dicklesworthstone__coding_agent_session_search/.github/workflows/qualify-ann-shards.yml

# One-shot source-only landing. The connector retires this writer after success.
name: Wire retained multi-shard ANN
on:
  push:
    branches: [main]
    paths: [.github/workflows/qualify-ann-shards.yml]
permissions:
  contents: write
concurrency:
  group: cass-ann-shards-qualification
  cancel-in-progress: false
jobs:
  qualify:
    runs-on: ubuntu-24.04
    timeout-minutes: 25
    env:
      CARGO_BUILD_JOBS: '2'
      CARGO_PROFILE_TEST_DEBUG: '0'
      CARGO_PROFILE_DEV_DEBUG: '0'
      CARGO_TERM_COLOR: never
      RUST_MIN_STACK: '134217728'
    defaults:
      run:
        working-directory: cass-source
        shell: bash
    steps:
      - uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
        with:
          ref: main
          path: cass-source
      - uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
        with:
          repository: Dicklesworthstone/frankensearch
          ref: 6852e79a830c402ea4e55eb79eda3373c618dfc3
          path: fs-source
          persist-credentials: false
      - name: Guard and wire the complete native shard cohort
        run: |
          set -euo pipefail
          test "$(git branch --show-current)" = main
          test -z "$(git status --porcelain)"
          reviewed=5401c9fa93f6351e25b350663ccd1823c843bfc8
          git fetch --depth=1 origin "$reviewed"
          test ! -e src/search/query/ann_shards_integration.rs
          test ! -L src/search/query/ann_shards_integration.rs
          git show "$reviewed:src/search/query/ann_shards_integration.rs" > src/search/query/ann_shards_integration.rs
          git show "$reviewed:scripts/patches/ann-shards-reference.patch" > "$RUNNER_TEMP/ann-shards-reference.patch"
          git show "$reviewed:scripts/patches/wire_ann_shards.py" > "$RUNNER_TEMP/wire_ann_shards.py"
          test "$(git hash-object src/search/query/ann_shards.rs)" = bb76ee87f8ea9fcca5ffa4b5870cc5934c3b381f
          test "$(git hash-object "$RUNNER_TEMP/ann-shards-reference.patch")" = 52646a1db1e3dcd91c4bfe5b91d7197d9dc71b58
          test "$(git hash-object "$RUNNER_TEMP/wire_ann_shards.py")" = 9239b3eb32cfc6dcfce3279fa962ee1d2ec888e9
          test "$(git hash-object src/search/query/ann_shards_integration.rs)" = 653a9c86afdccd07c58f9734f9d2e180837419ad
          git apply --check --whitespace=error-all "$RUNNER_TEMP/ann-shards-reference.patch"
          git apply --whitespace=error-all "$RUNNER_TEMP/ann-shards-reference.patch"
          python3 "$RUNNER_TEMP/wire_ann_shards.py"
          rustc --version
          rustfmt --edition 2024 src/search/query/ann_shards.rs src/search/query/ann_shards_integration.rs
          python3 - <<'PY'
          import importlib.util
          from pathlib import Path
          import subprocess
          spec = importlib.util.spec_from_file_location('extract', 'scripts/verify_message_topk_fsvi.py')
          extract = importlib.util.module_from_spec(spec)
          spec.loader.exec_module(extract)
          path = Path('src/search/query.rs')
          text = path.read_text()
          for name in ['ann_index', 'set_semantic_artifacts_context', 'search_semantic_candidates']:
              old = extract.function(text, name, '    ')
              formatted = subprocess.run(['rustfmt', '--edition', '2024', '--emit', 'stdout'], input='impl SearchClient {\n' + old + '}\n', text=True, capture_output=True, check=True).stdout
              new = extract.function(formatted, name, '    ')
              assert text.count(old) == 1
              text = text.replace(old, new, 1)
          path.write_text(text)
          PY
          rustfmt --edition 2024 --config skip_children=true --emit stdout src/search/query.rs > "$RUNNER_TEMP/query-parsed.rs"
          git diff --check
      - name: Execute all native-shard and exact-FSVI component tests
        run: |
          set -euo pipefail
          python3 scripts/verify_ann_shards.py "$PWD" "$GITHUB_WORKSPACE/fs-source" "$RUNNER_TEMP/cass-ann-shards"
          export CARGO_TARGET_DIR="$RUNNER_TEMP/cass-ann-shards-target"
          cd "$RUNNER_TEMP/cass-ann-shards"
          cargo generate-lockfile
          sha256sum Cargo.lock
          cargo test --locked -- --nocapture --test-threads=2 2>&1 | tee "$RUNNER_TEMP/ann-shards-tests.log"
          grep -Eq 'test result: ok\. 31 passed; 0 failed; 0 ignored;' "$RUNNER_TEMP/ann-shards-tests.log"
      - name: Attempt the complete CASS routing tests separately
        id: full
        run: |
          set -uo pipefail
          export CARGO_TARGET_DIR="$RUNNER_TEMP/cass-full-target"
          set +e
          timeout 90 cargo test --locked --lib search::query::ann_shards -- --nocapture > "$RUNNER_TEMP/full-cass.log" 2>&1
          status=$?
