Add slowest test issue CI/CD (#1125)

Author: williambdean

.github/workflows/slow-tests-issue.yml

@@ -0,0 +1,31 @@
+# Taken from https://github.com/pymc-labs/pymc-marketing/tree/main/.github/workflows/slow-tests-issue.yml
+# See the scripts in the `scripts/slowest_tests` directory for more information
+---
+name: Slow Tests Issue Body
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '0 */6 * * *'
+
+permissions:
+  issues: write
+
+jobs:
+  update-comment:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Install ZSH
+        run: sudo apt-get update && sudo apt-get install -y zsh
+      - name: Checkout code
+        uses: actions/checkout@v4
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - name: Trigger the script
+        working-directory: scripts/slowest_tests
+        shell: zsh {0}
+        run: source update-slowest-times-issue.sh
+        env:
+          GITHUB_TOKEN: ${{ github.token }}

scripts/slowest_tests/extract-slow-tests.py

@@ -0,0 +1,80 @@
+"""This script parses the GitHub action log for test times.
+
+Taken from https://github.com/pymc-labs/pymc-marketing/tree/main/scripts/slowest_tests/extract-slow-tests.py
+
+"""
+
+import re
+import sys
+from pathlib import Path
+
+
+start_pattern = re.compile(r"==== slow")
+separator_pattern = re.compile(r"====")
+time_pattern = re.compile(r"(\d+\.\d+)s ")
+
+
+def extract_lines(lines: list[str]) -> list[str]:
+    times = []
+
+    in_section = False
+    for line in lines:
+        detect_start = start_pattern.search(line)
+        detect_end = separator_pattern.search(line)
+
+        if detect_start:
+            in_section = True
+
+        if in_section:
+            times.append(line)
+
+        if not detect_start and in_section and detect_end:
+            break
+
+    return times
+
+
+def trim_up_to_match(pattern, string: str) -> str:
+    match = pattern.search(string)
+    if not match:
+        return ""
+
+    return string[match.start() :]
+
+
+def trim(pattern, lines: list[str]) -> list[str]:
+    return [trim_up_to_match(pattern, line) for line in lines]
+
+
+def strip_ansi(text: str) -> str:
+    ansi_escape = re.compile(r"\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])")
+    return ansi_escape.sub("", text)
+
+
+def format_times(times: list[str]) -> list[str]:
+    return (
+        trim(separator_pattern, times[:1])
+        + trim(time_pattern, times[1:-1])
+        + [strip_ansi(line) for line in trim(separator_pattern, times[-1:])]
+    )
+
+
+def read_lines_from_stdin():
+    return sys.stdin.read().splitlines()
+
+
+def read_from_file(file: Path):
+    """For testing purposes."""
+    return file.read_text().splitlines()
+
+
+def main(read_lines):
+    lines = read_lines()
+    times = extract_lines(lines)
+    parsed_times = format_times(times)
+    print("\n".join(parsed_times))
+
+
+if __name__ == "__main__":
+    read_lines = read_lines_from_stdin
+    main(read_lines)

scripts/slowest_tests/update-slowest-times-issue.sh

@@ -0,0 +1,113 @@
+#!/bin/zsh
+
+DRY_RUN=false
+
+owner=pymc-devs
+repo=pytensor
+issue_number=1124
+title="Speed up test times :rocket:"
+workflow=Tests
+latest_id=$(gh run list --workflow $workflow --status success --limit 1 --json databaseId --jq '.[0].databaseId')
+jobs=$(gh api /repos/$owner/$repo/actions/runs/$latest_id/jobs --jq '.jobs | map({name: .name, run_id: .run_id, id: .id, started_at: .started_at, completed_at: .completed_at})')
+
+# Skip 3.10, float32, and Benchmark tests
+function skip_job() {
+    name=$1
+    if [[ $name == *"py3.10"* ]]; then
+        return 0
+    fi
+
+    if [[ $name == *"float32 1"* ]]; then
+        return 0
+    fi
+
+    if [[ $name == *"Benchmark"* ]]; then
+        return 0
+    fi
+
+    return 1
+}
+
+# Remove common prefix from the name
+function remove_prefix() {
+    name=$1
+    echo $name | sed -e 's/^ubuntu-latest test py3.12 : fast-compile 0 : float32 0 : //'
+}
+
+function human_readable_time() {
+    started_at=$1
+    completed_at=$2
+
+    start_seconds=$(date -d "$started_at" +%s)
+    end_seconds=$(date -d "$completed_at" +%s)
+
+    seconds=$(($end_seconds - $start_seconds))
+
+    if [ $seconds -lt 60 ]; then
+        echo "$seconds seconds"
+    else
+        echo "$(date -u -d @$seconds +'%-M minutes %-S seconds')"
+    fi
+}
+
+all_times=""
+echo "$jobs" | jq -c '.[]' | while read -r job; do
+    id=$(echo $job | jq -r '.id')
+    name=$(echo $job | jq -r '.name')
+    run_id=$(echo $job | jq -r '.run_id')
+    started_at=$(echo $job | jq -r '.started_at')
+    completed_at=$(echo $job | jq -r '.completed_at')
+
+    if skip_job $name; then
+        echo "Skipping $name"
+        continue
+    fi
+
+    echo "Processing job: $name (ID: $id, Run ID: $run_id)"
+    times=$(gh run view --job $id --log | python extract-slow-tests.py)
+
+    if [ -z "$times" ]; then
+        # Some of the jobs are non-test jobs, so we skip them
+        echo "No tests found for '$name', skipping"
+        continue
+    fi
+
+    echo $times
+
+    human_readable=$(human_readable_time $started_at $completed_at)
+    name=$(remove_prefix $name)
+
+    top="<details><summary>($human_readable) $name</summary>\n\n\n\`\`\`"
+    bottom="\`\`\`\n\n</details>"
+
+    formatted_times="$top\n$times\n$bottom"
+
+    if [ -n "$all_times" ]; then
+        all_times="$all_times\n$formatted_times"
+    else
+        all_times="$formatted_times"
+    fi
+done
+
+run_date=$(date +"%Y-%m-%d")
+body=$(cat << EOF
+If you are motivated to help speed up some tests, we would appreciate it!
+
+Here are some of the slowest test times:
+
+$all_times
+
+You can find more information on how to contribute [here](https://pytensor.readthedocs.io/en/latest/dev_start_guide.html)
+
+Automatically generated by [GitHub Action](https://github.com/pymc-devs/pytensor/blob/main/.github/workflows/slow-tests-issue.yml)
+Latest run date: $run_date
+EOF
+)
+
+if [ "$DRY_RUN" = true ]; then
+    echo "Dry run, not updating issue"
+    echo $body
+    exit
+fi
+echo $body | gh issue edit $issue_number --body-file - --title "$title"
+echo "Updated issue $issue_number with all times"