar {
size_t i, j;
Op* op;
};
auto modified_vars = std::make_shared<std::vector<ModifiedVar>>();
for (size_t i : arg_order) {
res_lists[i] = args[i]->pre_compile(code, lval_globs);
for (size_t j = 0; j < res_lists[i].size(); ++j) {
TmpVar& var = code.vars.at(res_lists[i][j]);
if (code.flags & CodeBlob::_AllowPostModification) {
if (!lval_globs && (var.cls & TmpVar::_Named)) {
Op *op = &code.emplace_back(nullptr, Op::_Let, std::vector<var_idx_t>(), std::vector<var_idx_t>());
op->flags |= Op::_Disabled;
var.on_modification.push_back([modified_vars, i, j, op, done = false](const SrcLocation &here) mutable {
if (!done) {
done = true;
modified_vars->push_back({i, j, op});
}
});
} else {
var.on_modification.push_back([](const SrcLocation &) {
});
}
} else {
var.on_modification.push_back([name = var.to_string()](const SrcLocation &here) {
throw ParseError{here, PSTRING() << "Modifying local variable " << name
<< " after using it in the same expression"};
});
}
}
}
for (const auto& list : res_lists) {
for (var_idx_t v : list) {
tolk_assert(!code.vars.at(v).on_modification.empty());
code.vars.at(v).on_modification.pop_back();
}
}
for (const ModifiedVar &m : *modified_vars) {
var_idx_t& v = res_lists[m.i][m.j];
var_idx_t v2 = code.create_tmp_var(code.vars[v].v_type, code.vars[v].where.get());
m.op->left = {v2};
m.op->right = {v};
m.op->flags &= ~Op::_Disabled;
v = v2;
}
std::vector<var_idx_t> res;
for (const auto& list : res_lists) {
res.insert(res.end(), list.cbegin(), list.cend());
}
return res;
}
std::vector<var_idx_t> Expr::pre_compile(CodeBlob& code, std::vector<std::pair<SymDef*, var_idx_t>>* lval_globs) const {
if (lval_globs && !(cls == _Tensor || cls == _Var || cls == _Hole || cls == _TypeApply || cls == _GlobVar)) {
std::cerr << "lvalue expression constructor is " << cls << std::endl;
throw Fatal{"cannot compile lvalue expression with unknown constructor"};
}
switch (cls) {
case _Tensor: {
return pre_compile_tensor(args, code, lval_globs, {});
}
case _Apply: {
tolk_assert(sym);
auto func = dynamic_cast<SymValFunc*>(sym->value);
std::vector<var_idx_t> res;
if (func && func->arg_order.size() == args.size() && !(code.flags & CodeBlob::_ComputeAsmLtr)) {
//std::cerr << "!!! reordering " << args.size() << " arguments of " << sym->name() << std::endl;
res = pre_compile_tensor(args, code, lval_globs, func->arg_order);
} else {
res = pre_compile_tensor(args, code, lval_globs, {});
}
auto rvect = new_tmp_vect(code);
auto& op = code.emplace_back(here, Op::_Call, rvect, std::move(res), sym);
if (flags & _IsImpure) {
op.flags |= Op::_Impure;
}
return rvect;
}
case _TypeApply:
return args[0]->pre_compile(code, lval_globs);
case _Var:
case _Hole:
if (val < 0) {
throw ParseError{here, "unexpected variable definition"};
}
return {val};
case _VarApply:
if (args[0]->cls == _Glob) {
auto res = args[1]->pre_compile(code);
auto rvect = new_tmp_vect(code);
auto& op = code.emplace_back(here, Op::_Call, rvect, std::move(res), args[0]->sym);
if (args[0]->flags & _IsImpure) {
op.flags |= Op::_Impure;
}
return rvect;
} else {
auto res = args[1]->pre_compile(code);
auto tfunc = args[0]->pre_compile(code);
if (tfunc.size() != 1) {
throw Fatal{"stack tuple used as a function"};
}
res.push_back(tfunc[0]);
auto rvect = new_tmp_vect(code);
code.emplace_back(here, Op::_CallInd, rvect, std::move(res));
return rvect;
}
case _Const: {
auto rvect = new_tmp_vect(code);
code.emplace_back(here, Op::_IntConst, rvect, intval);
return rvect;
}
case _Glob:
case _GlobVar: {
auto rvect = new_tmp_vect(code);
if (lval_globs) {
lval_globs->push_back({ sym, rvect[0] });
return rvect;
} else {
code.emplace_back(here, Op::_GlobVar, rvect, std::vector<var_idx_t>{}, sym);
return rvect;
}
}
case _Letop: {
return pre_compile_let(code, args.at(0), args.at(1), here);
}
case _LetFirst: {
auto rvect = new_tmp_vect(code);
auto right = args[1]->pre_compile(code);
std::vector<std::pair<SymDef*, var_idx_t>> local_globs;
if (!lval_globs) {
lval_globs = &local_globs;
}
auto left = args[0]->pre_compile(code, lval_globs);
left.push_back(rvect[0]);
for (var_idx_t v : left) {
code.on_var_modification(v, here);
}
code.emplace_back(here, Op::_Let, std::move(left), std::move(right));
add_set_globs(code, local_globs, here);
return rvect;
}
case _MkTuple: {
auto left = new_tmp_vect(code);
auto right = args[0]->pre_compile(code);
code.emplace_back(here, Op::_Tuple, left, std::move(right));
return left;
}
case _CondExpr: {
auto cond = args[0]->pre_compile(code);
tolk_assert(cond.size() == 1);
auto rvect = new_tmp_vect(code);
Op& if_op = code.emplace_back(here, Op::_If, cond);
code.push_set_cur(if_op.block0);
code.emplace_back(here, Op::_Let, rvect, args[1]->pre_compile(code));
code.close_pop_cur(args[1]->here);
code.push_set_cur(if_op.block1);
code.emplace_back(here, Op::_Let, rvect, args[2]->pre_compile(code));
code.close_pop_cur(args[2]->here);
return rvect;
}
case _SliceConst: {
auto rvect = new_tmp_vect(code);
code.emplace_back(here, Op::_SliceConst, rvect, strval);
return rvect;
}
default:
std::cerr << "expression constructor is " << cls << std::endl;
throw Fatal{"cannot compile expression with unknown constructor"};
}
}
} // namespace tolk

126
tolk/keywords.cpp Normal file
View file

@ -0,0 +1,126 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* KEYWORD DEFINITION
*
*/
void define_keywords() {
symbols.add_kw_char('+')
.add_kw_char('-')
.add_kw_char('*')
.add_kw_char('/')
.add_kw_char('%')
.add_kw_char('?')
.add_kw_char(':')
.add_kw_char(',')
.add_kw_char(';')
.add_kw_char('(')
.add_kw_char(')')
.add_kw_char('[')
.add_kw_char(']')
.add_kw_char('{')
.add_kw_char('}')
.add_kw_char('=')
.add_kw_char('_')
.add_kw_char('<')
.add_kw_char('>')
.add_kw_char('&')
.add_kw_char('|')
.add_kw_char('^')
.add_kw_char('~');
symbols.add_keyword("==", Keyword::_Eq)
.add_keyword("!=", Keyword::_Neq)
.add_keyword("<=", Keyword::_Leq)
.add_keyword(">=", Keyword::_Geq)
.add_keyword("<=>", Keyword::_Spaceship)
.add_keyword("<<", Keyword::_Lshift)
.add_keyword(">>", Keyword::_Rshift)
.add_keyword("~>>", Keyword::_RshiftR)
.add_keyword("^>>", Keyword::_RshiftC)
.add_keyword("~/", Keyword::_DivR)
.add_keyword("^/", Keyword::_DivC)
.add_keyword("~%", Keyword::_ModR)
.add_keyword("^%", Keyword::_ModC)
.add_keyword("/%", Keyword::_DivMod)
.add_keyword("+=", Keyword::_PlusLet)
.add_keyword("-=", Keyword::_MinusLet)
.add_keyword("*=", Keyword::_TimesLet)
.add_keyword("/=", Keyword::_DivLet)
.add_keyword("~/=", Keyword::_DivRLet)
.add_keyword("^/=", Keyword::_DivCLet)
.add_keyword("%=", Keyword::_ModLet)
.add_keyword("~%=", Keyword::_ModRLet)
.add_keyword("^%=", Keyword::_ModCLet)
.add_keyword("<<=", Keyword::_LshiftLet)
.add_keyword(">>=", Keyword::_RshiftLet)
.add_keyword("~>>=", Keyword::_RshiftRLet)
.add_keyword("^>>=", Keyword::_RshiftCLet)
.add_keyword("&=", Keyword::_AndLet)
.add_keyword("|=", Keyword::_OrLet)
.add_keyword("^=", Keyword::_XorLet);
symbols.add_keyword("return", Keyword::_Return)
.add_keyword("var", Keyword::_Var)
.add_keyword("repeat", Keyword::_Repeat)
.add_keyword("do", Keyword::_Do)
.add_keyword("while", Keyword::_While)
.add_keyword("until", Keyword::_Until)
.add_keyword("try", Keyword::_Try)
.add_keyword("catch", Keyword::_Catch)
.add_keyword("if", Keyword::_If)
.add_keyword("ifnot", Keyword::_Ifnot)
.add_keyword("then", Keyword::_Then)
.add_keyword("else", Keyword::_Else)
.add_keyword("elseif", Keyword::_Elseif)
.add_keyword("elseifnot", Keyword::_Elseifnot);
symbols.add_keyword("int", Keyword::_Int)
.add_keyword("cell", Keyword::_Cell)
.add_keyword("slice", Keyword::_Slice)
.add_keyword("builder", Keyword::_Builder)
.add_keyword("cont", Keyword::_Cont)
.add_keyword("tuple", Keyword::_Tuple)
.add_keyword("type", Keyword::_Type)
.add_keyword("->", Keyword::_Mapsto)
.add_keyword("forall", Keyword::_Forall);
symbols.add_keyword("extern", Keyword::_Extern)
.add_keyword("global", Keyword::_Global)
.add_keyword("asm", Keyword::_Asm)
.add_keyword("impure", Keyword::_Impure)
.add_keyword("inline", Keyword::_Inline)
.add_keyword("inline_ref", Keyword::_InlineRef)
.add_keyword("auto_apply", Keyword::_AutoApply)
.add_keyword("method_id", Keyword::_MethodId)
.add_keyword("operator", Keyword::_Operator)
.add_keyword("infix", Keyword::_Infix)
.add_keyword("infixl", Keyword::_Infixl)
.add_keyword("infixr", Keyword::_Infixr)
.add_keyword("const", Keyword::_Const);
symbols.add_keyword("#pragma", Keyword::_PragmaHashtag)
.add_keyword("#include", Keyword::_IncludeHashtag);
}
} // namespace tolk

