update

api/utils/api_utils.py

@@ -707,6 +707,7 @@ async def is_strong_enough(chat_model, embedding_model):
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Pressure test failed: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)

deepdoc/parser/pdf_parser.py

@@ -1150,6 +1150,7 @@ class RAGFlowPdfParser:
                 try:
                     await asyncio.gather(*tasks, return_exceptions=False)
                 except Exception as e:
+                    logging.error(f"Error in OCR: {e}")
                     for t in tasks:
                         t.cancel()
                     await asyncio.gather(*tasks, return_exceptions=True)

deepdoc/vision/t_ocr.py

@@ -15,6 +15,7 @@
 #
 
 import asyncio
+import logging
 import os
 import sys
 sys.path.insert(
@@ -78,6 +79,7 @@ def main(args):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error("OCR tasks failed: {}".format(e))
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

graphrag/entity_resolution.py

@@ -146,6 +146,7 @@ class EntityResolution(Extractor):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error resolving candidate pairs: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)
@@ -169,6 +170,7 @@ class EntityResolution(Extractor):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error merging nodes: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

graphrag/general/extractor.py

@@ -192,6 +192,7 @@ class Extractor:
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error merging nodes: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)
@@ -221,6 +222,7 @@ class Extractor:
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error during relationships merging: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

graphrag/general/mind_map_extractor.py

@@ -108,6 +108,7 @@ class MindMapExtractor(Extractor):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error processing document: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

graphrag/utils.py

@@ -479,6 +479,7 @@ async def set_graph(tenant_id: str, kb_id: str, embd_mdl, graph: nx.Graph, chang
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error while deleting edges: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

rag/flow/parser/parser.py

@@ -20,6 +20,7 @@ import random
 import re
 from functools import partial
 
+from litellm import logging
 import numpy as np
 from PIL import Image
 
@@ -819,6 +820,7 @@ class Parser(ProcessBase):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error("Error while parsing: %s" % e)
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

rag/flow/splitter/splitter.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 import asyncio
+import logging
 import random
 import re
 from copy import deepcopy
@@ -135,6 +136,7 @@ class Splitter(ProcessBase):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"error when splitting: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

rag/svr/task_executor.py

@@ -339,6 +339,7 @@ async def build_chunks(task, progress_callback):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error("Error in doc_keyword_extraction: {}".format(e))
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)
@@ -370,6 +371,7 @@ async def build_chunks(task, progress_callback):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error("Error in doc_question_proposal", exc_info=e)
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)
@@ -430,6 +432,7 @@ async def build_chunks(task, progress_callback):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error("Error tagging docs: {}".format(e))
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)