feat(crawler): expand pixelrag marketplace automation

tests/test_pchome_crawler_search.py

@@ -69,6 +69,32 @@ def test_pchome_search_scans_multiple_pages_until_limit(monkeypatch):
     assert [product.product_id for product in products] == ["A001", "A002", "A003"]
 
 
+def test_pchome_search_auto_queues_pixelrag_manifest_on_empty_result(monkeypatch, tmp_path):
+    from services import pixelrag_crawler_integration_service as pixelrag
+    from services.pchome_crawler import PChomeCrawler
+
+    monkeypatch.setattr(pixelrag, "DEFAULT_MANIFEST_QUEUE_ROOT", str(tmp_path))
+    crawler = PChomeCrawler(timeout=1, delay=0, max_retries=0)
+
+    class FakeSession:
+        headers = {}
+
+        def get(self, url, params=None, timeout=None):
+            return _FakeResponse({"Prods": []})
+
+    crawler.session = FakeSession()
+
+    success, message, products = crawler.search_products("找不到的商品", limit=3, max_pages=1)
+
+    assert success is False
+    assert message == "沒有找到符合的商品"
+    assert products == []
+    action = crawler.last_visual_evidence_action
+    assert action["status"] == "manifest_persisted"
+    assert action["manifest"]["capture_target"]["crawler"] == "PChomeCrawler.search_products"
+    assert action["persistence"]["path"].startswith(str(tmp_path))
+
+
 def test_pchome_get_retries_transient_timeout():
     from services.pchome_crawler import PChomeCrawler
 

tests/test_pixelrag_crawler_integration_service.py

@@ -124,6 +124,67 @@ def test_visual_fallback_selector_routes_parser_empty_and_low_confidence_cases()
     assert clean_parser["fallback_reason"] == "structured_evidence_sufficient"
 
 
+def test_pixelrag_ecommerce_expansion_plan_includes_shopee_and_coupang():
+    from services.pixelrag_crawler_integration_service import (
+        ECOMMERCE_EXPANSION_POLICY,
+        build_pixelrag_ecommerce_expansion_plan,
+    )
+
+    payload = build_pixelrag_ecommerce_expansion_plan()
+    platforms = {item["platform"]: item for item in payload["platforms"]}
+
+    assert payload["policy"] == ECOMMERCE_EXPANSION_POLICY
+    assert payload["status"] == "marketplace_profiles_ready"
+    assert platforms["shopee_tw"]["manifest_ready"] is True
+    assert platforms["coupang_tw"]["manifest_ready"] is True
+    assert "price" in platforms["shopee_tw"]["data_targets"]
+    assert payload["next_actions"][0]["target_platforms"] == ["shopee_tw", "coupang_tw"]
+    assert payload["controlled_apply"]["production_price_write"] is False
+
+
+def test_marketplace_search_manifest_builds_shopee_visual_capture_target():
+    from services.pixelrag_crawler_integration_service import build_pixelrag_marketplace_search_manifest
+
+    payload = build_pixelrag_marketplace_search_manifest(
+        platform="shopee_tw",
+        keyword="防曬乳",
+        page_size={"width": 1440, "height": 1024},
+        tile_size={"width": 512, "height": 512},
+    )
+
+    assert payload["success"] is True
+    assert payload["capture_target"]["platform"] == "shopee_tw"
+    assert payload["capture_target"]["url"].startswith("https://shopee.tw/search?keyword=")
+    assert "%E9%98%B2%E6%9B%AC%E4%B9%B3" in payload["capture_target"]["url"]
+    assert payload["marketplace_profile"]["display_name"] == "Shopee Taiwan"
+    assert payload["tile_plan"]["tile_count"] == 6
+
+
+def test_crawler_failure_action_persists_manifest_queue(tmp_path):
+    from services.pixelrag_crawler_integration_service import build_pixelrag_crawler_failure_action
+
+    payload = build_pixelrag_crawler_failure_action(
+        url="https://24h.pchome.com.tw/search/?q=test",
+        platform="pchome",
+        crawler="PChomeCrawler.search_products",
+        parser_success=False,
+        parsed_item_count=0,
+        missing_fields=["price", "product_id"],
+        failure_reason="search API returned no product IDs",
+        persist_manifest=True,
+        queue_root=tmp_path,
+    )
+
+    assert payload["success"] is True
+    assert payload["status"] == "manifest_persisted"
+    assert payload["next_machine_action"] == "run_pixelrag_visual_capture_worker"
+    assert payload["controlled_apply"]["artifact_write"] is True
+    manifest_path = payload["persistence"]["path"]
+    queued = json.loads(open(manifest_path, encoding="utf-8").read())
+    assert queued["capture_target"]["platform"] == "pchome"
+    assert queued["queue_policy"] == "read_only_pixelrag_crawler_failure_action_v1"
+
+
 def test_visual_evidence_manifest_builds_tile_plan_without_runtime_writes():
     from services.pixelrag_crawler_integration_service import (
         RESULT_MANIFEST_READY,
@@ -226,6 +287,17 @@ def test_pixelrag_ai_automation_route_returns_assessment_and_manifest():
     ):
         manifest_response = routes.ai_automation_pixelrag_crawler_integration_api.__wrapped__()
     manifest = manifest_response.get_json()
+    with app.test_request_context(
+        "/api/ai-automation/pixelrag-crawler-integration?expansion_plan=true"
+    ):
+        expansion_response = routes.ai_automation_pixelrag_crawler_integration_api.__wrapped__()
+    expansion = expansion_response.get_json()
+    with app.test_request_context(
+        "/api/ai-automation/pixelrag-crawler-integration"
+        "?platform=shopee_tw&marketplace_keyword=防曬乳"
+    ):
+        marketplace_response = routes.ai_automation_pixelrag_crawler_integration_api.__wrapped__()
+    marketplace = marketplace_response.get_json()
 
