natural-pdf 0.1.6__py3-none-any.whl → 0.1.8__py3-none-any.whl

natural_pdf/utils/text_extraction.py

@@ -2,7 +2,7 @@
 import logging
 from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple
 
-from pdfplumber.utils.geometry import get_bbox_overlap, merge_bboxes, objects_to_bbox
+from pdfplumber.utils.geometry import get_bbox_overlap, merge_bboxes, objects_to_bbox, cluster_objects
 from pdfplumber.utils.text import TEXTMAP_KWARGS, WORD_EXTRACTOR_KWARGS, chars_to_textmap
 
 if TYPE_CHECKING:
@@ -11,6 +11,57 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
+def _get_layout_kwargs(
+    layout_context_bbox: Optional[Tuple[float, float, float, float]] = None,
+    user_kwargs: Optional[Dict[str, Any]] = None,
+) -> Dict[str, Any]:
+    """
+    Prepares the keyword arguments for pdfplumber's chars_to_textmap based
+    on defaults, context bbox, and allowed user overrides.
+    """
+    # 1. Start with an empty dict for layout kwargs 
+    layout_kwargs = {}
+    
+    # Build allowed keys set without trying to copy the constants
+    allowed_keys = set(TEXTMAP_KWARGS) | set(WORD_EXTRACTOR_KWARGS)
+
+    # Add common, well-known default values
+    layout_kwargs.update({
+        'x_tolerance': 5,
+        'y_tolerance': 5,
+        'x_density': 7.25,
+        'y_density': 13,
+        'mode': 'box',
+        'min_words_vertical': 1,
+        'min_words_horizontal': 1,
+    })
+    
+    # 2. Apply context if provided
+    if layout_context_bbox:
+        ctx_x0, ctx_top, ctx_x1, ctx_bottom = layout_context_bbox
+        layout_kwargs["layout_width"] = ctx_x1 - ctx_x0
+        layout_kwargs["layout_height"] = ctx_bottom - ctx_top
+        layout_kwargs["x_shift"] = ctx_x0
+        layout_kwargs["y_shift"] = ctx_top
+        # Add layout_bbox itself
+        layout_kwargs["layout_bbox"] = layout_context_bbox
+
+    # 3. Apply user overrides (only for allowed keys)
+    if user_kwargs:
+        for key, value in user_kwargs.items():
+            if key in allowed_keys:
+                layout_kwargs[key] = value
+            elif key == 'layout': # Always allow layout flag
+                layout_kwargs[key] = value
+            else:
+                logger.warning(f"Ignoring unsupported layout keyword argument: '{key}'")
+
+    # 4. Ensure layout flag is present, defaulting to True
+    if 'layout' not in layout_kwargs:
+        layout_kwargs['layout'] = True
+
+    return layout_kwargs
+
 def filter_chars_spatially(
     char_dicts: List[Dict[str, Any]],
     exclusion_regions: List["Region"],
@@ -116,80 +167,60 @@ def filter_chars_spatially(
 
 def generate_text_layout(
     char_dicts: List[Dict[str, Any]],
-    layout_context_bbox: Tuple[float, float, float, float],
-    user_kwargs: Dict[str, Any],
+    layout_context_bbox: Optional[Tuple[float, float, float, float]] = None,
+    user_kwargs: Optional[Dict[str, Any]] = None,
 ) -> str:
     """
-    Takes a list of filtered character dictionaries and generates
-    text output using pdfplumber's layout engine.
+    Generates a string representation of text from character dictionaries,
+    attempting to reconstruct layout using pdfplumber's utilities.
 
     Args:
-        char_dicts: The final list of character dictionaries to include.
-        layout_context_bbox: The bounding box (x0, top, x1, bottom) to use for
-                             calculating default layout width/height/shifts.
-        user_kwargs: Dictionary of user-provided keyword arguments.
+        char_dicts: List of character dictionary objects.
+        layout_context_bbox: Optional bounding box for layout context.
+        user_kwargs: User-provided kwargs, potentially overriding defaults.
 
     Returns:
-        The formatted text string.
+        String representation of the text.
     """
-    if not char_dicts:
-        logger.debug("generate_text_layout: No characters provided.")
-        return ""
-
-    # Prepare layout kwargs, prioritizing user input
-    layout_kwargs = {}
-    allowed_keys = set(WORD_EXTRACTOR_KWARGS) | set(TEXTMAP_KWARGS)
-    for key, value in user_kwargs.items():
-        if key in allowed_keys:
-            layout_kwargs[key] = value
+    # --- Filter out invalid char dicts early ---
+    initial_count = len(char_dicts)
+    valid_char_dicts = [c for c in char_dicts if isinstance(c.get("text"), str)]
+    filtered_count = initial_count - len(valid_char_dicts)
+    if filtered_count > 0:
+        logger.debug(
+            f"generate_text_layout: Filtered out {filtered_count} char dicts with non-string/None text."
+        )
 
-    # Default to layout=True unless explicitly False
-    use_layout = layout_kwargs.get("layout", True)  # Default to layout if called
-    layout_kwargs["layout"] = use_layout
+    if not valid_char_dicts:  # Return empty if no valid chars remain
+        logger.debug("generate_text_layout: No valid character dicts found after filtering.")
+        return ""
 
-    if use_layout:
-        ctx_x0, ctx_top, ctx_x1, ctx_bottom = layout_context_bbox
-        ctx_width = ctx_x1 - ctx_x0
-        ctx_height = ctx_bottom - ctx_top
-
-        # Set layout defaults based on context_bbox if not overridden by user
-        if "layout_bbox" not in layout_kwargs:
-            layout_kwargs["layout_bbox"] = layout_context_bbox
-        # Only set default layout_width if neither width specifier is present
-        if "layout_width_chars" not in layout_kwargs and "layout_width" not in layout_kwargs:
-            layout_kwargs["layout_width"] = ctx_width
-        if "layout_height" not in layout_kwargs:
-            layout_kwargs["layout_height"] = ctx_height
-        # Adjust shift based on context's top-left corner
-        if "x_shift" not in layout_kwargs:
-            layout_kwargs["x_shift"] = ctx_x0
-        if "y_shift" not in layout_kwargs:
-            layout_kwargs["y_shift"] = ctx_top
+    # Prepare layout arguments
+    layout_kwargs = _get_layout_kwargs(layout_context_bbox, user_kwargs)
+    use_layout = layout_kwargs.pop("layout", True)  # Extract layout flag, default True
 
