Quality overhaul: buildable, tested, honest (v0.2.0)

.clang-format

@@ -0,0 +1,13 @@
+# Matches the prevailing style of the existing headers.
+BasedOnStyle: LLVM
+IndentWidth: 4
+ColumnLimit: 100
+AccessModifierOffset: -4
+AllowShortFunctionsOnASingleLine: Inline
+AlwaysBreakTemplateDeclarations: Yes
+BreakBeforeBraces: Attach
+PointerAlignment: Left
+SpaceAfterTemplateKeyword: false
+IndentPPDirectives: None
+NamespaceIndentation: None
+FixNamespaceComments: true

.clang-tidy

@@ -0,0 +1,63 @@
+# clang-tidy configuration.
+#
+# The bugprone-*, performance-*, and clang-analyzer-* families (the checks
+# that find defects) run as errors in CI. The modernize-* and readability-*
+# families are enabled too, but the specific style-tier checks that conflict
+# with the codebase's deliberate and consistent idioms are disabled below,
+# each with a reason. clang-format enforces layout; these checks would only
+# churn it. Anything genuinely bug-prone stays on and is fixed, not silenced.
+Checks: >
+  bugprone-*,
+  performance-*,
+  modernize-*,
+  readability-*,
+  clang-analyzer-*,
+  -modernize-use-trailing-return-type,
+  -modernize-use-default-member-init,
+  -modernize-use-nodiscard,
+  -modernize-use-emplace,
+  -modernize-use-auto,
+  -modernize-avoid-c-arrays,
+  -modernize-loop-convert,
+  -modernize-return-braced-init-list,
+  -modernize-use-designated-initializers,
+  -modernize-use-starts-ends-with,
+  -readability-identifier-length,
+  -readability-magic-numbers,
+  -readability-function-cognitive-complexity,
+  -readability-braces-around-statements,
+  -readability-implicit-bool-conversion,
+  -readability-else-after-return,
+  -readability-qualified-auto,
+  -readability-named-parameter,
+  -readability-redundant-member-init,
+  -readability-redundant-string-init,
+  -readability-redundant-casting,
+  -readability-convert-member-functions-to-static,
+  -readability-use-anyofallof,
+  -readability-container-contains,
+  -readability-container-size-empty,
+  -readability-simplify-boolean-expr,
+  -readability-avoid-unconditional-preprocessor-if,
+  -readability-avoid-nested-conditional-operator,
+  -readability-non-const-parameter,
+  -performance-enum-size,
+  -performance-no-int-to-ptr,
+  -performance-unnecessary-value-param,
+  -bugprone-easily-swappable-parameters,
+  -bugprone-branch-clone
+# Rationale for the notable disables:
+#   modernize-use-default-member-init / avoid-c-arrays / return-braced-init-list
+#     - the codebase uses constructor init-lists and constexpr C-array lookup
+#       tables deliberately and uniformly.
+#   readability-qualified-auto / named-parameter / *-member-init
+#     - pervasive stylistic preferences, not correctness.
+#   performance-no-int-to-ptr - the arena's alignment computation is an
+#     intentional, correct uintptr_t round-trip.
+#   bugprone-branch-clone - the parser/generator have many intentionally
+#     parallel branches (per-token, per-dialect) that read clearer apart.
+#   bugprone-easily-swappable-parameters - advisory, high-noise on
+#     (row, col) / (start, end) style signatures.
+WarningsAsErrors: ''
+HeaderFilterRegex: '(core|sql|mime)/include/libglot/.*'
+FormatStyle: file

.git-blame-ignore-revs

@@ -0,0 +1,2 @@
+# Commits to ignore in git blame (mechanical reformatting)
+53f969a60fdc20fec9afe71bea1efb6d6cc6286d

