docling-core 2.6.0__py3-none-any.whl → 2.7.0__py3-none-any.whl

docling_core/types/doc/document.py

@@ -37,8 +37,8 @@ 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.labels import DocItemLabel, GroupLabel
-from docling_core.types.legacy_doc.tokens import DocumentToken
-from docling_core.utils.file import relative_path
+from docling_core.types.doc.tokens import DocumentToken, TableToken
+from docling_core.types.doc.utils import relative_path
 
 Uint64 = typing.Annotated[int, Field(ge=0, le=(2**64 - 1))]
 LevelNumber = typing.Annotated[int, Field(ge=1, le=100)]
@@ -1008,7 +1008,6 @@ class TableItem(FloatingItem):
                 DeprecationWarning,
             )
 
-        body = ""
         nrows = self.data.num_rows
         ncols = self.data.num_cols
 
@@ -1065,6 +1064,99 @@ class TableItem(FloatingItem):
 
         return body
 
+    def export_to_otsl(
+        self,
+        doc: "DoclingDocument",
+        add_cell_location: bool = True,
+        add_cell_text: bool = True,
+        xsize: int = 100,
+        ysize: int = 100,
+    ) -> str:
+        """Export the table as OTSL."""
+        # Possible OTSL tokens...
+        #
+        # Empty and full cells:
+        # "ecel", "fcel"
+        #
+        # Cell spans (horisontal, vertical, 2d):
+        # "lcel", "ucel", "xcel"
+        #
+        # New line:
+        # "nl"
+        #
+        # Headers (column, row, section row):
+        # "ched", "rhed", "srow"
+
+        body = []
+        nrows = self.data.num_rows
+        ncols = self.data.num_cols
+        if len(self.data.table_cells) == 0:
+            return ""
+
+        page_no = 0
+        if len(self.prov) > 0:
+            page_no = self.prov[0].page_no
+
+        for i in range(nrows):
+            for j in range(ncols):
+                cell: TableCell = self.data.grid[i][j]
+                content = cell.text.strip()
+                rowspan, rowstart = (
+                    cell.row_span,
+                    cell.start_row_offset_idx,
+                )
+                colspan, colstart = (
+                    cell.col_span,
+                    cell.start_col_offset_idx,
+                )
+
+                if len(doc.pages.keys()):
+                    page_w, page_h = doc.pages[page_no].size.as_tuple()
+                cell_loc = ""
+                if cell.bbox is not None:
+                    cell_loc = DocumentToken.get_location(
+                        bbox=cell.bbox.to_bottom_left_origin(page_h).as_tuple(),
+                        page_w=page_w,
+                        page_h=page_h,
+                        xsize=xsize,
+                        ysize=ysize,
+                        page_i=page_no,
+                    )
+
+                if rowstart == i and colstart == j:
+                    if len(content) > 0:
+                        if cell.column_header:
+                            body.append(str(TableToken.OTSL_CHED.value))
+                        elif cell.row_header:
+                            body.append(str(TableToken.OTSL_RHED.value))
+                        elif cell.row_section:
+                            body.append(str(TableToken.OTSL_SROW.value))
+                        else:
+                            body.append(str(TableToken.OTSL_FCEL.value))
+                        if add_cell_location:
+                            body.append(str(cell_loc))
+                        if add_cell_text:
+                            body.append(str(content))
+                    else:
+                        body.append(str(TableToken.OTSL_ECEL.value))
+                else:
+                    add_cross_cell = False
+                    if rowstart != i:
+                        if colspan == 1:
+                            body.append(str(TableToken.OTSL_UCEL.value))
+                        else:
+                            add_cross_cell = True
+                    if colstart != j:
+                        if rowspan == 1:
+                            body.append(str(TableToken.OTSL_LCEL.value))
+                        else:
+                            add_cross_cell = True
+                    if add_cross_cell:
+                        body.append(str(TableToken.OTSL_XCEL.value))
+            body.append(str(TableToken.OTSL_NL.value))
+            body_str = "".join(body)
+        return body_str
+
     def export_to_document_tokens(
         self,
         doc: "DoclingDocument",

