perf: Parse HogQL with C++ for a huge speedup

.dockerignore

@@ -33,4 +33,4 @@
 !plugin-server/.prettierrc
 !share/GeoLite2-City.mmdb
 !hogvm/python
-!unit.json
+!unit.json

.github/actions/run-backend-tests/action.yml

@@ -49,11 +49,17 @@ runs:
               python-version: ${{ inputs.python-version }}
               token: ${{ inputs.token }}
 
+        - name: Determine if hogql-parser has changed compared to master
+          shell: bash
+          id: hogql-parser-diff
+          run: |
+              changed=$(git diff --quiet HEAD master -- hogql_parser/ && echo "false" || echo "true")
+              echo "::set-output name=changed::$changed"
+
         - name: Install SAML (python3-saml) dependencies
           shell: bash
           run: |
-              sudo apt-get update
-              sudo apt-get install libxml2-dev libxmlsec1-dev libxmlsec1-openssl
+              sudo apt-get update && sudo apt-get install libxml2-dev libxmlsec1-dev libxmlsec1-openssl
 
         - uses: syphar/restore-virtualenv@v1
           id: cache-backend-tests
@@ -63,12 +69,28 @@ runs:
         - uses: syphar/restore-pip-download-cache@v1
           if: steps.cache-backend-tests.outputs.cache-hit != 'true'
 
-        - name: Install python dependencies
+        - name: Install Python dependencies
           if: steps.cache-backend-tests.outputs.cache-hit != 'true'
           shell: bash
           run: |
-              python -m pip install -r requirements-dev.txt
-              python -m pip install -r requirements.txt
+              pip install -r requirements.txt -r requirements-dev.txt
+
+        - name: Install the working version of hogql-parser
+          if: steps.hogql-parser-diff.outputs.changed == 'true'
+          shell: bash
+          # This is not cached currently, as it's important to build the current HEAD version of hogql-parser if it has
+          # changed (as requirements.txt only has the published version)
+          run: |
+              sudo apt-get install libboost-all-dev unzip cmake curl uuid pkg-config
+              curl https://www.antlr.org/download/antlr4-cpp-runtime-4.13.0-source.zip --output antlr4-source.zip
+              unzip antlr4-source.zip -d antlr4-source && cd antlr4-source
+              cmake .
+              DESTDIR=out make install
+              sudo cp -r out/usr/local/include/antlr4-runtime /usr/include/
+              sudo cp out/usr/local/lib/libantlr4-runtime.so* /usr/lib/
+              sudo ldconfig
+              cd ..
+              pip install ./hogql_parser
 
         - name: Set up needed files
           shell: bash

.github/workflows/build-hogql-parser.yml

@@ -0,0 +1,130 @@
+name: Release hogql-parser
+
+on:
+    push:
+        branches:
+            - master
+        paths:
+            - hogql_parser/**
+            - .github/workflows/build-hogql-parser.yml
+    pull_request:
+        paths:
+            - hogql_parser/**
+            - .github/workflows/build-hogql-parser.yml
+
+concurrency:
+    group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
+    cancel-in-progress: true
+
+jobs:
+    check-version:
+        name: Check version legitimacy
+        runs-on: ubuntu-22.04
+        outputs:
+            parser_any_changed: ${{ steps.changed-files-yaml.outputs.parser_any_changed }}
+        steps:
+            - uses: actions/checkout@v4
+              with:
+                  fetch-depth: 0
+
+            - name: Check if hogql_parser/ has changed
+              id: changed-files-yaml
+              uses: tj-actions/changed-files@v39
+              with:
+                  since_last_remote_commit: true
+                  files_yaml: |
+                      parser:
+                      - hogql_parser/**
+
+            - name: Notify about release needed
+              if: steps.changed-files-yaml.outputs.parser_any_changed == 'true'
+              shell: bash
+              run: |
+                  published=$(curl -fSsl https://pypi.org/pypi/hogql-parser/json | jq -r '.info.version')
+                  local=$(python hogql_parser/setup.py --version)
+                  # TODO: Only comment if no comment alraedy exists for $local
+                  if [[ "$published" == "$local" ]]; then
+                    MESSAGE_BODY="It looks like the code of `hogql-parser` has changed since last push, but its version stayed the same at $local. 👀\nMake sure to resolve this in `hogql_parser/setup.py` before merging!"
+                    curl -s -u posthog-bot:${{ secrets.POSTHOG_BOT_GITHUB_TOKEN || secrets.GITHUB_TOKEN }} -X POST -d "{ \"body\": \"$MESSAGE_BODY\" }" "https://api.github.com/repos/${{ github.repository }}/issues/${{ github.event.pull_request.number }}/comments"
+                  fi
+
+    build-wheels:
+        name: Build wheels on ${{ matrix.os }}
+        needs: check-version
+        runs-on: ${{ matrix.os }}
+        timeout-minutes: 30
+        if: ${{ needs.check-version.outputs.parser_any_changed == 'true' }}
+        strategy:
+            matrix:
+                # As of October 2023, GitHub doesn't have ARM Actions runners… and ARM emulation is insanely slow
+                # (20x longer) on the Linux runners (while being reasonable on the macOS runners). Hence, we use
+                # BuildJet as a provider of ARM runners - this solution saves a lot of time and consequently some money.
+                os: [ubuntu-22.04, buildjet-2vcpu-ubuntu-2204-arm, macos-12]
+
+        steps:
+            - uses: actions/checkout@v4
+
+            - if: ${{ !endsWith(matrix.os, '-arm') }}
+              uses: actions/setup-python@v4
+              with:
+                  python-version: '3.11'
+
+            - if: ${{ endsWith(matrix.os, '-arm') }}
+              uses: deadsnakes/[email protected] # Unfortunately actions/setup-python@v4 just doesn't work on ARM! This does
+              with:
+                  python-version: '3.11'
+
+            - name: Build sdist
+              if: matrix.os == 'ubuntu-22.04' # Only build the sdist once
+              run: cd hogql_parser && python setup.py sdist
+
+            - name: Install cibuildwheel
+              run: python -m pip install cibuildwheel==2.16.*
+
+            - name: Build wheels
+              run: cd hogql_parser && python -m cibuildwheel --output-dir dist
+              env:
+                  MACOSX_DEPLOYMENT_TARGET: '12' # A modern target allows us to use C++20
+
+            - uses: actions/upload-artifact@v3
+              with:
+                  path: |
+                      hogql_parser/dist/*.whl
+                      hogql_parser/dist/*.tar.gz
+                  if-no-files-found: error
+
+    publish:
+        name: Publish on PyPI
+        needs: build-wheels
+        environment: pypi-hogql-parser
+        permissions:
+            id-token: write
+        runs-on: ubuntu-22.04
+        steps:
+            - name: Fetch wheels
+              uses: actions/download-artifact@v3
+              with:
+                  name: artifact
+                  path: dist/
+
+            - name: Publish package to PyPI
+              uses: pypa/gh-action-pypi-publish@release/v1
+
+            - uses: actions/checkout@v4
+              with:
+                  token: ${{ secrets.POSTHOG_BOT_GITHUB_TOKEN }}
+                  ref: ${{ github.event.pull_request.head.ref }}
+
+            - name: Update hogql-parser in requirements
+              shell: bash
+              run: |
+                  local=$(python hogql_parser/setup.py --version)
+                  sed -i "s/hogql-parser==.*/hogql-parser==${local}/g" requirements.in
+                  sed -i "s/hogql-parser==.*/hogql-parser==${local}/g" requirements.txt
+
+            - uses: EndBug/add-and-commit@v9
+              with:
+                  add: '["requirements.in", "requirements.txt"]'
+                  message: 'Use new hogql-parser version'
+                  default_author: github_actions
+                  github_token: ${{ secrets.POSTHOG_BOT_GITHUB_TOKEN }}