-        logger.debug(
-            f"generate_text_layout: Calling chars_to_textmap with {len(char_dicts)} chars and kwargs: {layout_kwargs}"
-        )
-        try:
-            # Sort final list by reading order before passing to textmap
-            # TODO: Make sorting key dynamic based on layout_kwargs directions?
-            char_dicts.sort(key=lambda c: (c.get("top", 0), c.get("x0", 0)))
-            textmap = chars_to_textmap(char_dicts, **layout_kwargs)
-            result = textmap.as_string
-        except Exception as e:
-            logger.error(
-                f"generate_text_layout: Error calling chars_to_textmap: {e}", exc_info=True
-            )
-            logger.warning(
-                "generate_text_layout: Falling back to simple character join due to layout error."
-            )
-            # Ensure chars are sorted before fallback join
-            fallback_chars = sorted(char_dicts, key=lambda c: (c.get("top", 0), c.get("x0", 0)))
-            result = "".join(c.get("text", "") for c in fallback_chars)
-    else:
+    if not use_layout:
         # Simple join if layout=False
-        logger.debug("generate_text_layout: Using simple join (layout=False).")
-        # Sort by document order for simple join as well
-        char_dicts.sort(key=lambda c: (c.get("page_number", 0), c.get("top", 0), c.get("x0", 0)))
-        result = "".join(c.get("text", "") for c in char_dicts)
+        logger.debug("generate_text_layout: Using simple join (layout=False requested).")
+        # Sort before joining if layout is off
+        valid_char_dicts.sort(key=lambda c: (c.get("top", 0), c.get("x0", 0)))
+        result = "".join(c.get("text", "") for c in valid_char_dicts)  # Use valid chars
+        return result
+
+    try:
+        # Sort chars primarily by top, then x0 before layout analysis
+        # This helps pdfplumber group lines correctly
+        valid_char_dicts.sort(key=lambda c: (c.get("top", 0), c.get("x0", 0)))
+        textmap = chars_to_textmap(valid_char_dicts, **layout_kwargs)
+        result = textmap.as_string
+    except Exception as e:
+        # Fallback to simple join on error
+        logger.error(f"generate_text_layout: Error calling chars_to_textmap: {e}", exc_info=False)
+        logger.warning(
+            "generate_text_layout: Falling back to simple character join due to layout error."
+        )
+        # Fallback already has sorted characters if layout was attempted
+        # Need to use the valid_char_dicts here too
+        result = "".join(c.get("text", "") for c in valid_char_dicts)
 
     return result

natural_pdf/utils/tqdm_utils.py

@@ -0,0 +1,43 @@
+import sys
+import os
+
+# Default to standard tqdm
+try:
+    from tqdm.std import tqdm as selected_tqdm
+except ImportError:
+    # Basic fallback if even std is missing (though unlikely)
+    def selected_tqdm(*args, **kwargs):
+        iterable = args[0] if args else None
+        if iterable:
+            return iterable
+        return None # Simple passthrough if no iterable
+
+# Try to detect notebook environment
+try:
+    # Check 1: Are we running in an IPython kernel?
+    from IPython import get_ipython
+    ipython = get_ipython()
+    if ipython and 'IPKernelApp' in ipython.config:
+        # Check 2: Is it likely a notebook UI (Jupyter Notebook/Lab, VSCode, etc.)?
+        # This checks for common indicators. Might not be foolproof.
+        if 'VSCODE_PID' in os.environ or ('ipykernel' in sys.modules and 'spyder' not in sys.modules):
+             # Check 3: Can we import notebook version?
+            try:
+                from tqdm.notebook import tqdm as notebook_tqdm
+                selected_tqdm = notebook_tqdm # Use notebook version
+            except ImportError:
+                pass # Stick with std if notebook version missing
+except ImportError:
+    pass # Stick with std if IPython not available
+
+def get_tqdm():
+    """Returns the tqdm class best suited for the detected environment."""
+    return selected_tqdm
+
+# Example usage (for testing):
+if __name__ == '__main__':
+    import time
+    tqdm_instance = get_tqdm()
+    print(f"Using tqdm class: {tqdm_instance}")
+    for i in tqdm_instance(range(10), desc="Testing tqdm"):
+        time.sleep(0.1) 

natural_pdf/utils/visualization.py

@@ -192,7 +192,7 @@ def merge_images_with_legend(
     if not legend:
         return image  # Return original image if legend is None or empty
 
-    bg_color = (255, 255, 255, 255) # Always use white for the merged background
+    bg_color = (255, 255, 255, 255)  # Always use white for the merged background
 
     if position == "right":
         # Create a new image with extra width for the legend