docling_core/types/doc/tokens.py

@@ -0,0 +1,202 @@
+#
+# Copyright IBM Corp. 2024 - 2024
+# SPDX-License-Identifier: MIT
+#
+
+"""Tokens used in the docling document model."""
+
+from enum import Enum
+from typing import Annotated, Tuple
+
+from pydantic import Field
+
+
+class TableToken(Enum):
+    """Class to represent an LLM friendly representation of a Table."""
+
+    CELL_LABEL_COLUMN_HEADER = "<column_header>"
+    CELL_LABEL_ROW_HEADER = "<row_header>"
+    CELL_LABEL_SECTION_HEADERE = "<section_header>"
+    CELL_LABEL_DATA = "<data>"
+
+    OTSL_ECEL = "<ecel>"  # empty cell
+    OTSL_FCEL = "<fcel>"  # cell with content
+    OTSL_LCEL = "<lcel>"  # left looking cell,
+    OTSL_UCEL = "<ucel>"  # up looking cell,
+    OTSL_XCEL = "<xcel>"  # 2d extension cell (cross cell),
+    OTSL_NL = "<nl>"  # new line,
+    OTSL_CHED = "<ched>"  # - column header cell,
+    OTSL_RHED = "<rhed>"  # - row header cell,
+    OTSL_SROW = "<srow>"  # - section row cell
+
+    @classmethod
+    def get_special_tokens(cls):
+        """Function to get all special document tokens."""
+        special_tokens = [token.value for token in cls]
+        return special_tokens
+
+    @staticmethod
+    def is_known_token(label):
+        """Function to check if label is in tokens."""
+        return label in TableToken.get_special_tokens()
+
+
+class DocumentToken(Enum):
+    """Class to represent an LLM friendly representation of a Document."""
+
+    BEG_DOCUMENT = "<document>"
+    END_DOCUMENT = "</document>"
+
+    BEG_TITLE = "<title>"
+    END_TITLE = "</title>"
+
+    BEG_ABSTRACT = "<abstract>"
+    END_ABSTRACT = "</abstract>"
+
+    BEG_DOI = "<doi>"
+    END_DOI = "</doi>"
+    BEG_DATE = "<date>"
+    END_DATE = "</date>"
+
+    BEG_AUTHORS = "<authors>"
+    END_AUTHORS = "</authors>"
+    BEG_AUTHOR = "<author>"
+    END_AUTHOR = "</author>"
+
+    BEG_AFFILIATIONS = "<affiliations>"
+    END_AFFILIATIONS = "</affiliations>"
+    BEG_AFFILIATION = "<affiliation>"
+    END_AFFILIATION = "</affiliation>"
+
+    BEG_HEADER = "<section-header>"
+    END_HEADER = "</section-header>"
+    BEG_TEXT = "<text>"
+    END_TEXT = "</text>"
+    BEG_PARAGRAPH = "<paragraph>"
+    END_PARAGRAPH = "</paragraph>"
+    BEG_TABLE = "<table>"
+    END_TABLE = "</table>"
+    BEG_FIGURE = "<figure>"
+    END_FIGURE = "</figure>"
+    BEG_CAPTION = "<caption>"
+    END_CAPTION = "</caption>"
+    BEG_EQUATION = "<equation>"
+    END_EQUATION = "</equation>"
+    BEG_LIST = "<list>"
+    END_LIST = "</list>"
+    BEG_LISTITEM = "<list-item>"
+    END_LISTITEM = "</list-item>"
+
+    BEG_LOCATION = "<location>"
+    END_LOCATION = "</location>"
+    BEG_GROUP = "<group>"
+    END_GROUP = "</group>"
+
+    @classmethod
+    def get_special_tokens(
+        cls,
+        max_rows: int = 100,
+        max_cols: int = 100,
+        max_pages: int = 1000,
+        page_dimension: Tuple[int, int] = (100, 100),
+    ):
+        """Function to get all special document tokens."""
+        special_tokens = [token.value for token in cls]
+
+        # Adding dynamically generated row and col tokens
+        for i in range(0, max_rows + 1):
+            special_tokens += [f"<row_{i}>", f"</row_{i}>"]
