feat: move scripts to xtasks

.github/workflows/build.yml

@@ -36,22 +36,22 @@ jobs:
         # When adding a new `target`:
         # 1. Define a new platform alias above
         # 2. Add a new record to a matrix map in `cli/npm/install.js`
-        - { platform: linux-arm64       , target: aarch64-unknown-linux-gnu           , os: ubuntu-latest  , use-cross: true    }
+        - { platform: linux-arm64       , target: aarch64-unknown-linux-gnu   , os: ubuntu-latest  , use-cross: true    }
-        - { platform: linux-arm         , target: arm-unknown-linux-gnueabi           , os: ubuntu-latest  , use-cross: true    }
+        - { platform: linux-arm         , target: arm-unknown-linux-gnueabi   , os: ubuntu-latest  , use-cross: true    }
-        - { platform: linux-x64         , target: x86_64-unknown-linux-gnu            , os: ubuntu-20.04   , cli_features: wasm } #2272
+        - { platform: linux-x64         , target: x86_64-unknown-linux-gnu    , os: ubuntu-20.04   , cli_features: wasm } #2272
-        - { platform: linux-x86         , target: i686-unknown-linux-gnu              , os: ubuntu-latest  , use-cross: true    }
+        - { platform: linux-x86         , target: i686-unknown-linux-gnu      , os: ubuntu-latest  , use-cross: true    }
-        - { platform: linux-powerpc64   , target: powerpc64-unknown-linux-gnu         , os: ubuntu-latest  , use-cross: true    }
+        - { platform: linux-powerpc64   , target: powerpc64-unknown-linux-gnu , os: ubuntu-latest  , use-cross: true    }
-        - { platform: windows-arm64     , target: aarch64-pc-windows-msvc             , os: windows-latest                      }
+        - { platform: windows-arm64     , target: aarch64-pc-windows-msvc     , os: windows-latest                      }
-        - { platform: windows-x64       , target: x86_64-pc-windows-msvc              , os: windows-latest , cli_features: wasm }
+        - { platform: windows-x64       , target: x86_64-pc-windows-msvc      , os: windows-latest , cli_features: wasm }
-        - { platform: windows-x86       , target: i686-pc-windows-msvc                , os: windows-latest                      }
+        - { platform: windows-x86       , target: i686-pc-windows-msvc        , os: windows-latest                      }
-        - { platform: macos-arm64       , target: aarch64-apple-darwin                , os: macos-14       , cli_features: wasm }
+        - { platform: macos-arm64       , target: aarch64-apple-darwin        , os: macos-14       , cli_features: wasm }
-        - { platform: macos-x64         , target: x86_64-apple-darwin                 , os: macos-12       , cli_features: wasm }
+        - { platform: macos-x64         , target: x86_64-apple-darwin         , os: macos-12       , cli_features: wasm }
 
         # Cross compilers for C library
-        - { platform: linux-arm64       , cc: aarch64-linux-gnu-gcc             , ar: aarch64-linux-gnu-ar           }
+        - { platform: linux-arm64       , cc: aarch64-linux-gnu-gcc           , ar: aarch64-linux-gnu-ar   }
-        - { platform: linux-arm         , cc: arm-linux-gnueabi-gcc             , ar: arm-linux-gnueabi-ar           }
+        - { platform: linux-arm         , cc: arm-linux-gnueabi-gcc           , ar: arm-linux-gnueabi-ar   }
-        - { platform: linux-x86         , cc: i686-linux-gnu-gcc                , ar: i686-linux-gnu-ar              }
+        - { platform: linux-x86         , cc: i686-linux-gnu-gcc              , ar: i686-linux-gnu-ar      }
-        - { platform: linux-powerpc64   , cc: powerpc64-linux-gnu-gcc           , ar: powerpc64-linux-gnu-ar         }
+        - { platform: linux-powerpc64   , cc: powerpc64-linux-gnu-gcc         , ar: powerpc64-linux-gnu-ar }
 
         # See #2041 tree-sitter issue
         - { platform: windows-x64   , rust-test-threads: 1 }
@@ -86,23 +86,30 @@ jobs:
 
     - name: Install cross
       if: ${{ matrix.use-cross }}
-      uses: taiki-e/install-action@v2
+      run: cargo install cross --git https://github.com/cross-rs/cross
-      with:
-        tool: cross
 
     - name: Build custom cross image
-      if: ${{ matrix.use-cross && matrix.os == 'ubuntu-latest' }}
+      if: ${{ matrix.use-cross }}
       run: |
         target="${{ matrix.target }}"
         image=ghcr.io/cross-rs/$target:custom
-        echo "CROSS_IMAGE=$image"                              >> $GITHUB_ENV
 
-        echo "[target.$target]"                                >> Cross.toml
+        echo "[target.$target]"                                                       >> Cross.toml
-        echo "image = \"$image\""                              >> Cross.toml
+        echo "image = \"$image\""                                                     >> Cross.toml
-        echo "CROSS_CONFIG=$PWD/Cross.toml"                    >> $GITHUB_ENV
+        echo "[build]"                                                                >> Cross.toml
+        echo "pre-build = ["                                                          >> Cross.toml
+        echo "  \"dpkg --add-architecture \$CROSS_DEB_ARCH\","                        >> Cross.toml
+        echo "  \"apt-get update && apt-get -y install libssl-dev\""                  >> Cross.toml
+        echo "]"                                                                      >> Cross.toml
+
+        echo "Cross.toml:"
+        cat Cross.toml
+
+        echo "CROSS_IMAGE=$image"           >> $GITHUB_ENV
+        echo "CROSS_CONFIG=$PWD/Cross.toml" >> $GITHUB_ENV
 
         echo "FROM ghcr.io/cross-rs/$target:edge"                            >> Dockerfile
-        echo "RUN curl -fsSL https://deb.nodesource.com/setup_16.x | bash -" >> Dockerfile
+        echo "RUN curl -fsSL https://deb.nodesource.com/setup_22.x | bash -" >> Dockerfile
         echo "RUN apt-get update && apt-get -y install nodejs"               >> Dockerfile
         docker build -t $image .
 
@@ -139,22 +146,22 @@ jobs:
 
     - name: Build wasm library
       if: ${{ !matrix.cli-only && !matrix.use-cross }} # No sense to build on the same Github runner hosts many times
-      run: script/build-wasm
+      run: $BUILD_CMD run --package xtask -- build-wasm
 
     - run: $BUILD_CMD build --release --target=${{ matrix.target }} --features=${{ matrix.cli_features }}
 
-    - run: script/fetch-fixtures
+    - run: $BUILD_CMD run --package xtask -- fetch-fixtures
 
     - uses: ./.github/actions/cache
       id: cache
 
     - name: Generate fixtures
       if: ${{ !matrix.cli-only && inputs.run_test && steps.cache.outputs.cache-hit != 'true' }} # Can't natively run CLI on Github runner's host
-      run: script/generate-fixtures
+      run: $BUILD_CMD run --package xtask -- generate-fixtures
 
     - name: Generate WASM fixtures
       if: ${{ !matrix.cli-only && !matrix.use-cross && inputs.run_test && steps.cache.outputs.cache-hit != 'true' }} # See comment for the "Build wasm library" step
-      run: script/generate-fixtures-wasm
+      run: $BUILD_CMD run --package xtask -- generate-fixtures --wasm
 
     - name: Run main tests
       if: ${{ !matrix.cli-only && inputs.run_test }} # Can't natively run CLI on Github runner's host
@@ -162,7 +169,7 @@ jobs:
 
     - name: Run wasm tests
       if: ${{ !matrix.cli-only && !matrix.use-cross && inputs.run_test }} # See comment for the "Build wasm library" step
-      run: script/test-wasm
+      run: $BUILD_CMD run --package xtask -- test-wasm
 
     - name: Run benchmarks
       if: ${{ !matrix.cli-only && !matrix.use-cross && inputs.run_test }} # Cross-compiled benchmarks make no sense

.github/workflows/release.yml

@@ -84,12 +84,12 @@ jobs:
 
       - name: Build wasm
         if: matrix.directory == 'lib/binding_web'
-        run: ./script/build-wasm
+        run: cargo xtask build-wasm
 
       - name: Setup Node
         uses: actions/setup-node@v4
         with:
-          node-version: 18
+          node-version: 20
           registry-url: "https://registry.npmjs.org"
 
       - name: Publish lib to npmjs.com

.github/workflows/sanitize.yml

@@ -24,13 +24,13 @@ jobs:
     - run: rustup toolchain install stable --profile minimal
     - uses: Swatinem/rust-cache@v2
     - run: cargo build --release
-    - run: script/fetch-fixtures
+    - run: cargo xtask fetch-fixtures
 
     - uses: ./.github/actions/cache
       id: cache
 
     - if: ${{ steps.cache.outputs.cache-hit != 'true' }}
-      run: script/generate-fixtures
+      run: cargo xtask generate-fixtures
 
     - name: Run main tests with undefined behaviour sanitizer (UBSAN)
       env:

Makefile

@@ -95,13 +95,13 @@ uninstall:
 ##### Dev targets #####
 
 test:
-	script/fetch-fixtures
+	cargo xtask fetch-fixtures
-	script/generate-fixtures
+	cargo xtask generate-fixtures
-	script/test
+	cargo xtask test
 
 test_wasm:
-	script/generate-fixtures-wasm
+	cargo xtask generate-fixtures-wasm
-	script/test-wasm
+	cargo xtask test-wasm
 
 lint:
 	cargo update --workspace --locked --quiet

script/benchmark

@@ -1,62 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-function usage {
-  cat <<EOF
-USAGE
-
-  $0  [-h] [-l language-name] [-e example-file-name] [-r repetition-count]
-
-OPTIONS
-
-  -h  print this message
-
-  -l  run only the benchmarks for the given language
-
-  -e  run only the benchmarks that parse the example file with the given name
-
-  -r  parse each sample the given number of times (default 5)
-
-  -g  debug
-
-EOF
-}
-
-mode=normal
-
-while getopts "hgl:e:r:" option; do
-  case ${option} in
-    h)
-      usage
-      exit
-      ;;
-    g)
-      mode=debug
-      ;;
-    e)
-      export TREE_SITTER_BENCHMARK_EXAMPLE_FILTER=${OPTARG}
-      ;;
-    l)
-      export TREE_SITTER_BENCHMARK_LANGUAGE_FILTER=${OPTARG}
-      ;;
-    r)
-      export TREE_SITTER_BENCHMARK_REPETITION_COUNT=${OPTARG}
-      ;;
-    *)
-      usage
-      exit 1
-      ;;
-  esac
-done
-
-if [[ $mode == debug ]]; then
-  test_binary=$(
-    cargo bench benchmark -p tree-sitter-cli --no-run --message-format=json 2> /dev/null |
-    jq -rs 'map(select(.target.name == "benchmark" and .executable))[0].executable'
-  )
-  env | grep TREE_SITTER
-  echo "$test_binary"
-else
-  exec cargo bench benchmark -p tree-sitter-cli
-fi

