name: ingest

# Fires on every push to main. Computes the diff against the parent
# commit, filters to supported extensions, then invokes `compass-ingest
# run --files <list>` inside a container pulled from
# ${{ vars.COMPASS_APP_IMAGE }}.
#
# The CLI talks directly to postgres-rag + the raw Graphiti Neo4j —
# there is no HTTP hop through the long-running app service.

on:
  push:
    branches:
      - main

jobs:
  ingest:
    runs-on: ubuntu-latest
    container:
      image: ${{ vars.COMPASS_APP_IMAGE }}
      # Same network override + volume allow-list as deploy.yml. See
      # scripts/runner-config.yaml for the rationale.
      network: ${{ vars.COMPASS_NETWORK || 'compass_default' }}
    env:
      # Pgvector chunk store.
      POSTGRES_RAG_DSN: ${{ secrets.POSTGRES_RAG_DSN }}
      # Raw Graphiti graph (7688). These are NOT the structured-graph
      # write creds — those only exist in compass/migrations.
      NEO4J_URI: ${{ secrets.NEO4J_RAW_URI }}
      NEO4J_USER: ${{ secrets.NEO4J_RAW_USER }}
      NEO4J_PASSWORD: ${{ secrets.NEO4J_RAW_PASSWORD }}

    steps:
      - name: Check out knowledge-base repo
        uses: actions/checkout@v4
        with:
          # Depth 2 gives us $GITHUB_SHA and its parent so `git diff`
          # can run. The initial-commit path (parent == 000…) is
          # handled below.
          fetch-depth: 2

      - name: Compute changed files
        id: changes
        shell: bash
        run: |
          set -euo pipefail

          BEFORE="${{ github.event.before }}"
          AFTER="${{ github.sha }}"

          # Gitea (like GitHub) reports 40 zeros as the "before" of an
          # initial push or a branch creation. Fall back to the full
          # file listing so we don't silently skip the first batch.
          if [ -z "$BEFORE" ] || [ "$BEFORE" = "0000000000000000000000000000000000000000" ]; then
            echo "initial commit or branch creation — ingesting full tree"
            git ls-files > /tmp/changed-all.txt
          else
            # `--diff-filter=d` drops deletions: we don't want to ingest
            # a file that no longer exists.
            git diff --name-only --diff-filter=d "$BEFORE".."$AFTER" > /tmp/changed-all.txt
          fi

          # Filter to supported extensions. Keep this list in sync with
          # the matcher in app/ingestion/pipeline.py.
          grep -Ei '\.(md|txt|pdf|docx|xlsx)$' /tmp/changed-all.txt > /tmp/changed.txt || true

          count=$(wc -l < /tmp/changed.txt | tr -d ' ')
          echo "changed_count=$count" >> "$GITHUB_OUTPUT"
          echo "changed files:"
          cat /tmp/changed.txt || true

      - name: Ingest changed files
        if: steps.changes.outputs.changed_count != '0'
        shell: bash
        run: |
          set -euo pipefail
          # xargs feed avoids "argument list too long" on large pushes
          # while still giving compass-ingest a single invocation so
          # the shared DB/graph pools are reused.
          mapfile -t files < /tmp/changed.txt
          compass-ingest run --files "${files[@]}"

      - name: No-op summary
        if: steps.changes.outputs.changed_count == '0'
        run: echo "No supported files changed — nothing to ingest."

      - name: Upload ingest log
        if: always()
        # v4 uses GHES Artifacts API v2 which Gitea doesn't implement yet.
        uses: actions/upload-artifact@v3
        with:
          name: ingest-${{ github.sha }}
          path: |
            /tmp/changed.txt