+
+        for i in range(0, max_cols + 1):
+            special_tokens += [f"<col_{i}>", f"</col_{i}>"]
+
+        for i in range(6):
+            special_tokens += [f"<section-header-{i}>", f"</section-header-{i}>"]
+
+        # FIXME: this is synonym of section header
+        for i in range(6):
+            special_tokens += [f"<subtitle-level-{i}>", f"</subtitle-level-{i}>"]
+
+        # Adding dynamically generated page-tokens
+        for i in range(0, max_pages + 1):
+            special_tokens.append(f"<page_{i}>")
+            special_tokens.append(f"</page_{i}>")
+
+        # Adding dynamically generated location-tokens
+        for i in range(0, max(page_dimension[0] + 1, page_dimension[1] + 1)):
+            special_tokens.append(f"<loc_{i}>")
+
+        return special_tokens
+
+    @staticmethod
+    def is_known_token(label):
+        """Function to check if label is in tokens."""
+        return label in DocumentToken.get_special_tokens()
+
+    @staticmethod
+    def get_row_token(row: int, beg=bool) -> str:
+        """Function to get page tokens."""
+        if beg:
+            return f"<row_{row}>"
+        else:
+            return f"</row_{row}>"
+
+    @staticmethod
+    def get_col_token(col: int, beg=bool) -> str:
+        """Function to get page tokens."""
+        if beg:
+            return f"<col_{col}>"
+        else:
+            return f"</col_{col}>"
+
+    @staticmethod
+    def get_page_token(page: int):
+        """Function to get page tokens."""
+        return f"<page_{page}>"
+
+    @staticmethod
+    def get_location_token(val: float, rnorm: int = 100):
+        """Function to get location tokens."""
+        val_ = round(rnorm * val)
+
+        if val_ < 0:
+            return "<loc_0>"
+
+        if val_ > rnorm:
+            return f"<loc_{rnorm}>"
+
+        return f"<loc_{val_}>"
+
+    @staticmethod
+    def get_location(
+        # bbox: Tuple[float, float, float, float],
+        bbox: Annotated[list[float], Field(min_length=4, max_length=4)],
+        page_w: float,
+        page_h: float,
+        xsize: int = 100,
+        ysize: int = 100,
+        page_i: int = -1,
+    ):
+        """Get the location string give bbox and page-dim."""
+        assert bbox[0] <= bbox[2], f"bbox[0]<=bbox[2] => {bbox[0]}<={bbox[2]}"
+        assert bbox[1] <= bbox[3], f"bbox[1]<=bbox[3] => {bbox[1]}<={bbox[3]}"
+
+        x0 = bbox[0] / page_w
+        y0 = bbox[1] / page_h
+        x1 = bbox[2] / page_w
+        y1 = bbox[3] / page_h
+
+        page_tok = ""
+        if page_i != -1:
+            page_tok = DocumentToken.get_page_token(page=page_i)
+
+        x0_tok = DocumentToken.get_location_token(val=min(x0, x1), rnorm=xsize)
+        y0_tok = DocumentToken.get_location_token(val=min(y0, y1), rnorm=ysize)
+        x1_tok = DocumentToken.get_location_token(val=max(x0, x1), rnorm=xsize)
+        y1_tok = DocumentToken.get_location_token(val=max(y0, y1), rnorm=ysize)
+
+        loc_str = f"{DocumentToken.BEG_LOCATION.value}"
+        loc_str += f"{page_tok}{x0_tok}{y0_tok}{x1_tok}{y1_tok}"
+        loc_str += f"{DocumentToken.END_LOCATION.value}"
+
+        return loc_str