script/benchmark.cmd

@@ -1,4 +0,0 @@
-@echo off
-
-cargo bench benchmark -p tree-sitter-cli
-exit /b %errorlevel%

script/build-fuzzers

@@ -1,76 +0,0 @@
-#!/usr/bin/env bash
-
-# shellcheck disable=SC2086
-
-set -e
-
-if [[ $(uname -s) != Linux ]]; then
-  printf 'Fuzzing is only supported on Linux\n' >&2
-  exit 1
-fi
-
-CC=${CC:-clang}
-CXX=${CXX:-clang++}
-
-default_fuzz_flags=-fsanitize=fuzzer,address,undefined
-
-export CFLAGS="$default_fuzz_flags $CFLAGS"
-export CXXFLAGS="$default_fuzz_flags $CXXFLAGS"
-
-make CC="$CC" CXX="$CXX" libtree-sitter.a
-
-if [[ -z $* ]]; then
-  mapfile -t languages < <(ls test/fixtures/grammars)
-else
-  languages=("$@")
-fi
-
-mkdir -p test/fuzz/out
-
-for lang in "${languages[@]}"; do
-  # skip typescript & php
-  if [[ $lang == typescript || $lang == php ]]; then
-    continue
-  fi
-  printf 'Building %s fuzzer...\n' "$lang"
-  lang_dir="test/fixtures/grammars/$lang"
-  lang_grammar="${lang_dir}/src/grammar.json"
-
-  # The following assumes each language is implemented as src/parser.c plus an
-  # optional scanner in src/scanner.c
-  objects=()
-
-  lang_scanner="${lang_dir}/src/scanner"
-  if [[ -f "${lang_scanner}.c" ]]; then
-    $CC $CFLAGS -std=c11 -g -O1 -I "${lang_dir}/src" -c "${lang_scanner}.c" -o "${lang_scanner}.o"
-    objects+=("${lang_scanner}.o")
-  fi
-
-  # Compiling with -O0 speeds up the build dramatically
-  $CC $CFLAGS -g -O0 -I "${lang_dir}/src" "${lang_dir}/src/parser.c" -c -o "${lang_dir}/src/parser.o"
-  objects+=("${lang_dir}/src/parser.o")
-
-  highlights_filename="${lang_dir}/queries/highlights.scm"
-  if [[ -f "${highlights_filename}" ]]; then
-    ts_lang_query_filename="${lang}.scm"
-    cp "${highlights_filename}" "test/fuzz/out/${ts_lang_query_filename}"
-  else
-    ts_lang_query_filename=""
-  fi
-
-  ts_lang="tree_sitter_$(jq -r .name "$lang_grammar")"
-  $CXX $CXXFLAGS -std=c++11 -Ilib/include \
-    -D TS_LANG="$ts_lang" \
-    -D TS_LANG_QUERY_FILENAME="\"${ts_lang_query_filename}\"" \
-    test/fuzz/fuzzer.cc \
-    "${objects[@]}" \
-    libtree-sitter.a \
-    -o "test/fuzz/out/${lang}_fuzzer"
-
-  jq '
-    [ ..
-      | if .type? == "STRING" or (.type? == "ALIAS" and .named? == false) then .value else empty end
-      | select(test("\\S") and length == utf8bytelength)
-    ] | unique | .[]
-  ' "$lang_grammar" | sort > "test/fuzz/out/${lang}.dict"
-done

script/build-wasm

@@ -1,147 +0,0 @@
-#!/usr/bin/env bash
-
-usage() {
-  cat <<EOF
-USAGE
-
-  $0 [--help] [--debug] [--docker]
-
-SUMMARY
-
-  Compile the Tree-sitter WASM library. This will create two files in the
-  \`lib/binding_web\` directory: \`tree-sitter.js\` and \`tree-sitter.wasm\`.
-
-REQUIREMENTS
-
-  You must have either the \`emcc\` command or the \`docker\` command
-  on your PATH for this to work.
-
-OPTIONS
-
-  --help:   Display this message.
-  --debug:  Compile the library more quickly, with fewer optimizations
-            and more runtime assertions.
-  --docker: Run emscripten using docker, even if \`emcc\` is installed.
-            By default, \`emcc\` will be run directly when available.
-
-EOF
-}
-
-set -e
-
-WEB_DIR=lib/binding_web
-SRC_DIR=lib/src
-EMSCRIPTEN_VERSION=$(< cli/loader/emscripten-version)
-
-verbose=0
-force_docker=0
-emscripten_flags=(-O3 --minify 0)
-
-while (($# > 0)); do
-  case "$1" in
-    --debug)
-      emscripten_flags=(-s ASSERTIONS=1 -s SAFE_HEAP=1 -O0 -g)
-      ;;
-
-    --help)
-      usage
-      exit 0
-      ;;
-
-    --docker)
-      force_docker=1
-      ;;
-
-    -v|--verbose)
-      verbose=1
-      ;;
-
-    *)
-      usage
-      printf "Unrecognized argument '%s'\n" "$1" >&2
-      exit 1
-      ;;
-  esac
-  shift
-done
-
-if [[ $verbose == 1 ]]; then
-  emscripten_flags+=(-s VERBOSE=1 -v)
-fi
-
-emcc=
-docker=
-if [[ $force_docker == 0 ]] && command -v emcc > /dev/null; then
-  emcc=emcc
-elif command -v docker > /dev/null; then
-  # detect which one to use
-  docker=docker
-elif command -v podman > /dev/null; then
-  docker=podman
-fi
-
-if [[ -z $emcc ]] && [[ -n $docker ]]; then
-  if [[ $docker == podman ]]; then
-    export PODMAN_USERNS=keep-id
-  fi
-  emcc="$docker run                       \
-    --rm                                  \
-    -v $PWD:/src:Z                        \
-    -u $UID                               \
-    emscripten/emsdk:$EMSCRIPTEN_VERSION  \
-    emcc"
-fi
-
-if [[ -z $emcc ]]; then
-  if [[ $force_docker == 1 ]]; then
-    # shellcheck disable=SC2016
-    printf 'You must have `docker` or `podman` in your PATH to run this script with --docker\n' >&2
-  else
-    # shellcheck disable=SC2016
-    printf 'You must have either `docker`, `podman`, or `emcc` in your PATH to run this script\n' >&2
-  fi
-  exit 1
-fi
-
-mkdir -p target/scratch
-
-runtime_methods=stringToUTF16,AsciiToString
-
-# Remove quotes, add leading underscores, remove newlines, remove trailing comma.
-exported_functions=$(
-    cat ${SRC_DIR}/wasm/stdlib-symbols.txt ${WEB_DIR}/exports.txt |
-    sed -e 's/"//g;s/^/_/g' | tr -d '\n' | sed -e 's/,$//'
-)
-
-# Use emscripten to generate `tree-sitter.js` and `tree-sitter.wasm`
-# in the `target/scratch` directory
-$emcc                                            \
-  -s WASM=1                                      \
-  -s INITIAL_MEMORY=33554432                     \
-  -s ALLOW_MEMORY_GROWTH=1                       \
-  -s SUPPORT_BIG_ENDIAN=1                        \
-  -s MAIN_MODULE=2                               \
-  -s FILESYSTEM=0                                \
-  -s NODEJS_CATCH_EXIT=0                         \
-  -s NODEJS_CATCH_REJECTION=0                    \
-  -s EXPORTED_FUNCTIONS="${exported_functions}"  \
-  -s EXPORTED_RUNTIME_METHODS=$runtime_methods   \
-  "${emscripten_flags[@]}"                       \
-  -fno-exceptions                                \
-  -std=c11                                       \
-  -D 'fprintf(...)='                             \
-  -D NDEBUG=                                     \
-  -D _POSIX_C_SOURCE=200112L                     \
-  -D _DEFAULT_SOURCE=                            \
-  -I ${SRC_DIR}                                  \
-  -I lib/include                                 \
-  --js-library ${WEB_DIR}/imports.js             \
-  --pre-js ${WEB_DIR}/prefix.js                  \
-  --post-js ${WEB_DIR}/binding.js                \
-  --post-js ${WEB_DIR}/suffix.js                 \
-  lib/src/lib.c                                  \
-  ${WEB_DIR}/binding.c                           \
-  -o target/scratch/tree-sitter.js
-
-mv target/scratch/tree-sitter.js ${WEB_DIR}/tree-sitter.js
-mv target/scratch/tree-sitter.wasm ${WEB_DIR}/tree-sitter.wasm

script/build-wasm-stdlib

@@ -1,28 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-declare -a EXPORT_FLAGS
-while read -r -d, function; do
-    EXPORT_FLAGS+=("-Wl,--export=${function:1:-1}")
-done < lib/src/wasm/stdlib-symbols.txt
-
-target/wasi-sdk-21.0/bin/clang-17   \
-    -o stdlib.wasm                  \
-    -Os                             \
-    -fPIC                           \
-    -Wl,--no-entry                  \
-    -Wl,--stack-first               \
-    -Wl,-z -Wl,stack-size=65536     \
-    -Wl,--import-undefined          \
-    -Wl,--import-memory             \
-    -Wl,--import-table              \
-    -Wl,--strip-debug               \
-    -Wl,--export=reset_heap         \
-    -Wl,--export=__wasm_call_ctors  \
-    -Wl,--export=__stack_pointer    \
-    "${EXPORT_FLAGS[@]}"            \
-    lib/src/wasm/stdlib.c
-
-xxd -C -i stdlib.wasm > lib/src/wasm/wasm-stdlib.h
-mv stdlib.wasm target/

script/check-mallocs

@@ -1,12 +0,0 @@
-#!/usr/bin/env bash
-
-src_dir=lib/src
-allocation_functions=(malloc calloc realloc free)
-
-for function in "${allocation_functions[@]}"; do
-  usages=$(grep -n -E "\b${function}\(" -r $src_dir --exclude alloc.c --exclude stdlib.c)
-  if [[ -n $usages ]]; then
-    printf 'The %s function should not be called directly, but is called here:\n%s\n' "$function" "$usages" >&2
-    exit 1
-  fi
-done

