fix: disabled checking max pages for images (#3473)

Added fix related to
#3431, which
disables checking max pages for images

Author: mackurzawa

CHANGELOG.md

@@ -1,4 +1,4 @@
-## 0.15.1-dev8
+## 0.15.1-dev9
 
 ### Enhancements
 

test_unstructured/partition/pdf_image/test_pdf.py

@@ -1372,14 +1372,18 @@ def test_analysis_artifacts_saved():
         ("pdf/layout-parser-paper-with-empty-pages.pdf", 3, True),
         ("pdf/reliance.pdf", 3, False),
         ("pdf/reliance.pdf", 2, True),
+        ("img/DA-1p.jpg", None, False),
+        ("img/DA-1p.jpg", 2, False),
     ],
 )
 def test_pdf_hi_res_max_pages_argument(filename, pdf_hi_res_max_pages, expected_error):
+    is_image = not Path(filename).suffix.endswith("pdf")
     if not expected_error:
         pdf.partition_pdf_or_image(
             filename=example_doc_path(filename),
             strategy=PartitionStrategy.HI_RES,
             pdf_hi_res_max_pages=pdf_hi_res_max_pages,
+            is_image=is_image,
         )
 
     else:
@@ -1388,4 +1392,5 @@ def test_pdf_hi_res_max_pages_argument(filename, pdf_hi_res_max_pages, expected_
                 filename=example_doc_path(filename),
                 strategy=PartitionStrategy.HI_RES,
                 pdf_hi_res_max_pages=pdf_hi_res_max_pages,
+                is_image=is_image,
             )

unstructured/__version__.py

@@ -1 +1 @@
-__version__ = "0.15.1-dev8"  # pragma: no cover
+__version__ = "0.15.1-dev9"  # pragma: no cover

unstructured/partition/pdf.py

@@ -518,7 +518,7 @@ def _get_pdf_page_number(
     elif filename:
         number_of_pages = PdfReader(filename).get_num_pages()
     else:
-        ValueError("Either 'file' or 'filename' must be provided.")
+        raise ValueError("Either 'file' or 'filename' must be provided.")
     return number_of_pages
 
 
@@ -575,9 +575,10 @@ def _partition_pdf_or_image_local(
         process_file_with_pdfminer,
     )
 
-    check_pdf_hi_res_max_pages_exceeded(
-        filename=filename, file=file, pdf_hi_res_max_pages=pdf_hi_res_max_pages
-    )
+    if not is_image:
+        check_pdf_hi_res_max_pages_exceeded(
+            filename=filename, file=file, pdf_hi_res_max_pages=pdf_hi_res_max_pages
+        )
 
     hi_res_model_name = hi_res_model_name or model_name or default_hi_res_model()
     if pdf_image_dpi is None: