from __future__ import annotations # ruff: noqa: E402 import hashlib import json import sqlite3 import sys from pathlib import Path import pytest ROOT = Path(__file__).resolve().parents[1] sys.path.insert(0, str(ROOT / "scripts" / "health")) sys.path.insert(0, str(ROOT / "tests")) from dashboard_v5.document_originals import MAX_ORIGINAL_BYTES, probe_original from dashboard_v5.read_api import APIError, dispatch_api from dashboard_v5.render import render from fixtures.dashboard_v5_fixture import build_dashboard_v5_fixture def _insert_document(connection, *, name, path, review, content, day): connection.execute( """INSERT INTO dokumente( datei_name,dateipfad,daten_typ,kategorie,institution,review_status, document_date,extrahierte_inhalte,groessekbytes,datei_hash ) VALUES(?,?,?,?,?,?,?,?,?,?)""", (name, path, "pdf", "SYNTHETIC_6E4B", "SYNTHETIC_CLINIC", review, day, content, 1, f"hash-{day}"), ) def test_document_history_projects_explicit_fail_closed_statuses(tmp_path, monkeypatch): database = tmp_path / "fixture.db" originals = tmp_path / "originals" originals.mkdir() (originals / "available.pdf").write_bytes(b"%PDF-1.7\nsynthetic") (originals / "unsupported.pdf").write_bytes(b"not-a-pdf") build_dashboard_v5_fixture(database) connection = sqlite3.connect(database) _insert_document(connection, name="available.pdf", path="available.pdf", review="geprueft", content="SYNTHETIC CONTENT", day="2026-07-01") _insert_document(connection, name="pending.pdf", path="available.pdf", review="nicht_geprueft", content="SYNTHETIC PENDING", day="2026-07-02") _insert_document(connection, name="unsupported.pdf", path="unsupported.pdf", review="geprueft", content="", day="2026-07-03") _insert_document(connection, name="missing.pdf", path="missing.pdf", review="geprueft", content="SYNTHETIC MISSING", day="2026-07-04") connection.commit() connection.close() monkeypatch.setenv("HEALTH_DASHBOARD_ORIGINAL_ROOTS", str(originals)) payload = dispatch_api(database, "/api/v1/documents", "category=SYNTHETIC_6E4B&limit=10") assert len(payload["documents"]) == 4 by_day = {item["document_date"]: item for item in payload["documents"]} assert (by_day["2026-07-01"]["content_status"], by_day["2026-07-01"]["original_status"]) == ("extracted", "available") assert (by_day["2026-07-02"]["extraction_status"], by_day["2026-07-02"]["original_status"]) == ("extracted", "available") assert (by_day["2026-07-03"]["content_status"], by_day["2026-07-03"]["original_status"]) == ("not_extracted", "unsupported") assert by_day["2026-07-04"]["original_status"] == "missing" assert by_day["2026-07-02"]["search_status"] == "extracted_not_reviewed" assert payload["coverage"] == { "total": 4, "original_available": 2, "original_missing": 1, "original_unsupported": 1, "original_blocked": 0, "original_too_large": 0, "original_unavailable": 2, "text_extracted": 3, "content_reviewed": 3, "fulltext_searchable": 0, } detail = dispatch_api(database, f"/api/v1/documents/{by_day['2026-07-01']['id']}", "") assert detail["content_status"] == "available" assert detail["document_date_status"] == "known" assert detail["original_status"] == "available" assert detail["original_available"] is True with pytest.raises(APIError) as pending: dispatch_api(database, f"/api/v1/documents/{by_day['2026-07-02']['id']}", "") assert pending.value.status == 404 preview = dispatch_api(database, f"/api/v1/documents/{by_day['2026-07-02']['id']}/extracted-preview", "") assert preview["notice"] == "Maschinell extrahierter Inhalt – noch nicht inhaltlich geprüft" assert preview["search_status"] == "extracted_not_reviewed" serialized = json.dumps(payload) assert str(originals) not in serialized assert not any(name in serialized for name in ("available.pdf", "pending.pdf", "unsupported.pdf", "missing.pdf")) def test_shared_original_probe_blocks_symlinks_roots_magic_and_size(tmp_path): root = tmp_path / "root" root.mkdir() pdf = root / "ok.pdf" png = root / "ok.png" bad = root / "bad.pdf" large = root / "large.pdf" outside = tmp_path / "outside.pdf" pdf.write_bytes(b"%PDF-1.7\nsynthetic") png.write_bytes(b"\x89PNG\r\n\x1a\nsynthetic") bad.write_bytes(b"wrong-magic") outside.write_bytes(b"%PDF-1.7\nsynthetic") with large.open("wb") as handle: handle.write(b"%PDF-") handle.truncate(MAX_ORIGINAL_BYTES + 1) symlink = root / "link.pdf" symlink.symlink_to(pdf) for candidate, expected in ((pdf, "available"), (png, "available"), (bad, "unsupported"), (large, "too_large"), (outside, "blocked"), (symlink, "blocked"), (root / "missing.pdf", "missing")): result = probe_original(candidate, reviewed=True, roots=(root,)) try: assert result.status == expected finally: result.close() independent = probe_original(pdf, reviewed=False, roots=(root,)) try: assert independent.status == "available" finally: independent.close() def test_v5_render_bundles_pinned_fullcalendar_and_reusable_components(tmp_path): from dashboard_v5.data_provider import build_bundle database = tmp_path / "render.db" build_dashboard_v5_fixture(database) connection = sqlite3.connect(database) connection.row_factory = sqlite3.Row try: bundle = build_bundle(connection, today="2026-07-16", generated_at="2026-07-16T12:00:00+00:00") finally: connection.close() html = render(bundle, explorer_6c=True, calendar_day_6d=True, health_record_6e=True) assert "/health-assets/fullcalendar-6.1.21.min.js" in html assert "data-calendar-range-from" in html and "data-calendar-range-to" in html assert "data-day-context" in html and "data-day-missing" in html assert "data-record-breadcrumb" in html css = (ROOT / "scripts/health/assets/health-assets/dashboard-v5.css").read_text(encoding="utf-8") for contract in ("--v5-space-1", ".v5-grid-12", ".button-primary", ".status-chip", ".v5-skeleton"): assert contract in css def test_fullcalendar_notice_is_pinned_mit_and_standard_only(): notice = (ROOT / "THIRD_PARTY_NOTICES.md").read_text(encoding="utf-8") assert "FullCalendar Standard 6.1.21" in notice assert "MIT" in notice assert "No FullCalendar Premium or Scheduler" in notice assets = ROOT / "scripts/health/assets/health-assets" metadata = json.loads((assets / "fullcalendar-6.1.21.metadata.json").read_text(encoding="utf-8")) bundle = assets / "fullcalendar-6.1.21.min.js" license_file = assets / metadata["license_file"] assert metadata["version"] == "6.1.21" assert metadata["license"] == "MIT" assert hashlib.sha256(bundle.read_bytes()).hexdigest() == metadata["sha256"] assert hashlib.sha256(license_file.read_bytes()).hexdigest() == metadata["license_sha256"] assert "MIT License" in license_file.read_text(encoding="utf-8")