mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-07-26 20:14:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

Make page.parsed_page the only source of truth for text cells

Browse Source 
Signed-off-by: Christoph Auer <cau@zurich.ibm.com>
This commit is contained in:
Christoph Auer 2025-06-10 19:55:49 +02:00
parent e310c5cff3
commit d73c9a2995
58 changed files with 349497 additions and 331004 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

147
docling/backend/docling_parse_backend.py
View File

@ -7,7 +7,13 @@ from typing import List, Optional, Union
import pypdfium2 as pdfium
from docling_core.types.doc import BoundingBox, CoordOrigin, Size
from docling_core.types.doc.page import BoundingRectangle, SegmentedPdfPage, TextCell
from docling_core.types.doc.page import (
BoundingRectangle,
PdfPageBoundaryType,
PdfPageGeometry,
SegmentedPdfPage,
TextCell,
)
from docling_parse.pdf_parsers import pdf_parser_v1
from PIL import Image, ImageDraw
from pypdfium2 import PdfPage
@ -36,6 +42,51 @@ class DoclingParsePageBackend(PdfPageBackend):
def is_valid(self) -> bool:
return self.valid
def _compute_text_cells(self) -> List[TextCell]:
"""Compute text cells from docling-parse data."""
cells: List[TextCell] = []
cell_counter = 0
if not self.valid:
return cells
page_size = self.get_size()
parser_width = self._dpage["width"]
parser_height = self._dpage["height"]
for i in range(len(self._dpage["cells"])):
rect = self._dpage["cells"][i]["box"]["device"]
x0, y0, x1, y1 = rect
if x1 < x0:
x0, x1 = x1, x0
if y1 < y0:
y0, y1 = y1, y0
text_piece = self._dpage["cells"][i]["content"]["rnormalized"]
cells.append(
TextCell(
index=cell_counter,
text=text_piece,
orig=text_piece,
from_ocr=False,
rect=BoundingRectangle.from_bounding_box(
BoundingBox(
l=x0 * page_size.width / parser_width,
b=y0 * page_size.height / parser_height,
r=x1 * page_size.width / parser_width,
t=y1 * page_size.height / parser_height,
coord_origin=CoordOrigin.BOTTOMLEFT,
)
).to_top_left_origin(page_size.height),
)
)
cell_counter += 1
return cells
def get_text_in_rect(self, bbox: BoundingBox) -> str:
if not self.valid:
return ""
@ -70,75 +121,45 @@ class DoclingParsePageBackend(PdfPageBackend):
return text_piece
def get_segmented_page(self) -> Optional[SegmentedPdfPage]:
return None
def get_text_cells(self) -> Iterable[TextCell]:
cells: List[TextCell] = []
cell_counter = 0
if not self.valid:
return cells
return None
page_size = self.get_size()
text_cells = self._compute_text_cells()
parser_width = self._dpage["width"]
parser_height = self._dpage["height"]
# Create page geometry
crop_bbox = BoundingBox(
l=0,
r=page_size.width,
t=0,
b=page_size.height,
coord_origin=CoordOrigin.TOPLEFT,
).to_bottom_left_origin(page_size.height)
for i in range(len(self._dpage["cells"])):
rect = self._dpage["cells"][i]["box"]["device"]
x0, y0, x1, y1 = rect
dimension = PdfPageGeometry(
angle=0.0,
rect=BoundingRectangle.from_bounding_box(crop_bbox),
boundary_type=PdfPageBoundaryType.CROP_BOX,
art_bbox=crop_bbox,
bleed_bbox=crop_bbox,
crop_bbox=crop_bbox,
media_bbox=crop_bbox,
trim_bbox=crop_bbox,
)
if x1 < x0:
x0, x1 = x1, x0
if y1 < y0:
y0, y1 = y1, y0
# Create SegmentedPdfPage
return SegmentedPdfPage(
dimension=dimension,
textline_cells=text_cells,
char_cells=[],
word_cells=[],
has_lines=len(text_cells) > 0,
has_words=False,
has_chars=False,
)
text_piece = self._dpage["cells"][i]["content"]["rnormalized"]
cells.append(
TextCell(
index=cell_counter,
text=text_piece,
orig=text_piece,
from_ocr=False,
rect=BoundingRectangle.from_bounding_box(
BoundingBox(
# l=x0, b=y0, r=x1, t=y1,
l=x0 * page_size.width / parser_width,
b=y0 * page_size.height / parser_height,
r=x1 * page_size.width / parser_width,
t=y1 * page_size.height / parser_height,
coord_origin=CoordOrigin.BOTTOMLEFT,
)
).to_top_left_origin(page_size.height),
)
)
cell_counter += 1
def draw_clusters_and_cells():
image = (
self.get_page_image()
) # make new image to avoid drawing on the saved ones
draw = ImageDraw.Draw(image)
for c in cells:
x0, y0, x1, y1 = c.rect.to_bounding_box().as_tuple()
cell_color = (
random.randint(30, 140),
random.randint(30, 140),
random.randint(30, 140),
)
draw.rectangle([(x0, y0), (x1, y1)], outline=cell_color)
image.show()
# before merge:
# draw_clusters_and_cells()
# cells = merge_horizontal_cells(cells)
# after merge:
# draw_clusters_and_cells()
return cells
def get_text_cells(self) -> Iterable[TextCell]:
return self._compute_text_cells()
def get_bitmap_rects(self, scale: float = 1) -> Iterable[BoundingBox]:
AREA_THRESHOLD = 0 # 32 * 32

