docling 2.41.0__py3-none-any.whl → 2.42.0__py3-none-any.whl

docling/backend/docx/latex/omml.py

@@ -260,7 +260,15 @@ class oMath2Latex(Tag2Method):
         the fraction object
         """
         c_dict = self.process_children_dict(elm)
-        pr = c_dict["fPr"]
+        pr = c_dict.get("fPr")
+        if pr is None:
+            # Handle missing fPr element gracefully
+            _log.debug("Missing fPr element in fraction, using default formatting")
+            latex_s = F_DEFAULT
+            return latex_s.format(
+                num=c_dict.get("num"),
+                den=c_dict.get("den"),
+            )
         latex_s = get_val(pr.type, default=F_DEFAULT, store=F)
         return pr.text + latex_s.format(num=c_dict.get("num"), den=c_dict.get("den"))
 

docling/backend/html_backend.py

@@ -379,6 +379,25 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
         else:
             _log.debug(f"list-item has no text: {element}")
 
+    @staticmethod
+    def _get_cell_spans(cell: Tag) -> tuple[int, int]:
+        """Extract colspan and rowspan values from a table cell tag.
+
+        This function retrieves the 'colspan' and 'rowspan' attributes from a given
+        table cell tag.
+        If the attribute does not exist or it is not numeric, it defaults to 1.
+        """
+        raw_spans: tuple[str, str] = (
+            str(cell.get("colspan", "1")),
+            str(cell.get("rowspan", "1")),
+        )
+        int_spans: tuple[int, int] = (
+            int(raw_spans[0]) if raw_spans[0].isnumeric() else 1,
+            int(raw_spans[1]) if raw_spans[0].isnumeric() else 1,
+        )
+
+        return int_spans
+
     @staticmethod
     def parse_table_data(element: Tag) -> Optional[TableData]:  # noqa: C901
         nested_tables = element.find("table")
@@ -398,10 +417,9 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
                 if not isinstance(row, Tag):
                     continue
                 cell_tag = cast(Tag, cell)
-                val = cell_tag.get("colspan", "1")
-                colspan = int(val) if (isinstance(val, str) and val.isnumeric()) else 1
-                col_count += colspan
-                if cell_tag.name == "td" or cell_tag.get("rowspan") is None:
+                col_span, row_span = HTMLDocumentBackend._get_cell_spans(cell_tag)
+                col_count += col_span
+                if cell_tag.name == "td" or row_span == 1:
                     is_row_header = False
             num_cols = max(num_cols, col_count)
             if not is_row_header:
@@ -428,10 +446,11 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
             row_header = True
             for html_cell in cells:
                 if isinstance(html_cell, Tag):
+                    _, row_span = HTMLDocumentBackend._get_cell_spans(html_cell)
                     if html_cell.name == "td":
                         col_header = False
                         row_header = False
-                    elif html_cell.get("rowspan") is None:
+                    elif row_span == 1:
                         row_header = False
             if not row_header:
                 row_idx += 1
@@ -456,18 +475,7 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
                 text = html_cell.text
 
                 # label = html_cell.name
-                col_val = html_cell.get("colspan", "1")
-                col_span = (
-                    int(col_val)
-                    if isinstance(col_val, str) and col_val.isnumeric()
-                    else 1
-                )
-                row_val = html_cell.get("rowspan", "1")
-                row_span = (
-                    int(row_val)
-                    if isinstance(row_val, str) and row_val.isnumeric()
-                    else 1
-                )
+                col_span, row_span = HTMLDocumentBackend._get_cell_spans(html_cell)
                 if row_header:
                     row_span -= 1
                 while (

docling/backend/xml/jats_backend.py

@@ -93,8 +93,8 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
 
         # Initialize the root of the document hierarchy
         self.root: Optional[NodeItem] = None
-
-        self.valid = False
+        self.hlevel: int = 0
+        self.valid: bool = False
         try:
             if isinstance(self.path_or_stream, BytesIO):
                 self.path_or_stream.seek(0)
@@ -147,6 +147,7 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 binary_hash=self.document_hash,
             )
             doc = DoclingDocument(name=self.file.stem or "file", origin=origin)
+            self.hlevel = 0
 
