Refa: only support MinerU-API now (#11977)

### What problem does this PR solve?

Only support MinerU-API now, still need to complete frontend for
pipeline to allow the configuration of MinerU options.

### Type of change

- [x] Refactoring

rag/flow/parser/parser.py

@@ -280,6 +280,7 @@ class Parser(ProcessBase):
                 binary=blob,
                 callback=self.callback,
                 parse_method=conf.get("mineru_parse_method", "raw"),
+                lang=conf.get("lang", "Chinese"),
             )
             bboxes = []
             for t, poss in lines:

rag/llm/embedding_model.py

@@ -398,7 +398,7 @@ class JinaMultiVecEmbed(Base):
 
                     ress.append(chunk_emb)
 
-                token_count +=total_token_count_from_response(res)
+                token_count += total_token_count_from_response(res)
             except Exception as _e:
                 log_exception(_e, response)
                 raise Exception(f"Error: {response}")

rag/llm/ocr_model.py

@@ -16,7 +16,7 @@
 import json
 import logging
 import os
-from typing import Any, Optional, Tuple
+from typing import Any, Optional
 
 from deepdoc.parser.mineru_parser import MinerUParser
 
@@ -25,7 +25,7 @@ class Base:
     def __init__(self, key: str | dict, model_name: str, **kwargs):
         self.model_name = model_name
 
-    def parse_pdf(self, filepath: str, binary=None, **kwargs) -> Tuple[Any, Any]:
+    def parse_pdf(self, filepath: str, binary=None, **kwargs) -> tuple[Any, Any]:
         raise NotImplementedError("Please implement parse_pdf!")
 
 
@@ -56,21 +56,22 @@ class MinerUOcrModel(Base, MinerUParser):
         self.mineru_backend = _resolve_config("mineru_backend", "MINERU_BACKEND", "pipeline")
         self.mineru_server_url = _resolve_config("mineru_server_url", "MINERU_SERVER_URL", "")
         self.mineru_delete_output = bool(int(_resolve_config("mineru_delete_output", "MINERU_DELETE_OUTPUT", 1)))
-        self.mineru_executable = os.environ.get("MINERU_EXECUTABLE", "mineru")
 
-        logging.info(f"Parsed MinerU config: {config}")
+        logging.info(
+            f"Parsed MinerU config: backend={self.mineru_backend} api={self.mineru_api} server_url={self.mineru_server_url} output_dir={self.mineru_output_dir} delete_output={self.mineru_delete_output}"
+        )
 
-        MinerUParser.__init__(self, mineru_path=self.mineru_executable, mineru_api=self.mineru_api, mineru_server_url=self.mineru_server_url)
+        MinerUParser.__init__(self, mineru_api=self.mineru_api, mineru_server_url=self.mineru_server_url)
 
-    def check_available(self, backend: Optional[str] = None, server_url: Optional[str] = None) -> Tuple[bool, str]:
+    def check_available(self, backend: Optional[str] = None, server_url: Optional[str] = None) -> tuple[bool, str]:
         backend = backend or self.mineru_backend
         server_url = server_url or self.mineru_server_url
         return self.check_installation(backend=backend, server_url=server_url)
 
-    def parse_pdf(self, filepath: str, binary=None, callback=None, parse_method: str = "raw",**kwargs):
+    def parse_pdf(self, filepath: str, binary=None, callback=None, parse_method: str = "raw", **kwargs):
         ok, reason = self.check_available()
         if not ok:
-            raise RuntimeError(f"MinerU not found or server not accessible: {reason}. Please install it via: pip install -U 'mineru[core]'.")
+            raise RuntimeError(f"MinerU server not accessible: {reason}")
 
         sections, tables = MinerUParser.parse_pdf(
             self,

rag/svr/task_executor.py

@@ -69,7 +69,6 @@ from common.signal_utils import start_tracemalloc_and_snapshot, stop_tracemalloc
 from common.exceptions import TaskCanceledException
 from common import settings
 from common.constants import PAGERANK_FLD, TAG_FLD, SVR_CONSUMER_GROUP_NAME
-from common.misc_utils import check_and_install_mineru
 
 BATCH_SIZE = 64
 
@@ -1169,7 +1168,6 @@ async def main():
     show_configs()
     settings.init_settings()
     settings.check_and_install_torch()
-    check_and_install_mineru()
     logging.info(f'default embedding config: {settings.EMBEDDING_CFG}')
     settings.print_rag_settings()
     if sys.platform != "win32":