{natural_pdf-0.1.6.dist-info → natural_pdf-0.1.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: natural-pdf
-Version: 0.1.6
+Version: 0.1.8
 Summary: A more intuitive interface for working with PDFs
 Author-email: Jonathan Soma <jonathan.soma@gmail.com>
 License-Expression: MIT
@@ -23,7 +23,6 @@ Provides-Extra: haystack
 Requires-Dist: haystack-ai; extra == "haystack"
 Requires-Dist: chroma-haystack; extra == "haystack"
 Requires-Dist: sentence-transformers; extra == "haystack"
-Requires-Dist: protobuf<4; extra == "haystack"
 Requires-Dist: natural-pdf[core-ml]; extra == "haystack"
 Provides-Extra: easyocr
 Requires-Dist: easyocr; extra == "easyocr"
@@ -45,6 +44,10 @@ Requires-Dist: natural-pdf[core-ml]; extra == "docling"
 Provides-Extra: llm
 Requires-Dist: openai>=1.0; extra == "llm"
 Requires-Dist: pydantic; extra == "llm"
+Provides-Extra: classification
+Requires-Dist: sentence-transformers; extra == "classification"
+Requires-Dist: timm; extra == "classification"
+Requires-Dist: natural-pdf[core-ml]; extra == "classification"
 Provides-Extra: test
 Requires-Dist: pytest; extra == "test"
 Provides-Extra: dev
@@ -71,15 +74,19 @@ Requires-Dist: natural-pdf[qa]; extra == "all"
 Requires-Dist: natural-pdf[ocr-export]; extra == "all"
 Requires-Dist: natural-pdf[docling]; extra == "all"
 Requires-Dist: natural-pdf[llm]; extra == "all"
+Requires-Dist: natural-pdf[classification]; extra == "all"
 Requires-Dist: natural-pdf[test]; extra == "all"
 Provides-Extra: core-ml
 Requires-Dist: torch; extra == "core-ml"
 Requires-Dist: torchvision; extra == "core-ml"
-Requires-Dist: transformers; extra == "core-ml"
+Requires-Dist: transformers[sentencepiece]; extra == "core-ml"
 Requires-Dist: huggingface_hub; extra == "core-ml"
 Provides-Extra: ocr-export
 Requires-Dist: ocrmypdf; extra == "ocr-export"
 Requires-Dist: pikepdf; extra == "ocr-export"
+Provides-Extra: export-extras
+Requires-Dist: jupytext; extra == "export-extras"
+Requires-Dist: nbformat; extra == "export-extras"
 Dynamic: license-file
 
