docling-core 2.36.0__py3-none-any.whl → 2.37.0__py3-none-any.whl

docling_core/transforms/serializer/html.py

@@ -340,7 +340,7 @@ class HTMLTableSerializer(BaseTableSerializer):
 
                     content = html.escape(cell.text.strip())
                     celltag = "td"
-                    if cell.column_header:
+                    if cell.column_header or cell.row_header or cell.row_section:
                         celltag = "th"
 
                     opening_tag = f"{celltag}"

docling_core/transforms/visualizer/table_visualizer.py

@@ -23,8 +23,23 @@ class TableVisualizer(BaseVisualizer):
 
         # show_Label: bool = False
         show_cells: bool = True
-        # show_rows: bool = False
-        # show_cols: bool = False
+        show_rows: bool = False
+        show_cols: bool = False
+
+        cell_color: tuple[int, int, int, int] = (256, 0, 0, 32)
+        cell_outline: tuple[int, int, int, int] = (256, 0, 0, 128)
+
+        row_color: tuple[int, int, int, int] = (256, 0, 0, 32)
+        row_outline: tuple[int, int, int, int] = (256, 0, 0, 128)
+
+        row_header_color: tuple[int, int, int, int] = (0, 256, 0, 32)
+        row_header_outline: tuple[int, int, int, int] = (0, 256, 0, 128)
+
+        col_color: tuple[int, int, int, int] = (0, 256, 0, 32)
+        col_outline: tuple[int, int, int, int] = (0, 256, 0, 128)
+
+        col_header_color: tuple[int, int, int, int] = (0, 0, 256, 32)
+        col_header_outline: tuple[int, int, int, int] = (0, 0, 256, 128)
 
     base_visualizer: Optional[BaseVisualizer] = None
     params: Params = Params()
@@ -45,7 +60,21 @@ class TableVisualizer(BaseVisualizer):
 
                 tl_bbox = cell.bbox.to_top_left_origin(page_height=page_height)
 
-                cell_color = (256, 0, 0, 32)  # Transparent black for cells
+                cell_color = self.params.cell_color  # Transparent black for cells
+                cell_outline = self.params.cell_outline
+                if cell.column_header:
+                    cell_color = (
+                        self.params.col_header_color
+                    )  # Transparent black for cells
+                    cell_outline = self.params.col_header_outline
+                if cell.row_header:
+                    cell_color = (
+                        self.params.row_header_color
+                    )  # Transparent black for cells
+                    cell_outline = self.params.row_header_outline
+                if cell.row_section:
+                    cell_color = self.params.row_header_color
+                    cell_outline = self.params.row_header_outline
 
                 cx0, cy0, cx1, cy1 = tl_bbox.as_tuple()
                 cx0 *= scale_x
@@ -55,10 +84,68 @@ class TableVisualizer(BaseVisualizer):
 
                 draw.rectangle(
                     [(cx0, cy0), (cx1, cy1)],
-                    outline=(256, 0, 0, 128),
+                    outline=cell_outline,
                     fill=cell_color,
                 )
 
+    def _draw_table_rows(
+        self,
+        table: TableItem,
+        page_image: Image,
+        page_height: float,
+        scale_x: float,
+        scale_y: float,
+    ):
+        """Draw individual table cells."""
+        draw = ImageDraw.Draw(page_image, "RGBA")
+
+        rows = table.data.get_row_bounding_boxes()
+
+        for rid, bbox in rows.items():
+
+            tl_bbox = bbox.to_top_left_origin(page_height=page_height)
+
+            cx0, cy0, cx1, cy1 = tl_bbox.as_tuple()
+            cx0 *= scale_x
+            cx1 *= scale_x
+            cy0 *= scale_y
+            cy1 *= scale_y
+
+            draw.rectangle(
+                [(cx0, cy0), (cx1, cy1)],
+                outline=self.params.row_outline,
+                fill=self.params.row_color,
+            )
+
+    def _draw_table_cols(
+        self,
+        table: TableItem,
+        page_image: Image,
+        page_height: float,
+        scale_x: float,
+        scale_y: float,
+    ):
+        """Draw individual table cells."""
+        draw = ImageDraw.Draw(page_image, "RGBA")
+
+        cols = table.data.get_column_bounding_boxes()
+
+        for cid, bbox in cols.items():
+
+            tl_bbox = bbox.to_top_left_origin(page_height=page_height)
+
+            cx0, cy0, cx1, cy1 = tl_bbox.as_tuple()
+            cx0 *= scale_x
+            cx1 *= scale_x
+            cy0 *= scale_y
+            cy1 *= scale_y
+
+            draw.rectangle(
+                [(cx0, cy0), (cx1, cy1)],
+                outline=self.params.col_outline,
+                fill=self.params.col_color,
+            )
+
     def _draw_doc_tables(
         self,
         doc: DoclingDocument,
@@ -108,6 +195,24 @@ class TableVisualizer(BaseVisualizer):
                             scale_y=image.height / doc.pages[page_nr].size.height,
                         )
 