script/fetch-emscripten

@@ -1,26 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-EMSDK_DIR=target/emsdk
-EMSCRIPTEN_VERSION=$(< cli/loader/emscripten-version)
-
-{
-  if [[ ! -f $EMSDK_DIR/emsdk ]]; then
-    printf 'Downloading emscripten SDK...\n'
-    git clone https://github.com/emscripten-core/emsdk.git $EMSDK_DIR
-  fi
-
-  cd $EMSDK_DIR
-
-  printf 'Updating emscripten SDK...\n'
-  git reset --hard
-  git pull
-  ./emsdk list
-
-  printf 'Installing emscripten...\n'
-  ./emsdk install "$EMSCRIPTEN_VERSION"
-
-  printf 'Activating emscripten...\n'
-  ./emsdk activate "$EMSCRIPTEN_VERSION"
-} >&2

script/fetch-fixtures

@@ -1,37 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-GRAMMARS_DIR="$PWD/test/fixtures/grammars"
-
-fetch_grammar() {
-  local grammar=$1
-  local ref=$2
-  local grammar_dir="${GRAMMARS_DIR}/${grammar}"
-  local grammar_url=https://github.com/tree-sitter/tree-sitter-${grammar}
-
-  printf 'Updating %s grammar...\n' "$grammar"
-
-  if [[ ! -d "$grammar_dir" ]]; then
-    git clone "$grammar_url" "$grammar_dir" --depth=1
-  fi
-
-  git -C "$grammar_dir" fetch origin "$ref" --depth=1
-  git -C "$grammar_dir" reset --hard FETCH_HEAD
-}
-
-fetch_grammar bash              master
-fetch_grammar c                 master
-fetch_grammar cpp               master
-fetch_grammar embedded-template master
-fetch_grammar go                master
-fetch_grammar html              master
-fetch_grammar java              master
-fetch_grammar javascript        master
-fetch_grammar jsdoc             master
-fetch_grammar json              master
-fetch_grammar php               master
-fetch_grammar python            master
-fetch_grammar ruby              master
-fetch_grammar rust              master
-fetch_grammar typescript        master

script/fetch-fixtures.cmd

@@ -1,32 +0,0 @@
-@echo off
-
-call:fetch_grammar bash              master
-call:fetch_grammar c                 master
-call:fetch_grammar cpp               master
-call:fetch_grammar embedded-template master
-call:fetch_grammar go                master
-call:fetch_grammar html              master
-call:fetch_grammar java              master
-call:fetch_grammar javascript        master
-call:fetch_grammar jsdoc             master
-call:fetch_grammar json              master
-call:fetch_grammar php               master
-call:fetch_grammar python            master
-call:fetch_grammar ruby              master
-call:fetch_grammar rust              master
-call:fetch_grammar typescript        master
-exit /B 0
-
-:fetch_grammar
-setlocal
-set grammar_dir=test\fixtures\grammars\%~1
-set grammar_url=https://github.com/tree-sitter/tree-sitter-%~1
-set grammar_branch=%~2
-@if not exist %grammar_dir% (
-  git clone %grammar_url% %grammar_dir% --depth=1
-)
-pushd %grammar_dir%
-git fetch origin %2 --depth=1
-git reset --hard FETCH_HEAD
-popd
-exit /B 0

script/generate-bindings

@@ -1,44 +0,0 @@
-#!/bin/bash
-
-output_path=lib/binding_rust/bindings.rs
-header_path=lib/include/tree_sitter/api.h
-no_derive_copy=(
-  TSInput
-  TSLanguage
-  TSLogger
-  TSLookaheadIterator
-  TSParser
-  TSTree
-  TSQuery
-  TSQueryCursor
-  TSQueryCapture
-  TSQueryMatch
-  TSQueryPredicateStep
-)
-no_copy=$(IFS='|'; echo "${no_derive_copy[*]}")
-
-file_version=$(head -n1 "$output_path" | cut -d' ' -f6)
-tool_version=$(bindgen --version | cut -d' ' -f2)
-higher_version=$(printf '%s\n' "$file_version" "$tool_version" | sort -V | tail -n1)
-
-if [[ "$higher_version" != "$tool_version" ]]; then
-  printf 'Latest used bindgen version was %s\n' "$file_version" >&2
-  printf 'Currently installed bindgen CLI version is %s\n\n' "$tool_version" >&2
-  # shellcheck disable=SC2016
-  printf 'You must upgrade bindgen CLI first with `cargo install bindgen-cli`\n' >&2
-  exit 1
-fi
-
-bindgen                                        \
-  --no-layout-tests                            \
-  --allowlist-type '^TS.*'                     \
-  --allowlist-function '^ts_.*'                \
-  --allowlist-var '^TREE_SITTER.*'             \
-  --blocklist-type '^__.*'                     \
-  --no-prepend-enum-name                       \
-  --no-copy "$no_copy"                         \
-  --use-core                                   \
-  "$header_path"                               \
-  --                                           \
-  -D TREE_SITTER_FEATURE_WASM                  \
-  > "$output_path"

script/generate-fixtures

@@ -1,27 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-ROOT_DIR="$PWD"
-GRAMMARS_DIR="$ROOT_DIR/test/fixtures/grammars"
-
-if [[ $CI == true ]]; then
-  set -x
-else
-  cargo build --release
-  TREE_SITTER="$ROOT_DIR/target/release/tree-sitter"
-fi
-
-filter_grammar_name="$1"
-
-while read -r grammar_file; do
-  grammar_dir="${grammar_file%/*}"
-  grammar_name="${grammar_dir##*/}"
-
-  if [[ -n $filter_grammar_name && "$filter_grammar_name" != "$grammar_name" ]]; then
-    continue
-  fi
-
-  printf 'Regenerating %s parser\n' "$grammar_name"
-  (cd "$grammar_dir" && "$TREE_SITTER" generate src/grammar.json --abi=latest)
-done < <(find "$GRAMMARS_DIR" -name grammar.js -not -path '*/node_modules/*')

script/generate-fixtures-wasm

@@ -1,33 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-ROOT_DIR="$PWD"
-GRAMMARS_DIR="$ROOT_DIR/test/fixtures/grammars"
-
-if [[ $CI == true ]]; then
-  set -x
-else
-  cargo build --release
-  TREE_SITTER="$ROOT_DIR/target/release/tree-sitter"
-fi
-
-build_wasm_args=
-if [[ $1 == --docker ]]; then
-  build_wasm_args=--docker
-  shift
-fi
-
-filter_grammar_name="$1"
-
-while read -r grammar_file; do
-  grammar_dir="${grammar_file%/*}"
-  grammar_name="${grammar_dir##*/}"
-
-  if [[ -n $filter_grammar_name && "$filter_grammar_name" != "$grammar_name" ]]; then
-    continue
-  fi
-
-  printf 'Compiling %s parser to wasm\n' "$grammar_name"
-  "$TREE_SITTER" build --wasm $build_wasm_args -o "target/release/tree-sitter-${grammar_name}.wasm" "$grammar_dir"
-done < <(find "$GRAMMARS_DIR" -name grammar.js -not -path '*/node_modules/*')

script/generate-fixtures.cmd

@@ -1,13 +0,0 @@
-@echo off
-
-setlocal EnableDelayedExpansion
-set tree_sitter="%cd%\target\release\tree-sitter"
-
-for /f "tokens=*" %%f in ('dir test\fixtures\grammars\grammar.js /b/s') do (
-  pushd "%%f\.."
-  echo Regenerating parser !cd!
-  %tree_sitter% generate src\grammar.json --abi=latest
-  popd
-)
-
-exit /B 0

script/heap-profile

@@ -1,36 +0,0 @@
-#!/usr/bin/env bash
-
-# Usage:
-#   script/heap-profile
-#
-# Parse an example source file and record memory usage
-#
-# Dependencies:
-#   * `pprof` executable:   https://github.com/google/pprof
-#   * `gperftools` package: https://github.com/gperftools/gperftools
-
-set -e
-
-GRAMMARS_DIR="$PWD/test/fixtures/grammars"
-
-# Build the library
-make libtree-sitter.a
-
-# Build the heap-profiling harness
-clang++                                  \
-  -Wno-reorder-init-list                 \
-  -Wno-c99-designator                    \
-  -I lib/include                         \
-  -I "$GRAMMARS_DIR"                     \
-  -D GRAMMARS_DIR="\"${GRAMMARS_DIR}/\"" \
-  test/profile/heap.cc                   \
-  -l tcmalloc                            \
-  libtree-sitter.a                       \
-  -o target/heap-profile
-
-# Run the harness with heap profiling enabled.
-export HEAPPROFILE="$PWD/profile"
-target/heap-profile "$@"
-
-# Extract statistics using pprof.
-pprof -top -cum profile.0001.heap

script/reproduce

@@ -1,35 +0,0 @@
-#!/bin/bash
-
-if (($# < 3)); then
-  echo "usage: $0 <language> <halt|recover> <testcase> [libFuzzer args...]" >&2
-  exit 1
-fi
-
-set -eu
-
-export ASAN_OPTIONS=quarantine_size_mb=10:detect_leaks=1:symbolize=1
-export UBSAN=print_stacktrace=1:halt_on_error=1:symbolize=1
-
-# check if CI env var exists
-if [[ -z ${CI:-} ]]; then
-  declare -A mode_config=(
-    [halt]='-timeout=1 -rss_limit_mb=2048'
-    [recover]='-timeout=10 -rss_limit_mb=2048'
-  )
-else
-  declare -A mode_config=(
-    [halt]='-max_total_time=120 -timeout=1 -rss_limit_mb=2048'
-    [recover]='-time=120 -timeout=10 -rss_limit_mb=2048'
-  )
-fi
-
-lang="$1"
-shift
-mode="$1"
-shift
-testcase="$1"
-shift
-# Treat remainder of arguments as libFuzzer arguments
-
-# shellcheck disable=SC2086
-test/fuzz/out/${lang}_fuzzer ${mode_config[$mode]} -runs=1 "$testcase" "$@"

script/run-fuzzer

@@ -1,42 +0,0 @@
-#!/usr/bin/env bash
-
-if (($# < 2)); then
-  echo "usage: $0 <language> <halt|recover> [libFuzzer args...]" >&2
-  exit 1
-fi
-
-set -eu
-
-export ASAN_OPTIONS=quarantine_size_mb=10:detect_leaks=1:symbolize=1
-export UBSAN=print_stacktrace=1:halt_on_error=1:symbolize=1
-
-# check if CI env var exists
-if [[ -z ${CI:-} ]]; then
-  declare -A mode_config=(
-    [halt]='-timeout=1 -rss_limit_mb=2048'
-    [recover]='-timeout=10 -rss_limit_mb=2048'
-  )
-else
-  declare -A mode_config=(
-    [halt]='-max_total_time=120 -timeout=1 -rss_limit_mb=2048'
-    [recover]='-time=120 -timeout=10 -rss_limit_mb=2048'
-  )
-fi
-
-lang="$1"
-shift
-mode="$1"
-shift
-# Treat remainder of arguments as libFuzzer arguments
-
-# Fuzzing logs and testcases are always written to `pwd`, so `cd` there first
-results="$PWD/test/fuzz/out/fuzz-results/${lang}"
-mkdir -p "${results}"
-cd "${results}"
-
-# Create a corpus directory, so new discoveries are stored on disk. These will
-# then be loaded on subsequent fuzzing runs
-mkdir -p corpus
-
-# shellcheck disable=SC2086
-../../${lang}_fuzzer -dict="../../${lang}.dict" -artifact_prefix=${lang}_ -max_len=2048 ${mode_config[$mode]} corpus "$@"