docling_core/types/doc/utils.py

@@ -0,0 +1,48 @@
+#
+# Copyright IBM Corp. 2024 - 2024
+# SPDX-License-Identifier: MIT
+#
+
+"""Utils for document types."""
+
+from pathlib import Path
+
+
+def relative_path(src: Path, target: Path) -> Path:
+    """Compute the relative path from `src` to `target`.
+
+    Args:
+        src (str | Path): The source directory or file path (must be absolute).
+        target (str | Path): The target directory or file path (must be absolute).
+
+    Returns:
+        Path: The relative path from `src` to `target`.
+
+    Raises:
+        ValueError: If either `src` or `target` is not an absolute path.
+    """
+    src = Path(src).resolve()
+    target = Path(target).resolve()
+
+    # Ensure both paths are absolute
+    if not src.is_absolute():
+        raise ValueError(f"The source path must be absolute: {src}")
+    if not target.is_absolute():
+        raise ValueError(f"The target path must be absolute: {target}")
+
+    # Find the common ancestor
+    common_parts = []
+    for src_part, target_part in zip(src.parts, target.parts):
+        if src_part == target_part:
+            common_parts.append(src_part)
+        else:
+            break
+
+    # Determine the path to go up from src to the common ancestor
+    up_segments = [".."] * (len(src.parts) - len(common_parts))
+
+    # Add the path from the common ancestor to the target
+    down_segments = target.parts[len(common_parts) :]
+
+    # Combine and return the result
+    return Path(*up_segments, *down_segments)

docling_core/utils/file.py

@@ -15,6 +15,7 @@ import requests
 from pydantic import AnyHttpUrl, TypeAdapter, ValidationError
 from typing_extensions import deprecated
 
+from docling_core.types.doc.utils import relative_path  # noqa
 from docling_core.types.io import DocumentStream
 
 
@@ -168,43 +169,3 @@ def resolve_file_source(
         source=source,
         headers=headers,
     )
-
-
-def relative_path(src: Path, target: Path) -> Path:
-    """Compute the relative path from `src` to `target`.
-
-    Args:
-        src (str | Path): The source directory or file path (must be absolute).
-        target (str | Path): The target directory or file path (must be absolute).
-
-    Returns:
-        Path: The relative path from `src` to `target`.
-
-    Raises:
-        ValueError: If either `src` or `target` is not an absolute path.
-    """
-    src = Path(src).resolve()
-    target = Path(target).resolve()
-
-    # Ensure both paths are absolute
-    if not src.is_absolute():
-        raise ValueError(f"The source path must be absolute: {src}")
-    if not target.is_absolute():
-        raise ValueError(f"The target path must be absolute: {target}")
-
-    # Find the common ancestor
-    common_parts = []
-    for src_part, target_part in zip(src.parts, target.parts):
-        if src_part == target_part:
-            common_parts.append(src_part)
-        else:
-            break
-
-    # Determine the path to go up from src to the common ancestor
-    up_segments = [".."] * (len(src.parts) - len(common_parts))
-
-    # Add the path from the common ancestor to the target
-    down_segments = target.parts[len(common_parts) :]
-
-    # Combine and return the result
-    return Path(*up_segments, *down_segments)

{docling_core-2.6.0.dist-info → docling_core-2.7.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.6.0
+Version: 2.7.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://ds4sd.github.io/
 License: MIT
@@ -29,7 +29,7 @@ Requires-Dist: jsonref (>=1.1.0,<2.0.0)
 Requires-Dist: jsonschema (>=4.16.0,<5.0.0)
 Requires-Dist: pandas (>=2.1.4,<3.0.0)
 Requires-Dist: pillow (>=10.3.0,<11.0.0)
-Requires-Dist: pydantic (>=2.6.0,<2.10)
+Requires-Dist: pydantic (>=2.6.0,<3.0.0,!=2.10.0,!=2.10.1,!=2.10.2)
 Requires-Dist: pyyaml (>=5.1,<7.0.0)
 Requires-Dist: tabulate (>=0.9.0,<0.10.0)
 Requires-Dist: typing-extensions (>=4.12.2,<5.0.0)

