feat: added new exception type to epub conversions (#4052)

Added UnprocessableEpubError to better handle the case when incoming
epub file is actually damanged which makes pandoc lib crash with exit
code 64.

Author: mateuszkuprowski

.gitignore

@@ -212,3 +212,5 @@ annotated/
 .aider*
 pcaps
 python-output
+
+.vs/

CHANGELOG.md

@@ -1,3 +1,12 @@
+## 0.18.6-dev0
+
+### Enhancements
+
+### Features
+
+### Fixes
+- **Improved epub partition errors** EPUB partition will now produce new type of error on unprocessable files.
+
 ## 0.18.5
 
 ### Enhancements

test_unstructured/partition/test_api.py

@@ -43,7 +43,9 @@ def test_partition_via_api_with_filename_correctly_calls_sdk(
 
     elements = partition_via_api(filename=example_doc_path("eml/fake-email.eml"))
 
-    partition_mock_.assert_called_once_with(*expected_call_)
+    partition_mock_.assert_called_once_with(
+        expected_call_[0], request=expected_call_[1], retries=expected_call_[2]
+    )
     assert isinstance(partition_mock_.call_args_list[0].args[0], General)
     assert len(elements) == 1
     assert elements[0] == NarrativeText("This is a test email to use for unit tests.")
@@ -66,7 +68,11 @@ def test_partition_via_api_with_file_correctly_calls_sdk(
     modified_expected_call = expected_call_[:]
     modified_expected_call[1].partition_parameters.files.content = f
 
-    partition_mock_.assert_called_once_with(*modified_expected_call)
+    partition_mock_.assert_called_once_with(
+        modified_expected_call[0],
+        request=modified_expected_call[1],
+        retries=modified_expected_call[2],
+    )
     assert isinstance(partition_mock_.call_args_list[0].args[0], General)
     assert len(elements) == 1
     assert elements[0] == NarrativeText("This is a test email to use for unit tests.")
@@ -87,7 +93,11 @@ def test_partition_via_api_warns_with_file_and_filename_and_calls_sdk(
     modified_expected_call = expected_call_[:]
     modified_expected_call[1].partition_parameters.files.content = f
 
-    partition_mock_.assert_called_once_with(*modified_expected_call)
+    partition_mock_.assert_called_once_with(
+        modified_expected_call[0],
+        request=modified_expected_call[1],
+        retries=modified_expected_call[2],
+    )
     assert "WARNING" in caplog.text
     assert "The file_filename kwarg will be deprecated" in caplog.text
 

unstructured/__version__.py

@@ -1 +1 @@
-__version__ = "0.18.5"  # pragma: no cover
+__version__ = "0.18.6-dev0"  # pragma: no cover

unstructured/errors.py

@@ -9,3 +9,9 @@ def __init__(self, document_pages: int, pdf_hi_res_max_pages: int):
             f"pages={document_pages}, maximum={pdf_hi_res_max_pages}."
         )
         super().__init__(self.message)
+
+
+class UnprocessableEntityError(Exception):
+    """Error raised when a file is not valid."""
+
+    pass

unstructured/file_utils/file_conversion.py

@@ -1,9 +1,11 @@
 from __future__ import annotations
 
 import os
+import re
 import tempfile
 from typing import IO
 
+from unstructured.errors import UnprocessableEntityError
 from unstructured.partition.common.common import exactly_one
 from unstructured.utils import requires_dependencies
 
@@ -14,7 +16,9 @@ def convert_file_to_text(filename: str, source_format: str, target_format: str)
     import pypandoc
 
     try:
-        text = pypandoc.convert_file(filename, target_format, format=source_format, sandbox=True)
+        text: str = pypandoc.convert_file(
+            filename, target_format, format=source_format, sandbox=True
+        )
     except FileNotFoundError as err:
         msg = (
             f"Error converting the file to text. Ensure you have the pandoc package installed on"
@@ -23,6 +27,14 @@ def convert_file_to_text(filename: str, source_format: str, target_format: str)
         )
         raise FileNotFoundError(msg)
     except RuntimeError as err:
+        err_str = str(err)
+        if source_format == "epub" and (
+            "Couldn't extract ePub file" in err_str
+            or "No entry on path" in err_str
+            or re.search(r"exitcode ['\"]?64['\"]?", err_str)
+        ):
+            raise UnprocessableEntityError(f"Invalid EPUB file: {err_str}")
+
         supported_source_formats, _ = pypandoc.get_pandoc_formats()
 
         if source_format == "rtf" and source_format not in supported_source_formats:

unstructured/partition/epub.py

@@ -46,7 +46,9 @@ def partition_epub(
     last_modified = get_last_modified_date(filename) if filename else None
 
     html_text = convert_file_to_html_text_using_pandoc(
-        source_format="epub", filename=filename, file=file
+        source_format="epub",
+        filename=filename,
+        file=file,
     )
 
     return partition_html(