script/serve-docs

@@ -1,29 +0,0 @@
-#!/bin/bash
-
-root=$PWD
-cd docs
-
-bundle exec jekyll serve "$@" &
-
-bundle exec ruby <<RUBY &
-  require "listen"
-
-  def copy_wasm_files
-    `cp $root/lib/binding_web/tree-sitter.{js,wasm} $root/docs/assets/js/`
-    `cp $root/target/release/*.wasm $root/docs/assets/js/`
-  end
-
-  puts "Copying WASM files to docs folder..."
-  copy_wasm_files
-
-  puts "Watching release directory"
-  listener = Listen.to("$root/lib/binding_web", only: /^tree-sitter\.(js|wasm)$/, wait_for_delay: 2) do
-    puts "WASM files updated. Copying new files to docs folder..."
-    copy_wasm_files
-  end
-
-  listener.start
-  sleep
-RUBY
-
-wait

script/show-symbol-sizes

@@ -1,50 +0,0 @@
-#!/usr/bin/env node
-
-const {statSync} = require('fs');
-const {execFileSync} = require('child_process');
-const libPath = process.argv[2];
-
-if (!libPath || libPath === '--help') {
-  console.log(`Usage: ${process.argv[1]} <dylib-path>`);
-  process.exit(0)
-}
-
-// Get total file size
-const totalSize = statSync(libPath).size
-
-// Dump symbols with addresses
-const output = execFileSync(
-  'nm',
-  ['-t', 'd', libPath],
-  {encoding: 'utf8'}
-);
-
-// Parse addresses
-const addressEntries = [];
-for (const line of output.split('\n')) {
-  const [address, _, name] = line.split(/\s+/);
-  if (address && name) {
-    addressEntries.push({name, address: parseInt(address)})
-  }
-}
-
-// Compute sizes by subtracting addresses
-addressEntries.sort((a, b) => a.address - b.address);
-const sizeEntries = addressEntries.map(({name, address}, i) => {
-  const next = addressEntries[i + 1] ? addressEntries[i + 1].address : totalSize;
-  const size = next - address;
-  return {name, size}
-})
-
-function formatSize(sizeInBytes) {
-  return sizeInBytes > 1024
-    ? `${(sizeInBytes / 1024).toFixed(1)} kb`
-    : `${sizeInBytes} b`
-}
-
-// Display sizes
-sizeEntries.sort((a, b) => b.size - a.size);
-console.log('total'.padEnd(64, ' '), '\t', formatSize(totalSize));
-for (const entry of sizeEntries) {
-  console.log(entry.name.padEnd(64, ' '), '\t', formatSize(entry.size));
-}

script/test

@@ -1,101 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-function usage {
-  cat <<EOF
-USAGE
-
-  $0 [-adDg] [-s SEED] [-l LANGUAGE] [-e EXAMPLE]
-
-OPTIONS
-
-  -h  Print this message
-
-  -a  Compile C code with the Clang address sanitizer
-
-  -e  Run only the corpus tests whose name contain the given string
-
-  -i  Run the given number of iterations of randomized tests (default 10)
-
-  -s  Set the seed used to control random behavior
-
-  -d  Print parsing log to stderr
-
-  -D  Generate an SVG graph of parsing logs
-
-  -g  Run the tests with a debugger
-
-EOF
-}
-
-export RUST_BACKTRACE=full
-
-mode=normal
-test_flags=()
-
-while getopts "adDghl:e:s:i:" option; do
-  case ${option} in
-    h)
-      usage
-      exit
-      ;;
-    a)
-      export CFLAGS=-fsanitize=undefined,address
-
-      # When the Tree-sitter C library is compiled with the address sanitizer, the address sanitizer
-      # runtime library needs to be linked into the final test executable. When using Xcode clang,
-      # the Rust linker doesn't know where to find that library, so we need to specify linker flags directly.
-      runtime_dir=$(cc -print-runtime-dir)
-      if [[ $runtime_dir == */Xcode.app/* ]]; then
-        export RUSTFLAGS="-C link-arg=-L${runtime_dir} -C link-arg=-lclang_rt.asan_osx_dynamic -C link-arg=-Wl,-rpath,${runtime_dir}"
-      fi
-
-      # Specify a `--target` explicitly. This is required for address sanitizer support.
-      toolchain=$(rustup show active-toolchain)
-      toolchain_regex='(stable|beta|nightly)-([_a-z0-9-]+).*'
-      if [[ $toolchain =~ $toolchain_regex ]]; then
-        release=${BASH_REMATCH[1]}
-        current_target=${BASH_REMATCH[2]}
-      else
-        printf "Failed to parse toolchain '%s'\n" "$toolchain" >&2
-      fi
-
-      test_flags+=("--target=$current_target")
-      ;;
-    e)
-      export TREE_SITTER_EXAMPLE=${OPTARG}
-      ;;
-    s)
-      export TREE_SITTER_SEED=${OPTARG}
-      ;;
-    i)
-      export TREE_SITTER_ITERATIONS=${OPTARG}
-      ;;
-    d)
-      export TREE_SITTER_LOG=1
-      ;;
-    D)
-      export TREE_SITTER_LOG_GRAPHS=1
-      ;;
-    g)
-      mode=debug
-      ;;
-    *)
-      usage
-      exit 1
-      ;;
-  esac
-done
-
-shift $((OPTIND - 1))
-
-if [[ ${mode} == debug ]]; then
-  test_binary=$(
-    cargo test "${test_flags[@]}" --no-run --message-format=json 2> /dev/null |
-    jq -rs 'map(select(.target.name == "tree-sitter-cli" and .executable))[0].executable'
-  )
-  lldb "${test_binary}" -- "$1"
-else
-  cargo test "${test_flags[@]}" "$1" -- --nocapture
-fi

script/test-wasm

@@ -1,12 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-cd lib/binding_web
-
-if [[ ! -d node_modules/chai ]] || [[ ! -d node_modules/mocha ]]; then
-  printf 'Installing test dependencies...\n'
-  npm install
-fi
-
-node_modules/.bin/mocha

script/test.cmd

@@ -1,10 +0,0 @@
-@echo off
-
-setlocal
-set RUST_TEST_THREADS=1
-set RUST_BACKTRACE=full
-cargo test "%~1"
-if %errorlevel% NEQ 0 (
-  exit /b %errorlevel%
-)
-endlocal

script/util/scan-build.sh

@@ -1,24 +0,0 @@
-function scan_build {
-  extra_args=()
-
-  # AFAICT, in the trusty travis container the scan-build tool is from the 3.4
-  # installation. Therefore, by default it will use clang-3.4 when analysing code
-  # which doesn't support the '-std=c++14' (it is available via '-std=c++1y').
-  # Use the system-wide installed clang instead which is 3.5 and does support
-  # '-std=c++14'.
-  extra_args+=("--use-analyzer=$(command -v clang)")
-
-  # scan-build will try to guess which CXX should be used to compile the actual
-  # code, which is usually g++ but we need g++5 in the CI. Explicitly pass
-  # $CC/$CXX to scan-build if they are set in the environment.
-
-  if [[ -n $CC ]]; then
-    extra_args+=("--use-cc=$CC")
-  fi
-
-  if [[ -n $CXX ]]; then
-    extra_args+=("--use-c++=$CXX")
-  fi
-
-  scan-build "${extra_args[@]}" --status-bugs -disable-checker deadcode.DeadStores "$@"
-}