149
docling/backend/docling_parse_v2_backend.py
View File

@ -7,7 +7,13 @@ from typing import TYPE_CHECKING, List, Optional, Union
import pypdfium2 as pdfium
from docling_core.types.doc import BoundingBox, CoordOrigin
from docling_core.types.doc.page import BoundingRectangle, SegmentedPdfPage, TextCell
from docling_core.types.doc.page import (
BoundingRectangle,
PdfPageBoundaryType,
PdfPageGeometry,
SegmentedPdfPage,
TextCell,
)
from docling_parse.pdf_parsers import pdf_parser_v2
from PIL import Image, ImageDraw
from pypdfium2 import PdfPage
@ -40,6 +46,55 @@ class DoclingParseV2PageBackend(PdfPageBackend):
def is_valid(self) -> bool:
return self.valid
def _compute_text_cells(self) -> List[TextCell]:
"""Compute text cells from docling-parse v2 data."""
cells: List[TextCell] = []
cell_counter = 0
if not self.valid:
return cells
page_size = self.get_size()
parser_width = self._dpage["sanitized"]["dimension"]["width"]
parser_height = self._dpage["sanitized"]["dimension"]["height"]
cells_data = self._dpage["sanitized"]["cells"]["data"]
cells_header = self._dpage["sanitized"]["cells"]["header"]
for i, cell_data in enumerate(cells_data):
x0 = cell_data[cells_header.index("x0")]
y0 = cell_data[cells_header.index("y0")]
x1 = cell_data[cells_header.index("x1")]
y1 = cell_data[cells_header.index("y1")]
if x1 < x0:
x0, x1 = x1, x0
if y1 < y0:
y0, y1 = y1, y0
text_piece = cell_data[cells_header.index("text")]
cells.append(
TextCell(
index=cell_counter,
text=text_piece,
orig=text_piece,
from_ocr=False,
rect=BoundingRectangle.from_bounding_box(
BoundingBox(
l=x0 * page_size.width / parser_width,
b=y0 * page_size.height / parser_height,
r=x1 * page_size.width / parser_width,
t=y1 * page_size.height / parser_height,
coord_origin=CoordOrigin.BOTTOMLEFT,
)
).to_top_left_origin(page_size.height),
)
)
cell_counter += 1
return cells
def get_text_in_rect(self, bbox: BoundingBox) -> str:
if not self.valid:
return ""
@ -81,73 +136,45 @@ class DoclingParseV2PageBackend(PdfPageBackend):
return text_piece
def get_segmented_page(self) -> Optional[SegmentedPdfPage]:
return None
def get_text_cells(self) -> Iterable[TextCell]:
cells: List[TextCell] = []
cell_counter = 0
if not self.valid:
return cells
return None
page_size = self.get_size()
