build(deps): version bumps for maintenance (#420)

### Summary

Version bumps for regular maintenance and to address moderate CVEs from
security scans. Also updates the `unstructured` extra from
`local-inference` to `all-docs` to keep up with latest best practices
for the `unstructured` library.

Includes an update for appropriately setting `pdf_infer_table_structure`
depending on the value of `skip_infer_table_types` and adds a test.

Author: MthwRobinson

CHANGELOG.md

@@ -1,3 +1,8 @@
+## 0.0.69
+
+* Bump to `unstructured` 0.14.4
+* Add handling for `pdf_infer_table_structure` to reflect the "tables off by default" behavior in `unstructured`.
+
 ## 0.0.68
 
 * Fix list params such as `extract_image_block_types` not working via the python/js clients
@@ -20,7 +25,7 @@
 * Bump unstructured to 0.12.4
 * Add support for both `list[str]` and `str` input formats for `ocr_languages` parameter
 * Adds support for additional MIME types from `unstructured`
-* Document the support for gzip files and add additional testing 
+* Document the support for gzip files and add additional testing
 
 ## 0.0.64
 

prepline_general/api/app.py

@@ -13,7 +13,7 @@
 app = FastAPI(
     title="Unstructured Pipeline API",
     summary="Partition documents with the Unstructured library",
-    version="0.0.68",
+    version="0.0.69",
     docs_url="/general/docs",
     openapi_url="/general/openapi.json",
     servers=[

prepline_general/api/general.py

@@ -361,7 +361,11 @@ 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)
+    pdf_infer_table_structure = _set_pdf_infer_table_structure(
+        pdf_infer_table_structure,
+        strategy,
+        skip_infer_table_types,
+    )
 
     # Parallel mode is set by env variable
     enable_parallel_mode = os.environ.get("UNSTRUCTURED_PARALLEL_MODE_ENABLED", "false")
@@ -441,9 +445,9 @@ def pipeline_api(
             )
         elif hi_res_model_name and hi_res_model_name in CHIPPER_MODEL_TYPES:
             with ChipperMemoryProtection():
-                elements = partition(**partition_kwargs)  # pyright: ignore[reportGeneralTypeIssues]
+                elements = partition(**partition_kwargs)  # type: ignore # pyright: ignore[reportGeneralTypeIssues]
         else:
-            elements = partition(**partition_kwargs)  # pyright: ignore[reportGeneralTypeIssues]
+            elements = partition(**partition_kwargs)  # type: ignore # pyright: ignore[reportGeneralTypeIssues]
 
     except OSError as e:
         if isinstance(e.args[0], str) and (
@@ -595,8 +599,13 @@ 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:
+def _set_pdf_infer_table_structure(
+    pdf_infer_table_structure: bool, strategy: str, skip_infer_table_types: Optional[List[str]]
+) -> bool:
     """Avoids table inference in "fast" and "ocr_only" runs."""
+    # NOTE(robinson) - line below is for type checking
+    skip_infer_table_types = [] if skip_infer_table_types is None else skip_infer_table_types
+    pdf_infer_table_structure = pdf_infer_table_structure and ("pdf" not in skip_infer_table_types)
     return strategy in ("hi_res", "auto") and pdf_infer_table_structure
 
 
@@ -704,7 +713,7 @@ def return_content_type(filename: str):
 
 
 @router.get("/general/v0/general", include_in_schema=False)
-@router.get("/general/v0.0.68/general", include_in_schema=False)
+@router.get("/general/v0.0.69/general", include_in_schema=False)
 async def handle_invalid_get_request():
     raise HTTPException(
         status_code=status.HTTP_405_METHOD_NOT_ALLOWED, detail="Only POST requests are supported."
@@ -719,7 +728,7 @@ async def handle_invalid_get_request():
     description="Description",
     operation_id="partition_parameters",
 )
-@router.post("/general/v0.0.68/general", include_in_schema=False)
+@router.post("/general/v0.0.69/general", include_in_schema=False)
 def general_partition(
     request: Request,
     # cannot use annotated type here because of a bug described here:

preprocessing-pipeline-family.yaml

@@ -1,2 +1,2 @@
 name: general
-version: 0.0.68
+version: 0.0.69

requirements/base.in

@@ -1,5 +1,5 @@
 -c constraints.in
-unstructured[local-inference]>=0.8.1
+unstructured[all-docs]>=0.8.1
 # Pinning click due to a unicode issue in black
 # can remove after black drops support for Python 3.6
 # ref: https://github.com/psf/black/issues/2964