335
tolk/lexer.cpp Normal file
View file

@ -0,0 +1,335 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "lexer.h"
#include "symtable.h"
#include <sstream>
#include <cassert>
namespace tolk {
/*
*
* LEXER
*
*/
std::string Lexem::lexem_name_str(int idx) {
if (idx == Eof) {
return "end of file";
} else if (idx == Ident) {
return "identifier";
} else if (idx == Number) {
return "number";
} else if (idx == String) {
return "string";
} else if (idx == Special) {
return "special";
} else if (symbols.get_keyword(idx)) {
return "`" + symbols.get_keyword(idx)->str + "`";
} else {
std::ostringstream os{"<unknown lexem of type "};
os << idx << ">";
return os.str();
}
}
std::string Lexem::name_str() const {
if (tp == Ident) {
return std::string{"identifier `"} + symbols.get_name(val) + "`";
} else if (tp == String) {
return std::string{"string \""} + str + '"';
} else {
return lexem_name_str(tp);
}
}
bool is_number(std::string str) {
auto st = str.begin(), en = str.end();
if (st == en) {
return false;
}
if (*st == '-') {
st++;
}
bool hex = false;
if (st + 1 < en && *st == '0' && st[1] == 'x') {
st += 2;
hex = true;
}
if (st == en) {
return false;
}
while (st < en) {
int c = *st;
if (c >= '0' && c <= '9') {
++st;
continue;
}
if (!hex) {
return false;
}
c |= 0x20;
if (c < 'a' || c > 'f') {
return false;
}
++st;
}
return true;
}
int Lexem::classify() {
if (tp != Unknown) {
return tp;
}
sym_idx_t i = symbols.lookup(str);
if (i) {
assert(str == symbols[i]->str);
str = symbols[i]->str;
sym_idx_t idx = symbols[i]->idx;
tp = (idx < 0 ? -idx : Ident);
val = i;
} else if (is_number(str)) {
tp = Number;
} else {
tp = 0;
}
if (tp == Unknown) {
tp = Ident;
val = symbols.lookup(str, 1);
}
return tp;
}
int Lexem::set(std::string _str, const SrcLocation& _loc, int _tp, int _val) {
str = _str;
loc = _loc;
tp = _tp;
val = _val;
return classify();
}
Lexer::Lexer(SourceReader& _src, bool init, std::string active_chars, std::string eol_cmts, std::string open_cmts,
std::string close_cmts, std::string quote_chars, std::string multiline_quote)
: src(_src), eof(false), lexem("", src.here(), Lexem::Undefined), peek_lexem("", {}, Lexem::Undefined),
multiline_quote(std::move(multiline_quote)) {
std::memset(char_class, 0, sizeof(char_class));
unsigned char activity = cc::active;
for (char c : active_chars) {
if (c == ' ') {
if (!--activity) {
activity = cc::allow_repeat;
}
} else if ((unsigned)c < 0x80) {
char_class[(unsigned)c] |= activity;
}
}
set_spec(eol_cmt, eol_cmts);
set_spec(cmt_op, open_cmts);
set_spec(cmt_cl, close_cmts);
for (int c : quote_chars) {
if (c > ' ' && c <= 0x7f) {
char_class[(unsigned)c] |= cc::quote_char;
}
}
if (init) {
next();
}
}
void Lexer::set_spec(std::array<int, 3>& arr, std::string setup) {
arr[0] = arr[1] = arr[2] = -0x100;
std::size_t n = setup.size(), i;
for (i = 0; i < n; i++) {
if (setup[i] == ' ') {
continue;
}
if (i == n - 1 || setup[i + 1] == ' ') {
arr[0] = setup[i];
} else if (i == n - 2 || (i < n - 2 && setup[i + 2] == ' ')) {
arr[1] = setup[i];
arr[2] = setup[++i];
} else {
while (i < n && setup[i] != ' ') {
i++;
}
}
}
}
bool Lexer::is_multiline_quote(const char* begin, const char* end) {
if (multiline_quote.empty()) {
return false;
}
for (const char& c : multiline_quote) {
if (begin == end || *begin != c) {
return false;
}
++begin;
}
return true;
}
void Lexer::expect(int exp_tp, const char* msg) {
if (tp() != exp_tp) {
throw ParseError{lexem.loc, (msg ? std::string{msg} : Lexem::lexem_name_str(exp_tp)) + " expected instead of " +
cur().name_str()};
}
next();
}
const Lexem& Lexer::next() {
if (peek_lexem.valid()) {
lexem = std::move(peek_lexem);
peek_lexem.clear({}, Lexem::Undefined);
eof = (lexem.tp == Lexem::Eof);
return lexem;
}
if (eof) {
return lexem.clear(src.here(), Lexem::Eof);
}
long long comm = 1;
while (!src.seek_eof()) {
int cc = src.cur_char(), nc = src.next_char();
if (cc == eol_cmt[0] || (cc == eol_cmt[1] && nc == eol_cmt[2])) {
src.load_line();
} else if (cc == cmt_op[1] && nc == cmt_op[2]) {
src.advance(2);
comm = comm * 2 + 1;
} else if (cc == cmt_op[0]) {
src.advance(1);
comm *= 2;
} else if (comm == 1) {
break;
} else if (cc == cmt_cl[1] && nc == cmt_cl[2]) {
if (!(comm & 1)) {
src.error(std::string{"a `"} + (char)cmt_op[0] + "` comment closed by `" + (char)cmt_cl[1] + (char)cmt_cl[2] +
"`");
}
comm >>= 1;
src.advance(2);
} else if (cc == cmt_cl[0]) {
if (!(comm & 1)) {
src.error(std::string{"a `"} + (char)cmt_op[1] + (char)cmt_op[2] + "` comment closed by `" + (char)cmt_cl[0] +
"`");
}
comm >>= 1;
src.advance(1);
} else {
src.advance(1);
}
if (comm < 0) {
src.error("too many nested comments");
}
}
if (src.seek_eof()) {
eof = true;
if (comm > 1) {
if (comm & 1) {
src.error(std::string{"`"} + (char)cmt_op[1] + (char)cmt_op[2] + "` comment extends past end of file");
} else {
src.error(std::string{"`"} + (char)cmt_op[0] + "` comment extends past end of file");
}
}
return lexem.clear(src.here(), Lexem::Eof);
}
if (is_multiline_quote(src.get_ptr(), src.get_end_ptr())) {
src.advance(multiline_quote.size());
const char* end = nullptr;
SrcLocation here = src.here();
std::string body;
while (!src.is_eof()) {
if (src.is_eoln()) {
body.push_back('\n');
src.load_line();
continue;
}
if (is_multiline_quote(src.get_ptr(), src.get_end_ptr())) {
end = src.get_ptr();
src.advance(multiline_quote.size());
break;
}
body.push_back(src.cur_char());
src.advance(1);
}
if (!end) {
src.error("string extends past end of file");
}
lexem.set(body, here, Lexem::String);
int c = src.cur_char();
if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')) {
lexem.val = c;
src.advance(1);
}
return lexem;
}
int c = src.cur_char();
const char* end = src.get_ptr();
if (is_quote_char(c) || c == '`') {
int qc = c;
++end;
while (end < src.get_end_ptr() && *end != qc) {
++end;
}
if (*end != qc) {
src.error(qc == '`' ? "a `back-quoted` token extends past end of line" : "string extends past end of line");
}
lexem.set(std::string{src.get_ptr() + 1, end}, src.here(), qc == '`' ? Lexem::Unknown : Lexem::String);
src.set_ptr(end + 1);
c = src.cur_char();
if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')) {
lexem.val = c;
src.set_ptr(end + 2);
}
// std::cerr << lexem.name_str() << ' ' << lexem.str << std::endl;
return lexem;
}
int len = 0, pc = -0x100;
while (end < src.get_end_ptr()) {
c = *end;
bool repeated = (c == pc && is_repeatable(c));
if (c == ' ' || c == 9 || (len && is_left_active(c) && !repeated)) {
break;
}
++len;
++end;
if (is_right_active(c) && !repeated) {
break;
}
pc = c;
}
lexem.set(std::string{src.get_ptr(), end}, src.here());
src.set_ptr(end);
// std::cerr << lexem.name_str() << ' ' << lexem.str << std::endl;
return lexem;
}
const Lexem& Lexer::peek() {
if (peek_lexem.valid()) {
return peek_lexem;
}
if (eof) {
return lexem.clear(src.here(), Lexem::Eof);
}
Lexem keep = std::move(lexem);
next();
peek_lexem = std::move(lexem);
lexem = std::move(keep);
eof = false;
return peek_lexem;
}
} // namespace tolk

113
tolk/lexer.h Normal file
View file

@ -0,0 +1,113 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "srcread.h"
#include <array>
#include <memory>
#include <cstring>
namespace tolk {
/*
*
* LEXER
*
*/
struct Lexem {
enum { Undefined = -2, Eof = -1, Unknown = 0, Ident = 0, Number = 1, Special = 2, String = 3 };
int tp;
int val;
std::string str;
SrcLocation loc;
int classify();
Lexem(std::string _str = "", const SrcLocation& _loc = {}, int _tp = Unknown, int _val = 0)
: tp(_tp), val(_val), str(_str), loc(_loc) {
classify();
}
int set(std::string _str = "", const SrcLocation& _loc = {}, int _tp = Unknown, int _val = 0);
Lexem& clear(const SrcLocation& _loc = {}, int _tp = Unknown, int _val = 0) {
tp = _tp;
val = _val;
loc = _loc;
str = "";
return *this;
}
bool valid() const {
return tp != Undefined;
}
std::string name_str() const;
void error(std::string _str) const {
throw ParseError{loc, _str};
}
void error_at(std::string str1, std::string str2) const {
error(str1 + str + str2);
}
static std::string lexem_name_str(int idx);
};
class Lexer {
SourceReader& src;
bool eof;
Lexem lexem, peek_lexem;
unsigned char char_class[128];
std::array<int, 3> eol_cmt, cmt_op, cmt_cl;
std::string multiline_quote;
enum cc { left_active = 2, right_active = 1, active = 3, allow_repeat = 4, quote_char = 8 };
public:
bool eof_found() const {
return eof;
}
Lexer(SourceReader& _src, bool init = false, std::string active_chars = ";,() ~.", std::string eol_cmts = ";;",
std::string open_cmts = "{-", std::string close_cmts = "-}", std::string quote_chars = "\"",
std::string multiline_quote = "\"\"\"");
const Lexem& next();
const Lexem& cur() const {
return lexem;
}
const Lexem& peek();
int tp() const {
return lexem.tp;
}
void expect(int exp_tp, const char* msg = 0);
int classify_char(unsigned c) const {
return c < 0x80 ? char_class[c] : 0;
}
bool is_active(int c) const {
return (classify_char(c) & cc::active) == cc::active;
}
bool is_left_active(int c) const {
return (classify_char(c) & cc::left_active);
}
bool is_right_active(int c) const {
return (classify_char(c) & cc::right_active);
}
bool is_repeatable(int c) const {
return (classify_char(c) & cc::allow_repeat);
}
bool is_quote_char(int c) const {
return (classify_char(c) & cc::quote_char);
}
private:
void set_spec(std::array<int, 3>& arr, std::string setup);
bool is_multiline_quote(const char* begin, const char* end);
};
} // namespace tolk