text_cells = self._compute_text_cells()
parser_width = self._dpage["sanitized"]["dimension"]["width"]
parser_height = self._dpage["sanitized"]["dimension"]["height"]
# Create page geometry
crop_bbox = BoundingBox(
l=0,
r=page_size.width,
t=0,
b=page_size.height,
coord_origin=CoordOrigin.TOPLEFT,
).to_bottom_left_origin(page_size.height)
cells_data = self._dpage["sanitized"]["cells"]["data"]
cells_header = self._dpage["sanitized"]["cells"]["header"]
dimension = PdfPageGeometry(
angle=0.0,
rect=BoundingRectangle.from_bounding_box(crop_bbox),
boundary_type=PdfPageBoundaryType.CROP_BOX,
art_bbox=crop_bbox,
bleed_bbox=crop_bbox,
crop_bbox=crop_bbox,
media_bbox=crop_bbox,
trim_bbox=crop_bbox,
)
for i, cell_data in enumerate(cells_data):
x0 = cell_data[cells_header.index("x0")]
y0 = cell_data[cells_header.index("y0")]
x1 = cell_data[cells_header.index("x1")]
y1 = cell_data[cells_header.index("y1")]
# Create SegmentedPdfPage
return SegmentedPdfPage(
dimension=dimension,
textline_cells=text_cells,
char_cells=[],
word_cells=[],
has_lines=len(text_cells) > 0,
has_words=False,
has_chars=False,
)
if x1 < x0:
x0, x1 = x1, x0
if y1 < y0:
y0, y1 = y1, y0
text_piece = cell_data[cells_header.index("text")]
cells.append(
TextCell(
index=cell_counter,
text=text_piece,
orig=text_piece,
from_ocr=False,
rect=BoundingRectangle.from_bounding_box(
BoundingBox(
# l=x0, b=y0, r=x1, t=y1,
l=x0 * page_size.width / parser_width,
b=y0 * page_size.height / parser_height,
r=x1 * page_size.width / parser_width,
t=y1 * page_size.height / parser_height,
coord_origin=CoordOrigin.BOTTOMLEFT,
)
).to_top_left_origin(page_size.height),
)
)
cell_counter += 1
def draw_clusters_and_cells():
image = (
self.get_page_image()
) # make new image to avoid drawing on the saved ones
draw = ImageDraw.Draw(image)
for c in cells:
x0, y0, x1, y1 = c.bbox.as_tuple()
cell_color = (
random.randint(30, 140),
random.randint(30, 140),
random.randint(30, 140),
)
draw.rectangle([(x0, y0), (x1, y1)], outline=cell_color)
image.show()
# draw_clusters_and_cells()
return cells
def get_text_cells(self) -> Iterable[TextCell]:
return self._compute_text_cells()
def get_bitmap_rects(self, scale: float = 1) -> Iterable[BoundingBox]:
AREA_THRESHOLD = 0 # 32 * 32

