docling-core 2.17.0__tar.gz → 2.17.2__tar.gz

{docling_core-2.17.0 → docling_core-2.17.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.17.0
+Version: 2.17.2
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://ds4sd.github.io/
 License: MIT

{docling_core-2.17.0 → docling_core-2.17.2}/docling_core/types/doc/document.py

@@ -5,6 +5,7 @@ import copy
 import hashlib
 import html
 import json
+import logging
 import mimetypes
 import os
 import re
@@ -20,6 +21,7 @@ from xml.etree.cElementTree import SubElement, tostring
 from xml.sax.saxutils import unescape
 
 import latex2mathml.converter
+import latex2mathml.exceptions
 import pandas as pd
 import yaml
 from PIL import Image as PILImage
@@ -42,7 +44,13 @@ from docling_core.types.doc import BoundingBox, Size
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.labels import CodeLanguageLabel, DocItemLabel, GroupLabel
 from docling_core.types.doc.tokens import DocumentToken, TableToken
-from docling_core.types.doc.utils import relative_path
+from docling_core.types.doc.utils import (
+    get_html_tag_with_text_direction,
+    get_text_direction,
+    relative_path,
+)
+
+_logger = logging.getLogger(__name__)
 
 Uint64 = typing.Annotated[int, Field(ge=0, le=(2**64 - 1))]
 LevelNumber = typing.Annotated[int, Field(ge=1, le=100)]
@@ -862,7 +870,9 @@ class PictureItem(FloatingItem):
 
         caption_text = ""
         if len(text) > 0:
-            caption_text = f"<figcaption>{text}</figcaption>"
+            caption_text = get_html_tag_with_text_direction(
+                html_tag="figcaption", text=text
+            )
 
         default_response = f"<figure>{caption_text}</figure>"
 
@@ -1086,15 +1096,28 @@ class TableItem(FloatingItem):
                 if colspan > 1:
                     opening_tag += f' colspan="{colspan}"'
 
+                text_dir = get_text_direction(content)
+                if text_dir == "rtl":
+                    opening_tag += f' dir="{dir}"'
+
                 body += f"<{opening_tag}>{content}</{celltag}>"
             body += "</tr>"
 
+        # dir = get_text_direction(text)
+
         if len(text) > 0 and len(body) > 0:
-            body = f"<table><caption>{text}</caption><tbody>{body}</tbody></table>"
+            caption_text = get_html_tag_with_text_direction(
+                html_tag="caption", text=text
+            )
+            body = f"<table>{caption_text}<tbody>{body}</tbody></table>"
+
         elif len(text) == 0 and len(body) > 0:
             body = f"<table><tbody>{body}</tbody></table>"
         elif len(text) > 0 and len(body) == 0:
-            body = f"<table><caption>{text}</caption></table>"
+            caption_text = get_html_tag_with_text_direction(
+                html_tag="caption", text=text
+            )
+            body = f"<table>{caption_text}</table>"
         else:
             body = "<table></table>"
 
@@ -2466,17 +2489,17 @@ class DoclingDocument(BaseModel):
                 continue
 
             elif isinstance(item, TextItem) and item.label in [DocItemLabel.TITLE]:
+                text_inner = _prepare_tag_content(item.text)
+                text = get_html_tag_with_text_direction(html_tag="h1", text=text_inner)
 
-                text = f"<h1>{_prepare_tag_content(item.text)}</h1>"
                 html_texts.append(text)
 
             elif isinstance(item, SectionHeaderItem):
 
                 section_level: int = min(item.level + 1, 6)
 
-                text = (
-                    f"<h{(section_level)}>"
-                    f"{_prepare_tag_content(item.text)}</h{(section_level)}>"
+                text = get_html_tag_with_text_direction(
+                    html_tag=f"h{section_level}", text=_prepare_tag_content(item.text)
                 )
                 html_texts.append(text)
 
@@ -2487,34 +2510,47 @@ class DoclingDocument(BaseModel):
                 )
                 text = ""
 
-                # If the formula is not processed correcty, use its image
-                if (
-                    item.text == ""
-                    and item.orig != ""
-                    and image_mode == ImageRefMode.EMBEDDED
-                    and len(item.prov) > 0
-                ):
+                def _image_fallback(item: TextItem):
                     item_image = item.get_image(doc=self)
                     if item_image is not None:
                         img_ref = ImageRef.from_pil(item_image, dpi=72)
-                        text = (
+                        return (
                             "<figure>"
                             f'<img src="{img_ref.uri}" alt="{item.orig}" />'
                             "</figure>"
                         )
 
