Add get_page_range form util helper

Author: awalker4

_test_unstructured_client/unit/test_split_pdf_hook.py

@@ -9,6 +9,7 @@
 from unstructured_client._hooks.custom.form_utils import (
     PARTITION_FORM_CONCURRENCY_LEVEL_KEY,
     PARTITION_FORM_STARTING_PAGE_NUMBER_KEY,
+    PARTITION_FORM_PAGE_RANGE_KEY,
 )
 from unstructured_client._hooks.custom.split_pdf_hook import (
     DEFAULT_CONCURRENCY_LEVEL,
@@ -398,3 +399,25 @@ def test_unit_get_starting_page_number(starting_page_number, expected_result):
         fallback_value=DEFAULT_STARTING_PAGE_NUMBER,
     )
     assert result == expected_result
+
+
+@pytest.mark.parametrize(
+    "page_range, expected_result",
+    [
+        (["2", "5"], (2, 5)),  # Valid range
+        (["2", "100"], (2, 20)), # End too high
+        (["-50", "5"], (1, 5)), # Start too low
+        (None, (1, 20)), # Range not specified
+        (["foo", "foo"], (1, 20)), # Parse error
+    ],
+)
+def test_unit_get_page_range_returns_valid_range(page_range, expected_result):
+    """Test get_page_range method with different inputs.
+    Ranges that are out of bounds for a 20 page doc will be adjusted."""
+    form_data = {"split_pdf_page_range[]": page_range}
+    result = form_utils.get_page_range(
+        form_data,
+        key=PARTITION_FORM_PAGE_RANGE_KEY,
+        max_pages=20,
+    )
+    assert result == expected_result

src/unstructured_client/_hooks/custom/form_utils.py

@@ -13,10 +13,48 @@
 
 PARTITION_FORM_FILES_KEY = "files"
 PARTITION_FORM_SPLIT_PDF_PAGE_KEY = "split_pdf_page"
+PARTITION_FORM_PAGE_RANGE_KEY = "split_pdf_page_range[]"
 PARTITION_FORM_STARTING_PAGE_NUMBER_KEY = "starting_page_number"
 PARTITION_FORM_CONCURRENCY_LEVEL_KEY = "split_pdf_concurrency_level"
 
 
+def get_page_range(form_data: FormData, key: str, max_pages: int) -> tuple[int, int]:
+    """Retrieves the split page range from the given form data.
+
+    If the range is invalid or outside the bounds of the page count,
+    returns (1, num_pages), i.e. the full range.
+
+    Args:
+        form_data: The form data containing the page range
+        key: The key to look for in the form data.
+
+    Returns:
+        The range of pages to send in the request in the form (start, end)
+    """
+    try:
+        _page_range = form_data.get(key)
+
+        if _page_range is not None:
+            page_range = (int(_page_range[0]), int(_page_range[1]))
+        else:
+            page_range = (1, max_pages)
+
+    except (ValueError, IndexError):
+        logger.warning(
+            "'%s' is not a valid page range. Selecting default range (1 to %d).",
+            _page_range,
+            max_pages,
+        )
+        page_range = (1, max_pages)
+
+    if page_range[0] < 1 or page_range[1] > max_pages:
+        new_page_range = (max(page_range[0], 1), min(page_range[1], max_pages))
+        logger.warning(f"Page range {page_range} is out of bounds, setting to {new_page_range}.")
+        page_range = new_page_range
+
+    return page_range
+
+
 def get_starting_page_number(form_data: FormData, key: str, fallback_value: int) -> int:
     """Retrieves the starting page number from the given form data.