652
tolk/optimize.cpp Normal file
View file

@ -0,0 +1,652 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* PEEPHOLE OPTIMIZER
*
*/
void Optimizer::set_code(AsmOpConsList code) {
code_ = std::move(code);
unpack();
}
void Optimizer::unpack() {
int i = 0, j = 0;
for (AsmOpCons *p = code_.get(); p && i < n; p = p->cdr.get(), ++j) {
if (p->car->is_very_custom()) {
break;
}
if (p->car->is_comment()) {
continue;
}
op_cons_[i] = p;
op_[i] = std::move(p->car);
offs_[i] = j;
++i;
}
l_ = i;
indent_ = (i ? op_[0]->indent : 0);
}
void Optimizer::pack() {
for (int i = 0; i < l_; i++) {
op_cons_[i]->car = std::move(op_[i]);
op_cons_[i] = nullptr;
}
l_ = 0;
}
void Optimizer::apply() {
if (!p_ && !q_) {
return;
}
tolk_assert(p_ > 0 && p_ <= l_ && q_ >= 0 && q_ <= n && l_ <= n);
for (int i = p_; i < l_; i++) {
tolk_assert(op_[i]);
op_cons_[i]->car = std::move(op_[i]);
op_cons_[i] = nullptr;
}
for (int c = offs_[p_ - 1]; c >= 0; --c) {
code_ = std::move(code_->cdr);
}
for (int j = q_ - 1; j >= 0; j--) {
tolk_assert(oq_[j]);
oq_[j]->indent = indent_;
code_ = AsmOpCons::cons(std::move(oq_[j]), std::move(code_));
}
l_ = 0;
}
AsmOpConsList Optimizer::extract_code() {
pack();
return std::move(code_);
}
void Optimizer::show_head() const {
if (!debug_) {
return;
}
std::cerr << "optimizing";
for (int i = 0; i < l_; i++) {
if (op_[i]) {
std::cerr << ' ' << *op_[i] << ' ';
} else {
std::cerr << " (null) ";
}
}
std::cerr << std::endl;
}
void Optimizer::show_left() const {
if (!debug_) {
return;
}
std::cerr << "// *** rewriting";
for (int i = 0; i < p_; i++) {
if (op_[i]) {
std::cerr << ' ' << *op_[i] << ' ';
} else {
std::cerr << " (null) ";
}
}
}
void Optimizer::show_right() const {
if (!debug_) {
return;
}
std::cerr << "->";
for (int i = 0; i < q_; i++) {
if (oq_[i]) {
std::cerr << ' ' << *oq_[i] << ' ';
} else {
std::cerr << " (null) ";
}
}
std::cerr << std::endl;
}
bool Optimizer::say(std::string str) const {
if (debug_) {
std::cerr << str << std::endl;
}
return true;
}
bool Optimizer::find_const_op(int* op_idx, int cst) {
for (int i = 0; i < l2_; i++) {
if (op_[i]->is_gconst() && tr_[i].get(0) == cst) {
*op_idx = i;
return true;
}
}
return false;
}
bool Optimizer::is_push_const(int* i, int* c) const {
return pb_ >= 3 && pb_ <= l2_ && tr_[pb_ - 1].is_push_const(i, c);
}
// PUSHCONST c ; PUSH s(i+1) ; SWAP -> PUSH s(i) ; PUSHCONST c
bool Optimizer::rewrite_push_const(int i, int c) {
p_ = pb_;
q_ = 2;
int idx = -1;
if (!(p_ >= 2 && find_const_op(&idx, c) && idx < p_)) {
return false;
}
show_left();
oq_[1] = std::move(op_[idx]);
oq_[0] = std::move(op_[!idx]);
*oq_[0] = AsmOp::Push(i);
show_right();
return true;
}
bool Optimizer::is_const_rot(int* c) const {
return pb_ >= 3 && pb_ <= l2_ && tr_[pb_ - 1].is_const_rot(c);
}
bool Optimizer::rewrite_const_rot(int c) {
p_ = pb_;
q_ = 2;
int idx = -1;
if (!(p_ >= 2 && find_const_op(&idx, c) && idx < p_)) {
return false;
}
show_left();
oq_[0] = std::move(op_[idx]);
oq_[1] = std::move(op_[!idx]);
*oq_[1] = AsmOp::Custom("ROT", 3, 3);
show_right();
return true;
}
bool Optimizer::is_const_pop(int* c, int* i) const {
return pb_ >= 3 && pb_ <= l2_ && tr_[pb_ - 1].is_const_pop(c, i);
}
bool Optimizer::rewrite_const_pop(int c, int i) {
p_ = pb_;
q_ = 2;
int idx = -1;
if (!(p_ >= 2 && find_const_op(&idx, c) && idx < p_)) {
return false;
}
show_left();
oq_[0] = std::move(op_[idx]);
oq_[1] = std::move(op_[!idx]);
*oq_[1] = AsmOp::Pop(i);
show_right();
return true;
}
bool Optimizer::is_const_push_xchgs() {
if (!(pb_ >= 2 && pb_ <= l2_ && op_[0]->is_gconst())) {
return false;
}
StackTransform t;
int pos = 0, i;
for (i = 1; i < pb_; i++) {
int a, b;
if (op_[i]->is_xchg(&a, &b)) {
if (pos == a) {
pos = b;
} else if (pos == b) {
pos = a;
} else {
t.apply_xchg(a - (a > pos), b - (b > pos));
}
} else if (op_[i]->is_push(&a)) {
if (pos == a) {
return false;
}
t.apply_push(a - (a > pos));
++pos;
} else {
return false;
}
}
if (pos) {
return false;
}
t.apply_push_newconst();
if (t <= tr_[i - 1]) {
p_ = i;
return true;
} else {
return false;
}
}
bool Optimizer::rewrite_const_push_xchgs() {
if (!p_) {
return false;
}
show_left();
auto c_op = std::move(op_[0]);
tolk_assert(c_op->is_gconst());
StackTransform t;
q_ = 0;
int pos = 0;
for (int i = 1; i < p_; i++) {
int a, b;
if (op_[i]->is_xchg(&a, &b)) {
if (a == pos) {
pos = b;
} else if (b == pos) {
pos = a;
} else {
oq_[q_] = std::move(op_[i]);
if (a > pos) {
oq_[q_]->a = a - 1;
}
if (b > pos) {
oq_[q_]->b = b - 1;
}
tolk_assert(apply_op(t, *oq_[q_]));
++q_;
}
} else {
tolk_assert(op_[i]->is_push(&a));
tolk_assert(a != pos);
oq_[q_] = std::move(op_[i]);
if (a > pos) {
oq_[q_]->a = a - 1;
}
tolk_assert(apply_op(t, *oq_[q_]));
++q_;
++pos;
}
}
tolk_assert(!pos);
t.apply_push_newconst();
tolk_assert(t <= tr_[p_ - 1]);
oq_[q_++] = std::move(c_op);
show_right();
return true;
}
bool Optimizer::rewrite(int p, AsmOp&& new_op) {
tolk_assert(p > 0 && p <= l_);
p_ = p;
q_ = 1;
show_left();
oq_[0] = std::move(op_[0]);
*oq_[0] = new_op;
show_right();
return true;
}
bool Optimizer::rewrite(int p, AsmOp&& new_op1, AsmOp&& new_op2) {
tolk_assert(p > 1 && p <= l_);
p_ = p;
q_ = 2;
show_left();
oq_[0] = std::move(op_[0]);
*oq_[0] = new_op1;
oq_[1] = std::move(op_[1]);
*oq_[1] = new_op2;
show_right();
return true;
}
bool Optimizer::rewrite(int p, AsmOp&& new_op1, AsmOp&& new_op2, AsmOp&& new_op3) {
tolk_assert(p > 2 && p <= l_);
p_ = p;
q_ = 3;
show_left();
oq_[0] = std::move(op_[0]);
*oq_[0] = new_op1;
oq_[1] = std::move(op_[1]);
*oq_[1] = new_op2;
oq_[2] = std::move(op_[2]);
*oq_[2] = new_op3;
show_right();
return true;
}
bool Optimizer::rewrite_nop() {
tolk_assert(p_ > 0 && p_ <= l_);
q_ = 0;
show_left();
show_right();
return true;
}
bool Optimizer::is_pred(const std::function<bool(const StackTransform&)>& pred, int min_p) {
min_p = std::max(min_p, pb_);
for (int p = l2_; p >= min_p; p--) {
if (pred(tr_[p - 1])) {
p_ = p;
return true;
}
}
return false;
}
bool Optimizer::is_same_as(const StackTransform& trans, int min_p) {
return is_pred([&trans](const auto& t) { return t >= trans; }, min_p);
}
// s1 s3 XCHG ; s0 s2 XCHG -> 2SWAP
bool Optimizer::is_2swap() {
static const StackTransform t_2swap{2, 3, 0, 1, 4};
return is_same_as(t_2swap);
}
// s3 PUSH ; s3 PUSH -> 2OVER
bool Optimizer::is_2over() {
static const StackTransform t_2over{2, 3, 0};
return is_same_as(t_2over);
}
bool Optimizer::is_2dup() {
static const StackTransform t_2dup{0, 1, 0};
return is_same_as(t_2dup);
}
bool Optimizer::is_tuck() {
static const StackTransform t_tuck{0, 1, 0, 2};
return is_same_as(t_tuck);
}
bool Optimizer::is_2drop() {
static const StackTransform t_2drop{2};
return is_same_as(t_2drop);
}
bool Optimizer::is_rot() {
return is_pred([](const auto& t) { return t.is_rot(); });
}
bool Optimizer::is_rotrev() {
return is_pred([](const auto& t) { return t.is_rotrev(); });
}
bool Optimizer::is_nop() {
return is_pred([](const auto& t) { return t.is_id(); }, 1);
}
bool Optimizer::is_xchg(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_xchg(i, j) && ((*i < 16 && *j < 16) || (!*i && *j < 256)); });
}
bool Optimizer::is_xchg_xchg(int* i, int* j, int* k, int* l) {
return is_pred([i, j, k, l](const auto& t) {
return t.is_xchg_xchg(i, j, k, l) && (*i < 2 && *j < (*i ? 16 : 256) && *k < 2 && *l < (*k ? 16 : 256));
}) &&
(!(p_ == 2 && op_[0]->is_xchg(*i, *j) && op_[1]->is_xchg(*k, *l)));
}
bool Optimizer::is_push(int* i) {
return is_pred([i](const auto& t) { return t.is_push(i) && *i < 256; });
}
bool Optimizer::is_pop(int* i) {
return is_pred([i](const auto& t) { return t.is_pop(i) && *i < 256; });
}
bool Optimizer::is_pop_pop(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_pop_pop(i, j) && *i < 256 && *j < 256; }, 3);
}
bool Optimizer::is_push_rot(int* i) {
return is_pred([i](const auto& t) { return t.is_push_rot(i) && *i < 16; }, 3);
}
bool Optimizer::is_push_rotrev(int* i) {