+                # If the formula is not processed correcty, use its image
+                if (
+                    item.text == ""
+                    and item.orig != ""
+                    and image_mode == ImageRefMode.EMBEDDED
+                    and len(item.prov) > 0
+                ):
+                    text = _image_fallback(item)
+
                 # Building a math equation in MathML format
                 # ref https://www.w3.org/TR/wai-aria-1.1/#math
                 elif formula_to_mathml:
-                    mathml_element = latex2mathml.converter.convert_to_element(
-                        math_formula, display="block"
-                    )
-                    annotation = SubElement(
-                        mathml_element, "annotation", dict(encoding="TeX")
-                    )
-                    annotation.text = math_formula
-                    mathml = unescape(tostring(mathml_element, encoding="unicode"))
-                    text = f"<div>{mathml}</div>"
+                    try:
+                        mathml_element = latex2mathml.converter.convert_to_element(
+                            math_formula, display="block"
+                        )
+                        annotation = SubElement(
+                            mathml_element, "annotation", dict(encoding="TeX")
+                        )
+                        annotation.text = math_formula
+                        mathml = unescape(tostring(mathml_element, encoding="unicode"))
+                        text = f"<div>{mathml}</div>"
+                    except Exception as err:
+                        _logger.warning(
+                            "Malformed formula cannot be rendered. "
+                            f"Error {err.__class__.__name__}, formula={math_formula}"
+                        )
+                        if image_mode == ImageRefMode.EMBEDDED and len(item.prov) > 0:
+                            text = _image_fallback(item)
+                        else:
+                            text = f"<pre>{math_formula}</pre>"
 
                 elif math_formula != "":
                     text = f"<pre>{math_formula}</pre>"
@@ -2527,13 +2563,15 @@ class DoclingDocument(BaseModel):
                     )
 
             elif isinstance(item, ListItem):
-
-                text = f"<li>{_prepare_tag_content(item.text)}</li>"
+                text = get_html_tag_with_text_direction(
+                    html_tag="li", text=_prepare_tag_content(item.text)
+                )
                 html_texts.append(text)
 
             elif isinstance(item, TextItem) and item.label in [DocItemLabel.LIST_ITEM]:
-
-                text = f"<li>{_prepare_tag_content(item.text)}</li>"
+                text = get_html_tag_with_text_direction(
+                    html_tag="li", text=_prepare_tag_content(item.text)
+                )
                 html_texts.append(text)
 
             elif isinstance(item, CodeItem):
@@ -2545,8 +2583,11 @@ class DoclingDocument(BaseModel):
 
             elif isinstance(item, TextItem):
 
-                text = f"<p>{_prepare_tag_content(item.text)}</p>"
+                text = get_html_tag_with_text_direction(
+                    html_tag="p", text=_prepare_tag_content(item.text)
+                )
                 html_texts.append(text)
+
             elif isinstance(item, TableItem):
 
                 text = item.export_to_html(doc=self, add_caption=True)

{docling_core-2.17.0 → docling_core-2.17.2}/docling_core/types/doc/utils.py

@@ -5,6 +5,7 @@
 
 """Utils for document types."""
 
+import unicodedata
 from pathlib import Path
 
 
@@ -46,3 +47,29 @@ def relative_path(src: Path, target: Path) -> Path:
 
     # Combine and return the result
     return Path(*up_segments, *down_segments)
+
+
+def get_html_tag_with_text_direction(html_tag: str, text: str) -> str:
+    """Form the HTML element with tag, text, and optional dir attribute."""
+    text_dir = get_text_direction(text)
+
+    if text_dir == "ltr":
+        return f"<{html_tag}>{text}</{html_tag}>"
+    else:
+        return f'<{html_tag} dir="{text_dir}">{text}</{html_tag}>'
+
+
+def get_text_direction(text: str) -> str:
+    """Determine the text direction of a given string as LTR or RTL script."""
+    if not text:
+        return "ltr"  # Default for empty input
+
+    rtl_scripts = {"R", "AL"}
+    rtl_chars = sum(unicodedata.bidirectional(c) in rtl_scripts for c in text)
+
+    return (
+        "rtl"
+        if unicodedata.bidirectional(text[0]) in rtl_scripts
+        or rtl_chars > len(text) / 2
+        else "ltr"
+    )

{docling_core-2.17.0 → docling_core-2.17.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "docling-core"
-version = "2.17.0"
+version = "2.17.2"
 description = "A python library to define and validate data types in Docling."
 license = "MIT"
 authors = [