.github/workflows/ci.yml

@@ -0,0 +1,206 @@
+name: CI
+
+on:
+  push:
+    branches: [master]
+  pull_request:
+
+jobs:
+  build-and-test:
+    name: ${{ matrix.compiler.name }} ${{ matrix.build_type }}${{ matrix.sanitize && ' + sanitizers' || '' }}
+    runs-on: ubuntu-24.04
+    strategy:
+      fail-fast: false
+      matrix:
+        compiler:
+          - { name: GCC, cc: gcc-14, cxx: g++-14 }
+          - { name: Clang, cc: clang-18, cxx: clang++-18 }
+        build_type: [Debug, Release]
+        sanitize: [false]
+        include:
+          - compiler: { name: GCC, cc: gcc-14, cxx: g++-14 }
+            build_type: Debug
+            sanitize: true
+          - compiler: { name: Clang, cc: clang-18, cxx: clang++-18 }
+            build_type: Debug
+            sanitize: true
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install toolchain
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y ninja-build ${{ matrix.compiler.cc == 'gcc-14' && 'gcc-14 g++-14' || 'clang-18' }}
+
+      - name: Configure
+        run: |
+          cmake -S . -B build -G Ninja \
+            -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
+            -DCMAKE_C_COMPILER=${{ matrix.compiler.cc }} \
+            -DCMAKE_CXX_COMPILER=${{ matrix.compiler.cxx }} \
+            -DLIBGLOT_WERROR=ON \
+            -DLIBGLOT_ENABLE_ASAN=${{ matrix.sanitize && 'ON' || 'OFF' }}
+
+      - name: Build
+        run: cmake --build build -j"$(nproc)"
+
+      - name: Test
+        run: ctest --test-dir build --output-on-failure -j"$(nproc)"
+
+  fuzz-smoke:
+    name: fuzzers (60s smoke per target)
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y ninja-build clang-18
+      - name: Configure
+        run: |
+          cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_C_COMPILER=clang-18 -DCMAKE_CXX_COMPILER=clang++-18 \
+            -DBUILD_TESTING=OFF -DLIBGLOT_BUILD_FUZZERS=ON
+      - name: Build fuzzers
+        run: cmake --build build -j"$(nproc)"
+      - name: Run fuzzers
+        run: |
+          mkdir -p corpus/sql corpus/mime
+          printf 'SELECT a, b FROM t WHERE x BETWEEN 1 AND 10 ORDER BY a LIMIT 5' > corpus/sql/seed1
+          printf 'WITH c AS (SELECT 1) INSERT INTO t (a) SELECT * FROM c' > corpus/sql/seed2
+          printf 'Content-Type: multipart/mixed; boundary="b"\r\n\r\n--b\r\n\r\nhi\r\n--b--\r\n' > corpus/mime/seed1
+          ./build/fuzz/fuzz_sql_parser    -max_total_time=60 -timeout=10 corpus/sql
+          ./build/fuzz/fuzz_sql_roundtrip -max_total_time=60 -timeout=10 corpus/sql
+          ./build/fuzz/fuzz_mime_parser   -max_total_time=60 -timeout=10 corpus/mime
+
+  coverage:
+    name: coverage report
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y ninja-build g++-14 gcovr
+      - name: Configure
+        run: |
+          cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Debug \
+            -DCMAKE_CXX_COMPILER=g++-14 \
+            -DCMAKE_CXX_FLAGS="--coverage" -DCMAKE_EXE_LINKER_FLAGS="--coverage"
+      - name: Build and test
+        run: |
+          cmake --build build -j"$(nproc)"
+          ctest --test-dir build -j"$(nproc)" --output-on-failure
+      - name: Report
+        run: |
+          gcovr --root . --filter 'core/include/' --filter 'sql/include/' --filter 'mime/include/' \
+            --exclude '.*/_deps/.*' --gcov-ignore-errors=no_working_dir_found \
+            --gcov-executable gcov-14 --print-summary --xml coverage.xml
+      - name: Upload coverage artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: coverage-xml
+          path: coverage.xml
+
+  mime-corpus:
+    name: MIME corpus gate
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y ninja-build g++-14
+      - name: Build corpus runner
+        run: |
+          cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_CXX_COMPILER=g++-14 -DBUILD_TESTING=OFF \
+            -DLIBGLOT_BUILD_EXAMPLES=OFF -DLIBGLOT_BUILD_SQL=OFF
+          cmake --build build -j"$(nproc)" --target mime_corpus
+      - name: Committed sample corpus (must be 100%)
+        run: ./build/tools/mime_corpus --min-success 1.0 tests/corpus/mime
+      - name: SpamAssassin public corpus (best effort, reported not gated)
+        continue-on-error: true
+        run: |
+          mkdir -p sa && cd sa
+          for f in 20021010_easy_ham 20030228_hard_ham 20050311_spam_2; do
+            curl -fsSL "https://spamassassin.apache.org/old/publiccorpus/${f}.tar.bz2" -o "$f.tar.bz2" \
+              && tar xjf "$f.tar.bz2" || echo "skip $f (download failed)"
+          done
+          cd ..
+          ./build/tools/mime_corpus --min-success 0.0 --quiet sa || true
+
+  mime-differential:
+    name: MIME differential vs Python email
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y ninja-build g++-14
+      - name: Build mime_dump
+        run: |
+          cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_CXX_COMPILER=g++-14 -DBUILD_TESTING=OFF \
+            -DLIBGLOT_BUILD_EXAMPLES=OFF -DLIBGLOT_BUILD_SQL=OFF
+          cmake --build build -j"$(nproc)" --target mime_dump
+      - name: Committed corpus must agree 100% with Python's email module
+        run: |
+          python3 scripts/mime_diff.py --tool build/tools/mime_dump \
+            --corpus tests/corpus/mime --fail-under 1.0 --verbose
+
+  clang-tidy:
+    name: clang-tidy
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y clang-tidy-18
+      - name: Run clang-tidy over first-party TUs
+        run: |
+          # Each example/tool/fuzz TU transitively includes the whole public
+          # header surface; HeaderFilterRegex in .clang-tidy scopes the checks
+          # to first-party headers, and any warning fails the job.
+          INC="-std=c++20 -Icore/include -Isql/include -Imime/include"
+          status=0
+          for f in $(git ls-files 'examples/*.cpp' 'tools/*.cpp' 'fuzz/*.cpp'); do
+            echo "::group::clang-tidy $f"
+            clang-tidy-18 --warnings-as-errors='*' "$f" -- $INC || status=1
+            echo "::endgroup::"
+          done
+          exit $status
+
+  install-package:
+    name: install + find_package smoke test
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install toolchain
+        run: sudo apt-get update && sudo apt-get install -y ninja-build g++-14
+      - name: Build and install
+        run: |
+          cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_CXX_COMPILER=g++-14 -DBUILD_TESTING=OFF \
+            -DCMAKE_INSTALL_PREFIX="$PWD/install"
+          cmake --build build -j"$(nproc)"
+          cmake --install build
+      - name: Consume via find_package
+        run: |
+          mkdir consumer && cd consumer
+          cat > CMakeLists.txt <<'EOF'
+          cmake_minimum_required(VERSION 3.28)
+          project(consumer CXX)
+          find_package(libglot REQUIRED)
+          add_executable(smoke smoke.cpp)
+          target_link_libraries(smoke PRIVATE libglot::sql libglot::mime)
+          EOF
+          cat > smoke.cpp <<'EOF'
+          #include <libglot/sql/parser.h>
+          #include <libglot/sql/generator.h>
+          #include <libglot/mime/parser.h>
+          #include <cassert>
+          int main() {
+              libglot::Arena arena;
+              libglot::sql::SQLParser parser(arena, "SELECT 1");
+              assert(parser.parse_top_level() != nullptr);
+              return 0;
+          }
+          EOF
+          cmake -S . -B build -G Ninja -DCMAKE_CXX_COMPILER=g++-14 \
+            -DCMAKE_PREFIX_PATH="$GITHUB_WORKSPACE/install"
+          cmake --build build
+          ./build/smoke