return is_pred([i](const auto& t) { return t.is_push_rotrev(i) && *i < 16; }, 3);
}
bool Optimizer::is_push_xchg(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_push_xchg(i, j, k) && *i < 16 && *j < 16 && *k < 16; }) &&
!(p_ == 2 && op_[0]->is_push() && op_[1]->is_xchg());
}
bool Optimizer::is_xchg2(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_xchg2(i, j) && *i < 16 && *j < 16; });
}
bool Optimizer::is_xcpu(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_xcpu(i, j) && *i < 16 && *j < 16; });
}
bool Optimizer::is_puxc(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_puxc(i, j) && *i < 16 && *j < 15; });
}
bool Optimizer::is_push2(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_push2(i, j) && *i < 16 && *j < 16; });
}
bool Optimizer::is_xchg3(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_xchg3(i, j, k) && *i < 16 && *j < 16 && *k < 16; });
}
bool Optimizer::is_xc2pu(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_xc2pu(i, j, k) && *i < 16 && *j < 16 && *k < 16; });
}
bool Optimizer::is_xcpuxc(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_xcpuxc(i, j, k) && *i < 16 && *j < 16 && *k < 15; });
}
bool Optimizer::is_xcpu2(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_xcpu2(i, j, k) && *i < 16 && *j < 16 && *k < 16; });
}
bool Optimizer::is_puxc2(int* i, int* j, int* k) {
return is_pred(
[i, j, k](const auto& t) { return t.is_puxc2(i, j, k) && *i < 16 && *j < 15 && *k < 15 && *j + *k != -1; });
}
bool Optimizer::is_puxcpu(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_puxcpu(i, j, k) && *i < 16 && *j < 15 && *k < 15; });
}
bool Optimizer::is_pu2xc(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_pu2xc(i, j, k) && *i < 16 && *j < 15 && *k < 14; });
}
bool Optimizer::is_push3(int* i, int* j, int* k) {
return is_pred([i, j, k](const auto& t) { return t.is_push3(i, j, k) && *i < 16 && *j < 16 && *k < 16; });
}
bool Optimizer::is_blkswap(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_blkswap(i, j) && *i > 0 && *j > 0 && *i <= 16 && *j <= 16; });
}
bool Optimizer::is_blkpush(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_blkpush(i, j) && *i > 0 && *i < 16 && *j < 16; });
}
bool Optimizer::is_blkdrop(int* i) {
return is_pred([i](const auto& t) { return t.is_blkdrop(i) && *i > 0 && *i < 16; });
}
bool Optimizer::is_blkdrop2(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_blkdrop2(i, j) && *i > 0 && *i < 16 && *j > 0 && *j < 16; });
}
bool Optimizer::is_reverse(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_reverse(i, j) && *i >= 2 && *i <= 17 && *j < 16; });
}
bool Optimizer::is_nip_seq(int* i, int* j) {
return is_pred([i, j](const auto& t) { return t.is_nip_seq(i, j) && *i >= 3 && *i <= 15; });
}
bool Optimizer::is_pop_blkdrop(int* i, int* k) {
return is_pred([i, k](const auto& t) { return t.is_pop_blkdrop(i, k) && *i >= *k && *k >= 2 && *k <= 15; }, 3);
}
bool Optimizer::is_2pop_blkdrop(int* i, int* j, int* k) {
return is_pred(
[i, j, k](const auto& t) { return t.is_2pop_blkdrop(i, j, k) && *i >= *k && *j >= *k && *k >= 2 && *k <= 15; },
3);
}
bool Optimizer::compute_stack_transforms() {
StackTransform trans;
for (int i = 0; i < l_; i++) {
if (!apply_op(trans, *op_[i])) {
l2_ = i;
return true;
}
tr_[i] = trans;
}
l2_ = l_;
return true;
}
bool Optimizer::show_stack_transforms() const {
show_head();
// slow version
/*
StackTransform trans2;
std::cerr << "id = " << trans2 << std::endl;
for (int i = 0; i < l_; i++) {
StackTransform op;
if (!apply_op(op, *op_[i])) {
std::cerr << "* (" << *op_[i] << " = invalid)\n";
break;
}
trans2 *= op;
std::cerr << "* " << *op_[i] << " = " << op << " -> " << trans2 << std::endl;
}
*/
// fast version
StackTransform trans;
for (int i = 0; i < l_; i++) {
std::cerr << trans << std::endl << *op_[i] << " -> ";
if (!apply_op(trans, *op_[i])) {
std::cerr << " <not-applicable>" << std::endl;
return true;
}
}
std::cerr << trans << std::endl;
return true;
}
bool Optimizer::find_at_least(int pb) {
p_ = q_ = 0;
pb_ = pb;
// show_stack_transforms();
int i, j, k, l, c;
return (is_push_const(&i, &c) && rewrite_push_const(i, c)) || (is_nop() && rewrite_nop()) ||
(!(mode_ & 1) && is_const_rot(&c) && rewrite_const_rot(c)) ||
(is_const_push_xchgs() && rewrite_const_push_xchgs()) || (is_const_pop(&c, &i) && rewrite_const_pop(c, i)) ||
(is_xchg(&i, &j) && rewrite(AsmOp::Xchg(i, j))) || (is_push(&i) && rewrite(AsmOp::Push(i))) ||
(is_pop(&i) && rewrite(AsmOp::Pop(i))) || (is_pop_pop(&i, &j) && rewrite(AsmOp::Pop(i), AsmOp::Pop(j))) ||
(is_xchg_xchg(&i, &j, &k, &l) && rewrite(AsmOp::Xchg(i, j), AsmOp::Xchg(k, l))) ||
(!(mode_ & 1) &&
((is_rot() && rewrite(AsmOp::Custom("ROT", 3, 3))) || (is_rotrev() && rewrite(AsmOp::Custom("-ROT", 3, 3))) ||
(is_2dup() && rewrite(AsmOp::Custom("2DUP", 2, 4))) ||
(is_2swap() && rewrite(AsmOp::Custom("2SWAP", 2, 4))) ||
(is_2over() && rewrite(AsmOp::Custom("2OVER", 2, 4))) ||
(is_tuck() && rewrite(AsmOp::Custom("TUCK", 2, 3))) ||
(is_2drop() && rewrite(AsmOp::Custom("2DROP", 2, 0))) || (is_xchg2(&i, &j) && rewrite(AsmOp::Xchg2(i, j))) ||
(is_xcpu(&i, &j) && rewrite(AsmOp::XcPu(i, j))) || (is_puxc(&i, &j) && rewrite(AsmOp::PuXc(i, j))) ||
(is_push2(&i, &j) && rewrite(AsmOp::Push2(i, j))) || (is_blkswap(&i, &j) && rewrite(AsmOp::BlkSwap(i, j))) ||
(is_blkpush(&i, &j) && rewrite(AsmOp::BlkPush(i, j))) || (is_blkdrop(&i) && rewrite(AsmOp::BlkDrop(i))) ||
(is_push_rot(&i) && rewrite(AsmOp::Push(i), AsmOp::Custom("ROT"))) ||
(is_push_rotrev(&i) && rewrite(AsmOp::Push(i), AsmOp::Custom("-ROT"))) ||
(is_push_xchg(&i, &j, &k) && rewrite(AsmOp::Push(i), AsmOp::Xchg(j, k))) ||
(is_reverse(&i, &j) && rewrite(AsmOp::BlkReverse(i, j))) ||
(is_blkdrop2(&i, &j) && rewrite(AsmOp::BlkDrop2(i, j))) ||
(is_nip_seq(&i, &j) && rewrite(AsmOp::Xchg(i, j), AsmOp::BlkDrop(i))) ||
(is_pop_blkdrop(&i, &k) && rewrite(AsmOp::Pop(i), AsmOp::BlkDrop(k))) ||
(is_2pop_blkdrop(&i, &j, &k) && (k >= 3 && k <= 13 && i != j + 1 && i <= 15 && j <= 14
? rewrite(AsmOp::Xchg2(j + 1, i), AsmOp::BlkDrop(k + 2))
: rewrite(AsmOp::Pop(i), AsmOp::Pop(j), AsmOp::BlkDrop(k)))) ||
(is_xchg3(&i, &j, &k) && rewrite(AsmOp::Xchg3(i, j, k))) ||
(is_xc2pu(&i, &j, &k) && rewrite(AsmOp::Xc2Pu(i, j, k))) ||
(is_xcpuxc(&i, &j, &k) && rewrite(AsmOp::XcPuXc(i, j, k))) ||
(is_xcpu2(&i, &j, &k) && rewrite(AsmOp::XcPu2(i, j, k))) ||
(is_puxc2(&i, &j, &k) && rewrite(AsmOp::PuXc2(i, j, k))) ||
(is_puxcpu(&i, &j, &k) && rewrite(AsmOp::PuXcPu(i, j, k))) ||
(is_pu2xc(&i, &j, &k) && rewrite(AsmOp::Pu2Xc(i, j, k))) ||
(is_push3(&i, &j, &k) && rewrite(AsmOp::Push3(i, j, k)))));
}
bool Optimizer::find() {
if (!compute_stack_transforms()) {
return false;
}
for (int pb = l_; pb > 0; --pb) {
if (find_at_least(pb)) {
return true;
}
}
return false;
}
bool Optimizer::optimize() {
bool f = false;
while (find()) {
f = true;
apply();
unpack();
}
return f;
}
AsmOpConsList optimize_code_head(AsmOpConsList op_list, int mode) {
Optimizer opt(std::move(op_list), op_rewrite_comments, mode);
opt.optimize();
return opt.extract_code();
}
AsmOpConsList optimize_code(AsmOpConsList op_list, int mode) {
std::vector<std::unique_ptr<AsmOp>> v;
while (op_list) {
if (!op_list->car->is_comment()) {
op_list = optimize_code_head(std::move(op_list), mode);
}
if (op_list) {
v.push_back(std::move(op_list->car));
op_list = std::move(op_list->cdr);
}
}
for (auto it = v.rbegin(); it < v.rend(); ++it) {
op_list = AsmOpCons::cons(std::move(*it), std::move(op_list));
}
return std::move(op_list);
}
void optimize_code(AsmOpList& ops) {
AsmOpConsList op_list;
for (auto it = ops.list_.rbegin(); it < ops.list_.rend(); ++it) {
op_list = AsmOpCons::cons(std::make_unique<AsmOp>(std::move(*it)), std::move(op_list));
}
for (int mode : {1, 1, 1, 1, 0, 0, 0, 0}) {
op_list = optimize_code(std::move(op_list), mode);
}
ops.list_.clear();
while (op_list) {
ops.list_.push_back(std::move(*(op_list->car)));
op_list = std::move(op_list->cdr);
}
}
} // namespace tolk