             # Get metadata XML components
             xml_components: XMLComponents = self._parse_metadata()
@@ -304,7 +305,9 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
             title: str = abstract["label"] or DEFAULT_HEADER_ABSTRACT
             if not text:
                 continue
-            parent = doc.add_heading(parent=self.root, text=title)
+            parent = doc.add_heading(
+                parent=self.root, text=title, level=self.hlevel + 1
+            )
             doc.add_text(
                 parent=parent,
                 text=text,
@@ -637,7 +640,10 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 elif child.tag == "ack":
                     text = DEFAULT_HEADER_ACKNOWLEDGMENTS
                 if text:
-                    new_parent = doc.add_heading(text=text, parent=parent)
+                    self.hlevel += 1
+                    new_parent = doc.add_heading(
+                        text=text, parent=parent, level=self.hlevel
+                    )
             elif child.tag == "list":
                 new_parent = doc.add_group(
                     label=GroupLabel.LIST, name="list", parent=parent
@@ -694,6 +700,8 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 new_text = self._walk_linear(doc, new_parent, child)
                 if not (node.getparent().tag == "p" and node.tag in flush_tags):
                     node_text += new_text
+                if child.tag in ("sec", "ack") and text:
+                    self.hlevel -= 1
 
             # pick up the tail text
             node_text += child.tail.replace("\n", " ") if child.tail else ""

docling/datamodel/pipeline_options.py

@@ -217,7 +217,7 @@ smolvlm_picture_description = PictureDescriptionVlmOptions(
 
 # GraniteVision
 granite_picture_description = PictureDescriptionVlmOptions(
-    repo_id="ibm-granite/granite-vision-3.2-2b-preview",
+    repo_id="ibm-granite/granite-vision-3.3-2b",
     prompt="What is shown in this image?",
 )
 
@@ -279,6 +279,9 @@ class LayoutOptions(BaseModel):
     """Options for layout processing."""
 
     create_orphan_clusters: bool = True  # Whether to create clusters for orphaned cells
+    keep_empty_clusters: bool = (
+        False  # Whether to keep clusters that contain no text cells
+    )
     model_spec: LayoutModelConfig = DOCLING_LAYOUT_V2
 
 

docling/document_converter.py

@@ -1,6 +1,7 @@
 import hashlib
 import logging
 import sys
+import threading
 import time
 from collections.abc import Iterable, Iterator
 from functools import partial
@@ -49,6 +50,7 @@ from docling.pipeline.standard_pdf_pipeline import StandardPdfPipeline
 from docling.utils.utils import chunkify
 
 _log = logging.getLogger(__name__)
+_PIPELINE_CACHE_LOCK = threading.Lock()
 
 
 class FormatOption(BaseModel):
@@ -315,17 +317,18 @@ class DocumentConverter:
         # Use a composite key to cache pipelines
         cache_key = (pipeline_class, options_hash)
 
-        if cache_key not in self.initialized_pipelines:
-            _log.info(
-                f"Initializing pipeline for {pipeline_class.__name__} with options hash {options_hash}"
-            )
-            self.initialized_pipelines[cache_key] = pipeline_class(
-                pipeline_options=pipeline_options
-            )
-        else:
-            _log.debug(
-                f"Reusing cached pipeline for {pipeline_class.__name__} with options hash {options_hash}"
-            )
+        with _PIPELINE_CACHE_LOCK:
+            if cache_key not in self.initialized_pipelines:
+                _log.info(
+                    f"Initializing pipeline for {pipeline_class.__name__} with options hash {options_hash}"
+                )
+                self.initialized_pipelines[cache_key] = pipeline_class(
+                    pipeline_options=pipeline_options
+                )
+            else:
+                _log.debug(
+                    f"Reusing cached pipeline for {pipeline_class.__name__} with options hash {options_hash}"
+                )
 
         return self.initialized_pipelines[cache_key]
 

docling/models/picture_description_vlm_model.py

@@ -65,6 +65,7 @@ class PictureDescriptionVlmModel(
                 self.processor = AutoProcessor.from_pretrained(artifacts_path)
                 self.model = AutoModelForVision2Seq.from_pretrained(
                     artifacts_path,
+                    device_map=self.device,
                     torch_dtype=torch.bfloat16,
                     _attn_implementation=(
                         "flash_attention_2"
@@ -72,7 +73,7 @@ class PictureDescriptionVlmModel(
                         and accelerator_options.cuda_use_flash_attention2
                         else "eager"
                     ),
-                ).to(self.device)
+                )
 
             self.provenance = f"{self.options.repo_id}"
 

docling/utils/layout_postprocessor.py

@@ -267,8 +267,9 @@ class LayoutPostprocessor:
         # Initial cell assignment
         clusters = self._assign_cells_to_clusters(clusters)
 