.gitignore

@@ -1,53 +1,36 @@
-# Build directories
-build/
-build-release/
-Testing/
+# Build directories (out-of-source builds only; never build in the source tree)
+build*/
 cmake-build-*/
 out/
 
-# CMake generated files
+# In-source build debris, in case one happens anyway
 CMakeCache.txt
 CMakeFiles/
 cmake_install.cmake
+CTestTestfile.cmake
+Testing/
+_deps/
 Makefile
-*.cmake
-!CMakeLists.txt
-!CMakePresets.json
-!cmake/
+!/Makefile.dist
 
-# Dependencies
-_deps/
+# Installed package artifacts
 vcpkg_installed/
 
-# Backup and temporary files
-*.backup
-*.bak
-*.old
-*.tmp
-*.log
-*~
-.DS_Store
-
 # Python bytecode
 __pycache__/
-*.pyc
-*.pyo
+*.py[co]
 
-# IDE and editor files
+# Editor and OS files
 .vscode/
 .idea/
+.vs/
 *.swp
 *.swo
-.vs/
-
-# Test artifacts
-test_*_bin
-*_debug.cpp
-debug_*.cpp
-
-# Scripts (keeping source scripts, ignoring generated ones)
-*.sh
+*~
+.DS_Store
 
-# Text verification files
-*.txt
-!CMakeLists.txt
+# Coverage and profiling output
+*.gcda
+*.gcno
+*.profraw
+coverage/