ci: run only affected integration targets

.github/workflows/ci.yml

@@ -31,6 +31,14 @@ jobs:
       - name: Install workspace dependencies
         run: dart pub get
 
+      # Integration target selection treats an unchanged lockfile as unchanged resolution.
+      - name: Check committed lockfile
+        run: |
+          if ! git diff --exit-code pubspec.lock; then
+            echo "::error file=pubspec.lock::dart pub get changed pubspec.lock. Run dart pub get and commit pubspec.lock."
+            exit 1
+          fi
+
       - name: Set PR diff base
         if: github.event_name == 'pull_request'
         run: |

.github/workflows/integration-tests.yml

@@ -4,13 +4,17 @@ on:
   pull_request:
   push:
     branches: [main]
+  schedule:
+    # Weekly full regression run; pushes to main only run affected targets.
+    - cron: '23 19 * * 0'
   workflow_dispatch:
 
 permissions:
   contents: read
 
 concurrency:
-  group: integration-${{ github.event.pull_request.number || github.ref }}
+  # Only PR runs share a group: a replaced pending run on main would leave its pushed changes untested.
+  group: integration-${{ github.event.pull_request.number || github.run_id }}
   cancel-in-progress: ${{ github.event_name == 'pull_request' }}
 
 jobs:
@@ -30,13 +34,16 @@ jobs:
         id: targets
         env:
           EVENT: ${{ github.event_name }}
-          BASE: ${{ github.event.pull_request.base.sha }}
-          HEAD: ${{ github.event.pull_request.head.sha }}
+          PR_BASE: ${{ github.event.pull_request.base.sha }}
+          PR_HEAD: ${{ github.event.pull_request.head.sha }}
+          PUSH_BEFORE: ${{ github.event.before }}
         run: |
-          args=()
-          if [[ "$EVENT" == "pull_request" ]]; then
-            args+=(--base "$BASE" --head "$HEAD")
-          fi
+          case "$EVENT" in
+            pull_request) args=(--base "$PR_BASE" --head "$PR_HEAD") ;;
+            push) args=(--base "$PUSH_BEFORE" --head "$GITHUB_SHA") ;;
+            # schedule and workflow_dispatch run every target.
+            *) args=() ;;
+          esac
           python3 -B tool/ci/select_integration_targets.py "${args[@]}" --output "$GITHUB_OUTPUT"
 
   integration:

tool/ci/select_integration_targets.py

@@ -5,6 +5,7 @@
 from pathlib import Path
 import re
 import subprocess
+import sys
 
 
 ROOT = Path(__file__).resolve().parents[2]
@@ -61,6 +62,16 @@ def load_targets(root):
     return targets
 
 
+def diffable_base(root, base):
+    """Return base when it names a local commit, or None when a push diff cannot be computed.
+
+    A new branch reports an all-zero base, and a force push can report a base
+    that is no longer in the history.
+    """
+    exists = subprocess.run(['git', 'cat-file', '-e', f'{base}^{{commit}}'], cwd=root, capture_output=True)
+    return base if exists.returncode == 0 else None
+
+
 def changed_files(root, base, head):
     # Include both sides of moves so moving a file out of a package still tests it.
     output = subprocess.check_output(
@@ -70,29 +81,94 @@ def changed_files(root, base, head):
     return output.decode().split('\0')[:-1]
 