-        # Remove clusters with no cells
-        clusters = [cluster for cluster in clusters if cluster.cells]
+        # Remove clusters with no cells (if keep_empty_clusters is False)
+        if not self.options.keep_empty_clusters:
+            clusters = [cluster for cluster in clusters if cluster.cells]
 
         # Handle orphaned cells
         unassigned = self._find_unassigned_cells(clusters)

{docling-2.41.0.dist-info → docling-2.42.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling
-Version: 2.41.0
+Version: 2.42.0
 Summary: SDK and CLI for parsing PDF, DOCX, HTML, and more, to a unified document representation for powering downstream workflows such as gen AI applications.
 Author-email: Christoph Auer <cau@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Maxim Lysak <mly@zurich.ibm.com>, Nikos Livathinos <nli@zurich.ibm.com>, Ahmed Nassar <ahn@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 License-Expression: MIT
@@ -50,6 +50,7 @@ Requires-Dist: tqdm<5.0.0,>=4.65.0
 Requires-Dist: pluggy<2.0.0,>=1.0.0
 Requires-Dist: pylatexenc<3.0,>=2.10
 Requires-Dist: scipy<2.0.0,>=1.6.0
+Requires-Dist: accelerate<2,>=1.0.0
 Provides-Extra: tesserocr
 Requires-Dist: tesserocr<3.0.0,>=2.7.1; extra == "tesserocr"
 Provides-Extra: ocrmac

{docling-2.41.0.dist-info → docling-2.42.0.dist-info}/RECORD

@@ -1,5 +1,5 @@
 docling/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-docling/document_converter.py,sha256=3jWywP_TLy-1PMvjJBUlnTM9FNzpBLRCHYA1RKFvGR4,14333
+docling/document_converter.py,sha256=9aH8B30_jOYN4P_ySCCvtgEb3GoIpec15r7lEAFlMDU,14469
 docling/exceptions.py,sha256=K1WnCS1leK2JtMB5ewZWKkb0EaijFgl-tRzrO9ntgPM,134
 docling/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 docling/backend/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -9,7 +9,7 @@ docling/backend/csv_backend.py,sha256=2g9famYG2W-ID9jEdZPxc6O8QGv1vWQfjN8pL-QMBE
 docling/backend/docling_parse_backend.py,sha256=9rUo1vPxX6QLzGqF-2B2iEYglZg6YQ3Uea00XrLluTg,7918
 docling/backend/docling_parse_v2_backend.py,sha256=3ckTfke8IICjaImlIzc3TRhG7KDuxDDba0AuCEcjA-M,9500
 docling/backend/docling_parse_v4_backend.py,sha256=qR_WRVq9JGtRioWCw6MnLWgbvXbC6Y1yds7Ol1-E6UQ,6550
-docling/backend/html_backend.py,sha256=Z959dzqYQO2pPE4xgPRxC5MR9j3nFGtiD6_F_osQ2iI,20670
+docling/backend/html_backend.py,sha256=xyCbCGR3vYNl-wSP2YJRgSCy9kIIMKKu28AUylPEUq8,20959
 docling/backend/md_backend.py,sha256=mfwGj8g2hGC-Q_HREtl_Web65uMVXD-Ie1nRqWTXzF0,21013
 docling/backend/msexcel_backend.py,sha256=cq8MQ2RSh6pqCiVrldjOerSww7dOPTWmCQoCBI57i6w,18579
 docling/backend/mspowerpoint_backend.py,sha256=wJgB2JStEPfD7MPpWQlpPN7bffPxaHFUnKD4wj8SLxU,15114
@@ -20,11 +20,11 @@ docling/backend/pypdfium2_backend.py,sha256=8dVniLHgiTdJuDbYr66kPp6Ccv5ZDlqDMEbA
 docling/backend/docx/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 docling/backend/docx/latex/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 docling/backend/docx/latex/latex_dict.py,sha256=tFJp4ScT_AkY2ON7nLEa560p601Jq2glcZvMKxxjn7w,6593
-docling/backend/docx/latex/omml.py,sha256=nEpcfyyrOucJyj6cD7wfThrIa-q0CQCoqMb3dkrhCRg,12094
+docling/backend/docx/latex/omml.py,sha256=4vh9FCbXh-Tb6KJGqNwzlMUMYEnnJgBtBI24dwy6t2U,12416
 docling/backend/json/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 docling/backend/json/docling_json_backend.py,sha256=LlFMVoZrrCfVwbDuRbNN4Xg96Lujh4xxrTBt9jGhY9I,1984
 docling/backend/xml/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-docling/backend/xml/jats_backend.py,sha256=ghGi9bHjx3BvaOtmzLw86-wZy4UxpQPOPQL4e73-BI8,24927
+docling/backend/xml/jats_backend.py,sha256=LPj33EFdi2MRCakkLWrRLlUAc-B-949f8zp5gKNvBcg,25238
 docling/backend/xml/uspto_backend.py,sha256=nyAMr5ht7dclxkVDwsKNeiOhLQrUtRLS8JdscB2AVJg,70924
 docling/chunking/__init__.py,sha256=h83TDs0AuOV6oEPLAPrn9dpGKiU-2Vg6IRNo4cv6GDA,346
 docling/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -37,7 +37,7 @@ docling/datamodel/asr_model_specs.py,sha256=Wg7z3zm_wXIWu122iPVy0RMECsA_JCFHrlFF
 docling/datamodel/base_models.py,sha256=9FslHkGUNmBp264LpLL_2JTfDAdaikldYs3SiQOHb5A,11828
 docling/datamodel/document.py,sha256=CA_dgt4V_phze5HXpfgfKNBKd1cPC1o3WE_IENX63EM,16252
 docling/datamodel/layout_model_specs.py,sha256=GSkJ-Z_0PVgwWGi7C7TsxbzRjlrWS9ZrHJjHumv-Z5U,2339
-docling/datamodel/pipeline_options.py,sha256=aMwpbyEMbAC-xGJnjQp8iw2ocpSU4eiD8D73gHf7T4U,10033
+docling/datamodel/pipeline_options.py,sha256=nlejeQjnJx2RBMkCukDECHGuVEOol9hbsSLUi2ee9hY,10134
 docling/datamodel/pipeline_options_asr_model.py,sha256=7X068xl-qpbyPxC7-TwX7Q6tLyZXGT5h1osZ_xLNLM0,1454
