natural-pdf 0.2.15__py3-none-any.whl → 0.2.17__py3-none-any.whl

natural_pdf/core/page_collection.py

@@ -789,6 +789,7 @@ class PageCollection(TextMixin, Generic[P], ApplyMixin, ShapeDetectionMixin, Vis
                                 start_element.page,
                                 (0, top, start_element.page.width, bottom),
                             )
+                            section._boundary_exclusions = include_boundaries
                         else:  # horizontal
                             left = start_element.x0
                             right = end_element.x1
@@ -821,6 +822,7 @@ class PageCollection(TextMixin, Generic[P], ApplyMixin, ShapeDetectionMixin, Vis
                                 start_element.page,
                                 (left, 0, right, start_element.page.height),
                             )
+                            section._boundary_exclusions = include_boundaries
                         section.start_element = start_element
                         section.boundary_element_found = end_element
                     else:
@@ -865,6 +867,10 @@ class PageCollection(TextMixin, Generic[P], ApplyMixin, ShapeDetectionMixin, Vis
                                 start_element.page, (0, top, start_element.page.width, bottom)
                             )
                             section.start_element = start_element
+                            section.end_element = (
+                                next_start  # The next start is the end of this section
+                            )
+                            section._boundary_exclusions = include_boundaries
                             sections.append(section)
                     else:  # horizontal
                         # Determine horizontal bounds
@@ -882,6 +888,10 @@ class PageCollection(TextMixin, Generic[P], ApplyMixin, ShapeDetectionMixin, Vis
                                 start_element.page, (left, 0, right, start_element.page.height)
                             )
                             section.start_element = start_element
+                            section.end_element = (
+                                next_start  # The next start is the end of this section
+                            )
+                            section._boundary_exclusions = include_boundaries
                             sections.append(section)
                 else:
                     # Cross-page section - create from current_start to the end of its page
@@ -982,6 +992,71 @@ class PageCollection(TextMixin, Generic[P], ApplyMixin, ShapeDetectionMixin, Vis
 
         return ElementCollection(sections)
 
+    def split(self, divider, **kwargs) -> "ElementCollection[Region]":
+        """
+        Divide this page collection into sections based on the provided divider elements.
+
+        Args:
+            divider: Elements or selector string that mark section boundaries
+            **kwargs: Additional parameters passed to get_sections()
+                - include_boundaries: How to include boundary elements (default: 'start')
+                - orientation: 'vertical' or 'horizontal' (default: 'vertical')
+                - new_section_on_page_break: Whether to split at page boundaries (default: False)
+
+        Returns:
+            ElementCollection of Region objects representing the sections
+
+        Example:
+            # Split a PDF by chapter titles
+            chapters = pdf.pages.split("text[size>20]:contains('CHAPTER')")
+
+            # Split by page breaks
+            page_sections = pdf.pages.split(None, new_section_on_page_break=True)
+
+            # Split multi-page document by section headers
+            sections = pdf.pages[10:20].split("text:bold:contains('Section')")
+        """
+        # Default to 'start' boundaries for split (include divider at start of each section)
+        if "include_boundaries" not in kwargs:
+            kwargs["include_boundaries"] = "start"
+
+        sections = self.get_sections(start_elements=divider, **kwargs)
+
+        # Add initial section if there's content before the first divider
+        if sections and divider is not None:
+            # Get all elements across all pages
+            all_elements = []
+            for page in self.pages:
+                all_elements.extend(page.get_elements())
+
+            if all_elements:
+                # Find first divider
+                if isinstance(divider, str):
+                    # Search for first matching element
+                    first_divider = None
+                    for page in self.pages:
+                        match = page.find(divider)
+                        if match:
+                            first_divider = match
+                            break
+                else:
+                    # divider is already elements
+                    first_divider = divider[0] if hasattr(divider, "__getitem__") else divider
+
+                if first_divider and all_elements[0] != first_divider:
+                    # There's content before the first divider
+                    # Get section from start to first divider
+                    initial_sections = self.get_sections(
+                        start_elements=None,
+                        end_elements=[first_divider],
+                        include_boundaries="none",
+                        orientation=kwargs.get("orientation", "vertical"),
+                    )
+                    if initial_sections:
+                        sections = ElementCollection([initial_sections[0]] + list(sections))
+
+        return sections
+
     def _gather_analysis_data(
         self,
         analysis_keys: List[str],

natural_pdf/core/pdf.py

@@ -1333,6 +1333,39 @@ class PDF(
             orientation=orientation,
         )
 