{docling_core-2.6.0.dist-info → docling_core-2.7.0.dist-info}/RECORD

@@ -21,8 +21,10 @@ docling_core/types/__init__.py,sha256=MVRSgsk5focwGyAplh_TRR3dEecIXpd98g_u3zZ5HX
 docling_core/types/base.py,sha256=PusJskRVL19y-hq0BgXr5e8--QEqSqLnFNJ8UbOqW88,8318
 docling_core/types/doc/__init__.py,sha256=bEL4zKVOG7Wxm6xQrgF58mu-Teds9aSavuEAKVNhrTU,639
 docling_core/types/doc/base.py,sha256=_ttU8QI8wXDTQRUnN5n7L6D9wYFVLSAibxlFoMbgAsk,4557
-docling_core/types/doc/document.py,sha256=K6ixUeB0vyrnd3_ljM0Ed_8JBdltLPCsrGz7IoLgjUI,87094
+docling_core/types/doc/document.py,sha256=LXmDD0qZiB34WTWSTklcdWndetOqumMFN3yJEqifb8M,90500
 docling_core/types/doc/labels.py,sha256=A8vWP82VAeXO1rlCO0oDKo_Hb8uDeQe0myOTY3P03hk,1596
+docling_core/types/doc/tokens.py,sha256=uU_MYW_p7ypf7eYICFBvxdnVaPZ7CQnvZmbJ6oPrtEA,6134
+docling_core/types/doc/utils.py,sha256=YDOh_ZD1Y7OmCEDdCLJ_MO5K3HA67nc_acfhOK6WztU,1439
 docling_core/types/gen/__init__.py,sha256=C6TuCfvpSnSL5XDOFMcYHUY2-i08vvfOGRcdu6Af0pI,124
 docling_core/types/gen/generic.py,sha256=l4CZ4_Lb8ONG36WNJWbKX5hGKvTh_yU-hXp5hsm7uVU,844
 docling_core/types/io/__init__.py,sha256=7QYvFRaDE0AzBg8e7tvsVNlLBbCbAbQ9rP2TU8aXR1k,350
@@ -45,13 +47,13 @@ docling_core/types/rec/statement.py,sha256=YwcV4CbVaAbzNwh14yJ_6Py3Ww0XnUJrEEUiK
 docling_core/types/rec/subject.py,sha256=PRCERGTMs4YhR3_Ne6jogkm41zYg8uUWb1yFpM7atm4,2572
 docling_core/utils/__init__.py,sha256=VauNNpWRHG0_ISKrsy5-gTxicrdQZSau6qMfuMl3iqk,120
 docling_core/utils/alias.py,sha256=B6Lqvss8CbaNARHLR4qSmNh9OkB6LvqTpxfsFmkLAFo,874
-docling_core/utils/file.py,sha256=B1Iu8buqk_Yz4bhrGf7NyFIiYlsa_MC37vZLwQHqKLU,6876
+docling_core/utils/file.py,sha256=GzX0pclvewwPoqHJSaVUuULzSJwJgkCUwgKgJ7G5ohQ,5628
 docling_core/utils/generate_docs.py,sha256=BdKAoduWXOc7YMvcmlhjoJOFlUxij1ybxglj6LZDtC8,2290
 docling_core/utils/generate_jsonschema.py,sha256=uNX1O5XnjyB5nA66XqZXTt3YbGuR2tyi_OhHepHYtZg,1654
 docling_core/utils/validate.py,sha256=3FmnxnKTDZC5J9OGxCL3U3DGRl0t0bBV1NcySXswdas,2031
 docling_core/utils/validators.py,sha256=azcrndLzhNkTWnbFSu9shJ5D3j_znnLrIFA5R8hzmGU,2798
-docling_core-2.6.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.6.0.dist-info/METADATA,sha256=LhnsqU5AgndZllazTDXe_acmPWQ6NuMuH_b6-d4K1gM,5519
-docling_core-2.6.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-docling_core-2.6.0.dist-info/entry_points.txt,sha256=jIxlWv3tnO04irlZc0zfhqJIgz1bg9Hha4AkaLWSdUA,177
-docling_core-2.6.0.dist-info/RECORD,,
+docling_core-2.7.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-2.7.0.dist-info/METADATA,sha256=ht4UM23KfXIPp2aeUjSr9AUruTANa-kSt9kDwHQyeNk,5547
+docling_core-2.7.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+docling_core-2.7.0.dist-info/entry_points.txt,sha256=jIxlWv3tnO04irlZc0zfhqJIgz1bg9Hha4AkaLWSdUA,177
+docling_core-2.7.0.dist-info/RECORD,,