script/util/valgrind.supp

@@ -1,256 +0,0 @@
-# Errors
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN6option6Parser9workhorseEbPKNS_10DescriptorEiPPKcRNS0_6ActionEbbi
-   fun:_ZN6option6Parser5parseEbPKNS_10DescriptorEiPPKcPNS_6OptionES8_ibi
-   fun:_ZN6option6ParserC1EPKNS_10DescriptorEiPPcPNS_6OptionES7_ibi
-   fun:_ZN6bandit6detail7optionsC1EiPPc
-   fun:_ZN6bandit3runEiPPc
-   fun:main
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN6option6Parser9workhorseEbPKNS_10DescriptorEiPPKcRNS0_6ActionEbbi
-   fun:_ZN6option5Stats3addEbPKNS_10DescriptorEiPPKcib
-   fun:_ZN6option5StatsC1EPKNS_10DescriptorEiPPcib
-   fun:_ZN6bandit6detail7optionsC1EiPPc
-   fun:_ZN6bandit3runEiPPc
-   fun:main
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN6option6Parser9workhorseEbPKNS_10DescriptorEiPPKcRNS0_6ActionEbbi
-   fun:_ZN6bandit6detail7optionsC2EiPPc
-   fun:_ZN6bandit3runEiPPc
-   fun:main
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Value8
-   fun:_platform_memcmp
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_ENKUlvE_clEvEUlvE0_NS7_ISD_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_EUlvE_NS7_ISC_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Addr1
-   fun:_platform_memcmp
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_ENKUlvE_clEvEUlvE0_NS7_ISD_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_EUlvE_NS7_ISC_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_ENKUlvE_clEvEUlvE0_NS7_ISD_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE9ts_parserE_clES9_SA_EUlvE_NS7_ISC_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlvE3_clEvEUlvE0_NS_9allocatorIS5_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE3_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Value8
-   fun:_platform_memcmp
-   fun:_ZNK9snowhouse16EqualsConstraintINSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEEEclIS7_EEbRKT_
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlvE3_clEvEUlvE0_NS_9allocatorIS5_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE3_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE1_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcI3$_0NS_9allocatorIS2_EEFvvEEclEv
-   fun:_ZN6bandit3runERKNS_6detail7optionsERKNSt3__14listINS4_8functionIFvvEEENS4_9allocatorIS8_EEEERNS4_5dequeIPNS0_7contextENS9_ISG_EEEERNS0_8listenerE
-   fun:_ZN6bandit3runEiPPc
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Value8
-   fun:_platform_memcmp
-   fun:_ZNK9snowhouse16EqualsConstraintINSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEEEclIS7_EEbRKT_
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE1_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcI3$_0NS_9allocatorIS2_EEFvvEEclEv
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlvE4_clEvEUlvE0_NS_9allocatorIS5_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE4_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Value8
-   fun:_platform_memcmp
-   fun:_ZNK9snowhouse16EqualsConstraintINSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEEEclIS7_EEbRKT_
-   fun:_ZN9snowhouse6Assert4ThatINSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS_16EqualsConstraintIS8_EEEEvRKT_RKT0_PKci
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlvE4_clEvEUlvE0_NS_9allocatorIS5_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZNK3$_0clEvENKUlvE_clEvEUlvE4_NS_9allocatorIS4_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Cond
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_ENKUlvE_clEvEUlvE0_NS7_ISF_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_EUlvE_NS7_ISE_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZNK3$_0clEvEUlvE_NS_9allocatorIS3_EEFvvEEclEv
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Addr1
-   fun:_platform_memcmp
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_ENKUlvE_clEvEUlvE0_NS7_ISF_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_EUlvE_NS7_ISE_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Value8
-   fun:_platform_memcmp
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEmmPKcm
-   fun:_ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7compareEPKc
-   fun:_ZN9snowhouse6Assert4ThatIPKcNS_16EqualsConstraintINSt3__112basic_stringIcNS5_11char_traitsIcEENS5_9allocatorIcEEEEEEEEvRKT_RKT0_S3_i
-   fun:_ZNSt3__110__function6__funcIZZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_ENKUlvE_clEvEUlvE0_NS7_ISF_EEFvvEEclEv
-   fun:_ZNSt3__110__function6__funcIZN6bandit2itEPKcNS_8functionIFvvEEERNS2_6detail8listenerERNS_5dequeIPNS8_7contextENS_9allocatorISD_EEEERNS2_8adapters17assertion_adapterERKNS8_10run_policyEEUlvE1_NSE_ISO_EES6_EclEv
-   fun:_ZN6bandit8adapters17snowhouse_adapter16adapt_exceptionsENSt3__18functionIFvvEEE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEERNS_8adapters17assertion_adapterERKNS6_10run_policyE
-   fun:_ZN6bandit2itEPKcNSt3__18functionIFvvEEE
-   fun:_ZNSt3__110__function6__funcIZZZNK3$_0clEvENKUlvE_clEvENKUlNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPF9ts_parservEE_clES9_SC_EUlvE_NS7_ISE_EEFvvEEclEv
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEERNS_6detail8listenerERNS2_5dequeIPNS6_7contextENS2_9allocatorISB_EEEEb
-   fun:_ZN6bandit8describeEPKcNSt3__18functionIFvvEEE
-}
-
-# Leaks
-
-{
-   <insert_a_suppression_name_here>
-   Memcheck:Leak
-   match-leak-kinds: possible
-   fun:malloc_zone_malloc
-   fun:_objc_copyClassNamesForImage
-   fun:_ZL9protocolsv
-   fun:_Z9readClassP10objc_classbb
-   fun:gc_init
-   fun:_ZL33objc_initializeClassPair_internalP10objc_classPKcS0_S0_
-   fun:layout_string_create
-   fun:_ZL12realizeClassP10objc_class
-   fun:_ZL22copySwiftV1MangledNamePKcb
-   fun:_ZL22copySwiftV1MangledNamePKcb
-   fun:_ZL22copySwiftV1MangledNamePKcb
-   fun:_ZL22copySwiftV1MangledNamePKcb
-}

xtask/Cargo.toml

@@ -11,10 +11,19 @@ keywords.workspace = true
 categories.workspace = true
 publish = false
 
+[lints]
+workspace = true
+
 [dependencies]
+anstyle.workspace = true
+anyhow.workspace = true
+bindgen = { version = "0.70.1" }
+cc.workspace = true
+clap.workspace = true
 git2.workspace = true
 indoc.workspace = true
 toml.workspace = true
+regex.workspace = true
 semver.workspace = true
 serde.workspace = true
 serde_json.workspace = true

xtask/src/benchmark.rs

@@ -0,0 +1,75 @@
+use anyhow::Result;
+
+use crate::{bail_on_err, Benchmark};
+
+pub fn run(args: &Benchmark) -> Result<()> {
+    if let Some(ref example) = args.example_file_name {
+        std::env::set_var("TREE_SITTER_BENCHMARK_EXAMPLE_FILTER", example);
+    }
+
+    if let Some(ref language) = args.language {
+        std::env::set_var("TREE_SITTER_BENCHMARK_LANGUAGE_FILTER", language);
+    }
+
+    if args.repetition_count != 5 {
+        std::env::set_var(
+            "TREE_SITTER_BENCHMARK_REPETITION_COUNT",
+            args.repetition_count.to_string(),
+        );
+    }
+
+    if args.debug {
+        let output = std::process::Command::new("cargo")
+            .arg("bench")
+            .arg("benchmark")
+            .arg("-p")
+            .arg("tree-sitter-cli")
+            .arg("--no-run")
+            .arg("--message-format=json")
+            .spawn()?
+            .wait_with_output()?;
+
+        bail_on_err(&output, "Failed to run `cargo bench`")?;
+
+        let json_output = serde_json::from_slice::<serde_json::Value>(&output.stdout)?;
+
+        let test_binary = json_output
+            .as_array()
+            .ok_or_else(|| anyhow::anyhow!("Invalid JSON output"))?
+            .iter()
+            .find_map(|message| {
+                if message
+                    .get("target")
+                    .and_then(|target| target.get("name"))
+                    .and_then(|name| name.as_str())
+                    .is_some_and(|name| name == "benchmark")
+                    && message
+                        .get("executable")
+                        .and_then(|executable| executable.as_str())
+                        .is_some()
+                {
+                    message
+                        .get("executable")
+                        .and_then(|executable| executable.as_str())
+                } else {
+                    None
+                }
+            })
+            .ok_or_else(|| anyhow::anyhow!("Failed to find benchmark executable"))?;
+
+        println!("{test_binary}");
+    } else {
+        let status = std::process::Command::new("cargo")
+            .arg("bench")
+            .arg("benchmark")
+            .arg("-p")
+            .arg("tree-sitter-cli")
+            .status()?;
+
+        if !status.success() {
+            anyhow::bail!("Failed to run `cargo bench`");
+        }
+    }
+
+    Ok(())
+}

xtask/src/build_wasm.rs

@@ -0,0 +1,228 @@
+use std::{
+    ffi::{OsStr, OsString},
+    fmt::Write,
+    fs,
+    process::Command,
+};
+
+use anyhow::{anyhow, Result};
+
+use crate::{bail_on_err, BuildWasm, EMSCRIPTEN_TAG};
+
+#[derive(PartialEq, Eq)]
+enum EmccSource {
+    Native,
+    Docker,
+    Podman,
+}
+
+pub fn run_wasm(args: &BuildWasm) -> Result<()> {
+    let mut emscripten_flags = vec!["-O3", "--minify", "0"];
+
+    if args.debug {
+        emscripten_flags.extend(["-s", "ASSERTIONS=1", "-s", "SAFE_HEAP=1", "-O0", "-g"]);
+    }
+
+    if args.verbose {
+        emscripten_flags.extend(["-s", "VERBOSE=1", "-v"]);
+    }
+
+    let emcc_name = if cfg!(windows) { "emcc.bat" } else { "emcc" };
+
+    // Order of preference: emscripten > docker > podman > error
+    let source = if !args.docker && Command::new(emcc_name).output().is_ok() {
+        EmccSource::Native
+    } else if Command::new("docker")
+        .arg("info")
+        .output()
+        .map_or(false, |out| out.status.success())
+    {
+        EmccSource::Docker
+    } else if Command::new("podman")
+        .arg("--version")
+        .output()
+        .map_or(false, |out| out.status.success())
+    {
+        EmccSource::Podman
+    } else {
+        return Err(anyhow!(
+            "You must have either emcc, docker, or podman on your PATH to run this command"
+        ));
+    };
+
+    let mut command = match source {
+        EmccSource::Native => Command::new(emcc_name),
+        EmccSource::Docker | EmccSource::Podman => {
+            let mut command = match source {
+                EmccSource::Docker => Command::new("docker"),
+                EmccSource::Podman => Command::new("podman"),
+                _ => unreachable!(),
+            };
+            command.args(["run", "--rm"]);
+
+            // Mount the root directory as a volume, which is the repo root
+            let mut volume_string = OsString::from(std::env::current_dir().unwrap());
+            volume_string.push(":/src:Z");
+            command.args([OsStr::new("--volume"), &volume_string]);
+
+            // In case `docker` is an alias to `podman`, ensure that podman
+            // mounts the current directory as writable by the container
+            // user which has the same uid as the host user. Setting the
+            // podman-specific variable is more reliable than attempting to
+            // detect whether `docker` is an alias for `podman`.
+            // see https://docs.podman.io/en/latest/markdown/podman-run.1.html#userns-mode
+            command.env("PODMAN_USERNS", "keep-id");
+
+            // Get the current user id so that files created in the docker container will have
+            // the same owner.
+            #[cfg(unix)]
+            {
+                #[link(name = "c")]
+                extern "C" {
+                    fn getuid() -> u32;
+                }
+                // don't need to set user for podman since PODMAN_USERNS=keep-id is already set
+                if source == EmccSource::Docker {
+                    let user_id = unsafe { getuid() };
+                    command.args(["--user", &user_id.to_string()]);
+                }
+            };
+
+            // Run `emcc` in a container using the `emscripten-slim` image
+            command.args([EMSCRIPTEN_TAG, "emcc"]);
+            command
+        }
+    };
+
+    fs::create_dir_all("target/scratch").unwrap();
+
+    let exported_functions = concat!(
+        include_str!("../../lib/src/wasm/stdlib-symbols.txt"),
+        include_str!("../../lib/binding_web/exports.txt")
+    )
+    .replace('"', "")
+    .lines()
+    .fold(String::new(), |mut output, line| {
+        let _ = write!(output, "_{line}");
+        output
+    })
+    .trim_end_matches(',')
+    .to_string();
+
+    let exported_functions = format!("EXPORTED_FUNCTIONS={exported_functions}");
+    let exported_runtime_methods = "EXPORTED_RUNTIME_METHODS=stringToUTF16,AsciiToString";
+
+    emscripten_flags.extend([
+        "-s",
+        "WASM=1",
+        "-s",
+        "INITIAL_MEMORY=33554432",
+        "-s",
+        "ALLOW_MEMORY_GROWTH=1",
+        "-s",
+        "SUPPORT_BIG_ENDIAN=1",
+        "-s",
+        "MAIN_MODULE=2",
+        "-s",
+        "FILESYSTEM=0",
+        "-s",
+        "NODEJS_CATCH_EXIT=0",
+        "-s",
+        "NODEJS_CATCH_REJECTION=0",
+        "-s",
+        &exported_functions,
+        "-s",
+        exported_runtime_methods,
+        "-fno-exceptions",
+        "-std=c11",
+        "-D",
+        "fprintf(...)=",
+        "-D",
+        "NDEBUG=",
+        "-D",
+        "_POSIX_C_SOURCE=200112L",
+        "-D",
+        "_DEFAULT_SOURCE=",
+        "-I",
+        "lib/src",
+        "-I",
+        "lib/include",
+        "--js-library",
+        "lib/binding_web/imports.js",
+        "--pre-js",
+        "lib/binding_web/prefix.js",
+        "--post-js",
+        "lib/binding_web/binding.js",
+        "--post-js",
+        "lib/binding_web/suffix.js",
+        "lib/src/lib.c",
+        "lib/binding_web/binding.c",
+        "-o",
+        "target/scratch/tree-sitter.js",
+    ]);
+
+    bail_on_err(
+        &command.args(emscripten_flags).spawn()?.wait_with_output()?,
+        "Failed to compile the Tree-sitter WASM library",
+    )?;
+
+    fs::rename(
+        "target/scratch/tree-sitter.js",
+        "lib/binding_web/tree-sitter.js",
+    )?;
+
+    fs::rename(
+        "target/scratch/tree-sitter.wasm",
+        "lib/binding_web/tree-sitter.wasm",
+    )?;
+
+    Ok(())
+}
+
+pub fn run_wasm_stdlib() -> Result<()> {
+    let export_flags = include_str!("../../lib/src/wasm/stdlib-symbols.txt")
+        .lines()
+        .map(|line| format!("-Wl,--export={}", &line[1..line.len() - 1]))
+        .collect::<Vec<String>>();
+
+    let mut command = Command::new("target/wasi-sdk-21.0/bin/clang-17");
+
+    let output = command
+        .args([
+            "-o",
+            "stdlib.wasm",
+            "-Os",
+            "-fPIC",
+            "-Wl,--no-entry",
+            "-Wl,--stack-first",
+            "-Wl,-z",
+            "-Wl,stack-size=65536",
+            "-Wl,--import-undefined",
+            "-Wl,--import-memory",
+            "-Wl,--import-table",
+            "-Wl,--strip-debug",
+            "-Wl,--export=reset_heap",
+            "-Wl,--export=__wasm_call_ctors",
+            "-Wl,--export=__stack_pointer",
+        ])
+        .args(export_flags)
+        .arg("lib/src/wasm/stdlib.c")
+        .output()?;
+
+    bail_on_err(&output, "Failed to compile the Tree-sitter WASM stdlib")?;
+
+    let xxd = Command::new("xxd")
+        .args(["-C", "-i", "stdlib.wasm"])
+        .output()?;
+
+    bail_on_err(
+        &xxd,
+        "Failed to run xxd on the compiled Tree-sitter WASM stdlib",
+    )?;
+
+    fs::write("lib/src/wasm/wasm-stdlib.h", xxd.stdout)?;
+
+    fs::rename("stdlib.wasm", "target/stdlib.wasm")?;
+
+    Ok(())
+}