1809
tolk/parse-tolk.cpp Normal file
View file

File diff suppressed because it is too large Load diff

228
tolk/srcread.cpp Normal file
View file

@ -0,0 +1,228 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "srcread.h"
#include <algorithm>
namespace tolk {
/*
*
* SOURCE FILE READER
*
*/
std::ostream& operator<<(std::ostream& os, const FileDescr* fdescr) {
return os << (fdescr ? (fdescr->is_stdin ? "stdin" : fdescr->filename) : "unknown-location");
}
std::ostream& operator<<(std::ostream& os, const Fatal& fatal) {
return os << fatal.get_msg();
}
const char* FileDescr::convert_offset(long offset, long* line_no, long* line_pos, long* line_size) const {
long lno = 0, lpos = -1, lsize = 0;
const char* lstart = nullptr;
if (offset >= 0 && offset < (long)text.size()) {
auto it = std::upper_bound(line_offs.begin(), line_offs.end(), offset);
lno = it - line_offs.begin();
if (lno && it != line_offs.end()) {
lsize = it[0] - it[-1];
lpos = offset - it[-1];
lstart = text.data() + it[-1];
}
} else {
lno = (long)line_offs.size();
}
if (line_no) {
*line_no = lno;
}
if (line_pos) {
*line_pos = lpos;
}
if (line_size) {
*line_size = lsize;
}
return lstart;
}
const char* FileDescr::push_line(std::string new_line) {
if (line_offs.empty()) {
line_offs.push_back(0);
}
std::size_t cur_size = text.size();
text += new_line;
text += '\0';
line_offs.push_back((long)text.size());
return text.data() + cur_size;
}
void SrcLocation::show(std::ostream& os) const {
os << fdescr;
long line_no, line_pos;
if (fdescr && convert_pos(&line_no, &line_pos)) {
os << ':' << line_no;
if (line_pos >= 0) {
os << ':' << (line_pos + 1);
}
}
}
bool SrcLocation::show_context(std::ostream& os) const {
long line_no, line_pos, line_size;
if (!fdescr || !convert_pos(&line_no, &line_pos, &line_size)) {
return false;
}
bool skip_left = (line_pos > 200), skip_right = (line_pos + 200u < line_size);
const char* here = fdescr->text.data() + char_offs;
const char* base = here - line_pos;
const char* start = skip_left ? here - 100 : base;
const char* end = skip_right ? here + 100 : base + line_size;
os << " ";
if (skip_left) {
os << "... ";
}
for (const char* ptr = start; ptr < end; ptr++) {
os << (char)*ptr;
}
if (skip_right) {
os << " ...";
}
os << std::endl;
os << " ";
if (skip_left) {
os << "... ";
}
for (const char* ptr = start; ptr < here; ptr++) {
char c = *ptr;
os << (c == 9 || c == 10 ? c : ' ');
}
os << '^' << std::endl;
return true;
}
std::ostream& operator<<(std::ostream& os, const SrcLocation& loc) {
loc.show(os);
return os;
}
void SrcLocation::show_gen_error(std::ostream& os, std::string message, std::string err_type) const {
show(os);
if (!err_type.empty()) {
os << ": " << err_type;
}
os << ": " << message << std::endl;
show_context(os);
}
std::ostream& operator<<(std::ostream& os, const Error& error) {
error.show(os);
return os;
}
void ParseError::show(std::ostream& os) const {
os << where << ": error: " << message << std::endl;
where.show_context(os);
}
SourceReader::SourceReader(std::istream* _is, FileDescr* _fdescr)
: ifs(_is), fdescr(_fdescr), loc(_fdescr), eof(false), cur_line_len(0), start(0), cur(0), end(0) {
load_line();
}
void SourceReader::set_eof() {
if (!eof) {
eof = true;
start = cur = end = 0;
}
}
int SourceReader::skip_spc() {
if (!cur) {
return 0;
}
const char* ptr = cur;
int res = 0;
while (*ptr == ' ' || *ptr == 9) {
++ptr;
++res;
}
set_ptr(ptr);
return res;
}
bool SourceReader::seek_eof() {
while (seek_eoln()) {
if (!load_line()) {
return true;
}
}
return false;
}
const char* SourceReader::set_ptr(const char* ptr) {
if (ptr != cur) {
if (ptr < cur || ptr > end) {
error("parsing position went outside of line");
}
loc.char_offs += ptr - cur;
cur = ptr;
}
return ptr;
}
bool SourceReader::load_line() {
if (eof) {
return false;
}
loc.set_eof();
if (ifs->eof()) {
set_eof();
return false;
}
std::getline(*ifs, cur_line);
if (ifs->fail()) {
set_eof();
if (!ifs->eof()) {
error("cannot read line from source stream");
}
return false;
}
std::size_t len = cur_line.size();
if (len > 0xffffff) {
set_eof();
error("line too long");
return false;
}
if (len && cur_line.back() == '\r') {
// CP/M line breaks support
cur_line.pop_back();
--len;
}
cur_line_len = (int)len;
if (fdescr) {
cur = start = fdescr->push_line(std::move(cur_line));
end = start + len;
loc.char_offs = (std::size_t)(cur - fdescr->text.data());
cur_line.clear();
} else {
cur = start = cur_line.c_str();
end = start + cur_line_len;
}
return true;
}
} // namespace tolk

162
tolk/srcread.h Normal file
View file

@ -0,0 +1,162 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <string>
#include <vector>
#include <iostream>
namespace tolk {
/*
*
* SOURCE FILE READER
*
*/
struct FileDescr {
std::string filename;
std::string text;
std::vector<long> line_offs;
bool is_stdin;
bool is_main = false;
FileDescr(std::string _fname, bool _stdin = false) : filename(std::move(_fname)), is_stdin(_stdin) {
}
const char* push_line(std::string new_line);
const char* convert_offset(long offset, long* line_no, long* line_pos, long* line_size = nullptr) const;
};
struct Fatal {
std::string message;
Fatal(std::string _msg) : message(std::move(_msg)) {
}
std::string get_msg() const {
return message;
}
};
std::ostream& operator<<(std::ostream& os, const Fatal& fatal);
struct SrcLocation {
const FileDescr* fdescr;
long char_offs;
SrcLocation() : fdescr(nullptr), char_offs(-1) {
}
SrcLocation(const FileDescr* _fdescr, long offs = -1) : fdescr(_fdescr), char_offs(-1) {
}
bool defined() const {
return fdescr;
}
bool eof() const {
return char_offs == -1;
}
void set_eof() {
char_offs = -1;
}
const char* convert_pos(long* line_no, long* line_pos, long* line_size = nullptr) const {
return defined() ? fdescr->convert_offset(char_offs, line_no, line_pos, line_size) : nullptr;
}
void show(std::ostream& os) const;
bool show_context(std::ostream& os) const;
void show_gen_error(std::ostream& os, std::string message, std::string err_type = "") const;
void show_note(std::string err_msg) const {
show_gen_error(std::cerr, err_msg, "note");
}
void show_warning(std::string err_msg) const {
show_gen_error(std::cerr, err_msg, "warning");
}
void show_error(std::string err_msg) const {
show_gen_error(std::cerr, err_msg, "error");
}
};
std::ostream& operator<<(std::ostream& os, const SrcLocation& loc);
struct Error {
virtual ~Error() = default;
virtual void show(std::ostream& os) const = 0;
};
std::ostream& operator<<(std::ostream& os, const Error& error);
struct ParseError : Error {
SrcLocation where;
std::string message;
ParseError(const SrcLocation& _where, std::string _msg) : where(_where), message(_msg) {
}
ParseError(const SrcLocation* _where, std::string _msg) : message(_msg) {
if (_where) {
where = *_where;
}
}
~ParseError() override = default;
void show(std::ostream& os) const override;
};
class SourceReader {
std::istream* ifs;
FileDescr* fdescr;
SrcLocation loc;
bool eof;
std::string cur_line;
int cur_line_len;
void set_eof();
const char *start, *cur, *end;
public:
SourceReader(std::istream* _is, FileDescr* _fdescr);
bool load_line();
bool is_eof() const {
return eof;
}
int is_eoln() const {
return cur == end;
}
int skip_spc();
bool seek_eoln() {
skip_spc();
return is_eoln();
}
bool seek_eof();
const char* cur_line_cstr() const {
return cur_line.c_str();
}
const SrcLocation& here() const {
return loc;
}
char cur_char() const {
return *cur;
}
char next_char() const {
return cur[1];
}
const char* get_ptr() const {
return cur;
}
const char* get_end_ptr() const {
return end;
}
const char* set_ptr(const char* ptr);
void advance(int n) {
set_ptr(get_ptr() + n);
}
void error(std::string err_msg) {
throw ParseError{loc, err_msg};
}
};
} // namespace tolk

1054
tolk/stack-transform.cpp Normal file
View file

File diff suppressed because it is too large Load diff

179
tolk/symtable.cpp Normal file
View file

@ -0,0 +1,179 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "symtable.h"
#include <sstream>
#include <cassert>
namespace tolk {
/*
*
* SYMBOL VALUES (DECLARED)
*
*/
int scope_level;
SymTable<100003> symbols;
SymDef* sym_def[symbols.hprime + 1];
SymDef* global_sym_def[symbols.hprime + 1];
std::vector<std::pair<int, SymDef>> symbol_stack;
std::vector<SrcLocation> scope_opened_at;
std::string Symbol::unknown_symbol_name(sym_idx_t i) {
if (!i) {
return "_";
} else {
std::ostringstream os;
os << "SYM#" << i;
return os.str();
}
}
sym_idx_t SymTableBase::gen_lookup(std::string str, int mode, sym_idx_t idx) {
unsigned long long h1 = 1, h2 = 1;
for (char c : str) {
h1 = ((h1 * 239) + (unsigned char)(c)) % p;
h2 = ((h2 * 17) + (unsigned char)(c)) % (p - 1);
}
++h2;
++h1;
while (true) {
if (sym_table[h1]) {
if (sym_table[h1]->str == str) {
return (mode & 2) ? not_found : sym_idx_t(h1);
}
h1 += h2;
if (h1 > p) {
h1 -= p;
}
} else {
if (!(mode & 1)) {
return not_found;
}
if (def_sym >= ((long long)p * 3) / 4) {
throw SymTableOverflow{def_sym};
}
sym_table[h1] = std::make_unique<Symbol>(str, idx <= 0 ? sym_idx_t(h1) : -idx);
++def_sym;
return sym_idx_t(h1);
}
}
}
SymTableBase& SymTableBase::add_keyword(std::string str, sym_idx_t idx) {
if (idx <= 0) {
idx = ++def_kw;
}
sym_idx_t res = gen_lookup(str, -1, idx);
if (!res) {
throw SymTableKwRedef{str};
}
if (idx < max_kw_idx) {
keywords[idx] = res;
}
return *this;
}
void open_scope(Lexer& lex) {
++scope_level;
scope_opened_at.push_back(lex.cur().loc);
}
void close_scope(Lexer& lex) {
if (!scope_level) {
throw Fatal{"cannot close the outer scope"};
}
while (!symbol_stack.empty() && symbol_stack.back().first == scope_level) {
SymDef old_def = symbol_stack.back().second;
auto idx = old_def.sym_idx;
symbol_stack.pop_back();
SymDef* cur_def = sym_def[idx];
assert(cur_def);
assert(cur_def->level == scope_level && cur_def->sym_idx == idx);
//std::cerr << "restoring local symbol `" << old_def.name << "` of level " << scope_level << " to its previous level " << old_def.level << std::endl;
if (cur_def->value) {
//std::cerr << "deleting value of symbol " << old_def.name << ":" << old_def.level << " at " << (const void*) it->second.value << std::endl;
delete cur_def->value;
}
if (!old_def.level && !old_def.value) {
delete cur_def; // ??? keep the definition always?
sym_def[idx] = nullptr;
} else {
cur_def->value = std::move(old_def.value);
cur_def->level = old_def.level;
}
old_def.value = nullptr;
}
--scope_level;
scope_opened_at.pop_back();
}
SymDef* lookup_symbol(sym_idx_t idx, int flags) {
if (!idx) {
return nullptr;
}
if ((flags & 1) && sym_def[idx]) {
return sym_def[idx];
}
if ((flags & 2) && global_sym_def[idx]) {
return global_sym_def[idx];
}
return nullptr;
}
SymDef* lookup_symbol(std::string name, int flags) {
return lookup_symbol(symbols.lookup(name), flags);
}
SymDef* define_global_symbol(sym_idx_t name_idx, bool force_new, const SrcLocation& loc) {
if (!name_idx) {
return nullptr;
}
auto found = global_sym_def[name_idx];
if (found) {
return force_new && found->value ? nullptr : found;
}
return global_sym_def[name_idx] = new SymDef(0, name_idx, loc);
}
SymDef* define_symbol(sym_idx_t name_idx, bool force_new, const SrcLocation& loc) {
if (!name_idx) {
return nullptr;
}
if (!scope_level) {
return define_global_symbol(name_idx, force_new, loc);
}
auto found = sym_def[name_idx];
if (found) {
if (found->level < scope_level) {
symbol_stack.push_back(std::make_pair(scope_level, *found));
found->level = scope_level;
} else if (found->value && force_new) {
return nullptr;
}
found->value = 0;
found->loc = loc;
return found;
}
found = sym_def[name_idx] = new SymDef(scope_level, name_idx, loc);
symbol_stack.push_back(std::make_pair(scope_level, SymDef{0, name_idx}));
return found;
}
} // namespace tolk

