natural-pdf 0.1.12__py3-none-any.whl → 0.1.14__py3-none-any.whl

natural_pdf/classification/manager.py

@@ -25,7 +25,7 @@ except ImportError:
     AutoModelForSequenceClassification = object
     torch = None
 
-from natural_pdf.utils.tqdm_utils import get_tqdm
+from tqdm.auto import tqdm
 
 # Import result classes
 from .results import CategoryScore, ClassificationResult
@@ -343,8 +343,7 @@ class ClassificationManager:
             total_items = len(item_contents)
             if progress_bar:
                 # Get the appropriate tqdm class
-                tqdm_class = get_tqdm()
-                results_iterator = tqdm_class(
+                results_iterator = tqdm(
                     results_iterator,
                     total=total_items,
                     desc=f"Classifying batch ({model_id})",

natural_pdf/collections/pdf_collection.py

@@ -25,14 +25,12 @@ from typing import (
 )
 
 from PIL import Image
-from tqdm import tqdm
-from tqdm.auto import tqdm as auto_tqdm
-from tqdm.notebook import tqdm as notebook_tqdm
+from tqdm.auto import tqdm
 
-from natural_pdf.utils.tqdm_utils import get_tqdm
+from natural_pdf.exporters.base import FinetuneExporter
 
-# Get the appropriate tqdm class once
-tqdm = get_tqdm()
+# Need to import this utility
+from natural_pdf.utils.identifiers import generate_short_path_hash
 
 # Set up logger early
 # Configure logging to include thread information
@@ -67,8 +65,10 @@ except ImportError as e:
 from natural_pdf.collections.mixins import ApplyMixin
 from natural_pdf.search.searchable_mixin import SearchableMixin  # Import the new mixin
 
+from natural_pdf.analyzers.shape_detection_mixin import ShapeDetectionMixin
 
-class PDFCollection(SearchableMixin, ApplyMixin, ExportMixin):  # Add ExportMixin
+
+class PDFCollection(SearchableMixin, ApplyMixin, ExportMixin, ShapeDetectionMixin):  # Add ExportMixin and ShapeDetectionMixin
     def __init__(
         self,
         source: Union[str, Iterable[Union[str, "PDF"]]],
@@ -119,16 +119,8 @@ class PDFCollection(SearchableMixin, ApplyMixin, ExportMixin):  # Add ExportMixi
     @staticmethod
     def _get_pdf_class():
         """Helper method to dynamically import the PDF class."""
-        try:
-            # Import needs to resolve path correctly
-            from natural_pdf.core.pdf import PDF
-
-            return PDF
-        except ImportError as e:
-            logger.error(
-                "Could not import PDF class from natural_pdf.core.pdf. Ensure it exists and there are no circular imports at runtime."
-            )
-            raise ImportError("PDF class is required but could not be imported.") from e
+        from natural_pdf.core.pdf import PDF
+        return PDF
 
     # --- Internal Helpers ---
 
@@ -141,16 +133,13 @@ class PDFCollection(SearchableMixin, ApplyMixin, ExportMixin):  # Add ExportMixi
     def _execute_glob(self, pattern: str) -> Set[str]:
         """Glob for paths and return a set of valid PDF paths."""
         found_paths = set()
-        try:
-            # Use iglob for potentially large directories/matches
-            paths_iter = py_glob.iglob(pattern, recursive=self._recursive)
-            for path_str in paths_iter:
-                # Use Path object for easier checking
-                p = Path(path_str)
-                if p.is_file() and p.suffix.lower() == ".pdf":
-                    found_paths.add(str(p.resolve()))  # Store resolved absolute path
-        except Exception as e:
-            logger.error(f"Error processing glob pattern '{pattern}': {e}")
+        # Use iglob for potentially large directories/matches
+        paths_iter = py_glob.iglob(pattern, recursive=self._recursive)
+        for path_str in paths_iter:
+            # Use Path object for easier checking
+            p = Path(path_str)
+            if p.is_file() and p.suffix.lower() == ".pdf":
+                found_paths.add(str(p.resolve()))  # Store resolved absolute path
         return found_paths
 
     def _resolve_sources_to_paths(self, source: Union[str, Iterable[str]]) -> List[str]:
@@ -534,19 +523,10 @@ class PDFCollection(SearchableMixin, ApplyMixin, ExportMixin):  # Add ExportMixi
             **kwargs: Additional arguments passed to create_correction_task_package
                       (e.g., image_render_scale, overwrite).
         """
-        try:
-            from natural_pdf.utils.packaging import create_correction_task_package
+        from natural_pdf.utils.packaging import create_correction_task_package
 