 # Natural PDF

{natural_pdf-0.1.6.dist-info → natural_pdf-0.1.8.dist-info}/RECORD

@@ -1,4 +1,4 @@
-docs/index.md,sha256=P1kXZc8aefnxH0bBjvBgj1o3puRiezjUiBLqS4bcUhM,4889
+docs/index.md,sha256=FG4MYQs-gUR16NQ4XF0AVoQeLuykLeY8XxNwW3h-qUM,5572
 docs/api/index.md,sha256=4bn8nYklWJuNDrnY-Kt7sf7IejeAEDhcnqYmjH9GJTA,22405
 docs/assets/favicon.png,sha256=nxca8jM2Y4GxZKzkmagUHO1GpUREK-GRA5LEFue9OOU,284
 docs/assets/favicon.svg,sha256=nxca8jM2Y4GxZKzkmagUHO1GpUREK-GRA5LEFue9OOU,284
@@ -8,16 +8,19 @@ docs/assets/social-preview.png,sha256=AvyzzM8dC0j5SPFF63bvQrxU4GE1f9j-GUNUv0oA9t
 docs/assets/social-preview.svg,sha256=AvyzzM8dC0j5SPFF63bvQrxU4GE1f9j-GUNUv0oA9ts,1085
 docs/assets/javascripts/custom.js,sha256=0NVHGprwiLPFYdYunJcHjOphzk_EhBSNuOUz5Uzdv_k,594
 docs/assets/stylesheets/custom.css,sha256=PbTp3k77gzUBUQQ01pDXzpNwo4wUv3aJD-SMBQvQItY,1156
+docs/categorizing-documents/index.md,sha256=tgKfv3DidZysrFhaOEM-FiIVDAzNPPnK02sKaE5pE2I,8196
+docs/data-extraction/index.md,sha256=LwQ2MJVI5u5ELI51Iq0WUdDo5sl_s18GWG_cBABI8fQ,3430
 docs/document-qa/index.ipynb,sha256=MXJoFhi8TUKK6ZnRFiUBglLGpMbzwdb7LJYfzw8Gp48,528713
 docs/document-qa/index.md,sha256=mwuO4tothg0OzBXewnj73QEJu46Udq7f1pQBYrKOHwM,2131
-docs/element-selection/index.ipynb,sha256=-7PwKw1RbPlZ4stzN1Rd1GJ8mwjOD4ySsLcpqVX7chc,1193628
-docs/element-selection/index.md,sha256=_1P8vI64Y0aSVwUzdRJD4ayb80BJWBLED9TvVpveFx8,6979
+docs/element-selection/index.ipynb,sha256=WuKd3bTTOnzBDfbuzkxmJxO6EzM9RAkFXoF0U3-8qRA,1223398
+docs/element-selection/index.md,sha256=ZUkOD6VVK11K6WQ86FPnTeeco27PrFWtkObKw8j6Fok,7867
+docs/finetuning/index.md,sha256=Ur3zqSaR0X8PvBCSyI7cFiDv5qZ6Jtv4omBKXCKAzEk,9200
 docs/installation/index.md,sha256=nd4RZrQFR8_vv7Xm3xAzp7z-CQQr9ffAcGa7yuEYn2U,1594
 docs/interactive-widget/index.ipynb,sha256=zY1rz5N34OUW-OtgcbI6iiOjlIJqXjVcx9OoNWMjuyU,293111
 docs/interactive-widget/index.md,sha256=tZbq0uYI7Zwo9mLbhXpqeBriuAjazkIyEJeP-jasJ-Q,259
 docs/layout-analysis/index.ipynb,sha256=dkS_-cu-KGir5G2LGRcxBThKnW0dfA5nPPnwpoYGFtU,1869093
 docs/layout-analysis/index.md,sha256=ZnH5yd7B_eOLgGxW_4rNlzQs4Tn3Xx1cK3jX43CSpSM,5390
-docs/ocr/index.md,sha256=uuzTqcAgUmMN7jZVq8VkVcbRDHn8Yg2nJVvHJ-bDK-Y,8177
+docs/ocr/index.md,sha256=BR8a3_X6zng5yAo8O8isOBhb2Gm9hM9FIasc58aYF78,11137
 docs/pdf-navigation/index.ipynb,sha256=h6yew0HePXK1_c5FmETqzjBQceUBT0MU-vnXx_y91mo,8018
 docs/pdf-navigation/index.md,sha256=P3b3tsmOcmRtnfRxpsMeTgwm7vApnH_4le_QIwJd51M,2391
 docs/regions/index.ipynb,sha256=5A-N5A4v4lcXNptOAeI4i7i9Gx66To-Yus8B816dHBk,1303347
@@ -28,38 +31,38 @@ docs/text-analysis/index.ipynb,sha256=iaup8pcQXGp0ZK3IWi-HHssQLdIzWYGYfvZK5i8yjj
 docs/text-analysis/index.md,sha256=02pfZemOgV37izV7H-XzKmHu7AedDKLidQ-sKhYaMVw,3527
 docs/text-extraction/index.ipynb,sha256=809y9ZamXT3bc3GhwwFyoDnlyEpO-kUZ3tIsZZWyrj8,2537087
 docs/text-extraction/index.md,sha256=b1KfQpvIEelc8cPbFETUnK92az7iB4b7-LqK2DRH8vw,6985
-docs/tutorials/01-loading-and-extraction.ipynb,sha256=-9hFAVQtHmuXsR9Ge3A80wKr-t9wxouAAlW4_Iotdwo,544610
+docs/tutorials/01-loading-and-extraction.ipynb,sha256=2vGLM1_2_Xcpn32HvMLXj_Ro8w4HPofSZNpxZ1qPtL8,520140
 docs/tutorials/01-loading-and-extraction.md,sha256=g40J8GhKz-ikM2URj5MqIatKKj4l5kTFozHeVjxDJQA,2191
-docs/tutorials/02-finding-elements.ipynb,sha256=k1CSz47_atA9D6DXfQzVS64t5-L-KjssU2VuFvdy7oU,524374
+docs/tutorials/02-finding-elements.ipynb,sha256=yVW3B578mKXkFUWJQnBaDB0SlnNodROjemMbdx-LWBw,524506
 docs/tutorials/02-finding-elements.md,sha256=qOkjcWUzem05of54aKzKvy-MMzRX_S4CyZisVV-73QM,4162
-docs/tutorials/03-extracting-blocks.ipynb,sha256=1UjdP0j3kPCE3aU8p1jBCBqflG-xRLli2Ltx80DhOVk,260729
+docs/tutorials/03-extracting-blocks.ipynb,sha256=qifBv5bsKcZIQVQAHtl84GqD6Wy-IZiUMkSXURCu3ug,263329
