polytext 0.2.3__tar.gz → 0.2.4__tar.gz

{polytext-0.2.3 → polytext-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: polytext
-Version: 0.2.3
+Version: 0.2.4
 Summary: Python utilities to simplify document files management
 Home-page: https://github.com/docsity/polytext
 Author: Matteo Senardi

{polytext-0.2.3 → polytext-0.2.4}/polytext/converter/audio_to_text.py

@@ -46,7 +46,7 @@ AUDIO_TAIL_REPETITION_THRESHOLD = float(os.getenv("AUDIO_TAIL_REPETITION_THRESHO
 AUDIO_FALLBACK_SOURCE_PATTERN = os.getenv("AUDIO_FALLBACK_SOURCE_PATTERN", "flash-lite")
 AUDIO_FALLBACK_MODEL = os.getenv("AUDIO_FALLBACK_MODEL", "gemini-3-flash-preview")
 AUDIO_FALLBACK_TEMPERATURE = float(os.getenv("AUDIO_FALLBACK_TEMPERATURE", "1.0"))
-AUDIO_FINAL_FALLBACK_MODEL = os.getenv("AUDIO_FINAL_FALLBACK_MODEL", "gemini-2.0-flash")
+AUDIO_FINAL_FALLBACK_MODEL = os.getenv("AUDIO_FINAL_FALLBACK_MODEL", "gemini-3.5-flash")
 AUDIO_FILE_UPLOAD_THRESHOLD_BYTES = 20 * 1024 * 1024
 NO_HUMAN_SPEECH_MARKER = "no human speech detected"
 

{polytext-0.2.3 → polytext-0.2.4}/polytext/converter/ocr_to_text.py

@@ -33,7 +33,7 @@ OCR_TAIL_REPETITION_THRESHOLD = float(os.getenv("OCR_TAIL_REPETITION_THRESHOLD",
 OCR_FALLBACK_SOURCE_PATTERN = os.getenv("OCR_FALLBACK_SOURCE_PATTERN", "flash-lite-preview")
 OCR_FALLBACK_MODEL = os.getenv("OCR_FALLBACK_MODEL", "gemini-3-flash-preview")
 OCR_FALLBACK_TEMPERATURE = float(os.getenv("OCR_FALLBACK_TEMPERATURE", "1.0"))
-OCR_FINAL_FALLBACK_MODEL = os.getenv("OCR_FINAL_FALLBACK_MODEL", "gemini-2.0-flash")
+OCR_FINAL_FALLBACK_MODEL = os.getenv("OCR_FINAL_FALLBACK_MODEL", "gemini-3.5-flash")
 
 
 def compress_and_convert_image(input_path: str, target_size=1):

{polytext-0.2.3 → polytext-0.2.4}/polytext/loader/base.py

@@ -55,6 +55,18 @@ def _read_bool_env(name: str, default: bool = False) -> bool:
     return value.strip().lower() in {"1", "true", "yes", "y", "on"}
 
 
+def _capture_exception_for_sentry(error: Exception) -> None:
+    try:
+        import sentry_sdk
+    except ImportError:
+        return
+
+    try:
+        sentry_sdk.capture_exception(error)
+    except Exception:
+        return
+
+
 class BaseLoader:
     def __init__(self, markdown_output=True, llm_api_key=None, provider: str = "google", temp_dir: str = "temp",
                  ocr_model: str = "gpt-5-mini", timeout_minutes: int | None = None,
@@ -153,19 +165,14 @@ class BaseLoader:
         try:
             response = self.run_loader_class(loader_class=loader_class, input_list=input_list)
         except EmptyDocument as e:
-            logger.info(f"Empty document encountered: {e.message}")
             if e.code in LLM_OUTPUT_ERROR_CODES:
-                logger.exception(
-                    "Raising LoaderError: status=422 code=%s original_empty_document_code=%s message=%s",
-                    LLM_OUTPUT_ERROR_CODES[e.code],
-                    e.code,
-                    e.message,
-                )
+                _capture_exception_for_sentry(e)
                 raise LoaderError(
                     message=e.message,
                     status=422,
                     code=LLM_OUTPUT_ERROR_CODES[e.code],
                 ) from e
+            logger.info(f"Empty document encountered: {e.message}")
             if self.fallback_ocr:
                 loader_class = self.init_loader_class(input=first_file_url, storage_client=storage_client,
                                                       llm_api_key=self.llm_api_key, is_document_fallback=True, **kwargs)

{polytext-0.2.3 → polytext-0.2.4}/polytext.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: polytext
-Version: 0.2.3
+Version: 0.2.4
 Summary: Python utilities to simplify document files management
 Home-page: https://github.com/docsity/polytext
 Author: Matteo Senardi