.github/workflows/ci-backend.yml

@@ -98,7 +98,6 @@ jobs:
             - uses: actions/checkout@v3
               with:
                   fetch-depth: 1
-                  path: 'current/'
 
             - name: Set up Python
               uses: actions/setup-python@v4
@@ -119,38 +118,30 @@ jobs:
                   sudo apt-get update
                   sudo apt-get install libxml2-dev libxmlsec1 libxmlsec1-dev libxmlsec1-openssl
 
-            - name: Install python dependencies
+            - name: Install Python dependencies
               if: steps.cache-backend-tests.outputs.cache-hit != 'true'
               run: |
-                  cd current
                   python -m pip install -r requirements.txt -r requirements-dev.txt
 
             - name: Check for syntax errors, import sort, and code style violations
               run: |
-                  cd current
                   ruff .
 
             - name: Check formatting
               run: |
-                  cd current
                   black --exclude posthog/hogql/grammar --check .
 
             - name: Check static typing
               run: |
-                  cd current
                   mypy -p posthog --exclude bin/migrate_kafka_data.py --exclude posthog/hogql/grammar/HogQLParser.py --exclude gunicorn.config.py --enable-recursive-aliases
 
             - name: Check if "schema.py" is up to date
               run: |
-                  cd current
                   npm run schema:build:python && git diff --exit-code
 
-            - name: Check if antlr definitions are up to date
+            - name: Check if ANTLR definitions are up to date
               run: |
-                  # Installing a version of ant compatible with what we use in development from homebrew (4.13)
-                  # "apt-get install antlr" would install 4.7 which is incompatible with our grammar.
-                  export ANTLR_VERSION=4.13.0
-                  # java version doesn't matter
+                  cd ..
                   sudo apt-get install default-jre
                   mkdir antlr
                   cd antlr
@@ -162,9 +153,13 @@ jobs:
                   export CLASSPATH=".:$PWD/antlr.jar:$CLASSPATH"
                   export PATH="$PWD:$PATH"
 
-                  cd ../current
+                  cd ../posthog
                   antlr | grep "Version"
                   npm run grammar:build && git diff --exit-code
+              env:
+                  # Installing a version of ANTLR compatible with what's Homebrew as of October 2023 (version 4.13),
+                  # as apt-get is quite out of date. The same version as here is in hogql_parser/pyproject.toml
+                  ANTLR_VERSION: '4.13.0'
 
     check-migrations:
         needs: changes

hogql_parser/.clang-format

@@ -0,0 +1,3 @@
+BasedOnStyle: Chromium
+ColumnLimit: 120
+AlignAfterOpenBracket: BlockIndent

hogql_parser/.gitignore

@@ -0,0 +1,5 @@
+# Build
+build/
+*.egg-info
+*.so
+dist/