 docs/tutorials/03-extracting-blocks.md,sha256=_kqvhk6rSL7cGp2MSwTJk8LYlJGbK_r_umnCSBdR8XU,1665
-docs/tutorials/04-table-extraction.ipynb,sha256=u92Wppw1qHG__Mx3ZKtETm4AWuGF8X-Ln3kvmF8zCSo,3973
+docs/tutorials/04-table-extraction.ipynb,sha256=Jj2OzN32I5z1_gfMVgdr2GGyEgbWTgI7harwMWfHxYc,4089
 docs/tutorials/04-table-extraction.md,sha256=4q4v17VX8K-ZBtWYy0nbWPccyqB_ybd5Vl_IROmxz6Q,2130
-docs/tutorials/05-excluding-content.ipynb,sha256=oSg8ll_nuWOfQHGLp0fNKVeyYyn_L8a-F7HJADjjdq8,336857
+docs/tutorials/05-excluding-content.ipynb,sha256=EaZwfDJK3BUghY1iwQ4qR8Z9nXf9e8QUeHxvJmZ3xsw,336933
 docs/tutorials/05-excluding-content.md,sha256=U52SPlc5knqxiyhRokmxrj06T54r2ENyTfP7BMGykhY,3907
-docs/tutorials/06-document-qa.ipynb,sha256=Facyqns8jw2bTvsOSbNnsLskFH8kg1JTz4kmJ16dpcE,10303
+docs/tutorials/06-document-qa.ipynb,sha256=sGesxP26CMSD2GD-47dXq7EnqK3tlEDzM-uu7sZVR2E,10421
 docs/tutorials/06-document-qa.md,sha256=PzPPgw0Rkkfe6sfz3XyKD9S9JbQ40qf4bDzCBvwH1P0,3026
-docs/tutorials/07-layout-analysis.ipynb,sha256=tdNnMro1V66YPx0h96HZnujSm-zDpy7o78euQix4lyU,559517
+docs/tutorials/07-layout-analysis.ipynb,sha256=DgyocqPRt9Rxsz-Sjgi153MvvnoDF3Vpsyhq27N72sE,571321
 docs/tutorials/07-layout-analysis.md,sha256=NAYVzJTecDnXjo_isbPCSUBSn3c-xM1tELct1Zn5GmI,2533
-docs/tutorials/07-working-with-regions.ipynb,sha256=s4BFKKbKUemmURCpg6j91rNI8eFFOJUgxY4QN4alK4I,69584
+docs/tutorials/07-working-with-regions.ipynb,sha256=JMUnjQ_tCBqs4dWIyZ2jNHQCnJkwAzTJuxQVRGBqLqI,67945
 docs/tutorials/07-working-with-regions.md,sha256=oanbTFSQ-topAVd9kjfkaPiMjHcx6Y8cqyxVbmxLhgs,4365
-docs/tutorials/08-spatial-navigation.ipynb,sha256=jfwF6OHLvrMvaaknp-9AfUvr-pPXjPljUyGnFKF9wsw,194523
+docs/tutorials/08-spatial-navigation.ipynb,sha256=Q0N-az8ZiaMmS42HXMnpDYp97Z_9YPXfM-azC9Sf_f8,186624
 docs/tutorials/08-spatial-navigation.md,sha256=IMbOYBjayXKE7pHfBjApTxOoKRD8WYj7opf8fsJCtzA,4855
-docs/tutorials/09-section-extraction.ipynb,sha256=Aqcy08oXTJ1pkJCmVVumndje-4WXnbkl_QfJPhps7f8,1100736
+docs/tutorials/09-section-extraction.ipynb,sha256=CPBXw28Y7WjWE3HY5SJlUnGlOFQQQ0ZUB65c_uVissA,1101081
 docs/tutorials/09-section-extraction.md,sha256=Jy_be8ftAl_VPBWl5nEv7_5sKSZPx22DLUcBVHMD3Nc,7832
-docs/tutorials/10-form-field-extraction.ipynb,sha256=yyopvBoS5vkKKtUQ6rZ4Kyo5E0Olp2WYnmunhfzSQkQ,281491
+docs/tutorials/10-form-field-extraction.ipynb,sha256=S0S5cdnrioweeKVjdRQnZptUEG-b0VvgrROkOygjAzk,268148
 docs/tutorials/10-form-field-extraction.md,sha256=t9tPlW36vJEhDrKIsHGg_f3P_MK62DT4-ZK1thKFs4Y,5494
-docs/tutorials/11-enhanced-table-processing.ipynb,sha256=BWpVUhtjaAX7r4OOdiy5gQgrSqREaoB0L5TuHqoHEn8,1278
+docs/tutorials/11-enhanced-table-processing.ipynb,sha256=2i8gQRwkLDH14Yie56-3K5YIhdaR83XbL7m-8pQ5cJU,1394
 docs/tutorials/11-enhanced-table-processing.md,sha256=2HK-r1UwU7FLn7zWr_pMG7iLk-i0L4U4-t6ubOEeduc,282
-docs/tutorials/12-ocr-integration.ipynb,sha256=xurkoPwgk2p6mhmPdCehy9ccuYHrAhBCb1zGnjRbZ7Y,26724
-docs/tutorials/12-ocr-integration.md,sha256=wU90sfnm1R6BoMFq-orbGpl8OUVcm-wEBTlK0bLgJC4,4572
-docs/tutorials/13-semantic-search.ipynb,sha256=5h806AIal3EwXPVuXJESbXwdUImCx7fo0mo5-f3Dj44,42817
+docs/tutorials/12-ocr-integration.ipynb,sha256=DB1pWJG1vW4aNVdQ2g5w42a71TFThmzObaVQs8h63U0,194084
+docs/tutorials/12-ocr-integration.md,sha256=-IW4wqLb10eOIWC00NHTGXwtD6jDv7Tp7d-UCOk9SuE,5057
+docs/tutorials/13-semantic-search.ipynb,sha256=BwFepMsOuHrWTFqczvxikPgTh5o97sYX4uleylnOBmc,54126
 docs/tutorials/13-semantic-search.md,sha256=nsNjv0ipYUC3YPSqT5d6dga9ZjObEc04Mc8c0-gsRnU,2914
 docs/visual-debugging/index.ipynb,sha256=MJ92u3Q9sfRCyDAQM4KWmCrs4QhKwIagbn6ytPF83L4,2175800
 docs/visual-debugging/index.md,sha256=ueGD2kNFhEAgIHt7qxCfrLRLjHcR7NTD3AU9okBhX9k,4176
 docs/visual-debugging/region.png,sha256=ULAJs3ZTxMjpD9F4w1DKaZXmhxga3KRq3NrUsXgw28s,67835