          set -e
          cat "$RUNNER_TEMP/full-cass.log"
          echo "status=$status" >> "$GITHUB_OUTPUT"
          printf 'FULL_CASS_EXIT=%s; component success is not full-binary qualification\n' "$status"
      - name: Commit tested source-only changes directly to main
        env:
          FULL_CASS_EXIT: ${{ steps.full.outputs.status }}
        run: |
          set -euo pipefail
          git add -- src/search/query.rs src/search/query/ann_shards.rs src/search/query/ann_shards_integration.rs
          python3 - <<'PY'
          import subprocess
          paths = set(subprocess.check_output(['git', 'diff', '--cached', '--name-only'], text=True).splitlines())
          assert paths == {'src/search/query.rs', 'src/search/query/ann_shards.rs', 'src/search/query/ann_shards_integration.rs'}, paths
          PY
          git diff --cached --check
          git hash-object src/search/query.rs src/search/query/ann_shards.rs src/search/query/ann_shards_integration.rs
          git -c user.name='github-actions[bot]' -c user.email='41898282+github-actions[bot]@users.noreply.github.com' commit \
            -m 'feat(search): route approximate retrieval across every retained native shard' \
            -m 'Replace the single-shard restriction in context admission, cached graph ownership and candidate execution. Keep session-scoped exact routing and all exact/progressive paths unchanged. A failed optional sidecar rejects the entire ANN cohort, preserving complete exact fallback. Add four SearchClient integration regressions for cache reuse, missing-sidecar fallback, scoped bypass and context invalidation.' \
            -m "Executed 31 component tests with pinned real FSVI/native HNSW, including 11 shard tests and 20 existing exact tests. Preserve the degenerate duplicate-vector corpus as native-union differential evidence: ANN recall is approximate, not an exhaustive guarantee. Full CASS routing test attempt exit=$FULL_CASS_EXIT; these four full-graph tests, hydration, CLI, RCH/UBS and archive-scale qualification remain separate."
          for attempt in 1 2 3; do
            if GIT_TERMINAL_PROMPT=0 git push origin HEAD:refs/heads/main; then
              git rev-parse HEAD
              exit 0
            fi
            parent=$(git rev-parse HEAD^)
            git fetch --depth=100 origin main
            test -z "$(git diff --name-only "$parent" FETCH_HEAD -- src/search/query.rs src/search/vector_index.rs src/search/query/ann_shards.rs src/search/query/ann_shards_integration.rs src/search/query/session_scope.rs src/search/ann_index.rs scripts/verify_ann_shards.py scripts/verify_message_topk_fsvi.py Cargo.toml Cargo.lock)"
            git -c user.name='github-actions[bot]' -c user.email='41898282+github-actions[bot]@users.noreply.github.com' rebase FETCH_HEAD
          done
          echo 'Direct main push failed; no force-push attempted' >&2
          exit 1