breakpointsoftware/document-parser

1name: Call Orchestrator Endpoint2 3on:4  workflow_dispatch:5    inputs:6      model:7        description: OpenAI model name8        required: false9        default: gpt-4o10      include_subfolders:11        description: Include subfolders in Drive scan12        required: false13        default: "true"14      send_to_sheet:15        description: Send rows to Google Sheets16        required: false17        default: "false"18  schedule:19    - cron: '*/5 * * * *'20jobs:21  call-orchestrator:22    runs-on: ubuntu-latest23    timeout-minutes: 124    env:25      ORCHESTRATOR_BASE_URL: ${{ secrets.ORCHESTRATOR_BASE_URL }}26      ORCHESTRATOR_API_KEY: ${{ secrets.ORCHESTRATOR_API_KEY }}27      MODEL_INPUT: ${{ github.event.inputs.model || 'gpt-4o' }}28      INCLUDE_SUBFOLDERS_INPUT: ${{ github.event.inputs.include_subfolders || 'true' }}29      SEND_TO_SHEET_INPUT: ${{ github.event.inputs.send_to_sheet || 'false' }}30 31    steps:32      - name: Validate required secrets33        run: |34          set -euo pipefail35          if [ -z "${ORCHESTRATOR_BASE_URL}" ]; then36            echo "Missing secret: ORCHESTRATOR_BASE_URL"37            exit 138          fi39          if [ -z "${ORCHESTRATOR_API_KEY}" ]; then40            echo "Missing secret: ORCHESTRATOR_API_KEY"41            exit 142          fi43 44      - name: Start orchestrator job (POST)45        id: start46        shell: bash47        run: |48          set -euo pipefail49          python - <<'PY'50          import json51          import os52          import urllib.request53 54          base_url = os.environ["ORCHESTRATOR_BASE_URL"].rstrip("/")55          payload = {56              "data": [57                  os.environ["MODEL_INPUT"],58                  os.environ["INCLUDE_SUBFOLDERS_INPUT"].lower() == "true",59                  os.environ["SEND_TO_SHEET_INPUT"].lower() == "true",60                  os.environ["ORCHESTRATOR_API_KEY"],61              ]62          }63 64          req = urllib.request.Request(65              url=f"{base_url}/gradio_api/call/orchestrate_drive_documents",66              data=json.dumps(payload).encode("utf-8"),67              headers={"Content-Type": "application/json"},68              method="POST",69          )70 71          with urllib.request.urlopen(req, timeout=60) as resp:72              body = resp.read().decode("utf-8")73 74          parsed = json.loads(body)75          event_id = parsed.get("event_id")76          if not event_id:77              raise SystemExit(f"POST succeeded but event_id missing. Response: {body}")78 79          with open(os.environ["GITHUB_OUTPUT"], "a", encoding="utf-8") as fh:80              fh.write(f"event_id={event_id}\n")81 82          print("Orchestrator job started.")83          PY84 85      - name: Read orchestrator stream result86        shell: bash87        run: |88          set -euo pipefail89          STREAM_URL="${ORCHESTRATOR_BASE_URL%/}/gradio_api/call/orchestrate_drive_documents/${{ steps.start.outputs.event_id }}"90          curl -sS --max-time 900 -N "$STREAM_URL" > stream.txt91          echo "Stream captured in stream.txt"92        id: stream93 94      - name: Validate orchestrator response95        shell: bash96        run: |97          set -euo pipefail98          python - <<'PY'99          import json100 101          with open("stream.txt", "r", encoding="utf-8") as fh:102              lines = [line.rstrip("\n") for line in fh]103 104          data_lines = [line[6:] for line in lines if line.startswith("data: ")]105          non_null = [line for line in data_lines if line and line != "null"]106          if not non_null:107              raise SystemExit("No non-null data payload found in stream.")108 109          payload = json.loads(non_null[-1])110          if not isinstance(payload, list) or not payload:111              raise SystemExit(f"Unexpected payload shape: {payload}")112 113          result = payload[0]114          ok = bool(result.get("ok")) if isinstance(result, dict) else False115          if not ok:116              raise SystemExit(f"Orchestrator failed: {json.dumps(result, ensure_ascii=False)}")117 118          summary = {119              "ok": result.get("ok"),120              "scanned": result.get("scanned"),121              "to_process": result.get("to_process"),122              "skipped": result.get("skipped"),123              "parsed": result.get("parsed"),124              "modified": result.get("modified"),125              "sent": result.get("sent"),126              "errors_count": len(result.get("errors", [])) if isinstance(result.get("errors"), list) else None,127          }128          print(json.dumps(summary, ensure_ascii=False))129          PY130