175
tolk/symtable.h Normal file
View file

@ -0,0 +1,175 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "srcread.h"
#include "lexer.h"
#include <vector>
namespace tolk {
/*
*
* SYMBOL VALUES (DECLARED)
*
*/
typedef int var_idx_t;
struct SymValBase {
enum { _Param, _Var, _Func, _Typename, _GlobVar, _Const };
int type;
int idx;
SymValBase(int _type, int _idx) : type(_type), idx(_idx) {
}
virtual ~SymValBase() = default;
};
/*
*
* SYMBOL TABLE
*
*/
// defined outside this module (by the end user)
int compute_symbol_subclass(std::string str); // return 0 if unneeded
typedef int sym_idx_t;
struct Symbol {
std::string str;
sym_idx_t idx;
int subclass;
Symbol(std::string _str, sym_idx_t _idx, int _sc) : str(_str), idx(_idx), subclass(_sc) {
}
Symbol(std::string _str, sym_idx_t _idx) : str(_str), idx(_idx) {
subclass = compute_symbol_subclass(std::move(_str));
}
static std::string unknown_symbol_name(sym_idx_t i);
};
class SymTableBase {
unsigned p;
std::unique_ptr<Symbol>* sym_table;
sym_idx_t def_kw, def_sym;
static constexpr int max_kw_idx = 10000;
sym_idx_t keywords[max_kw_idx];
public:
SymTableBase(unsigned p_, std::unique_ptr<Symbol>* sym_table_)
: p(p_), sym_table(sym_table_), def_kw(0x100), def_sym(0) {
std::memset(keywords, 0, sizeof(keywords));
}
static constexpr sym_idx_t not_found = 0;
SymTableBase& add_keyword(std::string str, sym_idx_t idx = 0);
SymTableBase& add_kw_char(char c) {
return add_keyword(std::string{c}, c);
}
sym_idx_t lookup(std::string str, int mode = 0) {
return gen_lookup(str, mode);
}
sym_idx_t lookup_add(std::string str) {
return gen_lookup(str, 1);
}
Symbol* operator[](sym_idx_t i) const {
return sym_table[i].get();
}
bool is_keyword(sym_idx_t i) const {
return sym_table[i] && sym_table[i]->idx < 0;
}
std::string get_name(sym_idx_t i) const {
return sym_table[i] ? sym_table[i]->str : Symbol::unknown_symbol_name(i);
}
int get_subclass(sym_idx_t i) const {
return sym_table[i] ? sym_table[i]->subclass : 0;
}
Symbol* get_keyword(int i) const {
return ((unsigned)i < (unsigned)max_kw_idx) ? sym_table[keywords[i]].get() : nullptr;
}
protected:
sym_idx_t gen_lookup(std::string str, int mode = 0, sym_idx_t idx = 0);
};
template <unsigned pp>
class SymTable : public SymTableBase {
public:
static constexpr int hprime = pp;
static int size() {
return pp + 1;
}
private:
std::unique_ptr<Symbol> sym[pp + 1];
public:
SymTable() : SymTableBase(pp, sym) {
}
SymTable& add_keyword(std::string str, sym_idx_t idx = 0) {
SymTableBase::add_keyword(str, idx);
return *this;
}
SymTable& add_kw_char(char c) {
return add_keyword(std::string{c}, c);
}
};
struct SymTableOverflow {
int sym_def;
SymTableOverflow(int x) : sym_def(x) {
}
};
struct SymTableKwRedef {
std::string kw;
SymTableKwRedef(std::string _kw) : kw(_kw) {
}
};
extern SymTable<100003> symbols;
extern int scope_level;
struct SymDef {
int level;
sym_idx_t sym_idx;
SymValBase* value;
SrcLocation loc;
SymDef(int lvl, sym_idx_t idx, const SrcLocation& _loc = {}, SymValBase* val = 0)
: level(lvl), sym_idx(idx), value(val), loc(_loc) {
}
bool has_name() const {
return sym_idx;
}
std::string name() const {
return symbols.get_name(sym_idx);
}
};
extern SymDef* sym_def[symbols.hprime + 1];
extern SymDef* global_sym_def[symbols.hprime + 1];
extern std::vector<std::pair<int, SymDef>> symbol_stack;
extern std::vector<SrcLocation> scope_opened_at;
void open_scope(Lexer& lex);
void close_scope(Lexer& lex);
SymDef* lookup_symbol(sym_idx_t idx, int flags = 3);
SymDef* lookup_symbol(std::string name, int flags = 3);
SymDef* define_global_symbol(sym_idx_t name_idx, bool force_new = false, const SrcLocation& loc = {});
SymDef* define_symbol(sym_idx_t name_idx, bool force_new = false, const SrcLocation& loc = {});
} // namespace tolk

122
tolk/tolk-main.cpp Normal file
View file

@ -0,0 +1,122 @@
/*
This file is part of TON Blockchain source code.
TON Blockchain is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
TON Blockchain is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with TON Blockchain. If not, see <http://www.gnu.org/licenses/>.
In addition, as a special exception, the copyright holders give permission
to link the code of portions of this program with the OpenSSL library.
You must obey the GNU General Public License in all respects for all
of the code used other than OpenSSL. If you modify file(s) with this
exception, you may extend this exception to your version of the file(s),
but you are not obligated to do so. If you do not wish to do so, delete this
exception statement from your version. If you delete this exception statement
from all source files in the program, then also delete it here.
*/
#include "tolk.h"
#include <getopt.h>
#include <fstream>
#include "git.h"
void usage(const char* progname) {
std::cerr
<< "usage: " << progname
<< " [-vIAPSR][-O<level>][-i<indent-spc>][-o<output-filename>][-W<boc-filename>] {<filename.tolk> ...}\n"
"\tGenerates Fift TVM assembler code from a Tolk source\n"
"-I\tEnables interactive mode (parse stdin)\n"
"-o<fift-output-filename>\tWrites generated code into specified file instead of stdout\n"
"-v\tIncreases verbosity level (extra information output into stderr)\n"
"-i<indent>\tSets indentation for the output code (in two-space units)\n"
"-A\tPrefix code with `\"Asm.fif\" include` preamble\n"
"-O<level>\tSets optimization level (2 by default)\n"
"-P\tEnvelope code into PROGRAM{ ... }END>c\n"
"-S\tInclude stack layout comments in the output code\n"
"-R\tInclude operation rewrite comments in the output code\n"
"-W<output-boc-file>\tInclude Fift code to serialize and save generated code into specified BoC file. Enables "
"-A and -P.\n"
"\t-s\tOutput semantic version of Tolk and exit\n"
"\t-V<version>\tShow Tolk build information\n";
std::exit(2);
}
int main(int argc, char* const argv[]) {
int i;
std::string output_filename;
while ((i = getopt(argc, argv, "Ahi:Io:O:PRsSvW:V")) != -1) {
switch (i) {
case 'A':
tolk::asm_preamble = true;
break;
case 'I':
tolk::interactive = true;
break;
case 'i':
tolk::indent = std::max(0, atoi(optarg));
break;
case 'o':
output_filename = optarg;
break;
case 'O':
tolk::opt_level = std::max(0, atoi(optarg));
break;
case 'P':
tolk::program_envelope = true;
break;
case 'R':
tolk::op_rewrite_comments = true;
break;
case 'S':
tolk::stack_layout_comments = true;
break;
case 'v':
++tolk::verbosity;
break;
case 'W':
tolk::boc_output_filename = optarg;
tolk::asm_preamble = tolk::program_envelope = true;
break;
case 's':
std::cout << tolk::tolk_version << "\n";
std::exit(0);
case 'V':
std::cout << "Tolk semantic version: v" << tolk::tolk_version << "\n";
std::cout << "Build information: [ Commit: " << GitMetadata::CommitSHA1() << ", Date: " << GitMetadata::CommitDate() << "]\n";
std::exit(0);
case 'h':
default:
usage(argv[0]);
}
}
std::ostream *outs = &std::cout;
std::unique_ptr<std::fstream> fs;
if (!output_filename.empty()) {
fs = std::make_unique<std::fstream>(output_filename, std::fstream::trunc | std::fstream::out);
if (!fs->is_open()) {
std::cerr << "failed to create output file " << output_filename << '\n';
return 2;
}
outs = fs.get();
}
std::vector<std::string> sources;
while (optind < argc) {
sources.push_back(std::string(argv[optind++]));
}
tolk::read_callback = tolk::fs_read_callback;
return tolk::tolk_proceed(sources, *outs, std::cerr);
}