 
-def select_targets(targets, paths):
+def read_pubspec(text):
+    """Read the package name, workspace members, and direct dependency names from pubspec text.
+
+    Only the top-level keys this selector needs are parsed, line by line, because
+    the standard library has no YAML parser.
+    """
+    pubspec = {'name': None, 'workspace': [], 'dependencies': set()}
+    section = child_indent = None
+    for line in text.splitlines():
+        content = line.split('#', 1)[0].rstrip()
+        if not content.strip():
+            continue
+        indent = len(content) - len(content.lstrip())
+        if indent == 0:
+            section, _, value = content.partition(':')
+            child_indent = None
+            if section == 'name':
+                pubspec['name'] = value.strip().strip('\'"')
+            continue
+        if child_indent is None:
+            child_indent = indent
+        if section == 'workspace' and content.lstrip().startswith('- '):
+            pubspec['workspace'].append(content.lstrip()[2:].strip().strip('\'"').rstrip('/'))
+        elif section in {'dependencies', 'dev_dependencies'} and indent == child_indent:
+            match = re.fullmatch(r'\s*([A-Za-z_][A-Za-z0-9_]*)\s*:.*', content)
+            if match:
+                pubspec['dependencies'].add(match.group(1))
+    return pubspec
+
+
+def load_workspace(root):
+    """Map each workspace member directory to its parsed pubspec."""
+    members = read_pubspec((root / 'pubspec.yaml').read_text())['workspace']
+    return {member: read_pubspec((root / member / 'pubspec.yaml').read_text()) for member in members}
+
+
+def base_manifest_targets(root, base, head):
+    """Return the registered targets at the PR merge base, or none when the manifest did not exist."""
+    merge_base = subprocess.check_output(['git', 'merge-base', base, head], cwd=root).decode().strip()
+    revision = f'{merge_base}:.github/integration-tests.json'
+    if subprocess.run(['git', 'cat-file', '-e', revision], cwd=root, capture_output=True).returncode:
+        return []
+    return json.loads(subprocess.check_output(['git', 'show', revision], cwd=root))['targets']
+
+
+def select_targets(targets, paths, base_targets=(), workspace=None):
     if paths is None:
         return targets
-    shared = any(
-        path in {'pubspec.yaml', 'pubspec.lock', '.github/integration-tests.json'}
-        or path.startswith(('.github/workflows/', 'tool/ci/'))
-        or (path.startswith('packages/') and path.endswith('/pubspec.yaml'))
+    # Lockfile and CI changes can affect every job.
+    if any(path == 'pubspec.lock' or path.startswith(('.github/workflows/', 'tool/ci/')) for path in paths):
+        return targets
+    workspace = workspace or {}
+    packages = {path.split('/')[1] for path in paths if path.startswith('packages/') and path.count('/') > 1}
+    # A member pubspec change also affects workspace members that depend on it, directly or transitively.
+    names = {
+        workspace[path.removesuffix('/pubspec.yaml')]['name']
         for path in paths
-    )
-    return [target for target in targets if shared or any(
-        path.startswith(f'packages/{target["package"]}/') for path in paths
+        if path.endswith('/pubspec.yaml') and path.removesuffix('/pubspec.yaml') in workspace
+    }
+    while dependents := {
+        pubspec['name'] for pubspec in workspace.values() if pubspec['dependencies'] & names
+    } - names:
+        names |= dependents
+    packages |= {
+        member.split('/')[1]
+        for member, pubspec in workspace.items()
+        if member.startswith('packages/') and pubspec['name'] in names
+    }
+    # A manifest change affects only added or edited targets; removed targets need no job.
+    manifest_changed = '.github/integration-tests.json' in paths
+    return [target for target in targets if target['package'] in packages or (
+        manifest_changed and target not in base_targets
     )]
 
 
 def main():
     parser = argparse.ArgumentParser(description=__doc__)
-    parser.add_argument('--base', help='PR base SHA; omit to select all targets')
-    parser.add_argument('--head', default='HEAD', help='PR head SHA')
+    parser.add_argument('--base', help='PR base or pushed-before SHA; omit to select all targets')
+    parser.add_argument('--head', default='HEAD', help='PR head or pushed SHA')
     parser.add_argument('--output', type=Path, help='GitHub Actions output file')
     args = parser.parse_args()
     targets = load_targets(ROOT)
-    paths = changed_files(ROOT, args.base, args.head) if args.base else None
-    selected = select_targets(targets, paths)
+    base = args.base and diffable_base(ROOT, args.base)
+    if args.base and not base:
+        print(f'Cannot diff from {args.base}; selecting all targets.', file=sys.stderr)
+    paths = changed_files(ROOT, base, args.head) if base else None
+    base_targets = base_manifest_targets(ROOT, base, args.head) if base else []
+    selected = select_targets(targets, paths, base_targets, load_workspace(ROOT))
     matrix = json.dumps({'include': selected}, separators=(',', ':'))
     print(matrix)
     if args.output: