fix(ai): fallback PixelRAG platform probe runtime gaps

services/pixelrag_platform_probe_worker_service.py

@@ -238,14 +238,40 @@ def _run_capture_subprocess(
         payload = json.loads(completed.stdout)
     except json.JSONDecodeError:
         payload = {}
+    stderr_excerpt = completed.stderr[:500]
+    if len(completed.stderr) > 500:
+        stderr_excerpt = f"{completed.stderr[:250]}...{completed.stderr[-250:]}"
     return {
         "returncode": completed.returncode,
         "stdout_json": payload,
         "stdout_excerpt": completed.stdout[:500],
-        "stderr_excerpt": completed.stderr[:500],
+        "stderr_excerpt": stderr_excerpt,
     }
 
 
+def _capture_runtime_unavailable(result: Mapping[str, Any]) -> bool:
+    haystack = " ".join(
+        [
+            str(result.get("stdout_excerpt") or ""),
+            str(result.get("stderr_excerpt") or ""),
+        ]
+    ).lower()
+    return (
+        "modulenotfounderror" in haystack
+        and "playwright" in haystack
+    ) or "no module named 'playwright'" in haystack or (
+        "from playwright.sync_api" in haystack
+        and "traceback" in haystack
+    ) or (
+        "playwright" in haystack
+        and (
+            "executable doesn't exist" in haystack
+            or "please run" in haystack
+            or "browsertype.launch" in haystack
+        )
+    )
+
+
 def _capture_item(
     item: Mapping[str, Any],
     *,
@@ -302,8 +328,25 @@ def _capture_item(
         and capture_payload.get("success") is True
         and capture_payload.get("status") == "captured"
     )
+    runtime_missing = not captured_ok and _capture_runtime_unavailable(result)
+    structured_fallback_available = bool(
+        (item.get("structured_source_fallback") or {}).get("available")
+    )
+    capture_fallback = not captured_ok and structured_fallback_available
     base.update({
-        "worker_status": "executed_capture_ok" if captured_ok else "capture_worker_error",
+        "worker_status": (
+            "executed_capture_ok"
+            if captured_ok
+            else (
+                (
+                    "capture_runtime_unavailable_structured_fallback_package"
+                    if runtime_missing
+                    else "capture_failed_structured_fallback_package"
+                )
+                if capture_fallback
+                else "capture_worker_error"
+            )
+        ),
         "ready_for_execution": True,
         "capture_execution_ready": True,
         "capture_manifest_id": manifest.get("manifest_id"),
@@ -314,14 +357,39 @@ def _capture_item(
         else 0,
         "capture_tile_file_count": tile_file_count,
         "capture_receipt_path": receipt_path,
-        "network_call_performed": True,
+        "network_call_performed": bool(not runtime_missing),
         "artifact_write_performed": bool(captured_ok),
+        "structured_source_fallback": item.get("structured_source_fallback") or {},
+        "structured_source_package": (
+            {
+                "adapter_code": (item.get("structured_source_fallback") or {}).get("adapter_code"),
+                "available": True,
+                "source_count": len(list((item.get("structured_source_fallback") or {}).get("sources") or [])),
+                "network_request_allowed": bool(
+                    (item.get("structured_source_fallback") or {}).get("network_request_allowed")
+                ),
+                "database_write_allowed": False,
+                "dry_run_only": True,
+                "blocked_page_not_product_data": True,
+                "capture_runtime_unavailable": runtime_missing,
+            }
+            if capture_fallback
+            else None
+        ),
         "capture_stdout_excerpt": result["stdout_excerpt"] if not captured_ok else "",
         "capture_stderr_excerpt": result["stderr_excerpt"] if not captured_ok else "",
         "next_machine_action": (
             "run_pixelrag_rag_candidate_replay_after_probe_capture"
             if captured_ok
-            else "repair_pixelrag_platform_probe_worker_capture_runtime_or_use_structured_source"
+            else (
+                (
+                    "run_structured_source_or_install_pixelrag_capture_runtime"
+                    if runtime_missing
+                    else "run_structured_source_or_retry_platform_capture"
+                )
+                if capture_fallback
+                else "repair_pixelrag_platform_probe_worker_capture_runtime_or_use_structured_source"
+            )
         ),
     })
     if write_receipt:
@@ -408,7 +476,7 @@ def run_pixelrag_platform_probe_worker(
     ready_count = sum(1 for item in probe_items if item.get("probe_ready"))
     dry_run_count = sum(1 for item in worker_items if str(item.get("worker_status") or "").startswith("dry_run_"))
     structured_count = sum(
-        1 for item in worker_items if "structured_source_fallback" in str(item.get("worker_status") or "")
+        1 for item in worker_items if "structured" in str(item.get("worker_status") or "")
     )
     capture_ready_count = sum(1 for item in worker_items if item.get("capture_execution_ready"))
     capture_execute_count = sum(
@@ -417,7 +485,10 @@ def run_pixelrag_platform_probe_worker(
     capture_ok_count = sum(1 for item in worker_items if item.get("worker_status") == "executed_capture_ok")
     capture_error_count = sum(1 for item in worker_items if item.get("worker_status") == "capture_worker_error")
     executed_structured_count = sum(
-        1 for item in worker_items if item.get("worker_status") == "executed_structured_source_fallback_package"
+        1
+        for item in worker_items
+        if "structured" in str(item.get("worker_status") or "")
+        and not str(item.get("worker_status") or "").startswith("dry_run_")
     )
     executed_count = capture_execute_count + executed_structured_count
     skipped_count = sum(1 for item in worker_items if str(item.get("worker_status") or "").startswith("skipped_"))