{polytext-0.2.3 → polytext-0.2.4}/setup.py

@@ -51,7 +51,7 @@ def get_requirements(*requirements_file):
 
 setup(
     name='polytext',
-    version='0.2.3',
+    version='0.2.4',
     url='https://github.com/docsity/polytext',
     # download_url='https://github.com/pualien/py-polytext/archive/0.1.23.tar.gz',
     license='MIT',

{polytext-0.2.3 → polytext-0.2.4}/tests/test_base_loader_error_mapping.py

@@ -1,5 +1,5 @@
 import unittest
-from unittest.mock import patch
+from unittest.mock import Mock, patch
 
 from polytext.exceptions import EmptyDocument, LoaderError
 from polytext.loader.base import BaseLoader
@@ -43,19 +43,21 @@ class TestBaseLoaderErrorMapping(unittest.TestCase):
                     )
                 )
 
-                with patch("polytext.loader.base.logger.exception") as mock_exception:
-                    with self.assertRaises(LoaderError) as error_context:
-                        loader.get_text(["dummy.txt"])
+                sentry_sdk = Mock()
+                with patch("polytext.loader.base.logger.info") as mock_info:
+                    with patch("polytext.loader.base.logger.exception") as mock_exception:
+                        with patch.dict("sys.modules", {"sentry_sdk": sentry_sdk}):
+                            with self.assertRaises(LoaderError) as error_context:
+                                loader.get_text(["dummy.txt"])
 
                 error = error_context.exception
                 self.assertEqual(error.status, 422)
                 self.assertEqual(error.code, expected_loader_code)
                 self.assertEqual(error.message, f"diagnostic failure {empty_document_code}")
-                mock_exception.assert_called_once()
-                self.assertIn("Raising LoaderError", mock_exception.call_args.args[0])
-                self.assertEqual(mock_exception.call_args.args[1], expected_loader_code)
-                self.assertEqual(mock_exception.call_args.args[2], empty_document_code)
-                self.assertEqual(mock_exception.call_args.args[3], f"diagnostic failure {empty_document_code}")
+                mock_info.assert_not_called()
+                mock_exception.assert_not_called()
+                sentry_sdk.capture_exception.assert_called_once()
+                self.assertIs(sentry_sdk.capture_exception.call_args.args[0], error.__cause__)
 
     def test_empty_or_too_short_documents_still_return_empty_response(self):
         loader = _FakeBaseLoader(

{polytext-0.2.3 → polytext-0.2.4}/tests/test_get_ocr_from_image.py

@@ -38,7 +38,7 @@ def main():
 
     # local_file_path = "/Users/marcodelgiudice/Projects/polytext/IMG_9695.jpg"
     # local_file_path = "/Users/marcodelgiudice/Projects/polytext/IMG_9701.jpg"
-    local_file_path = "/Users/marcodelgiudice/Projects/polytext/chimicaformula.png"
+    local_file_path = "/Users/marcodelgiudice/Projects/polytext/gm1.png"
 
     try:
         start = time.time()

{polytext-0.2.3 → polytext-0.2.4}/tests/test_youtube_transcript.py

@@ -32,9 +32,9 @@ url = 'https://www.youtube.com/watch?v=L4as3tks4Js'  # basement alberto angela
 
 # url = 'https://www.youtube.com/watch?v=UabBYexBD4k'  # INM RAG 11 minuti, completato in 26 secondi con successo con gemini-3.1-flash-lite
 
-url = 'https://www.youtube.com/watch?v=96jN2OCOfLs'  # Vibe coding 30 minuti, completato in 150 secondi con successo con gemini-3-flash-preview (160k token in input, 7k in output), 3.1-flash-lite ha raggiunto i max tokens in output (50k) probabile repetition
+#url = 'https://www.youtube.com/watch?v=96jN2OCOfLs'  # Vibe coding 30 minuti, completato in 150 secondi con successo con gemini-3-flash-preview (160k token in input, 7k in output), 3.1-flash-lite ha raggiunto i max tokens in output (50k) probabile repetition
 
-# url = 'https://www.youtube.com/watch?v=HGfsGvmRaaw'  # barbero2 50 minuti, fallito, RECITATION in tutti e 3 i modelli (275k token in input)
+url = 'https://www.youtube.com/watch?v=HGfsGvmRaaw'  # barbero2 50 minuti, fallito, RECITATION in tutti e 3 i modelli (275k token in input)
 
 # url = 'https://www.youtube.com/watch?v=CM2CkNU9xR0'  # google antigravity 27 minuti, completato in 39 secondi con successo con gemini-3.1-flash-lite (146k token in input, 6k token in output)