+                    if self.params.show_rows:
+                        self._draw_table_rows(
+                            table=elem,
+                            page_height=doc.pages[page_nr].size.height,
+                            page_image=image,
+                            scale_x=image.width / doc.pages[page_nr].size.width,
+                            scale_y=image.height / doc.pages[page_nr].size.height,
+                        )
+
+                    if self.params.show_cols:
+                        self._draw_table_cols(
+                            table=elem,
+                            page_height=doc.pages[page_nr].size.height,
+                            page_image=image,
+                            scale_x=image.width / doc.pages[page_nr].size.width,
+                            scale_y=image.height / doc.pages[page_nr].size.height,
+                        )
+
                 else:
                     raise RuntimeError(f"Cannot visualize page-image for {page_nr}")
 

docling_core/types/doc/document.py

@@ -38,7 +38,7 @@ from typing_extensions import Annotated, Self, deprecated
 from docling_core.search.package import VERSION_PATTERN
 from docling_core.types.base import _JSON_POINTER_REGEX
 from docling_core.types.doc import BoundingBox, Size
-from docling_core.types.doc.base import ImageRefMode
+from docling_core.types.doc.base import CoordOrigin, ImageRefMode
 from docling_core.types.doc.labels import (
     CodeLanguageLabel,
     DocItemLabel,
@@ -372,6 +372,119 @@ class TableData(BaseModel):  # TBD
 
         return table_data
 
+    def get_row_bounding_boxes(self) -> dict[int, BoundingBox]:
+        """Get the minimal bounding box for each row in the table.
+
+        Returns:
+        List[Optional[BoundingBox]]: A list where each element is the minimal
+        bounding box that encompasses all cells in that row, or None if no
+        cells in the row have bounding boxes.
+        """
+        coords = []
+        for cell in self.table_cells:
+            if cell.bbox is not None:
+                coords.append(cell.bbox.coord_origin)
+
+        if len(set(coords)) > 1:
+            raise ValueError(
+                "All bounding boxes must have the same \
+                CoordOrigin to compute their union."
+            )
+
+        row_bboxes: dict[int, BoundingBox] = {}
+
+        for row_idx in range(self.num_rows):
+            row_cells_with_bbox: dict[int, list[BoundingBox]] = {}
+
+            # Collect all cells in this row that have bounding boxes
+            for cell in self.table_cells:
+
+                if (
+                    cell.bbox is not None
+                    and cell.start_row_offset_idx <= row_idx < cell.end_row_offset_idx
+                ):
+
+                    row_span = cell.end_row_offset_idx - cell.start_row_offset_idx
+                    if row_span in row_cells_with_bbox:
+                        row_cells_with_bbox[row_span].append(cell.bbox)
+                    else:
+                        row_cells_with_bbox[row_span] = [cell.bbox]
+
+            # Calculate the enclosing bounding box for this row
+            if len(row_cells_with_bbox) > 0:
+                min_row_span = min(row_cells_with_bbox.keys())
+                row_bbox: BoundingBox = BoundingBox.enclosing_bbox(
+                    row_cells_with_bbox[min_row_span]
+                )
+
+                for rspan, bboxs in row_cells_with_bbox.items():
+                    for bbox in bboxs:
+                        row_bbox.l = min(row_bbox.l, bbox.l)
+                        row_bbox.r = max(row_bbox.r, bbox.r)
+
+                row_bboxes[row_idx] = row_bbox
+
+        return row_bboxes
+
+    def get_column_bounding_boxes(self) -> dict[int, BoundingBox]:
+        """Get the minimal bounding box for each column in the table.
+
+        Returns:
+            List[Optional[BoundingBox]]: A list where each element is the minimal
+            bounding box that encompasses all cells in that column, or None if no
+            cells in the column have bounding boxes.
+        """
+        coords = []
+        for cell in self.table_cells:
+            if cell.bbox is not None:
+                coords.append(cell.bbox.coord_origin)
+
+        if len(set(coords)) > 1:
+            raise ValueError(
+                "All bounding boxes must have the same \
+                CoordOrigin to compute their union."
+            )
+
+        col_bboxes: dict[int, BoundingBox] = {}
+
+        for col_idx in range(self.num_cols):
+            col_cells_with_bbox: dict[int, list[BoundingBox]] = {}
+
+            # Collect all cells in this row that have bounding boxes
+            for cell in self.table_cells:
+
+                if (
+                    cell.bbox is not None
+                    and cell.start_col_offset_idx <= col_idx < cell.end_col_offset_idx
+                ):
+
+                    col_span = cell.end_col_offset_idx - cell.start_col_offset_idx
+                    if col_span in col_cells_with_bbox:
+                        col_cells_with_bbox[col_span].append(cell.bbox)
+                    else:
+                        col_cells_with_bbox[col_span] = [cell.bbox]
+
+            # Calculate the enclosing bounding box for this row
+            if len(col_cells_with_bbox) > 0:
+                min_col_span = min(col_cells_with_bbox.keys())
+                col_bbox: BoundingBox = BoundingBox.enclosing_bbox(
+                    col_cells_with_bbox[min_col_span]
+                )
+
+                for rspan, bboxs in col_cells_with_bbox.items():
+                    for bbox in bboxs:
+                        if bbox.coord_origin == CoordOrigin.TOPLEFT:
+                            col_bbox.b = max(col_bbox.b, bbox.b)
+                            col_bbox.t = min(col_bbox.t, bbox.t)
+
+                        elif bbox.coord_origin == CoordOrigin.BOTTOMLEFT:
+                            col_bbox.b = min(col_bbox.b, bbox.b)
+                            col_bbox.t = max(col_bbox.t, bbox.t)
+
+                col_bboxes[col_idx] = col_bbox
+
+        return col_bboxes
+
 
 class PictureTabularChartData(PictureChartData):
     """Base class for picture chart data.

{docling_core-2.36.0.dist-info → docling_core-2.37.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling-core
-Version: 2.36.0
+Version: 2.37.0
 Summary: A python library to define and validate data types in Docling.
 Author-email: Cesar Berrospi Ramis <ceb@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 Maintainer-email: Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>, Cesar Berrospi Ramis <ceb@zurich.ibm.com>