xtask/src/bump.rs

@@ -1,11 +1,14 @@
 use std::{cmp::Ordering, path::Path};
 
+use anyhow::{anyhow, Result};
 use git2::{DiffOptions, Repository};
 use indoc::indoc;
 use semver::{BuildMetadata, Prerelease, Version};
 use toml::Value;
 
-pub fn get_latest_tag(repo: &Repository) -> Result<String, Box<dyn std::error::Error>> {
+use crate::BumpVersion;
+
+pub fn get_latest_tag(repo: &Repository) -> Result<String> {
     let mut tags = repo
         .tag_names(None)?
         .into_iter()
@@ -23,10 +26,10 @@ pub fn get_latest_tag(repo: &Repository) -> Result<String, Box<dyn std::error::E
 
     tags.last()
         .map(std::string::ToString::to_string)
-        .ok_or_else(|| "No tags found".into())
+        .ok_or_else(|| anyhow!("No tags found"))
 }
 
-pub fn bump_versions() -> Result<(), Box<dyn std::error::Error>> {
+pub fn run(args: BumpVersion) -> Result<()> {
     let repo = Repository::open(".")?;
     let latest_tag = get_latest_tag(&repo)?;
     let current_version = Version::parse(&latest_tag)?;
@@ -104,35 +107,39 @@ pub fn bump_versions() -> Result<(), Box<dyn std::error::Error>> {
         }
     }
 
-    let mut version = current_version.clone();
+    let next_version = if let Some(version) = args.version {
-    if should_increment_minor {
+        version
-        version.minor += 1;
-        version.patch = 0;
-        version.pre = Prerelease::EMPTY;
-        version.build = BuildMetadata::EMPTY;
-    } else if should_increment_patch {
-        version.patch += 1;
-        version.pre = Prerelease::EMPTY;
-        version.build = BuildMetadata::EMPTY;
     } else {
-        return Err(format!("No source code changed since {current_version}").into());
+        let mut next_version = current_version.clone();
-    }
+        if should_increment_minor {
+            next_version.minor += 1;
+            next_version.patch = 0;
+            next_version.pre = Prerelease::EMPTY;
+            next_version.build = BuildMetadata::EMPTY;
+        } else if should_increment_patch {
+            next_version.patch += 1;
+            next_version.pre = Prerelease::EMPTY;
+            next_version.build = BuildMetadata::EMPTY;
+        } else {
+            return Err(anyhow!(format!(
+                "No source code changed since {current_version}"
+            )));
+        }
+        next_version
+    };
 
-    println!("Bumping from {current_version} to {version}");
+    println!("Bumping from {current_version} to {next_version}");
-    update_crates(&current_version, &version)?;
+    update_crates(&current_version, &next_version)?;
-    update_makefile(&version)?;
+    update_makefile(&next_version)?;
-    update_cmake(&version)?;
+    update_cmake(&next_version)?;
-    update_npm(&version)?;
+    update_npm(&next_version)?;
-    update_zig(&version)?;
+    update_zig(&next_version)?;
-    tag_next_version(&repo, &version)?;
+    tag_next_version(&repo, &next_version)?;
 
     Ok(())
 }
 
-fn tag_next_version(
+fn tag_next_version(repo: &Repository, next_version: &Version) -> Result<()> {
-    repo: &Repository,
-    next_version: &Version,
-) -> Result<(), Box<dyn std::error::Error>> {
     // first add the manifests
 
     let mut index = repo.index()?;
@@ -184,7 +191,7 @@ fn tag_next_version(
     Ok(())
 }
 
-fn update_makefile(next_version: &Version) -> Result<(), Box<dyn std::error::Error>> {
+fn update_makefile(next_version: &Version) -> Result<()> {
     let makefile = std::fs::read_to_string("Makefile")?;
     let makefile = makefile
         .lines()
@@ -204,7 +211,7 @@ fn update_makefile(next_version: &Version) -> Result<(), Box<dyn std::error::Err
     Ok(())
 }
 
-fn update_cmake(next_version: &Version) -> Result<(), Box<dyn std::error::Error>> {
+fn update_cmake(next_version: &Version) -> Result<()> {
     let cmake = std::fs::read_to_string("lib/CMakeLists.txt")?;
     let cmake = cmake
         .lines()
@@ -230,10 +237,7 @@ fn update_cmake(next_version: &Version) -> Result<(), Box<dyn std::error::Error>
     Ok(())
 }
 
-fn update_crates(
+fn update_crates(current_version: &Version, next_version: &Version) -> Result<()> {
-    current_version: &Version,
-    next_version: &Version,
-) -> Result<(), Box<dyn std::error::Error>> {
     let mut cmd = std::process::Command::new("cargo");
     cmd.arg("workspaces").arg("version");
 
@@ -253,20 +257,20 @@ fn update_crates(
     let status = cmd.status()?;
 
     if !status.success() {
-        return Err("Failed to update crates".into());
+        return Err(anyhow!("Failed to update crates"));
     }
 
     Ok(())
 }
 
-fn update_npm(next_version: &Version) -> Result<(), Box<dyn std::error::Error>> {
+fn update_npm(next_version: &Version) -> Result<()> {
     for path in ["lib/binding_web/package.json", "cli/npm/package.json"] {
         let package_json =
             serde_json::from_str::<serde_json::Value>(&std::fs::read_to_string(path)?)?;
 
         let mut package_json = package_json
             .as_object()
-            .ok_or("Invalid package.json")?
+            .ok_or_else(|| anyhow!("Invalid package.json"))?
             .clone();
         package_json.insert(
             "version".to_string(),
@@ -281,7 +285,7 @@ fn update_npm(next_version: &Version) -> Result<(), Box<dyn std::error::Error>>
     Ok(())
 }
 
-fn update_zig(next_version: &Version) -> Result<(), Box<dyn std::error::Error>> {
+fn update_zig(next_version: &Version) -> Result<()> {
     let zig = std::fs::read_to_string("build.zig.zon")?;
 
     let zig = zig
@@ -303,7 +307,7 @@ fn update_zig(next_version: &Version) -> Result<(), Box<dyn std::error::Error>>
 }
 
 /// read Cargo.toml and get the version
-fn fetch_workspace_version() -> Result<String, Box<dyn std::error::Error>> {
+fn fetch_workspace_version() -> Result<String> {
     let cargo_toml = toml::from_str::<Value>(&std::fs::read_to_string("Cargo.toml")?)?;
 