128
docling/backend/pypdfium2_backend.py
View File

@ -8,7 +8,13 @@ from typing import TYPE_CHECKING, List, Optional, Union
import pypdfium2 as pdfium
import pypdfium2.raw as pdfium_c
from docling_core.types.doc import BoundingBox, CoordOrigin, Size
from docling_core.types.doc.page import BoundingRectangle, SegmentedPdfPage, TextCell
from docling_core.types.doc.page import (
BoundingRectangle,
PdfPageBoundaryType,
PdfPageGeometry,
SegmentedPdfPage,
TextCell,
)
from PIL import Image, ImageDraw
from pypdfium2 import PdfTextPage
from pypdfium2._helpers.misc import PdfiumError
@ -41,38 +47,8 @@ class PyPdfiumPageBackend(PdfPageBackend):
def is_valid(self) -> bool:
return self.valid
def get_bitmap_rects(self, scale: float = 1) -> Iterable[BoundingBox]:
AREA_THRESHOLD = 0 # 32 * 32
page_size = self.get_size()
with pypdfium2_lock:
for obj in self._ppage.get_objects(filter=[pdfium_c.FPDF_PAGEOBJ_IMAGE]):
pos = obj.get_pos()
cropbox = BoundingBox.from_tuple(
pos, origin=CoordOrigin.BOTTOMLEFT
).to_top_left_origin(page_height=page_size.height)
if cropbox.area() > AREA_THRESHOLD:
cropbox = cropbox.scaled(scale=scale)
yield cropbox
def get_text_in_rect(self, bbox: BoundingBox) -> str:
with pypdfium2_lock:
if not self.text_page:
self.text_page = self._ppage.get_textpage()
if bbox.coord_origin != CoordOrigin.BOTTOMLEFT:
bbox = bbox.to_bottom_left_origin(self.get_size().height)
with pypdfium2_lock:
text_piece = self.text_page.get_text_bounded(*bbox.as_tuple())
return text_piece
def get_segmented_page(self) -> Optional[SegmentedPdfPage]:
return None
def get_text_cells(self) -> Iterable[TextCell]:
def _compute_text_cells(self) -> List[TextCell]:
"""Compute text cells from pypdfium."""
with pypdfium2_lock:
if not self.text_page:
self.text_page = self._ppage.get_textpage()
@ -203,30 +179,76 @@ class PyPdfiumPageBackend(PdfPageBackend):
return merged_cells
def draw_clusters_and_cells():
image = (
self.get_page_image()
) # make new image to avoid drawing on the saved ones
draw = ImageDraw.Draw(image)
for c in cells:
x0, y0, x1, y1 = c.rect.to_bounding_box().as_tuple()
cell_color = (
random.randint(30, 140),
random.randint(30, 140),
random.randint(30, 140),
)
draw.rectangle([(x0, y0), (x1, y1)], outline=cell_color)
image.show()
return merge_horizontal_cells(cells)
# before merge:
# draw_clusters_and_cells()
def get_bitmap_rects(self, scale: float = 1) -> Iterable[BoundingBox]:
AREA_THRESHOLD = 0 # 32 * 32
page_size = self.get_size()
with pypdfium2_lock:
for obj in self._ppage.get_objects(filter=[pdfium_c.FPDF_PAGEOBJ_IMAGE]):
pos = obj.get_pos()
cropbox = BoundingBox.from_tuple(
pos, origin=CoordOrigin.BOTTOMLEFT
).to_top_left_origin(page_height=page_size.height)
cells = merge_horizontal_cells(cells)
if cropbox.area() > AREA_THRESHOLD:
cropbox = cropbox.scaled(scale=scale)
# after merge:
# draw_clusters_and_cells()
yield cropbox
return cells
def get_text_in_rect(self, bbox: BoundingBox) -> str:
with pypdfium2_lock:
if not self.text_page:
self.text_page = self._ppage.get_textpage()
if bbox.coord_origin != CoordOrigin.BOTTOMLEFT:
bbox = bbox.to_bottom_left_origin(self.get_size().height)
with pypdfium2_lock:
text_piece = self.text_page.get_text_bounded(*bbox.as_tuple())
return text_piece
def get_segmented_page(self) -> Optional[SegmentedPdfPage]:
if not self.valid:
return None
page_size = self.get_size()
text_cells = self._compute_text_cells()
# Create page geometry
crop_bbox = BoundingBox(
l=0,
r=page_size.width,
t=0,
b=page_size.height,
coord_origin=CoordOrigin.TOPLEFT,
).to_bottom_left_origin(page_size.height)
dimension = PdfPageGeometry(
angle=0.0,
rect=BoundingRectangle.from_bounding_box(crop_bbox),
boundary_type=PdfPageBoundaryType.CROP_BOX,
art_bbox=crop_bbox,
bleed_bbox=crop_bbox,
crop_bbox=crop_bbox,
media_bbox=crop_bbox,
trim_bbox=crop_bbox,
)
# Create SegmentedPdfPage
return SegmentedPdfPage(
dimension=dimension,
textline_cells=text_cells,
char_cells=[],
word_cells=[],
has_lines=len(text_cells) > 0,
has_words=False,
has_chars=False,
)
def get_text_cells(self) -> Iterable[TextCell]:
return self._compute_text_cells()
def get_page_image(
self, scale: float = 1, cropbox: Optional[BoundingBox] = None