-natural_pdf/__init__.py,sha256=IjeAfnDU9fpkVc9YvU2DltNtBtIsZm8_DUyXGuyHGB8,2669
+natural_pdf/__init__.py,sha256=aCnIBTYZlUCL1j78sScPX8kXF88JnuQSHsErboTcjnM,2727
 natural_pdf/analyzers/__init__.py,sha256=dIXjsMqoxKmd9OOnSBzn12wvdIz7D7YNQRAnXslpJSM,142
 natural_pdf/analyzers/text_options.py,sha256=nE2E1pp4psDPpxmtarvNtEQsgozPkyFRjv0TVP2HTyU,2865
 natural_pdf/analyzers/text_structure.py,sha256=9h8hKRz0JWnr13xQr3b4FFr_-hDIjue07WvG7LmT8nc,12827
@@ -67,7 +70,7 @@ natural_pdf/analyzers/utils.py,sha256=Lgub1kYSTOnNxeLO1klStHLwH-GIuT4vpdqyVRF-Mc
 natural_pdf/analyzers/layout/__init__.py,sha256=oq1uJ5UkGGMbBKGirV1aRKK3hxAUyjTLywYkPCQH1f0,33
 natural_pdf/analyzers/layout/base.py,sha256=9dCR758mAuz7ExlHJ-gwnPnETaM4GZV3W1IRei_t13s,6815
 natural_pdf/analyzers/layout/docling.py,sha256=4BJYyNVR6VegZGxyisvNIBBRvVk6YKPyDVs7ZdVfzEU,12676
-natural_pdf/analyzers/layout/gemini.py,sha256=Dslt6rOar9S-QSahyjjXN7Z0KEp7CoVStAA4tvB06X8,10894
+natural_pdf/analyzers/layout/gemini.py,sha256=CzJPWyyEghuCNpu2CMb6OA6FtBGdGhXspHjsjy6I4JE,11195
 natural_pdf/analyzers/layout/layout_analyzer.py,sha256=6aed1qz5jpndOiakXCBRZAcnyG_waeXi3WPuP5fRvh4,14046
 natural_pdf/analyzers/layout/layout_manager.py,sha256=Vh8EKiszKqjELofxQ1eiVLKVjibyjBsZpLFzTf0_21E,11179
 natural_pdf/analyzers/layout/layout_options.py,sha256=s7xr4brE3OutE6aYNAi2PniRy1p2w8a342C2xGpvX2s,3777
@@ -75,32 +78,41 @@ natural_pdf/analyzers/layout/paddle.py,sha256=gTI9ZqNd5-t4H5IByGfL32WgcE6JrdchW6
 natural_pdf/analyzers/layout/surya.py,sha256=vhji6ynHPMyQLHuYRPQcplNi7m_lG4P4NYtWv6MzcME,13556
 natural_pdf/analyzers/layout/tatr.py,sha256=-GJhMy4d0yx6egkO9-ULAIdQkkQRyAKExoIta-b256U,12971
 natural_pdf/analyzers/layout/yolo.py,sha256=ANo2U4EZgeN2eYKM1bZIuysiuJLgwl4JeQchrRxOKwA,8388
-natural_pdf/collections/pdf_collection.py,sha256=Qfauo7fskmw3MSzG0C_OjIGLskyFSQxWQkbHrPDYpno,13240
+natural_pdf/classification/manager.py,sha256=pLcEDe1a5QARJCMimE5Ul_HKZD4jX-eREUCeUuniA0U,16445
+natural_pdf/classification/mixin.py,sha256=aySe0bEjkaI9qYDmSkQe536w0Xrxcg4j6k3JGPvj-cY,6737
+natural_pdf/classification/results.py,sha256=Hn-3xDSThR8x7XpoTlQLWpX6JE1VHVe2QpOeWNY2Ycw,2949
+natural_pdf/collections/mixins.py,sha256=BXk4o_PRrczSXjR7vorIEe4WyEKyms4_qYnY8ZAZd-A,2737
+natural_pdf/collections/pdf_collection.py,sha256=F_4Z-nrL9wFQ-mt4T4cJ2ERVUnkh2kyQdmOV8ASBgoM,27281
 natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,38
-natural_pdf/core/element_manager.py,sha256=rPTkppvU5nQ6Q1_ZtTbcY-KpcRBZV_uBKuNwllHBks0,22100
+natural_pdf/core/element_manager.py,sha256=KZ9yNtpFwuImDWmFUXgISAoWQdSib93E4t3ILUZzIic,24805
 natural_pdf/core/highlighting_service.py,sha256=CTVd7y-fpIreFSe70cTpMu1Pwl6HKMtTHp0bh2U7VXk,32609
-natural_pdf/core/page.py,sha256=Fw01hvK7ekj3iGpj9YbbsVygCmOtTC3qLNvzGBHS4iQ,84784
-natural_pdf/core/pdf.py,sha256=dq2bcxbrbY0zS8t3NKkvqi63Oj2nNzaCHUyHz81OoBI,46383
+natural_pdf/core/page.py,sha256=4iykmXdVwmSQOpGukTxfJYU-5XEgSafNbKsnIedVaGA,94051
+natural_pdf/core/pdf.py,sha256=yPAaOv5vNKZlC9oVk5sKsFxb4LdoRygz_Qkp2EaDtOY,43074
 natural_pdf/elements/__init__.py,sha256=S8XeiNWJ1WcgnyYKdYV1yxQlAxCCO3FfITT8MQwNbyk,41
-natural_pdf/elements/base.py,sha256=hgXpcftlvyUOMGZ8EHEu0WhJz2weqcLrCOFgMEhfm-w,35171
-natural_pdf/elements/collections.py,sha256=Fy5l1Rso-LMunZ6D6Y0n90Uk89h8dzJ2Js7r-8r8NSY,68885
+natural_pdf/elements/base.py,sha256=UtoSD-c_s0yiLpWZrIIJjeJ9MgGz_4R0UHYcsFWH6bc,35157
+natural_pdf/elements/collections.py,sha256=CCQVgglxWLfhuy4FZvVHXdmgiZxU27Ay7Myt8ttQYWg,79467
 natural_pdf/elements/line.py,sha256=7cow3xMUKhAj7zoQz7OaB1eIH2_a8B__LB7iGJ4Mb0o,4612
 natural_pdf/elements/rect.py,sha256=kiVa3e377ZnqIOXc89d9ZSY4EcmDxtccdtUw-HOQzpw,3796
-natural_pdf/elements/region.py,sha256=-mBZnWoYMB7cLm5RZhHWCnJkJjsfcgH-f2ESktl-lNk,69656
