markitdown-pro 1.3.4__tar.gz → 1.3.6__tar.gz

{markitdown_pro-1.3.4 → markitdown_pro-1.3.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: markitdown-pro
-Version: 1.3.4
+Version: 1.3.6
 Summary: A package that converts almost any file format to Markdown.
 Author: Developer
 Classifier: Programming Language :: Python :: 3

markitdown_pro-1.3.6/markitdown_pro/common/schemas.py

@@ -0,0 +1,17 @@
+import enum
+from io import BytesIO
+
+
+class NamedBytesIO(BytesIO):
+    def __init__(self, data: bytes, name: str):
+        super().__init__(data)
+        self.name = name
+
+
+class ExtensionCategory(enum.Enum):
+    IMAGE = "image"
+    AUDIO = "audio"
+    VIDEO = "video"
+    LIGHTWEIGHT_TEXT = "lightweight_text"
+    FORMATTED_DOCS = "formatted_docs"
+    OTHER = "other"

{markitdown_pro-1.3.4 → markitdown_pro-1.3.6}/markitdown_pro/conversion_pipeline.py

@@ -38,6 +38,7 @@ from typing import Optional
 import httpx
 
 from .common.logger import logger
+from .common.schemas import ExtensionCategory
 from .common.utils import clean_markdown, detect_extension, ensure_minimum_content
 from .converters.azure_doc_intel_wrapper import DocIntelligenceWrapper
 from .handlers.audio_handler import AudioHandler
@@ -91,73 +92,70 @@ class ConversionPipeline:
         self.markitdown_handler = MarkItDownHandler()
         self.doc_intel_wrapper = DocIntelligenceWrapper()
 
