Pass MinerU HTTP device field

@@ -3,6 +3,10 @@ WEKNORA_API_KEY=
 WEKNORA_KB_ID=
 WEKNORA_KB_ID=
 WEKNORA_KB_NAME=ragas-eval-pilot
 WEKNORA_KB_NAME=ragas-eval-pilot
 
 
+# MinerU HTTP parser. Use cpu, cuda, cuda:0, etc. according to the deployed
+# MinerU backend.
+MINERU_DEVICE=cpu
+
 # Ragas generation and judge models. These are evaluation-side models, not the
 # Ragas generation and judge models. These are evaluation-side models, not the
 # model configuration used by the WeKnora backend.
 # model configuration used by the WeKnora backend.
 RAGAS_LLM_API_KEY=replace-me
 RAGAS_LLM_API_KEY=replace-me

@@ -29,6 +29,8 @@ parsing:
     output_dir: "data/parsed_docs/mineru_raw"
     output_dir: "data/parsed_docs/mineru_raw"
     http_base_url: "http://172.23.184.9:8002"
     http_base_url: "http://172.23.184.9:8002"
     http_parse_path: "/file_parse"
     http_parse_path: "/file_parse"
+    http_form_fields:
+      device: "${MINERU_DEVICE:-cpu}"
     api_key: "mineru"
     api_key: "mineru"
     timeout_seconds: 600
     timeout_seconds: 600
     fallback_to_local: false
     fallback_to_local: false

@@ -131,16 +131,23 @@ def parse_pdf_with_http(
     endpoint = str(mineru_config.get("http_parse_path") or "/file_parse")
     endpoint = str(mineru_config.get("http_parse_path") or "/file_parse")
     if not endpoint.startswith("/"):
     if not endpoint.startswith("/"):
         endpoint = "/" + endpoint
         endpoint = "/" + endpoint
+    form_fields = {
+        str(key): str(value)
+        for key, value in (mineru_config.get("http_form_fields") or {}).items()
+        if value not in {None, ""}
+    }
 
 
     with target.open("rb") as file:
     with target.open("rb") as file:
         response = requests.post(
         response = requests.post(
             f"{base_url}{endpoint}",
             f"{base_url}{endpoint}",
             files=[("files", (target.name, file, "application/pdf"))],
             files=[("files", (target.name, file, "application/pdf"))],
+            data=form_fields,
             headers=headers,
             headers=headers,
             timeout=int(mineru_config.get("timeout_seconds", 600)),
             timeout=int(mineru_config.get("timeout_seconds", 600)),
         )
         )
     if response.status_code >= 400:
     if response.status_code >= 400:
-        raise MinerUParseError(f"MinerU HTTP failed with {response.status_code}: {response.text[:500]}")
+        error_detail = _mineru_error_detail(response)
+        raise MinerUParseError(f"MinerU HTTP failed with {response.status_code}: {error_detail}")
 
 
     payload = response.json()
     payload = response.json()
     contents = extract_mineru_contents(payload)
     contents = extract_mineru_contents(payload)
@@ -161,12 +168,29 @@ def parse_pdf_with_http(
                 source_file=target.name,
                 source_file=target.name,
                 file_type="pdf",
                 file_type="pdf",
                 content=content,
                 content=content,
-                metadata={"parser": "mineru:http", "mineru_endpoint": endpoint},
+                metadata={
+                    "parser": "mineru:http",
+                    "mineru_endpoint": endpoint,
+                    "mineru_form_fields": form_fields,
+                },
             )
             )
         )
         )
     return docs
     return docs
 
 
 
 
+def _mineru_error_detail(response: requests.Response) -> str:
+    try:
+        payload = response.json()
+    except ValueError:
+        return response.text[:1000]
+    error = payload.get("error")
+    if error:
+        task_id = payload.get("task_id")
+        status = payload.get("status")
+        return f"task_id={task_id} status={status} error={error}"
+    return response.text[:1000]
+
+
 def extract_mineru_contents(payload: Any) -> list[str]:
 def extract_mineru_contents(payload: Any) -> list[str]:
     contents: list[str] = []
     contents: list[str] = []
     _collect_text_values(payload, contents)
     _collect_text_values(payload, contents)