+    def split(self, divider, **kwargs) -> "ElementCollection":
+        """
+        Divide the PDF into sections based on the provided divider elements.
+
+        Args:
+            divider: Elements or selector string that mark section boundaries
+            **kwargs: Additional parameters passed to get_sections()
+                - include_boundaries: How to include boundary elements (default: 'start')
+                - orientation: 'vertical' or 'horizontal' (default: 'vertical')
+                - new_section_on_page_break: Whether to split at page boundaries (default: False)
+
+        Returns:
+            ElementCollection of Region objects representing the sections
+
+        Example:
+            # Split a PDF by chapter titles
+            chapters = pdf.split("text[size>20]:contains('Chapter')")
+
+            # Export each chapter to a separate file
+            for i, chapter in enumerate(chapters):
+                chapter_text = chapter.extract_text()
+                with open(f"chapter_{i+1}.txt", "w") as f:
+                    f.write(chapter_text)
+
+            # Split by horizontal rules/lines
+            sections = pdf.split("line[orientation=horizontal]")
+
+            # Split only by page breaks (no divider elements)
+            pages = pdf.split(None, new_section_on_page_break=True)
+        """
+        # Delegate to pages collection
+        return self.pages.split(divider, **kwargs)
+
     def save_searchable(self, output_path: Union[str, "Path"], dpi: int = 300, **kwargs):
         """
         DEPRECATED: Use save_pdf(..., ocr=True) instead.

natural_pdf/describe/base.py

@@ -272,17 +272,12 @@ def _get_columns_for_type(element_type: str, show_page_column: bool) -> List[str
                 "font_family",
                 "font_variant",
                 "size",
-                "bold",
-                "italic",
-                "strike",
-                "underline",
-                "highlight",
+                "styles",
                 "source",
                 "confidence",
+                "color",
             ]
         )
-        # Add foreground text colour too
-        columns.append("color")
     elif element_type == "rect":
         columns = base_columns + ["width", "height", "stroke", "fill", "stroke_width"]
     elif element_type == "line":
@@ -358,6 +353,52 @@ def _extract_element_value(element: "Element", column: str) -> Any:
                 return str(col_val)
             return ""
 
+        elif column == "styles":
+            # Collect all active text decorations
+            styles = []
+
+            if getattr(element, "bold", False):
+                styles.append("bold")
+            if getattr(element, "italic", False):
+                styles.append("italic")
+            if getattr(element, "strike", False):
+                styles.append("strike")
+            if getattr(element, "underline", False):
+                styles.append("underline")
+
+            # Handle highlight specially - include color if not default yellow
+            if getattr(element, "is_highlighted", False):
+                highlight_color = getattr(element, "highlight_color", None)
+                if highlight_color is not None:
+                    # Convert color to hex if needed
+                    if isinstance(highlight_color, (tuple, list)) and len(highlight_color) >= 3:
+                        try:
+                            r, g, b = [
+                                int(v * 255) if v <= 1 else int(v) for v in highlight_color[:3]
+                            ]
+                            hex_color = f"#{r:02x}{g:02x}{b:02x}"
+                            styles.append(f"highlight({hex_color})")
+                        except Exception:
+                            styles.append("highlight")
+                    elif isinstance(highlight_color, (int, float)):
+                        # Grayscale value
+                        try:
+                            gray = (
+                                int(highlight_color * 255)
+                                if highlight_color <= 1
+                                else int(highlight_color)
+                            )
+                            hex_color = f"#{gray:02x}{gray:02x}{gray:02x}"
+                            styles.append(f"highlight({hex_color})")
+                        except Exception:
+                            styles.append("highlight")
+                    else:
+                        styles.append("highlight")
+                else:
+                    styles.append("highlight")
+
+            return ", ".join(styles) if styles else ""
+
         elif column in ["stroke", "fill", "color"]:
             value = getattr(element, column, None)
             # If already a string (e.g. '#ff00aa' or 'red') return as is