{docling_core-2.36.0.dist-info → docling_core-2.37.0.dist-info}/RECORD

@@ -29,19 +29,19 @@ docling_core/transforms/serializer/__init__.py,sha256=CECQlMoCDUxkg4RAUdC3itA3I3
 docling_core/transforms/serializer/base.py,sha256=ZFIiZeplL-QbBs9EDUb1awqxapQ23PsApVetJtAs7Vs,6891
 docling_core/transforms/serializer/common.py,sha256=WP-qO-woidrKyvZ56m0vlKMysoLrMzzZtHSCIwsl3ek,19119
 docling_core/transforms/serializer/doctags.py,sha256=PuAExlP-2HxcDSP_R_phtYQU0yKBW94RrPgb85IUxck,19905
-docling_core/transforms/serializer/html.py,sha256=KiywrroYBS3yk07gQizlmk3oqkXg_NpFwE0VF31_Z-I,37112
+docling_core/transforms/serializer/html.py,sha256=SZgQa0QnknEoRwMFLdgmVsLQqLF2rQl3D7XyEZzUHCE,37151
 docling_core/transforms/serializer/html_styles.py,sha256=-jBwS4EU7yfKoz0GSoxhwx90OmIKieO6TwPw57IuxcA,4692
 docling_core/transforms/serializer/markdown.py,sha256=wfMNrjA4wMehWLCejAhEN1eQPRixUO1SyL6ojkKkzZY,20614
 docling_core/transforms/visualizer/__init__.py,sha256=gUfF25yiJ_KO46ZIUNqZQOZGy2PLx6gnnr6AZYxKHXI,35
 docling_core/transforms/visualizer/base.py,sha256=aEF7b3rHq6DVdX8zDYEPoq55BHDYe4Hh_97lBdcW4lY,555
 docling_core/transforms/visualizer/layout_visualizer.py,sha256=hpq7OnyBgGxt3iW3_aNy9KH_0kmKdgoiJIFPcA2SSHU,8040
 docling_core/transforms/visualizer/reading_order_visualizer.py,sha256=yBra_W33bb16BxrTqP-ABu5NfRplTEJgu3dKdew3zKA,5601
-docling_core/transforms/visualizer/table_visualizer.py,sha256=XlLMSROyRW2UtAjKTltcESSs_rdQNKjO3QvO7ET7uc0,4275
+docling_core/transforms/visualizer/table_visualizer.py,sha256=iJPjk-XQSSCH3oujcjPMz-redAwNNHseZ41lFyd-u3k,8097
 docling_core/types/__init__.py,sha256=MVRSgsk5focwGyAplh_TRR3dEecIXpd98g_u3zZ5HXo,260
 docling_core/types/base.py,sha256=PusJskRVL19y-hq0BgXr5e8--QEqSqLnFNJ8UbOqW88,8318
 docling_core/types/doc/__init__.py,sha256=bysJn2iwjAHwThSWDPXEdVUUij7p_ax12_nx2_0CMdg,653
 docling_core/types/doc/base.py,sha256=ndXquBrOKTFQApIJ5s2-zstj3xlVKRbJDSId0KOQnUg,14817
-docling_core/types/doc/document.py,sha256=elFR5J7O9FUWXiweNK2W7S-cPvAakdzkMls0Uh4ViU8,149361
+docling_core/types/doc/document.py,sha256=JIrCXTeTYSbjTM1wt6kAbXF6QZ1OepC9vG2C3rO0j8I,153808
 docling_core/types/doc/labels.py,sha256=JiciRK7_DOkebsrfQ6PVCvS__TsKgWn1ANk84BeB14k,7359
 docling_core/types/doc/page.py,sha256=1JMPwglaTITBvg959L_pcWPb-fXoDYGh-e_tGZMzVMQ,41060
 docling_core/types/doc/tokens.py,sha256=z22l9J81_sg9CYMvOuLmPuLsNT7h_s7wao2UT89DvI8,9278
@@ -74,9 +74,9 @@ docling_core/utils/generate_jsonschema.py,sha256=uNX1O5XnjyB5nA66XqZXTt3YbGuR2ty
 docling_core/utils/legacy.py,sha256=DrI3QGoL755ZCIoKHF74-pTWm8R0zfFo2C2vB5dT2aY,24463
 docling_core/utils/validate.py,sha256=aQ11UbFyl8iD_N7yTTZmm_VVeXz8KcCyn3GLXgkfYRM,2049
 docling_core/utils/validators.py,sha256=azcrndLzhNkTWnbFSu9shJ5D3j_znnLrIFA5R8hzmGU,2798
-docling_core-2.36.0.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.36.0.dist-info/METADATA,sha256=8CnZkQHylNT1mgEEs_lIB18f2NL96R3kFAl-rBYVR0U,6453
-docling_core-2.36.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-docling_core-2.36.0.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
-docling_core-2.36.0.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
-docling_core-2.36.0.dist-info/RECORD,,
+docling_core-2.37.0.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-2.37.0.dist-info/METADATA,sha256=B0hyQog06wYqrKsB2jbeiAZ-Rk3Pl_uy2JH7Rws-9EY,6453
+docling_core-2.37.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+docling_core-2.37.0.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
+docling_core-2.37.0.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
+docling_core-2.37.0.dist-info/RECORD,,