148
tolk/tolk-wasm.cpp Normal file
View file

@ -0,0 +1,148 @@
/*
This file is part of TON Blockchain source code.
TON Blockchain is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
TON Blockchain is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with TON Blockchain. If not, see <http://www.gnu.org/licenses/>.
In addition, as a special exception, the copyright holders give permission
to link the code of portions of this program with the OpenSSL library.
You must obey the GNU General Public License in all respects for all
of the code used other than OpenSSL. If you modify file(s) with this
exception, you may extend this exception to your version of the file(s),
but you are not obligated to do so. If you do not wish to do so, delete this
exception statement from your version. If you delete this exception statement
from all source files in the program, then also delete it here.
*/
#include "tolk.h"
#include "git.h"
#include "td/utils/JsonBuilder.h"
#include "fift/utils.h"
#include "td/utils/base64.h"
#include "td/utils/Status.h"
#include <sstream>
#include <iomanip>
td::Result<std::string> compile_internal(char *config_json) {
TRY_RESULT(input_json, td::json_decode(td::MutableSlice(config_json)))
auto &obj = input_json.get_object();
TRY_RESULT(opt_level, td::get_json_object_int_field(obj, "optLevel", false));
TRY_RESULT(sources_obj, td::get_json_object_field(obj, "sources", td::JsonValue::Type::Array, false));
auto &sources_arr = sources_obj.get_array();
std::vector<std::string> sources;
for (auto &item : sources_arr) {
sources.push_back(item.get_string().str());
}
tolk::opt_level = std::max(0, opt_level);
tolk::program_envelope = true;
tolk::verbosity = 0;
tolk::indent = 1;
std::ostringstream outs, errs;
auto compile_res = tolk::tolk_proceed(sources, outs, errs);
if (compile_res != 0) {
return td::Status::Error(std::string("Tolk compilation error: ") + errs.str());
}
TRY_RESULT(code_cell, fift::compile_asm(outs.str(), "/fiftlib/", false));
TRY_RESULT(boc, vm::std_boc_serialize(code_cell));
td::JsonBuilder result_json;
auto result_obj = result_json.enter_object();
result_obj("status", "ok");
result_obj("codeBoc", td::base64_encode(boc));
result_obj("fiftCode", outs.str());
result_obj("codeHashHex", code_cell->get_hash().to_hex());
result_obj.leave();
outs.clear();
errs.clear();
return result_json.string_builder().as_cslice().str();
}
/// Callback used to retrieve additional source files or data.
///
/// @param _kind The kind of callback (a string).
/// @param _data The data for the callback (a string).
/// @param o_contents A pointer to the contents of the file, if found. Allocated via malloc().
/// @param o_error A pointer to an error message, if there is one. Allocated via malloc().
///
/// The callback implementor must use malloc() to allocate storage for
/// contents or error. The callback implementor must use free() to free
/// said storage after tolk_compile returns.
///
/// If the callback is not supported, *o_contents and *o_error must be set to NULL.
typedef void (*CStyleReadFileCallback)(char const* _kind, char const* _data, char** o_contents, char** o_error);
tolk::ReadCallback::Callback wrapReadCallback(CStyleReadFileCallback _readCallback)
{
tolk::ReadCallback::Callback readCallback;
if (_readCallback) {
readCallback = [=](tolk::ReadCallback::Kind _kind, char const* _data) -> td::Result<std::string> {
char* contents_c = nullptr;
char* error_c = nullptr;
_readCallback(tolk::ReadCallback::kindString(_kind).data(), _data, &contents_c, &error_c);
if (!contents_c && !error_c) {
return td::Status::Error("Callback not supported");
}
if (contents_c) {
return contents_c;
}
return td::Status::Error(std::string(error_c));
};
}
return readCallback;
}
extern "C" {
const char* version() {
auto version_json = td::JsonBuilder();
auto obj = version_json.enter_object();
obj("tolkVersion", tolk::tolk_version);
obj("tolkFiftLibCommitHash", GitMetadata::CommitSHA1());
obj("tolkFiftLibCommitDate", GitMetadata::CommitDate());
obj.leave();
return strdup(version_json.string_builder().as_cslice().c_str());
}
const char *tolk_compile(char *config_json, CStyleReadFileCallback callback) {
if (callback) {
tolk::read_callback = wrapReadCallback(callback);
} else {
tolk::read_callback = tolk::fs_read_callback;
}
auto res = compile_internal(config_json);
if (res.is_error()) {
auto result = res.move_as_error();
auto error_res = td::JsonBuilder();
auto error_o = error_res.enter_object();
error_o("status", "error");
error_o("message", result.message().str());
error_o.leave();
return strdup(error_res.string_builder().as_cslice().c_str());
}
auto res_string = res.move_as_ok();
return strdup(res_string.c_str());
}
}

260
tolk/tolk.cpp Normal file
View file

@ -0,0 +1,260 @@
/*
This file is part of TON Blockchain source code.
TON Blockchain is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
TON Blockchain is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with TON Blockchain. If not, see <http://www.gnu.org/licenses/>.
In addition, as a special exception, the copyright holders give permission
to link the code of portions of this program with the OpenSSL library.
You must obey the GNU General Public License in all respects for all
of the code used other than OpenSSL. If you modify file(s) with this
exception, you may extend this exception to your version of the file(s),
but you are not obligated to do so. If you do not wish to do so, delete this
exception statement from your version. If you delete this exception statement
from all source files in the program, then also delete it here.
*/
#include "tolk.h"
#include "srcread.h"
#include "lexer.h"
#include <getopt.h>
#include "git.h"
#include <fstream>
#include "td/utils/port/path.h"
namespace tolk {
int verbosity, indent, opt_level = 2;
bool stack_layout_comments, op_rewrite_comments, program_envelope, asm_preamble;
bool interactive = false;
GlobalPragma pragma_allow_post_modification{"allow-post-modification"};
GlobalPragma pragma_compute_asm_ltr{"compute-asm-ltr"};
std::string generated_from, boc_output_filename;
ReadCallback::Callback read_callback;
td::Result<std::string> fs_read_callback(ReadCallback::Kind kind, const char* query) {
switch (kind) {
case ReadCallback::Kind::ReadFile: {
std::ifstream ifs{query};
if (ifs.fail()) {
auto msg = std::string{"cannot open source file `"} + query + "`";
return td::Status::Error(msg);
}
std::stringstream ss;
ss << ifs.rdbuf();
return ss.str();
}
case ReadCallback::Kind::Realpath: {
return td::realpath(td::CSlice(query));
}
default: {
return td::Status::Error("Unknown query kind");
}
}
}
/*
*
* OUTPUT CODE GENERATOR
*
*/
void generate_output_func(SymDef* func_sym, std::ostream &outs, std::ostream &errs) {
SymValCodeFunc* func_val = dynamic_cast<SymValCodeFunc*>(func_sym->value);
tolk_assert(func_val);
std::string name = symbols.get_name(func_sym->sym_idx);
if (verbosity >= 2) {
errs << "\n\n=========================\nfunction " << name << " : " << func_val->get_type() << std::endl;
}
if (!func_val->code) {
errs << "( function `" << name << "` undefined )\n";
throw ParseError(func_sym->loc, name);
} else {
CodeBlob& code = *(func_val->code);
if (verbosity >= 3) {
code.print(errs, 9);
}
code.simplify_var_types();
if (verbosity >= 5) {
errs << "after simplify_var_types: \n";
code.print(errs, 0);
}
code.prune_unreachable_code();
if (verbosity >= 5) {
errs << "after prune_unreachable: \n";
code.print(errs, 0);
}
code.split_vars(true);
if (verbosity >= 5) {
errs << "after split_vars: \n";
code.print(errs, 0);
}
for (int i = 0; i < 8; i++) {
code.compute_used_code_vars();
if (verbosity >= 4) {
errs << "after compute_used_vars: \n";
code.print(errs, 6);
}
code.fwd_analyze();
if (verbosity >= 5) {
errs << "after fwd_analyze: \n";
code.print(errs, 6);
}
code.prune_unreachable_code();
if (verbosity >= 5) {
errs << "after prune_unreachable: \n";
code.print(errs, 6);
}
}
code.mark_noreturn();
if (verbosity >= 3) {
code.print(errs, 15);
}
if (verbosity >= 2) {
errs << "\n---------- resulting code for " << name << " -------------\n";
}
bool inline_func = (func_val->flags & 1);
bool inline_ref = (func_val->flags & 2);
const char* modifier = "";
if (inline_func) {
modifier = "INLINE";
} else if (inline_ref) {
modifier = "REF";
}
outs << std::string(indent * 2, ' ') << name << " PROC" << modifier << ":<{\n";
int mode = 0;
if (stack_layout_comments) {
mode |= Stack::_StkCmt | Stack::_CptStkCmt;
}
if (opt_level < 2) {
mode |= Stack::_DisableOpt;
}
auto fv = dynamic_cast<const SymValCodeFunc*>(func_sym->value);
// Flags: 1 - inline, 2 - inline_ref
if (fv && (fv->flags & 1) && code.ops->noreturn()) {
mode |= Stack::_InlineFunc;
}
if (fv && (fv->flags & 3)) {
mode |= Stack::_InlineAny;
}
code.generate_code(outs, mode, indent + 1);
outs << std::string(indent * 2, ' ') << "}>\n";
if (verbosity >= 2) {
errs << "--------------\n";
}
}
}
int generate_output(std::ostream &outs, std::ostream &errs) {
if (asm_preamble) {
outs << "\"Asm.fif\" include\n";
}
outs << "// automatically generated from " << generated_from << std::endl;
if (program_envelope) {
outs << "PROGRAM{\n";
}
for (SymDef* func_sym : glob_func) {
SymValCodeFunc* func_val = dynamic_cast<SymValCodeFunc*>(func_sym->value);
tolk_assert(func_val);
std::string name = symbols.get_name(func_sym->sym_idx);
outs << std::string(indent * 2, ' ');
if (func_val->method_id.is_null()) {
outs << "DECLPROC " << name << "\n";
} else {
outs << func_val->method_id << " DECLMETHOD " << name << "\n";
}
}
for (SymDef* gvar_sym : glob_vars) {
tolk_assert(dynamic_cast<SymValGlobVar*>(gvar_sym->value));
std::string name = symbols.get_name(gvar_sym->sym_idx);
outs << std::string(indent * 2, ' ') << "DECLGLOBVAR " << name << "\n";
}
int errors = 0;
for (SymDef* func_sym : glob_func) {
try {
generate_output_func(func_sym, outs, errs);
} catch (Error& err) {
errs << "cannot generate code for function `" << symbols.get_name(func_sym->sym_idx) << "`:\n"
<< err << std::endl;
++errors;
}
}
if (program_envelope) {
outs << "}END>c\n";
}
if (!boc_output_filename.empty()) {
outs << "2 boc+>B \"" << boc_output_filename << "\" B>file\n";
}
return errors;
}
void output_inclusion_stack(std::ostream &errs) {
while (!inclusion_locations.empty()) {
SrcLocation loc = inclusion_locations.top();
inclusion_locations.pop();
if (loc.fdescr) {
errs << "note: included from ";
loc.show(errs);
errs << std::endl;
}
}
}
int tolk_proceed(const std::vector<std::string> &sources, std::ostream &outs, std::ostream &errs) {
if (program_envelope && !indent) {
indent = 1;
}
define_keywords();
define_builtins();
int ok = 0, proc = 0;
try {
for (auto src : sources) {
ok += parse_source_file(src.c_str(), {}, true);
proc++;
}
if (interactive) {
generated_from += "stdin ";
ok += parse_source_stdin();
proc++;
}
if (ok < proc) {
throw Fatal{"output code generation omitted because of errors"};
}
if (!proc) {
throw Fatal{"no source files, no output"};
}
pragma_allow_post_modification.check_enable_in_libs();
pragma_compute_asm_ltr.check_enable_in_libs();
return generate_output(outs, errs);
} catch (Fatal& fatal) {
errs << "fatal: " << fatal << std::endl;
output_inclusion_stack(errs);
return 2;
} catch (Error& error) {
errs << error << std::endl;
output_inclusion_stack(errs);
return 2;
} catch (UnifyError& unif_err) {
errs << "fatal: ";
unif_err.print_message(errs);
errs << std::endl;
output_inclusion_stack(errs);
return 2;
}
return 0;
}
} // namespace tolk

1785
tolk/tolk.h Normal file
View file

File diff suppressed because it is too large Load diff

429
tolk/unify-types.cpp Normal file
View file

@ -0,0 +1,429 @@
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tolk.h"
namespace tolk {
/*
*
* TYPE EXPRESSIONS
*
*/
int TypeExpr::holes = 0, TypeExpr::type_vars = 0; // not thread safe, but it is ok for now
void TypeExpr::compute_width() {
switch (constr) {
case te_Atomic:
case te_Map:
minw = maxw = 1;
break;
case te_Tensor:
minw = maxw = 0;
for (TypeExpr* arg : args) {
minw += arg->minw;
maxw += arg->maxw;
}
if (minw > w_inf) {
minw = w_inf;
}
if (maxw > w_inf) {
maxw = w_inf;
}
break;
case te_Tuple:
minw = maxw = 1;
for (TypeExpr* arg : args) {
arg->compute_width();
}
break;
case te_Indirect:
minw = args[0]->minw;
maxw = args[0]->maxw;
break;
default:
minw = 0;
maxw = w_inf;
break;
}
}
bool TypeExpr::recompute_width() {
switch (constr) {
case te_Tensor:
case te_Indirect: {
int min = 0, max = 0;
for (TypeExpr* arg : args) {
min += arg->minw;
max += arg->maxw;
}
if (min > maxw || max < minw) {
return false;
}
if (min > w_inf) {
min = w_inf;
}
if (max > w_inf) {
max = w_inf;
}
if (minw < min) {
minw = min;
}
if (maxw > max) {
maxw = max;
}
return true;
}
case te_Tuple: {
for (TypeExpr* arg : args) {
if (arg->minw > 1 || arg->maxw < 1 || arg->minw > arg->maxw) {
return false;
}
}
return true;
}
default:
return false;
}
}
int TypeExpr::extract_components(std::vector<TypeExpr*>& comp_list) {
if (constr != te_Indirect && constr != te_Tensor) {
comp_list.push_back(this);
return 1;
}
int res = 0;
for (TypeExpr* arg : args) {
res += arg->extract_components(comp_list);
}
return res;
}
TypeExpr* TypeExpr::new_map(TypeExpr* from, TypeExpr* to) {
return new TypeExpr{te_Map, std::vector<TypeExpr*>{from, to}};
}
void TypeExpr::replace_with(TypeExpr* te2) {
if (te2 == this) {
return;
}
constr = te_Indirect;
value = 0;
minw = te2->minw;
maxw = te2->maxw;
args.clear();
args.push_back(te2);
}
bool TypeExpr::remove_indirect(TypeExpr*& te, TypeExpr* forbidden) {
tolk_assert(te);
while (te->constr == te_Indirect) {
te = te->args[0];
}
if (te->constr == te_Unknown) {
return te != forbidden;
}
bool res = true;
for (auto& x : te->args) {
res &= remove_indirect(x, forbidden);
}
return res;
}
std::vector<TypeExpr*> TypeExpr::remove_forall(TypeExpr*& te) {
tolk_assert(te && te->constr == te_ForAll);
tolk_assert(te->args.size() >= 1);
std::vector<TypeExpr*> new_vars;
for (std::size_t i = 1; i < te->args.size(); i++) {
new_vars.push_back(new_hole(1));
}
TypeExpr* te2 = te;
// std::cerr << "removing universal quantifier in " << te << std::endl;
te = te->args[0];
remove_forall_in(te, te2, new_vars);
// std::cerr << "-> " << te << std::endl;
return new_vars;
}
bool TypeExpr::remove_forall_in(TypeExpr*& te, TypeExpr* te2, const std::vector<TypeExpr*>& new_vars) {
tolk_assert(te);
tolk_assert(te2 && te2->constr == te_ForAll);
if (te->constr == te_Var) {
for (std::size_t i = 0; i < new_vars.size(); i++) {
if (te == te2->args[i + 1]) {
te = new_vars[i];
return true;
}
}
return false;
}
if (te->constr == te_ForAll) {
return false;
}
if (te->args.empty()) {
return false;
}
auto te1 = new TypeExpr(*te);
bool res = false;
for (auto& arg : te1->args) {
res |= remove_forall_in(arg, te2, new_vars);
}
if (res) {
te = te1;
} else {
delete te1;
}
return res;
}
void TypeExpr::show_width(std::ostream& os) {
os << minw;
if (maxw != minw) {
os << "..";
if (maxw < w_inf) {
os << maxw;
}
}
}
std::ostream& operator<<(std::ostream& os, TypeExpr* type_expr) {
if (!type_expr) {
return os << "(null-type-ptr)";
}
return type_expr->print(os);
}
std::ostream& TypeExpr::print(std::ostream& os, int lex_level) {
switch (constr) {
case te_Unknown:
return os << "??" << value;
case te_Var:
if (value >= -26 && value < 0) {
return os << "_" << (char)(91 + value);
} else if (value >= 0 && value < 26) {
return os << (char)(65 + value);
} else {
return os << "TVAR" << value;
}
case te_Indirect:
return os << args[0];
case te_Atomic: {
switch (value) {
case _Int:
return os << "int";
case _Cell:
return os << "cell";
case _Slice:
return os << "slice";
case _Builder:
return os << "builder";
case _Cont:
return os << "cont";
case _Tuple:
return os << "tuple";
case _Type:
return os << "type";
default:
return os << "atomic-type-" << value;
}
}
case te_Tensor: {
if (lex_level > -127) {
os << "(";
}
auto c = args.size();
if (c) {
for (const auto& x : args) {
x->print(os);
if (--c) {
os << ", ";
}
}
}
if (lex_level > -127) {
os << ")";
}
return os;
}
case te_Tuple: {
os << "[";
auto c = args.size();
if (c == 1 && args[0]->constr == te_Tensor) {
args[0]->print(os, -127);
} else if (c) {
for (const auto& x : args) {
x->print(os);
if (--c) {
os << ", ";
}
}
}
return os << "]";
}
case te_Map: {
tolk_assert(args.size() == 2);
if (lex_level > 0) {
os << "(";
}
args[0]->print(os, 1);
os << " -> ";
args[1]->print(os);
if (lex_level > 0) {
os << ")";
}
return os;
}
case te_ForAll: {
tolk_assert(args.size() >= 1);
if (lex_level > 0) {
os << '(';
}
os << "Forall ";
for (std::size_t i = 1; i < args.size(); i++) {
os << (i > 1 ? ' ' : '(');
args[i]->print(os);
}
os << ") ";
args[0]->print(os);
if (lex_level > 0) {
os << ')';
}
return os;
}
default:
return os << "unknown-type-expr-" << constr;
}
}
void UnifyError::print_message(std::ostream& os) const {
os << "cannot unify type " << te1 << " with " << te2;
if (!msg.empty()) {
os << ": " << msg;
}
}
std::ostream& operator<<(std::ostream& os, const UnifyError& ue) {
ue.print_message(os);
return os;
}
std::string UnifyError::message() const {
std::ostringstream os;
print_message(os);
return os.str();
}
void check_width_compat(TypeExpr* te1, TypeExpr* te2) {
if (te1->minw > te2->maxw || te2->minw > te1->maxw) {
std::ostringstream os{"cannot unify types of widths ", std::ios_base::ate};
te1->show_width(os);
os << " and ";
te2->show_width(os);
throw UnifyError{te1, te2, os.str()};
}
}
void check_update_widths(TypeExpr* te1, TypeExpr* te2) {
check_width_compat(te1, te2);
te1->minw = te2->minw = std::max(te1->minw, te2->minw);
te1->maxw = te2->maxw = std::min(te1->maxw, te2->maxw);
tolk_assert(te1->minw <= te1->maxw);
}
void unify(TypeExpr*& te1, TypeExpr*& te2) {
tolk_assert(te1 && te2);
// std::cerr << "unify( " << te1 << " , " << te2 << " )\n";
while (te1->constr == TypeExpr::te_Indirect) {
te1 = te1->args[0];
}
while (te2->constr == TypeExpr::te_Indirect) {
te2 = te2->args[0];
}
if (te1 == te2) {
return;
}
if (te1->constr == TypeExpr::te_ForAll) {
TypeExpr* te = te1;
std::vector<TypeExpr*> new_vars = TypeExpr::remove_forall(te);
for (TypeExpr* t : new_vars) {
t->was_forall_var = true;
}
unify(te, te2);
for (TypeExpr* t : new_vars) {
t->was_forall_var = false;
}
return;
}
if (te2->constr == TypeExpr::te_ForAll) {
TypeExpr* te = te2;
std::vector<TypeExpr*> new_vars = TypeExpr::remove_forall(te);
for (TypeExpr* t : new_vars) {
t->was_forall_var = true;
}
unify(te1, te);
for (TypeExpr* t : new_vars) {
t->was_forall_var = false;
}
return;
}
if (te1->was_forall_var && te2->constr == TypeExpr::te_Tensor) {
throw UnifyError{te1, te2, "cannot unify generic type and tensor"};
}
if (te2->was_forall_var && te1->constr == TypeExpr::te_Tensor) {
throw UnifyError{te2, te1, "cannot unify generic type and tensor"};
}
if (te1->constr == TypeExpr::te_Unknown) {
if (te2->constr == TypeExpr::te_Unknown) {
tolk_assert(te1->value != te2->value);
}
if (!TypeExpr::remove_indirect(te2, te1)) {
throw UnifyError{te1, te2, "type unification results in an infinite cyclic type"};
}
check_update_widths(te1, te2);
te1->replace_with(te2);
te1 = te2;
return;
}
if (te2->constr == TypeExpr::te_Unknown) {
if (!TypeExpr::remove_indirect(te1, te2)) {
throw UnifyError{te2, te1, "type unification results in an infinite cyclic type"};
}
check_update_widths(te2, te1);
te2->replace_with(te1);
te2 = te1;
return;
}
if (te1->constr != te2->constr || te1->value != te2->value || te1->args.size() != te2->args.size()) {
throw UnifyError{te1, te2};
}
for (std::size_t i = 0; i < te1->args.size(); i++) {
unify(te1->args[i], te2->args[i]);
}
if (te1->constr == TypeExpr::te_Tensor) {
if (!te1->recompute_width()) {
throw UnifyError{te1, te2, "type unification incompatible with known width of first type"};
}
if (!te2->recompute_width()) {
throw UnifyError{te2, te1, "type unification incompatible with known width of first type"};
}
check_update_widths(te1, te2);
}
te1->replace_with(te2);
te1 = te2;
}
} // namespace tolk