-natural_pdf/elements/text.py,sha256=8PNKSLUgXUhEu9IFfbNbSSpuu0Slm11T6UH8jn4O6hQ,11078
-natural_pdf/exporters/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+natural_pdf/elements/region.py,sha256=f7ArCPizkosIei9ePixHYqedK3K6LBVJotwKZ-y33a0,74058
+natural_pdf/elements/text.py,sha256=ZpPluwZtAVfOkoeM4Fm2PDsN87BBZduURZaFWns03RM,11158
+natural_pdf/exporters/__init__.py,sha256=7MnvRLLQdwtg-ULu-8uK8C84GsKiJamyhRw_GgWhw7k,151
+natural_pdf/exporters/base.py,sha256=s-NpHoH81x80GQxs0oqjdhPGrzbUa8npjnK8apKOsHQ,2115
+natural_pdf/exporters/paddleocr.py,sha256=vyVetJ6RgEY46qS5Yl5mKl4cSJadwOxLWGGsdiDjico,16248
 natural_pdf/exporters/searchable_pdf.py,sha256=qsaPsnbOOaZHA_aplfZbwQnBoK9KghWm-wzbyRRomeY,16859
-natural_pdf/ocr/__init__.py,sha256=8EvoA2poYbtkS8Jmbkgn4hFEwOHbN5fIXdJc9tT4NzI,2477
-natural_pdf/ocr/engine.py,sha256=oTABzt20pEvKadTv1U7zOIv_jSTwv5dty2Q_6vf6HQs,8767
-natural_pdf/ocr/engine_easyocr.py,sha256=zki3-fd-u01E9VPdaBD0Rkel18Nk9xYBg477MXY0C8w,8487
-natural_pdf/ocr/engine_paddle.py,sha256=8OYbcYrjIaYop7h7W4R7bXWHKOCYhJQpqT2cVccNb5E,6209
-natural_pdf/ocr/engine_surya.py,sha256=PtzvDOqxNzhAEeh0rIhTwPXDAbTHMR2IRs0tXb-4pag,4693
-natural_pdf/ocr/ocr_factory.py,sha256=3fxIpgGi6NzxhfYpl5Kblufyx_-RRoT7dlRGktYjB_Y,4425
-natural_pdf/ocr/ocr_manager.py,sha256=sMQrpvAaA4PbjhZOf7G_KGW3KBhXhx1l26ig_Xqf_0g,9190
-natural_pdf/ocr/ocr_options.py,sha256=3njghlnnkcxOL4Td7l6mt-mTEFmL8Z-ryf_BX-EM3i4,3289
-natural_pdf/ocr/utils.py,sha256=AIBkGQExetjl2wyuQLpWJSy0HSxaOBmuCo2QsKmY7Rc,3404
+natural_pdf/extraction/manager.py,sha256=YH5dyUorMItGxuaZ-DhuJD5Sh_Ozjj0fa-WBMcQw1E0,4903
+natural_pdf/extraction/mixin.py,sha256=6CWYyutGcKCxFVYun8yXC4H1IZWLMXaeYZ-cWJRx5FE,11430
+natural_pdf/extraction/result.py,sha256=ihY1g_C2hsMACYqU7bcvAKRijuh-FHVtpnn0uoP--pk,1047
+natural_pdf/ocr/__init__.py,sha256=jKaDbo13CdCDcas1WiBmg5gjBvVeG-Z9uaeYxyzvaNY,2464
+natural_pdf/ocr/engine.py,sha256=ZBC1tZNM5EDbGDJJmZI9mNHr4nCMLEZvUFhiJq8GdF4,8741
+natural_pdf/ocr/engine_easyocr.py,sha256=9TbxJjmhWFrzM8mcNnZjoRtIDr6gwpuwKm4-Zfub2-8,9281
+natural_pdf/ocr/engine_paddle.py,sha256=2nIrvLBBAiZG1BxVo3eFVJulA6YGoOTXw_RN98p_BUk,6184
+natural_pdf/ocr/engine_surya.py,sha256=iySjG-Dahgh0cLICfbMtOcwUpRFcZjo-5Ed5Zwz-o5Y,4805
+natural_pdf/ocr/ocr_factory.py,sha256=IFccj0BB75YGV4hjcy4ECtGQX_JQzdptpvDFfeGxxgI,4391
+natural_pdf/ocr/ocr_manager.py,sha256=ivk4Aqr5gsDJWiCxP1-FLkhuvfJiQtilwbPtgIPm--4,13320
+natural_pdf/ocr/ocr_options.py,sha256=BcPVwJGYE3vMug7wsVh_ARUJlm_4emz9ynOAwYgwHBk,4257
+natural_pdf/ocr/utils.py,sha256=4b_A47hfynfV00iR8I9OWmXCzDzRvSdEkQhZLcSV4kQ,4394
 natural_pdf/qa/__init__.py,sha256=Pjo62JTnUNEjGNsC437mvsS5KQ5m7X_BibGvavR9AW0,108
-natural_pdf/qa/document_qa.py,sha256=W4E4vS_Eox_IBsYpVb0ifQbJb0FP-PYEIG93CU3rUkE,15246
+natural_pdf/qa/document_qa.py,sha256=Jw4yyq3Vifn57D0ANmOfUlZeG8CJjBkItZBV-8ZAmos,15111
 natural_pdf/search/__init__.py,sha256=EB_HRwlktJn5WGPVtSaRbOQNjLAZTxujeYf_eN-zd2U,4191
 natural_pdf/search/haystack_search_service.py,sha256=6RjTFWbTo3gaO-90IF6PEuo_9WRwOdj232eWn3OT0BQ,29270
 natural_pdf/search/haystack_utils.py,sha256=UI4eu3SVieGR_QnBtLhP8Fjtt2AJgeLgxrpa_dBmD6k,19289
@@ -108,24 +120,27 @@ natural_pdf/search/search_options.py,sha256=sq_e8_jSROicD94b_xtDtLnjEr_Zsy4icjzP
 natural_pdf/search/search_service_protocol.py,sha256=ybNcF_NoLZuIx0rb4XB1dsDl3o_LAaWR1fVVKld2TxI,6818
 natural_pdf/search/searchable_mixin.py,sha256=M2a6FaFVM0vcfh7FgjDH6BLhS-7ggeVpcfft4OOBDxY,26390
 natural_pdf/selectors/__init__.py,sha256=oZGeqSv53EqmIZOhcnawuaGGlRg1h79vArXuZCWKm4A,123
-natural_pdf/selectors/parser.py,sha256=59_GSsTApM6MFvtqhrrmbKaBfODPbGXMluvvQJcrqhE,15754
+natural_pdf/selectors/parser.py,sha256=AKXGv4MaZDiaWT_jSfn_vU-qVlECB8b-IxnyocXtaaE,22671
