Trigger GitHub CI with a comment on PR (#35211)

* fix

* fix

* comment

* final

* final

* final

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

.github/workflows/self-comment-ci.yml

@@ -0,0 +1,253 @@
+name: PR comment GitHub CI
+
+on:
+  issue_comment:
+    types:
+      - created
+    branches-ignore:
+      - main
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.issue.number }}-${{ startsWith(github.event.comment.body, 'run-slow') || startsWith(github.event.comment.body, 'run slow') || startsWith(github.event.comment.body, 'run_slow') }}
+  cancel-in-progress: true
+
+jobs:
+  get-pr-number:
+    runs-on: ubuntu-22.04
+    name: Get PR number
+    # For security: only allow team members to run
+    if: contains(fromJSON('["ydshieh", "ArthurZucker", "zucchini-nlp", "qubvel", "molbap", "gante", "LysandreJik", "Cyrilvallez"]'), github.actor)
+    outputs:
+      PR_NUMBER: ${{ steps.set_pr_number.outputs.PR_NUMBER }}
+    steps:
+      - name: Get PR number
+        shell: bash
+        run: |
+          if [[ "${{ github.event.issue.number }}" != "" && "${{ github.event.issue.pull_request }}" != "" ]]; then
+            echo "PR_NUMBER=${{ github.event.issue.number }}" >> $GITHUB_ENV
+          else
+            echo "PR_NUMBER=" >> $GITHUB_ENV
+          fi
+
+      - name: Check PR number
+        shell: bash
+        run: |
+          echo "${{ env.PR_NUMBER }}"
+
+      - name: Set PR number
+        id: set_pr_number
+        run: echo "PR_NUMBER=${{ env.PR_NUMBER }}" >> "$GITHUB_OUTPUT"
+
+  get-sha:
+    runs-on: ubuntu-22.04
+    needs: get-pr-number
+    if: ${{ needs.get-pr-number.outputs.PR_NUMBER != ''}}
+    outputs:
+      PR_HEAD_SHA: ${{ steps.get_sha.outputs.PR_HEAD_SHA }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: "0"
+          ref: "refs/pull/${{needs.get-pr-number.outputs.PR_NUMBER}}/merge"
+
+      - name: Get SHA
+        id: get_sha
+        env:
+          PR_NUMBER: ${{needs.get-pr-number.outputs.PR_NUMBER}}
+        run: |
+            git fetch origin refs/pull/$PR_NUMBER/head:refs/remotes/pull/$PR_NUMBER/head
+            git checkout refs/remotes/pull/$PR_NUMBER/head
+            echo "PR_HEAD_SHA: $(git log -1 --format=%H)"
+            echo "PR_HEAD_SHA=$(git log -1 --format=%H)" >> "$GITHUB_OUTPUT"
+
+  # use a python script to handle this complex logic
+  # case 1: `run-slow` (auto. infer with limited number of models, but in particular, new model)
+  # case 2: `run-slow model_1, model_2`
+  get-tests:
+    runs-on: ubuntu-22.04
+    needs: get-pr-number
+    if: ${{ needs.get-pr-number.outputs.PR_NUMBER != ''}}
+    permissions: write-all
+    outputs:
+      models: ${{ steps.models_to_run.outputs.models }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: "0"
+          ref: "refs/pull/${{needs.get-pr-number.outputs.PR_NUMBER}}/merge"
+
+      - name: Get models to test
+        env:
+          PR_COMMENT: ${{ github.event.comment.body }}
+        run: |
+          python -m pip install GitPython
+          python utils/pr_slow_ci_models.py --message "$PR_COMMENT" | tee output.txt
+          echo "models=$(tail -n 1 output.txt)" >> $GITHUB_ENV
+
+      - name: Show models to test
+        id: models_to_run
+        run: |
+          echo "${{ env.models }}"
+          echo "models=${{ env.models }}" >> $GITHUB_ENV
+          echo "models=${{ env.models }}" >> $GITHUB_OUTPUT
+
+      - name: Reply to the comment
+        if: ${{ env.models != '[]' }}
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          gh api \
+            --method POST \
+            -H "Accept: application/vnd.github+json" \
+            -H "X-GitHub-Api-Version: 2022-11-28" \
+            repos/${{ github.repository }}/issues/${{ needs.get-pr-number.outputs.PR_NUMBER }}/comments \
+            -f "body=This comment contains run-slow, running the specified jobs: ${{ env.models }} ..."
+
+  create_run:
+    name: Create run
+    if: ${{ needs.get-tests.outputs.models != '[]' }}
+    needs: [get-sha, get-tests]
+    permissions: write-all
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Create Run
+        id: create_run
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          # Create a commit status (pending) for a run of this workflow. The status has to be updated later in `update_run_status`.
+          # See https://docs.github.com/en/rest/commits/statuses?apiVersion=2022-11-28#create-a-commit-status
+          GITHUB_RUN_URL: https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}
+        run: |
+          gh api \
+            --method POST \
+            -H "Accept: application/vnd.github+json" \
+            -H "X-GitHub-Api-Version: 2022-11-28" \
+            repos/${{ github.repository }}/statuses/${{ needs.get-sha.outputs.PR_HEAD_SHA }} \
+            -f "target_url=$GITHUB_RUN_URL" -f "state=pending" -f "description=Slow CI job" -f "context=pytest/custom-tests"
+
+  run_models_gpu:
+      name: Run all tests for the model