     Ok(cargo_toml["workspace"]["package"]["version"]

xtask/src/clippy.rs

@@ -0,0 +1,33 @@
+use std::process::Command;
+
+use anyhow::Result;
+
+use crate::{bail_on_err, Clippy};
+
+pub fn run(args: &Clippy) -> Result<()> {
+    let mut clippy_command = Command::new("cargo");
+    clippy_command.arg("+nightly").arg("clippy");
+
+    if let Some(package) = args.package.as_ref() {
+        clippy_command.args(["--package", package]);
+    } else {
+        clippy_command.arg("--workspace");
+    }
+
+    clippy_command
+        .arg("--release")
+        .arg("--all-targets")
+        .arg("--all-features")
+        .arg("--")
+        .arg("-D")
+        .arg("warnings");
+
+    if args.fix {
+        clippy_command.arg("--fix");
+    }
+
+    bail_on_err(
+        &clippy_command.spawn()?.wait_with_output()?,
+        "Clippy failed",
+    )
+}

xtask/src/fetch.rs

@@ -0,0 +1,119 @@
+use std::{path::Path, process::Command};
+
+use anyhow::Result;
+
+use crate::{bail_on_err, EMSCRIPTEN_VERSION};
+
+pub fn run_fixtures() -> Result<()> {
+    let grammars_dir = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .parent()
+        .unwrap()
+        .join("test")
+        .join("fixtures")
+        .join("grammars");
+
+    [
+        ("bash", "master"),
+        ("c", "master"),
+        ("cpp", "master"),
+        ("embedded-template", "master"),
+        ("go", "master"),
+        ("html", "master"),
+        ("java", "master"),
+        ("javascript", "master"),
+        ("jsdoc", "master"),
+        ("json", "master"),
+        ("php", "master"),
+        ("python", "master"),
+        ("ruby", "master"),
+        ("rust", "master"),
+        ("typescript", "master"),
+    ]
+    .iter()
+    .try_for_each(|(grammar, r#ref)| {
+        let grammar_dir = grammars_dir.join(grammar);
+        let grammar_url = format!("https://github.com/tree-sitter/tree-sitter-{grammar}");
+
+        println!("Updating the {grammar} grammar...");
+
+        if !grammar_dir.exists() {
+            let mut command = Command::new("git");
+            command.args([
+                "clone",
+                "--depth",
+                "1",
+                &grammar_url,
+                &grammar_dir.to_string_lossy(),
+            ]);
+            bail_on_err(
+                &command.spawn()?.wait_with_output()?,
+                "Failed to clone the {grammar} grammar",
+            )?;
+        }
+
+        std::env::set_current_dir(&grammar_dir)?;
+
+        let mut command = Command::new("git");
+        command.args(["fetch", "origin", r#ref, "--depth", "1"]);
+        bail_on_err(
+            &command.spawn()?.wait_with_output()?,
+            "Failed to fetch the {grammar} grammar",
+        )?;
+
+        let mut command = Command::new("git");
+        command.args(["reset", "--hard", "FETCH_HEAD"]);
+        bail_on_err(
+            &command.spawn()?.wait_with_output()?,
+            "Failed to reset the {grammar} grammar",
+        )?;
+
+        Ok(())
+    })
+}
+
+pub fn run_emscripten() -> Result<()> {
+    let emscripten_dir = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .parent()
+        .unwrap()
+        .join("target")
+        .join("emsdk");
+
+    if emscripten_dir.exists() {
+        println!("Emscripten SDK already exists");
+        return Ok(());
+    }
+    println!("Cloning the Emscripten SDK...");
+
+    let mut command = Command::new("git");
+    command.args([
+        "clone",
+        "https://github.com/emscripten-core/emsdk.git",
+        &emscripten_dir.to_string_lossy(),
+    ]);
+    bail_on_err(
+        &command.spawn()?.wait_with_output()?,
+        "Failed to clone the Emscripten SDK",
+    )?;
+
+    std::env::set_current_dir(&emscripten_dir)?;
+
+    let emsdk = if cfg!(windows) {
+        "emsdk.bat"
+    } else {
+        "./emsdk"
+    };
+
+    let mut command = Command::new(emsdk);
+    command.args(["install", EMSCRIPTEN_VERSION]);
+    bail_on_err(
+        &command.spawn()?.wait_with_output()?,
+        "Failed to install Emscripten",
+    )?;
+
+    let mut command = Command::new(emsdk);
+    command.args(["activate", EMSCRIPTEN_VERSION]);
+    bail_on_err(
+        &command.spawn()?.wait_with_output()?,
+        "Failed to activate Emscripten",
+    )
+}

xtask/src/generate.rs

@@ -0,0 +1,118 @@
+use std::{ffi::OsStr, fs, process::Command};
+
+use anyhow::{Context, Result};
+
+use crate::{bail_on_err, GenerateFixtures};
+
+const HEADER_PATH: &str = "include/tree_sitter/api.h";
+
+pub fn run_fixtures(args: &GenerateFixtures) -> Result<()> {
+    let output = std::process::Command::new("cargo")
+        .args(["build", "--release"])
+        .spawn()?
+        .wait_with_output()?;
+    bail_on_err(&output, "Failed to run cargo build")?;
+
+    let tree_sitter_binary = std::env::current_dir()?
+        .join("target")
+        .join("release")
+        .join("tree-sitter");
+
+    let grammars_dir = std::env::current_dir()?
+        .join("test")
+        .join("fixtures")
+        .join("grammars");
+
+    for grammar_file in find_grammar_files(grammars_dir.to_str().unwrap()).flatten() {
+        let grammar_dir = grammar_file.parent().unwrap();
+        let grammar_name = grammar_dir.file_name().and_then(OsStr::to_str).unwrap();
+
+        println!(
+            "Regenerating {grammar_name} parser{}",
+            if args.wasm { " to wasm" } else { "" }
+        );
+
+        if args.wasm {
+            let mut cmd = Command::new(&tree_sitter_binary);
+            let cmd = cmd.args([
+                "build",
+                "--wasm",
+                "-o",
+                &format!("target/release/tree-sitter-{grammar_name}.wasm"),
+                grammar_dir.to_str().unwrap(),
+            ]);
+            if args.docker {
+                cmd.arg("--docker");
+            }
+            bail_on_err(
+                &cmd.spawn()?.wait_with_output()?,
+                &format!("Failed to regenerate {grammar_name} parser to wasm"),
+            )?;
+        } else {
+            let output = Command::new(&tree_sitter_binary)
+                .arg("generate")
+                .arg("src/grammar.json")
+                .arg("--abi=latest")
+                .current_dir(grammar_dir)
+                .spawn()?
+                .wait_with_output()?;
+            bail_on_err(
+                &output,
+                &format!("Failed to regenerate {grammar_name} parser"),
+            )?;
+        }
+    }
+
+    Ok(())
+}
+
+pub fn run_bindings() -> Result<()> {
+    let no_copy = [
+        "TSInput",
+        "TSLanguage",
+        "TSLogger",
+        "TSLookaheadIterator",
+        "TSParser",
+        "TSTree",
+        "TSQuery",
+        "TSQueryCursor",
+        "TSQueryCapture",
+        "TSQueryMatch",
+        "TSQueryPredicateStep",
+    ];
+
+    let bindings = bindgen::Builder::default()
+        .header(HEADER_PATH)
+        .layout_tests(false)
+        .allowlist_type("^TS.*")
+        .allowlist_function("^ts_.*")
+        .allowlist_var("^TREE_SITTER.*")
+        .no_copy(no_copy.join("|"))
+        .prepend_enum_name(false)
+        .use_core()
+        .clang_arg("-D TREE_SITTER_FEATURE_WASM")
+        .generate()
+        .expect("Failed to generate bindings");
+
+    bindings
+        .write_to_file("lib/binding_rust/bindings.rs")
+        .with_context(|| "Failed to write bindings")
+}
+
+fn find_grammar_files(
+    dir: &str,
+) -> impl Iterator<Item = Result<std::path::PathBuf, std::io::Error>> {
+    fs::read_dir(dir)
+        .expect("Failed to read directory")
+        .filter_map(Result::ok)
+        .flat_map(|entry| {
+            let path = entry.path();
+            if path.is_dir() && !path.to_string_lossy().contains("node_modules") {
+                Box::new(find_grammar_files(path.to_str().unwrap())) as Box<dyn Iterator<Item = _>>
+            } else if path.is_file() && path.file_name() == Some(OsStr::new("grammar.js")) {
+                Box::new(std::iter::once(Ok(path))) as _
+            } else {
+                Box::new(std::iter::empty()) as _
+            }
+        })
+}

xtask/src/main.rs

@@ -1,35 +1,235 @@
+mod benchmark;
+mod build_wasm;
 mod bump;
+mod clippy;
+mod fetch;
+mod generate;
+mod test;
 
-use bump::bump_versions;
+use anstyle::{AnsiColor, Color, Style};
+use anyhow::Result;
+use clap::{crate_authors, Args, Command, FromArgMatches as _, Subcommand};
+use semver::Version;
 
-fn print_help() {
+#[derive(Subcommand)]
-    println!(
+#[command(about="Run various tasks", author=crate_authors!("\n"), styles=get_styles())]
-        "
+enum Commands {
-xtask must specify a task to run.
+    /// Runs `cargo benchmark` with some optional environment variables set.
-
+    Benchmark(Benchmark),
-Usage: `cargo xtask <task>`
+    /// Compile the Tree-sitter WASM library. This will create two files in the
-
+    /// `lib/binding_web` directory: `tree-sitter.js` and `tree-sitter.wasm`.
-Tasks:
+    BuildWasm(BuildWasm),
-    bump-version
+    /// Compile the Tree-sitter WASM standard library.
-"
+    BuildWasmStdlib,
-    );
+    /// Bumps the version of the workspace.
+    BumpVersion(BumpVersion),
+    /// Runs `cargo clippy`.
+    Clippy(Clippy),
+    /// Fetches emscripten.
+    FetchEmscripten,
+    /// Fetches the fixtures for testing tree-sitter.
+    FetchFixtures,
+    /// Generate the Rust bindings from the C library.
+    GenerateBindings,
+    /// Generates the fixtures for testing tree-sitter.
+    GenerateFixtures(GenerateFixtures),
+    /// Run the test suite
+    Test(Test),
+    /// Run the WASM test suite
+    TestWasm,
 }
 
-fn main() -> Result<(), Box<dyn std::error::Error>> {
+#[derive(Args)]
-    let Some(task) = std::env::args().nth(1) else {
+struct Benchmark {
-        print_help();
+    /// The language to run the benchmarks for.
-        std::process::exit(0);
+    #[arg(long, short)]
-    };
+    language: Option<String>,
+    /// The example file to run the benchmarks for.
+    #[arg(long, short)]
+    example_file_name: Option<String>,
+    /// The number of times to parse each sample (default is 5).
+    #[arg(long, short, default_value = "5")]
+    repetition_count: u32,
+    /// Whether to run the benchmarks in debug mode.
+    #[arg(long, short = 'g')]
+    debug: bool,
+}
 