     assert assessment["policy"] == "read_only_pixelrag_crawler_integration_assessment_v1"
     assert assessment["target_platforms"] == ["momo"]
@@ -233,6 +305,10 @@ def test_pixelrag_ai_automation_route_returns_assessment_and_manifest():
     assert manifest["policy"] == "read_only_pixelrag_visual_evidence_manifest_v1"
     assert manifest["result"] == "PIXELRAG_VISUAL_EVIDENCE_MANIFEST_READY"
     assert manifest["capture_target"]["platform"] == "pchome"
+    assert expansion["policy"] == "read_only_pixelrag_ecommerce_expansion_plan_v1"
+    assert expansion["next_actions"][0]["target_platforms"] == ["shopee_tw", "coupang_tw"]
+    assert marketplace["capture_target"]["platform"] == "shopee_tw"
+    assert marketplace["marketplace_profile"]["keyword"] == "防曬乳"
 
 
 def _write_capture_receipt(root, *, platform="pchome", manifest_id="manifest-1"):
@@ -303,6 +379,30 @@ def test_pixelrag_visual_evidence_readback_reports_latest_receipt(tmp_path):
     assert payload["controlled_apply"]["db_write"] is False
 
 
+def test_pixelrag_visual_evidence_readback_warns_on_access_denied_page(tmp_path):
+    from services.pixelrag_crawler_integration_service import build_pixelrag_visual_evidence_readback
+
+    receipt_path = _write_capture_receipt(tmp_path, platform="coupang_tw", manifest_id="denied")
+    receipt = json.loads(receipt_path.read_text(encoding="utf-8"))
+    receipt["http_status"] = 403
+    receipt["page_metrics"] = {
+        "final_url": "https://www.tw.coupang.com/search?q=iphone",
+        "title": "Access Denied",
+    }
+    receipt_path.write_text(json.dumps(receipt), encoding="utf-8")
+
+    payload = build_pixelrag_visual_evidence_readback(
+        artifact_root=tmp_path,
+        platform="coupang_tw",
+        manifest_id="denied",
+    )
+
+    assert payload["status"] == "warning"
+    assert payload["summary"]["visual_barrier_detected"] is True
+    assert payload["summary"]["visual_barrier_reason"] == "http_status_403"
+    assert payload["next_machine_action"] == "run_platform_probe_or_use_structured_api"
+
+
 def test_pixelrag_visual_evidence_readback_warns_when_missing(tmp_path):
     from services.pixelrag_crawler_integration_service import build_pixelrag_visual_evidence_readback
 
@@ -467,3 +567,34 @@ def test_pixelrag_python_capture_worker_dry_run_plans_artifact_outputs(tmp_path)
     assert payload["controlled_apply"]["artifact_write"] is False
     assert payload["tile_plan"]["planned_tile_count"] == 4
     assert payload["capture_target"]["platform"] == "pchome"
+
+
+def test_pixelrag_python_capture_worker_accepts_shopee_manifest(tmp_path):
+    from services.pixelrag_crawler_integration_service import build_pixelrag_marketplace_search_manifest
+
+    manifest = build_pixelrag_marketplace_search_manifest(
+        platform="shopee_tw",
+        keyword="防曬乳",
+        page_size={"width": 1024, "height": 1024},
+        tile_size={"width": 512, "height": 512},
+    )
+    completed = subprocess.run(
+        [
+            sys.executable,
+            "scripts/ops/capture_pixelrag_visual_evidence.py",
+            "--manifest-json",
+            json.dumps(manifest),
+            "--output-dir",
+            str(tmp_path),
+            "--dry-run",
+        ],
+        capture_output=True,
+        check=False,
+        text=True,
+    )
+
+    assert completed.returncode == 0
+    payload = json.loads(completed.stdout)
+    assert payload["status"] == "dry_run_ready"
+    assert payload["capture_target"]["platform"] == "shopee_tw"
+    assert payload["tile_plan"]["planned_tile_count"] == 4