-            # Pass the collection itself (self) as the source
-            create_correction_task_package(source=self, output_zip_path=output_zip_path, **kwargs)
-        except ImportError:
-            logger.error(
-                "Failed to import 'create_correction_task_package'. Packaging utility might be missing."
-            )
-            # Or raise
-        except Exception as e:
-            logger.error(f"Failed to export correction task for collection: {e}", exc_info=True)
-            raise  # Re-raise the exception from the utility function
+        # Pass the collection itself (self) as the source
+        create_correction_task_package(source=self, output_zip_path=output_zip_path, **kwargs)
 
     # --- Mixin Required Implementation ---
     def get_indexable_items(self) -> Iterable[Indexable]:

natural_pdf/core/highlighting_service.py

@@ -215,21 +215,14 @@ class HighlightRenderer:
     def _render_ocr_text(self):
         """Renders OCR text onto the image. (Adapted from old HighlightManager)"""
         # Use the page reference to get OCR elements
-        try:
-            # Try finding first, then extracting if necessary
-            ocr_elements = self.page.find_all("text[source=ocr]")
-            if not ocr_elements:
-                # Don't run full OCR here, just extract if already run
-                ocr_elements = [
-                    el for el in self.page.words if getattr(el, "source", None) == "ocr"
-                ]
-                # Alternative: self.page.extract_ocr_elements() - but might be slow
-
-        except Exception as e:
-            logger.warning(
-                f"Could not get OCR elements for page {self.page.number}: {e}", exc_info=True
-            )
-            return  # Don't modify image if OCR elements aren't available
+        # Try finding first, then extracting if necessary
+        ocr_elements = self.page.find_all("text[source=ocr]")
+        if not ocr_elements:
+            # Don't run full OCR here, just extract if already run
+            ocr_elements = [
+                el for el in self.page.words if getattr(el, "source", None) == "ocr"
+            ]
+            # Alternative: self.page.extract_ocr_elements() - but might be slow
 
         if not ocr_elements:
             logger.debug(f"No OCR elements found for page {self.page.number} to render.")
@@ -293,20 +286,15 @@ class HighlightRenderer:
             )
 
             # Calculate text position (centered vertically, slightly offset from left)
-            try:
-                if hasattr(sized_font, "getbbox"):  # Modern PIL
-                    _, text_top_offset, _, text_bottom_offset = sized_font.getbbox(element.text)
-                    text_h = text_bottom_offset - text_top_offset
-                else:  # Older PIL approximation
-                    text_h = font_size
-                text_y = top_s + (box_h - text_h) / 2
-                # Adjust for vertical offset in some fonts
-                text_y -= text_top_offset if hasattr(sized_font, "getbbox") else 0
-                text_x = x0_s + padding  # Start near left edge with padding
-
-            except Exception:
-                # Fallback positioning
-                text_x, text_y = x0_s + padding, top_s + padding
+            if hasattr(sized_font, "getbbox"):  # Modern PIL
+                _, text_top_offset, _, text_bottom_offset = sized_font.getbbox(element.text)
+                text_h = text_bottom_offset - text_top_offset
+            else:  # Older PIL approximation
+                text_h = font_size
+            text_y = top_s + (box_h - text_h) / 2
+            # Adjust for vertical offset in some fonts
+            text_y -= text_top_offset if hasattr(sized_font, "getbbox") else 0
+            text_x = x0_s + padding  # Start near left edge with padding
 
             draw.text((text_x, text_y), element.text, fill=(0, 0, 0, 255), font=sized_font)
 
@@ -392,9 +380,6 @@ class HighlightingService:
             except ValueError:
                 logger.warning(f"Invalid color string: '{color_input}'")
                 return None
-            except Exception as e:
-                logger.error(f"Error processing color string '{color_input}': {e}")
-                return None
         else:
             logger.warning(f"Invalid color input type: {type(color_input)}")
             return None
@@ -677,9 +662,12 @@ class HighlightingService:
                     actual_scale_y = scale # Fallback
                 logger.debug(f"Calculated actual scales for page {page_index}: x={actual_scale_x:.2f}, y={actual_scale_y:.2f}")
 
-        except Exception as e:
-            logger.error(f"Error creating base image for page {page_index}: {e}", exc_info=True)
-            return None
+        except IOError as e:
+            logger.error(f"IOError creating base image for page {page_index}: {e}")
+            raise
+        except AttributeError as e:
+            logger.error(f"AttributeError creating base image for page {page_index}: {e}")
+            raise
         
         renderer_scale = actual_scale_x # Assuming aspect ratio maintained, use x_scale
 
@@ -865,8 +853,11 @@ class HighlightingService:
             else:
                 final_image = rendered_image
 
-        except Exception as e:
-            logger.error(f"Error rendering preview for page {page_index}: {e}", exc_info=True)
-            return None
+        except IOError as e:
+            logger.error(f"IOError rendering preview for page {page_index}: {e}")
+            raise
+        except AttributeError as e:
+            logger.error(f"AttributeError rendering preview for page {page_index}: {e}")
+            raise
 
         return final_image