9
docling/datamodel/base_models.py
View File

@ -232,7 +232,6 @@ class Page(BaseModel):
page_no: int
# page_hash: Optional[str] = None
size: Optional[Size] = None
cells: List[TextCell] = []
parsed_page: Optional[SegmentedPdfPage] = None
predictions: PagePredictions = PagePredictions()
assembled: Optional[AssembledUnit] = None
@ -245,6 +244,14 @@ class Page(BaseModel):
float, Image
] = {} # Cache of images in different scales. By default it is cleared during assembling.
@property
def cells(self) -> List[TextCell]:
"""Return text cells as a read-only view of parsed_page.textline_cells."""
if self.parsed_page is not None:
return self.parsed_page.textline_cells
else:
return []
def get_image(
self, scale: float = 1.0, cropbox: Optional[BoundingBox] = None
) -> Optional[Image]:

4
docling/datamodel/pipeline_options.py
View File

@ -292,7 +292,9 @@ class PdfPipelineOptions(PaginatedPipelineOptions):
),
)
generate_parsed_pages: bool = False
generate_parsed_pages: Literal[True] = (
True # Always True since parsed_page is now mandatory
)
class PdfPipeline(str, Enum):

27
docling/models/base_ocr_model.py
View File

@ -133,20 +133,19 @@ class BaseOcrModel(BasePageModel, BaseModelWithOptions):
def post_process_cells(self, ocr_cells, page):
r"""
Post-process the OCR cells and update the page object.
Treats page.parsed_page as authoritative when available, with page.cells for compatibility.
Updates parsed_page.textline_cells directly since page.cells is now read-only.
"""
# Get existing cells (prefer parsed_page, fallback to page.cells)
existing_cells = self._get_existing_cells(page)
# Get existing cells from the read-only property
existing_cells = page.cells
# Combine existing and OCR cells with overlap filtering
final_cells = self._combine_cells(existing_cells, ocr_cells)
# Update both structures efficiently
self._update_page_structures(page, final_cells)
assert page.parsed_page is not None
def _get_existing_cells(self, page):
"""Get existing cells, preferring parsed_page when available."""
return page.parsed_page.textline_cells if page.parsed_page else page.cells
# Update parsed_page.textline_cells directly
page.parsed_page.textline_cells = final_cells
page.parsed_page.has_lines = bool(final_cells)
def _combine_cells(self, existing_cells, ocr_cells):
"""Combine existing and OCR cells with filtering and re-indexing."""
@ -162,18 +161,6 @@ class BaseOcrModel(BasePageModel, BaseModelWithOptions):
return combined
def _update_page_structures(self, page, final_cells):
"""Update both page structures efficiently."""
if page.parsed_page:
# Update parsed_page as primary source
page.parsed_page.textline_cells = final_cells
page.parsed_page.has_lines = bool(final_cells)
# Sync to page.cells for compatibility
page.cells = final_cells
else:
# Legacy fallback: only page.cells available
page.cells = final_cells
def draw_ocr_rects_and_cells(self, conv_res, page, ocr_rects, show: bool = False):
image = copy.deepcopy(page.image)
scale_x = image.width / page.size.width

1
docling/models/layout_model.py
View File

@ -198,7 +198,6 @@ class LayoutModel(BasePageModel):
)
)
# page.cells is already updated by LayoutPostprocessor
page.predictions.layout = LayoutPrediction(
clusters=processed_clusters
)

9
docling/models/page_preprocessing_model.py
View File

@ -2,7 +2,7 @@ import re
import warnings
from collections.abc import Iterable
from pathlib import Path
from typing import Optional
from typing import Literal, Optional
import numpy as np
from PIL import ImageDraw
@ -17,7 +17,6 @@ from docling.utils.profiling import TimeRecorder
class PagePreprocessingOptions(BaseModel):
images_scale: Optional[float]
create_parsed_page: bool
class PagePreprocessingModel(BasePageModel):
@ -66,10 +65,8 @@ class PagePreprocessingModel(BasePageModel):
def _parse_page_cells(self, conv_res: ConversionResult, page: Page) -> Page:
assert page._backend is not None
page.cells = list(page._backend.get_text_cells())
if self.options.create_parsed_page:
page.parsed_page = page._backend.get_segmented_page()
page.parsed_page = page._backend.get_segmented_page()
assert page.parsed_page is not None
# Rate the text quality from the PDF parser, and aggregate on page
text_scores = []