+      if: ${{ needs.get-tests.outputs.models != '[]' }}
+      needs: [get-pr-number, get-tests, create_run]
+      strategy:
+        fail-fast: false
+        matrix:
+          folders: ${{ fromJson(needs.get-tests.outputs.models) }}
+          machine_type: [aws-g4dn-2xlarge-cache, aws-g4dn-12xlarge-cache]
+      runs-on:
+         group: '${{ matrix.machine_type }}'
+      container:
+        image: huggingface/transformers-all-latest-gpu
+        options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+      steps:
+      - name: Echo input and matrix info
+        shell: bash
+        run: |
+          echo "${{ matrix.folders }}"
+
+      - name: Echo folder ${{ matrix.folders }}
+        shell: bash
+        # For folders like `models/bert`, set an env. var. (`matrix_folders`) to `models_bert`, which will be used to
+        # set the artifact folder names (because the character `/` is not allowed).
+        run: |
+          echo "${{ matrix.folders }}"
+          matrix_folders=${{ matrix.folders }}
+          matrix_folders=${matrix_folders/'models/'/'models_'}
+          echo "$matrix_folders"
+          echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
+
+      - name: Checkout to PR merge commit
+        working-directory: /transformers
+        run: |
+            git fetch origin refs/pull/${{ needs.get-pr-number.outputs.PR_NUMBER }}/merge:refs/remotes/pull/${{ needs.get-pr-number.outputs.PR_NUMBER }}/merge
+            git checkout refs/remotes/pull/${{ needs.get-pr-number.outputs.PR_NUMBER }}/merge
+            git log -1 --format=%H
+
+      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
+        working-directory: /transformers
+        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
+
+      - name: NVIDIA-SMI
+        run: |
+          nvidia-smi
+
+      - name: Set `machine_type` for report and artifact names
+        working-directory: /transformers
+        shell: bash
+        run: |
+          echo "${{ matrix.machine_type }}"
+          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
+            machine_type=single-gpu
+          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+            machine_type=multi-gpu
+          else
+            machine_type=${{ matrix.machine_type }}
+          fi
+          echo "$machine_type"
+          echo "machine_type=$machine_type" >> $GITHUB_ENV
+
+      - name: Environment
+        working-directory: /transformers
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - name: Run all tests on GPU
+        working-directory: /transformers
+        run: |
+          export CUDA_VISIBLE_DEVICES="$(python3 utils/set_cuda_devices_for_ci.py --test_folder ${{ matrix.folders }})"
+          echo $CUDA_VISIBLE_DEVICES
+          python3 -m pytest -v -rsfE --make-reports=${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports tests/${{ matrix.folders }}
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/failures_short.txt
+
+      - name: Make sure report directory exists
+        shell: bash
+        run: |
+          mkdir -p /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports
+          echo "hello" > /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/hello.txt
+          echo "${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports"
+
+      - name: "Test suite reports artifacts: ${{ env.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports"
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ env.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports
+          path: /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports
+
+  update_run_status:
+    name: Update Check Run Status
+    needs: [get-sha, create_run, run_models_gpu]
+    permissions: write-all
+    if: ${{ always() && needs.create_run.result == 'success' }}
+    runs-on: ubuntu-22.04
+    env:
+      GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      GITHUB_RUN_URL: https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}
+    steps:
+      - name: Get `run_models_gpu` job status
+        run: |
+          echo "${{ needs.run_models_gpu.result }}"
+          if [ "${{ needs.run_models_gpu.result }}" = "cancelled" ]; then
+            echo "STATUS=failure" >> $GITHUB_ENV
+          elif [ "${{ needs.run_models_gpu.result }}" = "skipped" ]; then
+            echo "STATUS=success" >> $GITHUB_ENV
+          else
+            echo "STATUS=${{ needs.run_models_gpu.result }}" >> $GITHUB_ENV
+          fi
+
+      - name: Update PR commit statuses
+        run: |
+          echo "${{ needs.run_models_gpu.result }}"
+          echo "${{ env.STATUS }}"
+          gh api \
+            --method POST \
+            -H "Accept: application/vnd.github+json" \
+            -H "X-GitHub-Api-Version: 2022-11-28" \
+            repos/${{ github.repository }}/statuses/${{ needs.get-sha.outputs.PR_HEAD_SHA }} \
+            -f "target_url=$GITHUB_RUN_URL" -f "state=${{ env.STATUS }}" -f "description=Slow CI job" -f "context=pytest/custom-tests"