-    match task.as_str() {
+#[derive(Args)]
-        "bump-version" => {
+struct BuildWasm {
-            bump_versions()?;
+    /// Compile the library more quickly, with fewer optimizations
+    /// and more runtime assertions.
+    #[arg(long, short = '0')]
+    debug: bool,
+    /// Run emscripten using docker, even if \`emcc\` is installed.
+    /// By default, \`emcc\` will be run directly when available.
+    #[arg(long, short)]
+    docker: bool,
+    /// Run emscripten with verbose output.
+    #[arg(long, short)]
+    verbose: bool,
+}
+
+#[derive(Args)]
+struct BumpVersion {
+    /// The version to bump to.
+    #[arg(long, short)]
+    version: Option<Version>,
+}
+
+#[derive(Args)]
+struct Clippy {
+    /// Automatically apply lint suggestions (`clippy --fix`).
+    #[arg(long, short)]
+    fix: bool,
+    /// The package to run Clippy against (`cargo -p <PACKAGE> clippy`).
+    #[arg(long, short)]
+    package: Option<String>,
+}
+
+#[derive(Args)]
+struct GenerateFixtures {
+    /// Generates the parser to WASM
+    #[arg(long, short)]
+    wasm: bool,
+    /// Run emscripten via docker even if it is installed locally.
+    #[arg(long, short, requires = "wasm")]
+    docker: bool,
+}
+
+#[derive(Args)]
+struct Test {
+    /// Compile C code with the Clang address sanitizer.
+    #[arg(long, short)]
+    address_sanitizer: bool,
+    /// Run only the corpus tests whose name contain the given string.
+    #[arg(long, short)]
+    example: Option<String>,
+    /// Run the given number of iterations of randomized tests (default 10).
+    #[arg(long, short)]
+    iterations: Option<u32>,
+    /// Set the seed used to control random behavior.
+    #[arg(long, short)]
+    seed: Option<u32>,
+    /// Print parsing log to stderr.
+    #[arg(long, short)]
+    debug: bool,
+    /// Generate an SVG graph of parsing logs.
+    #[arg(long, short = 'D')]
+    debug_graph: bool,
+    /// Run the tests with a debugger.
+    #[arg(short)]
+    g: bool,
+    #[arg(trailing_var_arg = true)]
+    args: Vec<String>,
+    /// Don't capture the output
+    #[arg(long)]
+    nocapture: bool,
+}
+
+const BUILD_VERSION: &str = env!("CARGO_PKG_VERSION");
+const BUILD_SHA: Option<&str> = option_env!("BUILD_SHA");
+const EMSCRIPTEN_VERSION: &str = include_str!("../../cli/loader/emscripten-version");
+const EMSCRIPTEN_TAG: &str = concat!(
+    "docker.io/emscripten/emsdk:",
+    include_str!("../../cli/loader/emscripten-version")
+);
+
+fn main() {
+    let result = run();
+    if let Err(err) = &result {
+        // Ignore BrokenPipe errors
+        if let Some(error) = err.downcast_ref::<std::io::Error>() {
+            if error.kind() == std::io::ErrorKind::BrokenPipe {
+                return;
+            }
         }
-        _ => {
+        if !err.to_string().is_empty() {
-            println!("invalid task: {task}");
+            eprintln!("{err:?}");
-            std::process::exit(1);
         }
+        std::process::exit(1);
+    }
+}
+
+fn run() -> Result<()> {
+    let version = BUILD_SHA.map_or_else(
+        || BUILD_VERSION.to_string(),
+        |build_sha| format!("{BUILD_VERSION} ({build_sha})"),
+    );
+    let version: &'static str = Box::leak(version.into_boxed_str());
+
+    let cli = Command::new("xtask")
+        .help_template(
+            "\
+{before-help}{name} {version}
+{author-with-newline}{about-with-newline}
+{usage-heading} {usage}
+
+{all-args}{after-help}
+",
+        )
+        .version(version)
+        .subcommand_required(true)
+        .arg_required_else_help(true)
+        .disable_help_subcommand(true)
+        .disable_colored_help(false);
+    let command = Commands::from_arg_matches(&Commands::augment_subcommands(cli).get_matches())?;
+
+    match command {
+        Commands::Benchmark(benchmark_options) => benchmark::run(&benchmark_options)?,
+        Commands::BuildWasm(build_wasm_options) => build_wasm::run_wasm(&build_wasm_options)?,
+        Commands::BuildWasmStdlib => build_wasm::run_wasm_stdlib()?,
+        Commands::BumpVersion(bump_options) => bump::run(bump_options)?,
+        Commands::Clippy(clippy_options) => clippy::run(&clippy_options)?,
+        Commands::FetchEmscripten => fetch::run_emscripten()?,
+        Commands::FetchFixtures => fetch::run_fixtures()?,
+        Commands::GenerateBindings => generate::run_bindings()?,
+        Commands::GenerateFixtures(generate_fixtures_options) => {
+            generate::run_fixtures(&generate_fixtures_options)?;
+        }
+        Commands::Test(test_options) => test::run(&test_options)?,
+        Commands::TestWasm => test::run_wasm()?,
     }
 
     Ok(())
 }
+
+fn bail_on_err(output: &std::process::Output, prefix: &str) -> Result<()> {
+    if !output.status.success() {
+        let stderr = String::from_utf8_lossy(&output.stderr);
+        anyhow::bail!("{prefix}:\n{stderr}");
+    }
+    Ok(())
+}
+
+#[must_use]
+const fn get_styles() -> clap::builder::Styles {
+    clap::builder::Styles::styled()
+        .usage(
+            Style::new()
+                .bold()
+                .fg_color(Some(Color::Ansi(AnsiColor::Yellow))),
+        )
+        .header(
+            Style::new()
+                .bold()
+                .fg_color(Some(Color::Ansi(AnsiColor::Yellow))),
+        )
+        .literal(Style::new().fg_color(Some(Color::Ansi(AnsiColor::Green))))
+        .invalid(
+            Style::new()
+                .bold()
+                .fg_color(Some(Color::Ansi(AnsiColor::Red))),
+        )
+        .error(
+            Style::new()
+                .bold()
+                .fg_color(Some(Color::Ansi(AnsiColor::Red))),
+        )
+        .valid(
+            Style::new()
+                .bold()
+                .fg_color(Some(Color::Ansi(AnsiColor::Green))),
+        )
+        .placeholder(Style::new().fg_color(Some(Color::Ansi(AnsiColor::White))))
+}

xtask/src/test.rs

@@ -0,0 +1,122 @@
+use std::{
+    env,
+    path::Path,
+    process::{Command, Stdio},
+};
+
+use anyhow::{anyhow, Result};
+use regex::Regex;
+
+use crate::{bail_on_err, Test};
+
+pub fn run(args: &Test) -> Result<()> {
+    let test_flags = if args.address_sanitizer {
+        env::set_var("CFLAGS", "-fsanitize=undefined,address");
+
+        // When the Tree-sitter C library is compiled with the address sanitizer, the address
+        // sanitizer runtime library needs to be linked into the final test executable. When
+        // using Xcode clang, the Rust linker doesn't know where to find that library, so we
+        // need to specify linker flags directly.
+        let output = Command::new("cc").arg("-print-runtime-dir").output()?;
+        bail_on_err(&output, "Failed to get clang runtime dir")?;
+        let runtime_dir = String::from_utf8(output.stdout)?;
+        if runtime_dir.contains("/Xcode.app/") {
+            env::set_var(
+                "RUSTFLAGS",
+                format!(
+                    "-C link-arg=-L{runtime_dir} -C link-arg=-lclang_rt.asan_osx_dynamic -C link-arg=-Wl,-rpath,{runtime_dir}"
+                ),
+            );
+        }
+
+        // Specify a `--target` explicitly. This is required for address sanitizer support.
+        let output = Command::new("rustup")
+            .arg("show")
+            .arg("active-toolchain")
+            .output()?;
+        bail_on_err(&output, "Failed to get active Rust toolchain")?;
+        let toolchain = String::from_utf8(output.stdout)?;
+        let re = Regex::new(r"(stable|beta|nightly)-([_a-z0-9-]+).*")?;
+        let captures = re
+            .captures(&toolchain)
+            .ok_or_else(|| anyhow!("Failed to parse toolchain '{toolchain}'"))?;
+        let current_target = captures.get(2).unwrap().as_str();
+        format!("--target={current_target}")
+    } else {
+        String::new()
+    };
+    if let Some(example) = &args.example {
+        env::set_var("TREE_SITTER_EXAMPLE", example);
+    }
+    if let Some(seed) = args.seed {
+        env::set_var("TREE_SITTER_SEED", seed.to_string());
+    }
+    if let Some(iterations) = args.iterations {
+        env::set_var("TREE_SITTER_ITERATIONS", iterations.to_string());
+    }
+    if args.debug {
+        env::set_var("TREE_SITTER_LOG", "1");
+    }
+    if args.debug_graph {
+        env::set_var("TREE_SITTER_LOG_GRAPHS", "1");
+    }
+
+    if args.g {
+        let cargo_cmd = Command::new("cargo")
+            .arg("test")
+            .arg(test_flags)
+            .arg("--no-run")
+            .arg("--message-format=json")
+            .stdout(Stdio::piped())
+            .spawn()?;
+
+        let jq_cmd = Command::new("jq")
+            .arg("-rs")
+            .arg(r#"map(select(.target.name == "tree_sitter_cli" and .executable))[0].executable"#)
+            .stdin(cargo_cmd.stdout.unwrap())
+            .output()?;
+
+        let test_binary = String::from_utf8(jq_cmd.stdout)?;
+
+        let mut lldb_cmd = Command::new("lldb");
+        lldb_cmd.arg(test_binary.trim()).arg("--").args(&args.args);
+        bail_on_err(
+            &lldb_cmd.spawn()?.wait_with_output()?,
+            &format!("Failed to run {lldb_cmd:?}"),
+        )?;
+    } else {
+        let mut cargo_cmd = Command::new("cargo");
+        cargo_cmd.arg("test").arg(test_flags).args(&args.args);
+        if args.nocapture {
+            cargo_cmd.arg("--").arg("--nocapture");
+        }
+        bail_on_err(
+            &cargo_cmd.spawn()?.wait_with_output()?,
+            &format!("Failed to run {cargo_cmd:?}"),
+        )?;
+    }
+
+    Ok(())
+}
+
+pub fn run_wasm() -> Result<()> {
+    std::env::set_current_dir("lib/binding_web")?;
+
+    let node_modules_dir = Path::new("node_modules");
+    let npm = if cfg!(target_os = "windows") {
+        "npm.cmd"
+    } else {
+        "npm"
+    };
+
+    if !node_modules_dir.join("chai").exists() || !node_modules_dir.join("mocha").exists() {
+        println!("Installing test dependencies...");
+        let output = Command::new(npm).arg("install").output()?;
+        bail_on_err(&output, "Failed to install test dependencies")?;
+    }
+
+    let output = Command::new(npm).arg("test").output()?;
+    bail_on_err(&output, &format!("Failed to run {npm} test"))?;
+
+    Ok(())
+}