Feat: add context for figure and table (#11547)

### What problem does this PR solve?

Add context for figure table.



![demo_figure_table_context](https://github.com/user-attachments/assets/61b37fac-e22e-40a4-9665-9396c7b4103e)


`==================()` for demonstrating purpose. 
### Type of change

- [x] New Feature (non-breaking change which adds functionality)

api/db/db_models.py

@@ -749,7 +749,7 @@ class Knowledgebase(DataBaseModel):
 
     parser_id = CharField(max_length=32, null=False, help_text="default parser ID", default=ParserType.NAIVE.value, index=True)
     pipeline_id = CharField(max_length=32, null=True, help_text="Pipeline ID", index=True)
-    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]]})
+    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]], "table_context_size": 0, "image_context_size": 0})
     pagerank = IntegerField(default=0, index=False)
 
     graphrag_task_id = CharField(max_length=32, null=True, help_text="Graph RAG task ID", index=True)
@@ -774,7 +774,7 @@ class Document(DataBaseModel):
     kb_id = CharField(max_length=256, null=False, index=True)
     parser_id = CharField(max_length=32, null=False, help_text="default parser ID", index=True)
     pipeline_id = CharField(max_length=32, null=True, help_text="pipeline ID", index=True)
-    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]]})
+    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]], "table_context_size": 0, "image_context_size": 0})
     source_type = CharField(max_length=128, null=False, default="local", help_text="where dose this document come from", index=True)
     type = CharField(max_length=32, null=False, help_text="file extension", index=True)
     created_by = CharField(max_length=32, null=False, help_text="who created it", index=True)

api/db/services/document_service.py

@@ -923,7 +923,7 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
         ParserType.AUDIO.value: audio,
         ParserType.EMAIL.value: email
     }
-    parser_config = {"chunk_token_num": 4096, "delimiter": "\n!?;。；！？", "layout_recognize": "Plain Text"}
+    parser_config = {"chunk_token_num": 4096, "delimiter": "\n!?;。；！？", "layout_recognize": "Plain Text", "table_context_size": 0, "image_context_size": 0}
     exe = ThreadPoolExecutor(max_workers=12)
     threads = []
     doc_nm = {}

api/utils/api_utils.py

@@ -313,6 +313,10 @@ def get_parser_config(chunk_method, parser_config):
         chunk_method = "naive"
 
     # Define default configurations for each chunking method
+    base_defaults = {
+        "table_context_size": 0,
+        "image_context_size": 0,
+    }
     key_mapping = {
         "naive": {
             "layout_recognize": "DeepDOC",
@@ -365,16 +369,19 @@ def get_parser_config(chunk_method, parser_config):
 
     default_config = key_mapping[chunk_method]
 
-    # If no parser_config provided, return default
+    # If no parser_config provided, return default merged with base defaults
     if not parser_config:
-        return default_config
+        if default_config is None:
+            return deep_merge(base_defaults, {})
+        return deep_merge(base_defaults, default_config)
 
     # If parser_config is provided, merge with defaults to ensure required fields exist
     if default_config is None:
-        return parser_config
+        return deep_merge(base_defaults, parser_config)
 
     # Ensure raptor and graphrag fields have default values if not provided
-    merged_config = deep_merge(default_config, parser_config)
+    merged_config = deep_merge(base_defaults, default_config)
+    merged_config = deep_merge(merged_config, parser_config)
 
     return merged_config