fix defaults for table extraction

Author: MthwRobinson

prepline_general/api/general.py

@@ -295,7 +295,7 @@ def pipeline_api(
     hi_res_model_name: Optional[str] = None,
     include_page_breaks: bool = False,
     ocr_languages: Optional[List[str]] = None,
-    pdf_infer_table_structure: bool = True,
+    pdf_infer_table_structure: bool = False,
     skip_infer_table_types: Optional[List[str]] = None,
     strategy: str = "auto",
     xml_keep_tags: bool = False,
@@ -361,7 +361,6 @@ def pipeline_api(
 
     hi_res_model_name = _validate_hi_res_model_name(hi_res_model_name, coordinates)
     strategy = _validate_strategy(strategy)
-    pdf_infer_table_structure = _set_pdf_infer_table_structure(pdf_infer_table_structure, strategy)
 
     # Parallel mode is set by env variable
     enable_parallel_mode = os.environ.get("UNSTRUCTURED_PARALLEL_MODE_ENABLED", "false")
@@ -595,11 +594,6 @@ def _validate_chunking_strategy(chunking_strategy: Optional[str]) -> Optional[st
     return chunking_strategy
 
 
-def _set_pdf_infer_table_structure(pdf_infer_table_structure: bool, strategy: str) -> bool:
-    """Avoids table inference in "fast" and "ocr_only" runs."""
-    return strategy in ("hi_res", "auto") and pdf_infer_table_structure
-
-
 def get_validated_mimetype(file: UploadFile) -> Optional[str]:
     """The MIME-type of `file`.