 docling/datamodel/pipeline_options_vlm_model.py,sha256=z-pUqwRA8nJp6C3SEXZLem2zvSYdgavaAVYa8wkAIZY,2400
 docling/datamodel/settings.py,sha256=ajMz7Ao2m0ZGYkfArqTDDbiF89O408mtgeh06PUi0MA,1900
@@ -55,7 +55,7 @@ docling/models/page_assemble_model.py,sha256=TvN1naez7dUodLxpUUBzpuMCpqZBTf6YSpe
 docling/models/page_preprocessing_model.py,sha256=x8MI4mvjizqEqAb5511dtrNRCJSb-lSmwHw0tmHPFiI,5103
 docling/models/picture_description_api_model.py,sha256=o3EkV5aHW_6WzE_fdj_VRnNCrS_btclO_ZCLAUqrfl0,2377
 docling/models/picture_description_base_model.py,sha256=kLthLhdlgwhootQ4_xhhcAk6A-vso5-qcsFJ3TcYfO0,2991
-docling/models/picture_description_vlm_model.py,sha256=nAUt-eZOX2GvaCiV2BJO7VppxUbP7udVIF4oe_sEYXo,4000
+docling/models/picture_description_vlm_model.py,sha256=yfyAFOy8RjxQJrafPMSAMrrpaYu3anahjRX6tCnVcs0,4028
 docling/models/rapid_ocr_model.py,sha256=AMdc66s_iWO4p6nQ0LNjQMUYVxrDSxMyLNPpjPYt6N8,5916
 docling/models/readingorder_model.py,sha256=bZoXHaSwUsa8niSmJrbCuy784ixCeBXT-RQBUfgHJ4A,14925
 docling/models/table_structure_model.py,sha256=RFXo73f2q4XuKyaSqbxpznh7JVtlLcT0FsOWl9oZbSg,12518
@@ -83,7 +83,7 @@ docling/utils/accelerator_utils.py,sha256=DSajLxVx1JEVT0zt5de26llciLNlVfIDfSa2zY
 docling/utils/api_image_request.py,sha256=_CgdzmPqdsyXmyYUFGLZcXcoH586qC6A1p5vsNbj1Q0,1416
 docling/utils/export.py,sha256=VwVUnYDk3mhGmISDbVm306fwpGNnoojouStBD4UajXI,4673
 docling/utils/glm_utils.py,sha256=TKOWQqWAHsX_w4fvoAA7_2xCi_urhnp1DsmjY8_sk5w,12274
-docling/utils/layout_postprocessor.py,sha256=QuTZZq4LNs1eM_n_2gubVfAuLBMkJiozfs3hp-jUpK4,24399
+docling/utils/layout_postprocessor.py,sha256=m92UKjL-cIrOmOBi5Nuiby9FQWFyudcHigJKzud69-Q,24486
 docling/utils/locks.py,sha256=RzqQtD5UispgV71pGN_nU6GYfeN11BN0Sh_Dq9ycqGo,52
 docling/utils/model_downloader.py,sha256=3vijCsAIVwWqehGBDRxRq7mJ3yRb9-zBsG00iqjqegU,4076
 docling/utils/ocr_utils.py,sha256=nmresYyfin0raanpQc_GGeU3WoLsfExf6SEXNIQ7Djg,2325
@@ -91,9 +91,9 @@ docling/utils/orientation.py,sha256=jTyLxyT31FlOodZoBMlADHNQK2lAWKYVs5z7pXd_6Cg,
 docling/utils/profiling.py,sha256=YaMGoB9MMZpagF9mb5ndoHj8Lpb9aIdb7El-Pl7IcFs,1753
 docling/utils/utils.py,sha256=kJtIYuzXeOyJHYlxmLAo7dGM5rEsDa1i84qEsUj1nio,1908
 docling/utils/visualization.py,sha256=tY2ylE2aiQKkmzlSLnFW-HTfFyqUUMguW18ldd1PLfo,2868
-docling-2.41.0.dist-info/licenses/LICENSE,sha256=mBb7ErEcM8VS9OhiGHnQ2kk75HwPhr54W1Oiz3965MY,1088
-docling-2.41.0.dist-info/METADATA,sha256=KYqB0miKX2x2ESNy8tNHdAlyTCONqhwGLR2iag2PcQ0,10274
-docling-2.41.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-docling-2.41.0.dist-info/entry_points.txt,sha256=hzVlbeE0aMSTQ9S0-NTYN0Hmgsn6qL_EA2qX4UbkAuY,149
-docling-2.41.0.dist-info/top_level.txt,sha256=vkIywP-USjFyYo1AIRQbWQQaL3xB5jf8vkCYdTIfNic,8
-docling-2.41.0.dist-info/RECORD,,
+docling-2.42.0.dist-info/licenses/LICENSE,sha256=mBb7ErEcM8VS9OhiGHnQ2kk75HwPhr54W1Oiz3965MY,1088
+docling-2.42.0.dist-info/METADATA,sha256=jOwKrV5DDscuvMqHevJKC7-VA_hPOpDNz2lfJA6RAVE,10310
+docling-2.42.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+docling-2.42.0.dist-info/entry_points.txt,sha256=hzVlbeE0aMSTQ9S0-NTYN0Hmgsn6qL_EA2qX4UbkAuY,149
+docling-2.42.0.dist-info/top_level.txt,sha256=vkIywP-USjFyYo1AIRQbWQQaL3xB5jf8vkCYdTIfNic,8
+docling-2.42.0.dist-info/RECORD,,