 natural_pdf/templates/__init__.py,sha256=jYBxzfi73vew0f6yhIh1MlRxw4F_TVN2hKQR0YXOFe0,20
+natural_pdf/templates/finetune/fine_tune_paddleocr.md,sha256=H6Wmu3Nvi2qKK-rPwr8KUZfILzXz8VmWyCWYOTe6QCI,14764
 natural_pdf/templates/spa/index.html,sha256=6hLTp07OeV5Q4jUMp5Sgl-dwfBs3oPzBxqphG4kEs24,787
 natural_pdf/templates/spa/words.txt,sha256=vkGtl5Y7-Nq-3Vhx1daRWWF1Jp1UCVaw-ZZaiFwrurk,2493885
 natural_pdf/templates/spa/css/style.css,sha256=Qdl0U3L5HMyhBDNzyRPklfb3OxW6rMxCfQbzO8i8IW4,7643
 natural_pdf/templates/spa/js/app.js,sha256=Efb7NmcTN9RLdLwKpDcU6CG5Ix0laHtzRHmfUlDMJXw,19679
 natural_pdf/utils/__init__.py,sha256=s3M8FggaK1P3EBYn6R_-HgSDjNc9C73gyKe1hihtNWg,43
-natural_pdf/utils/debug.py,sha256=56dk0OcDUwjRbYU2g1k0_G_3hfHkAF9Z6vnMVt3myMU,992
+natural_pdf/utils/debug.py,sha256=lk_6qzxan8NagjEtJEZpZ2MS30SO8ce6iznBxmA0xgk,995
 natural_pdf/utils/highlighting.py,sha256=EIY6ihVGtUTS_DjWyxpnr_UXpcR4btC1KhSGQ9VUfKg,698
-natural_pdf/utils/identifiers.py,sha256=IvXj2q-NW1cmivWAedDOZBFPTRbVl0_J8BP4phREt9U,1092
-natural_pdf/utils/packaging.py,sha256=mNPEqCtc0CVWjKg1RcSldYTgspUWQOLAR_ZEllyKxs4,20175
+natural_pdf/utils/identifiers.py,sha256=n61viCQiMlf5-E_jsPLe-FkPBdKkMKv-gfs5tGqlKiw,1117
+natural_pdf/utils/locks.py,sha256=E_Fb6GnRNq-tF5aE7jnllkpidsNr8LXPhSaqgr56Ks4,215
+natural_pdf/utils/packaging.py,sha256=HSgpubpHICU75L4ZAZPU8iOjium055XWnklV9_YqoCA,21579
 natural_pdf/utils/reading_order.py,sha256=s3DsYq_3g_1YA07qhd4BGEjeIRTeyGtnwc_hNtSzwBY,7290
-natural_pdf/utils/text_extraction.py,sha256=VlbkXg14GlvwYTjRJWa8FVUigETY3Hq0v8NlIRnzYkM,8619
-natural_pdf/utils/visualization.py,sha256=ir5PgpptRuVuVeRT9IcdTsNeEpdOYD_69rByjHQ7JhI,8592
+natural_pdf/utils/text_extraction.py,sha256=qZfOuO57XeKg7p-Q7yzTBMTrpAvDRslYXjDSjiJLStI,9545
+natural_pdf/utils/tqdm_utils.py,sha256=bKWvsoAOl0lPOPLJC2hkTtkdxBf5f9aVtcA3DmUE19M,1570
+natural_pdf/utils/visualization.py,sha256=5GbhxtvZW-77ONVnICupg-s2D-OaxLZNqkKlOrQESK4,8593
 natural_pdf/widgets/__init__.py,sha256=O2fSDo604wDAP6UwUkmBq3eT91RSqHwBpAOQXq92S8s,214
 natural_pdf/widgets/viewer.py,sha256=Aiw6kuBc0WkhcZrPNKyLNzzWbmtmU6rvOmHV0IuXCBk,40862
 natural_pdf/widgets/frontend/viewer.js,sha256=w8ywfz_IOAAv2nP_qaf2VBUkF1KhjT3zorhJxM1-CfU,4371
-natural_pdf-0.1.6.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
+natural_pdf-0.1.8.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
 notebooks/Examples.ipynb,sha256=l4YMtMEx_DWBzWIjl9CmBkWTo0g_nK8l_XWOyzYooQM,4275170
 pdfs/.gitkeep,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pdfs/01-practice.pdf,sha256=dxWyJIa2cm7bALE3BWDJ2dg3inyFlo1n8ntVyy0hkTo,7906
@@ -135,7 +150,7 @@ pdfs/2014 Statistics.pdf,sha256=B-30OQVjqj_3718-G9cGUefNddnz-MosPdHAzfGfkcc,9559
 pdfs/2019 Statistics.pdf,sha256=reuSJxvAlx9_P-pW7IPqzox0jFCxSPbK1i1-WFu-uGA,511439
 pdfs/Atlanta_Public_Schools_GA_sample.pdf,sha256=PLBh_uWJQH0MnBaSm5ng5Ima63_m6Mi11CjdravB_S8,137689
 pdfs/needs-ocr.pdf,sha256=vusKiLxSOlELUTetfZfaotNU54RtMj9PCzGfLc2cuNs,139305
-natural_pdf-0.1.6.dist-info/METADATA,sha256=Yd92-6FRKc6KdEuTpCOWyxBcewnhAoYTAhtbq6qIB0Q,6668
-natural_pdf-0.1.6.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
-natural_pdf-0.1.6.dist-info/top_level.txt,sha256=7nDKUnpkN7B8cBI7DEpW5JM8S7OcOgHw3jXH-1iCX2o,32
-natural_pdf-0.1.6.dist-info/RECORD,,
+natural_pdf-0.1.8.dist-info/METADATA,sha256=Qz_ePmFWt4poceUJnVcldvhJoIRWuo2lEIEoVp-mnwE,7030
+natural_pdf-0.1.8.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
+natural_pdf-0.1.8.dist-info/top_level.txt,sha256=7nDKUnpkN7B8cBI7DEpW5JM8S7OcOgHw3jXH-1iCX2o,32
+natural_pdf-0.1.8.dist-info/RECORD,,

{natural_pdf-0.1.6.dist-info → natural_pdf-0.1.8.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (79.0.0)
+Generator: setuptools (79.0.1)
 Root-Is-Purelib: true
 Tag: py3-none-any