.github/workflows/self-pr-slow-ci.yml

@@ -1,151 +0,0 @@
-name: PR slow CI
-
-on:
-  pull_request:
-    paths:
-      - "src/transformers/models/*/modeling_*.py"
-      - "tests/**/test_*.py"
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
-  cancel-in-progress: true
-
-env:
-  HF_HOME: /mnt/cache
-  TRANSFORMERS_IS_CI: yes
-  OMP_NUM_THREADS: 8
-  MKL_NUM_THREADS: 8
-  RUN_SLOW: yes
-  # For gated repositories, we still need to agree to share information on the Hub repo. page in order to get access.
-  # This token is created under the bot `hf-transformers-bot`.
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
-  SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
-  TF_FORCE_GPU_ALLOW_GROWTH: true
-  RUN_PT_TF_CROSS_TESTS: 1
-  CUDA_VISIBLE_DEVICES: 0,1
-
-jobs:
-  find_models_to_run:
-      runs-on: ubuntu-22.04
-      name: Find models to run slow tests
-      # Triggered only if the required label `run-slow` is added
-      if: ${{ contains(github.event.pull_request.labels.*.name, 'run-slow') }}
-      outputs:
-        models: ${{ steps.models_to_run.outputs.models }}
-      steps:
-        - uses: actions/checkout@v4
-          with:
-            fetch-depth: "0"
-            ref: ${{ github.event.pull_request.head.sha }}
-
-        - name: Get commit message
-          run: |
-            echo "commit_message=$(git show -s --format=%s)" >> $GITHUB_ENV
-
-        - name: Get models to run slow tests
-          run: |
-            echo "${{ env.commit_message }}"
-            python -m pip install GitPython
-            python utils/pr_slow_ci_models.py --commit_message "${{ env.commit_message }}" | tee output.txt
-            echo "models=$(tail -n 1 output.txt)" >> $GITHUB_ENV
-
-        - name: Models to run slow tests
-          id: models_to_run
-          run: |
-            echo "${{ env.models }}"
-            echo "models=${{ env.models }}" >> $GITHUB_OUTPUT
-
-  run_models_gpu:
-      name: Run all tests for the model
-      # Triggered only `find_models_to_run` is triggered (label `run-slow` is added) which gives the models to run
-      # (either a new model PR or via a commit message)
-      if: ${{ needs.find_models_to_run.outputs.models != '[]' }}
-      needs: find_models_to_run
-      strategy:
-        fail-fast: false
-        matrix:
-          folders: ${{ fromJson(needs.find_models_to_run.outputs.models) }}
-          machine_type: [aws-g4dn-2xlarge-cache, aws-g4dn-12xlarge-cache]
-      runs-on:
-        group: '${{ matrix.machine_type }}'
-      container:
-        image: huggingface/transformers-all-latest-gpu
-        options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
-      steps:
-      - name: Echo input and matrix info
-        shell: bash
-        run: |
-          echo "${{ matrix.folders }}"
-
-      - name: Echo folder ${{ matrix.folders }}
-        shell: bash
-        # For folders like `models/bert`, set an env. var. (`matrix_folders`) to `models_bert`, which will be used to
-        # set the artifact folder names (because the character `/` is not allowed).
-        run: |
-          echo "${{ matrix.folders }}"
-          matrix_folders=${{ matrix.folders }}
-          matrix_folders=${matrix_folders/'models/'/'models_'}
-          echo "$matrix_folders"
-          echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
-
-      - name: Update clone
-        working-directory: /transformers
-        run: git fetch && git fetch origin pull/${{ github.event.pull_request.number }}/head:pull/${{ github.event.pull_request.number }}/merge && git checkout pull/${{ github.event.pull_request.number }}/merge
-
-      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
-        working-directory: /transformers
-        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e . && python3 -m pip install --upgrade torch torchaudio torchvision
-
-      - name: NVIDIA-SMI
-        run: |
-          nvidia-smi
-
-      - name: Set `machine_type` for report and artifact names
-        working-directory: /transformers
-        shell: bash
-        run: |
-          echo "${{ matrix.machine_type }}"
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
-            machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
-            machine_type=multi-gpu
-          else
-            machine_type=${{ matrix.machine_type }}
-          fi
-          echo "$machine_type"
-          echo "machine_type=$machine_type" >> $GITHUB_ENV    
-
-      - name: Environment
-        working-directory: /transformers
-        run: |
-          python3 utils/print_env.py
-
-      - name: Show installed libraries and their versions
-        working-directory: /transformers
-        run: pip freeze
-
-      - name: Run all tests on GPU
-        working-directory: /transformers
-        run: |
-          export CUDA_VISIBLE_DEVICES="$(python3 utils/set_cuda_devices_for_ci.py --test_folder ${{ matrix.folders }})"
-          echo $CUDA_VISIBLE_DEVICES
-          python3 -m pytest -v -rsfE --make-reports=${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports tests/${{ matrix.folders }}
-
-      - name: Failure short reports
-        if: ${{ failure() }}
-        continue-on-error: true
-        run: cat /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/failures_short.txt
-
-      - name: Make sure report directory exists
-        shell: bash
-        run: |
-          mkdir -p /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports
-          echo "hello" > /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/hello.txt
-          echo "${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports"
-
-      - name: "Test suite reports artifacts: ${{ env.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports"
-        if: ${{ always() }}
-        uses: actions/upload-artifact@v4
-        with:
-          name: ${{ env.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports
-          path: /transformers/reports/${{ env.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports

utils/pr_slow_ci_models.py

@@ -15,19 +15,20 @@
 """
 This script is used to get the models for which to run slow CI.
 
-A new model added in a pull request will be included, as well as models specified in a commit message with a prefix
+A new model added in a pull request will be included, as well as models specified in a GitHub pull request's comment
-`[run-slow]`, `[run_slow]` or `[run slow]`. For example, the commit message `[run_slow]bert, gpt2` will give `bert` and
+with a prefix `run-slow`, `run_slow` or `run slow`. For example, the commit message `run_slow: bert, gpt2` will give
-`gpt2`.
+`bert` and `gpt2`.
 
 Usage:
 
 ```bash
-python utils/pr_slow_ci_models.py.py
+python utils/pr_slow_ci_models.py
 ```
 """
 
 import argparse
 import re
+import string
 from pathlib import Path
 from typing import List
 
@@ -89,7 +90,7 @@ def get_new_python_files() -> List[str]:
 
 def get_new_model():
     new_files = get_new_python_files()
-    reg = re.compile(r"src/transformers/(models/.*)/modeling_.*\.py")
+    reg = re.compile(r"src/transformers/models/(.*)/modeling_.*\.py")
 
     new_model = ""
     for x in new_files:
@@ -101,45 +102,53 @@ def get_new_model():
     return new_model
 
 
-def parse_commit_message(commit_message: str) -> str:
+def parse_message(message: str) -> str:
     """
-    Parses the commit message to find the models specified in it to run slow CI.
+    Parses a GitHub pull request's comment to find the models specified in it to run slow CI.
 
     Args:
-        commit_message (`str`): The commit message of the current commit.
+        message (`str`): The body of a GitHub pull request's comment.
 
     Returns:
-        `str`: The substring in `commit_message` after `[run-slow]`, [run_slow]` or [run slow]`. If no such prefix is
+        `str`: The substring in `message` after `run-slow`, run_slow` or run slow`. If no such prefix is found, the
-         found, the empty string is returned.
+        empty string is returned.
     """
-    if commit_message is None:
+    if message is None:
         return ""
 
-    command_search = re.search(r"\[([^\]]*)\](.*)", commit_message)
+    message = message.strip().lower()
-    if command_search is None:
+
+    # run-slow: model_1, model_2
+    if not message.startswith(("run-slow", "run_slow", "run slow")):
         return ""
+    message = message[len("run slow") :]
+    # remove leading `:`
+    while message.strip().startswith(":"):
+        message = message.strip()[1:]
 
-    command = command_search.groups()[0]
+    return message
-    command = command.lower().replace("-", " ").replace("_", " ")
-    run_slow = command == "run slow"
-    if run_slow:
-        models = command_search.groups()[1].strip()
-        return models
-    else:
-        return ""
 
 
-def get_models(commit_message: str):
+def get_models(message: str):
-    models = parse_commit_message(commit_message)
+    models = parse_message(message)
-    return [f"models/{x}" for x in models.replace(",", " ").split()]
+    return models.replace(",", " ").split()
+
+
+def check_model_names(model_name: str):
+    allowed = string.ascii_letters + string.digits + "_"
+    return not (model_name.startswith("_") or model_name.endswith("_")) and all(c in allowed for c in model_name)
 
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--commit_message", type=str, default="", help="The commit message.")
+    parser.add_argument("--message", type=str, default="", help="The content of a comment.")
     args = parser.parse_args()
 
     new_model = get_new_model()
-    specified_models = get_models(args.commit_message)
+    specified_models = get_models(args.message)
     models = ([] if new_model == "" else [new_model]) + specified_models
+    # a guard for strange model names
+    models = [model for model in models if check_model_names(model)]
+    # Add "models/"
+    models = [f"models/{model}" for model in models]
     print(sorted(set(models)))