-        # Map extensions to handlers (normalized lowercase with dot)
-        self.handlers_mapping: dict[str, BaseHandler] = {
-            # region PDFs
-            ".pdf": self.pdf_handler,
-            # endregion
-            # region Audio
-            ".mp3": self.audio_handler,
-            ".wav": self.audio_handler,
-            # ".ogg": self.audio_handler,
-            # ".flac": self.audio_handler,
-            # ".m4a": self.audio_handler,
-            # ".aac": self.audio_handler,
-            # ".wma": self.audio_handler,
-            # ".webm": self.audio_handler,
-            # ".opus": self.audio_handler,
-            # endregion
-            # region Images
-            ".bmp": self.image_handler,
-            ".gif": self.image_handler,
-            # ".heic": self.image_handler,
-            ".jpeg": self.image_handler,
-            ".jpg": self.image_handler,
-            ".png": self.image_handler,
-            # ".prn": self.image_handler,
-            ".svg": self.image_handler,
-            ".tiff": self.image_handler,
-            ".webp": self.image_handler,
-            ".heif": self.image_handler,
-            # endregion
-            # region Plain/code text
-            ".txt": self.text_handler,
-            ".md": self.text_handler,
-            ".py": self.text_handler,
-            ".go": self.text_handler,
-            # endregion
-            # region Tabular (via MarkItDown)
-            ".csv": self.tabular_handler,
-            ".tsv": self.tabular_handler,
-            # endregion
-            # region Office/Spreadsheet/Slides
-            ".xls": self.tabular_handler,
-            ".xlsx": self.tabular_handler,
-            ".docx": self.office_handler,
-            # ".odt": self.office_handler,
-            # ".rtf": self.office_handler,
-            # ".ppt": self.office_handler,
-            ".pptx": self.office_handler,
-            # endregion
-            # region Markup/serialization
-            ".html": self.markup_handler,
-            ".htm": self.markup_handler,
-            ".xml": self.markup_handler,
-            ".json": self.markup_handler,
-            ".ndjson": self.markup_handler,
-            ".yaml": self.markup_handler,
-            ".yml": self.markup_handler,
-            # endregion
-            # region Others
-            ".epub": self.epub_handler,
-            ".eml": self.email_handler,
-            ".p7s": self.email_handler,
-            ".msg": self.email_handler,
-            ".pst": self.pst_handler,
-            ".ipynb": self.ipynb_handler,
-            # endregion
+        self.handlers_categories: dict[ExtensionCategory, dict[str, BaseHandler]] = {
+            ExtensionCategory.IMAGE: {
+                ".bmp": self.image_handler,
+                ".gif": self.image_handler,
+                # ".heic": self.image_handler,
+                ".jpeg": self.image_handler,
+                ".jpg": self.image_handler,
+                ".png": self.image_handler,
+                # ".prn": self.image_handler,
+                ".svg": self.image_handler,
+                ".tiff": self.image_handler,
+                ".webp": self.image_handler,
+                ".heif": self.image_handler,
+            },
+            ExtensionCategory.AUDIO: {
+                ".mp3": self.audio_handler,
+                ".wav": self.audio_handler,
+                # ".ogg": self.audio_handler,
+                # ".flac": self.audio_handler,
+                # ".m4a": self.audio_handler,
+                # ".aac": self.audio_handler,
+                # ".wma": self.audio_handler,
+                # ".webm": self.audio_handler,
+                # ".opus": self.audio_handler,
+            },
+            ExtensionCategory.LIGHTWEIGHT_TEXT: {
+                ".txt": self.text_handler,
+                ".md": self.text_handler,
+                ".py": self.text_handler,
+                ".go": self.text_handler,
+                ".csv": self.tabular_handler,
+                ".tsv": self.tabular_handler,
+                ".html": self.markup_handler,
+                ".htm": self.markup_handler,
+                ".xml": self.markup_handler,
+                ".json": self.markup_handler,
+                ".ndjson": self.markup_handler,
+                ".yaml": self.markup_handler,
+                ".yml": self.markup_handler,
+                ".epub": self.epub_handler,
+                ".eml": self.email_handler,
+                ".p7s": self.email_handler,
+                ".msg": self.email_handler,
+                ".pst": self.pst_handler,
+                ".ipynb": self.ipynb_handler,
+            },
+            ExtensionCategory.FORMATTED_DOCS: {
+                ".pdf": self.pdf_handler,
+                ".xls": self.tabular_handler,
+                ".xlsx": self.tabular_handler,
+                ".docx": self.office_handler,
+                # ".odt": self.office_handler,
+                # ".rtf": self.office_handler,
+                # ".ppt": self.office_handler,
+                ".pptx": self.office_handler,
+            },
         }
 
+        # Normalize map of extensions to handlers
+        self.handlers_mapping: dict[str, BaseHandler] = {}
+        for handlers in self.handlers_categories.values():
+            for ext, handler in handlers.items():
+                self.handlers_mapping[ext] = handler
+
     # ---------------------------------------------------------------------
     # Public APIs
     # ---------------------------------------------------------------------
@@ -313,3 +311,4 @@ class ConversionPipeline:
         else:
             logger.warning(f"ConversionPipeline: No page count support for extension '{extension}'")
             return None
+            return None

{markitdown_pro-1.3.4 → markitdown_pro-1.3.6}/markitdown_pro.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: markitdown-pro
-Version: 1.3.4
+Version: 1.3.6
 Summary: A package that converts almost any file format to Markdown.
 Author: Developer
 Classifier: Programming Language :: Python :: 3

{markitdown_pro-1.3.4 → markitdown_pro-1.3.6}/setup.py

@@ -2,7 +2,7 @@ from setuptools import find_packages, setup
 
 setup(
     name="markitdown-pro",
-    version="1.3.4",
+    version="1.3.6",
     author="Developer",
     description="A package that converts almost any file format to Markdown.",
     long_description=open("README.md").read(),

markitdown_pro-1.3.4/markitdown_pro/common/schemas.py

@@ -1,7 +0,0 @@
-from io import BytesIO
-
-
-class NamedBytesIO(BytesIO):
-    def __init__(self, data: bytes, name: str):
-        super().__init__(data)
-        self.name = name