2
docling/models/rapid_ocr_model.py
View File

@ -134,7 +134,7 @@ class RapidOcrModel(BaseOcrModel):
all_ocr_cells.extend(cells)
# Post-process the cells
page.cells = self.post_process_cells(all_ocr_cells, page.cells)
self.post_process_cells(all_ocr_cells, page.cells)
# DEBUG code:
if settings.debug.visualize_ocr:

1
docling/pipeline/standard_pdf_pipeline.py
View File

@ -72,7 +72,6 @@ class StandardPdfPipeline(PaginatedPipeline):
PagePreprocessingModel(
options=PagePreprocessingOptions(
images_scale=pipeline_options.images_scale,
create_parsed_page=pipeline_options.generate_parsed_pages,
)
),
# OCR

26
docling/utils/layout_postprocessor.py
View File

@ -196,8 +196,7 @@ class LayoutPostprocessor:
def __init__(self, page, clusters: List[Cluster]):
"""Initialize processor with page and clusters."""
# Get cells from best available source (prefer parsed_page)
self.cells = self._get_page_cells(page)
self.cells = page.cells
self.page = page
self.page_size = page.size
self.all_clusters = clusters
@ -215,24 +214,6 @@ class LayoutPostprocessor:
[c for c in self.special_clusters if c.label in self.WRAPPER_TYPES]
)
def _get_page_cells(self, page):
"""Get cells from best available source (prefer parsed_page)."""
return (
page.parsed_page.textline_cells
if page.parsed_page is not None
else page.cells
)
def _update_page_structures(self, final_cells):
"""Update both page structures efficiently."""
if self.page.parsed_page is not None:
# Update parsed_page as primary source
self.page.parsed_page.textline_cells = final_cells
self.page.parsed_page.has_lines = len(final_cells) > 0
# Legacy fallback: only page.cells available
self.page.cells = final_cells
def postprocess(self) -> Tuple[List[Cluster], List[TextCell]]:
"""Main processing pipeline."""
self.regular_clusters = self._process_regular_clusters()
@ -259,8 +240,9 @@ class LayoutPostprocessor:
for child in cluster.children:
child.cells = self._sort_cells(child.cells)
# Update page structures with processed cells
self._update_page_structures(self.cells)
assert self.page.parsed_page is not None
self.page.parsed_page.textline_cells = self.cells
self.page.parsed_page.has_lines = len(self.cells) > 0
return final_clusters, self.cells

101474
tests/data/groundtruth/docling_v1/2203.01017v2.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

89985
tests/data/groundtruth/docling_v1/2206.01062.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

5315
tests/data/groundtruth/docling_v1/2305.03393v1-pg9.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

56232
tests/data/groundtruth/docling_v1/2305.03393v1.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4965
tests/data/groundtruth/docling_v1/amt_handbook_sample.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4518
tests/data/groundtruth/docling_v1/code_and_formula.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

9633
tests/data/groundtruth/docling_v1/multi_page.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2570
tests/data/groundtruth/docling_v1/picture_classification.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

46512
tests/data/groundtruth/docling_v1/redp5110_sampled.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

1309
tests/data/groundtruth/docling_v1/right_to_left_01.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

8015
tests/data/groundtruth/docling_v1/right_to_left_02.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2809
tests/data/groundtruth/docling_v1/right_to_left_03.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

101474
tests/data/groundtruth/docling_v2/2203.01017v2.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

89985
tests/data/groundtruth/docling_v2/2206.01062.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

5315
tests/data/groundtruth/docling_v2/2305.03393v1-pg9.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

56232
tests/data/groundtruth/docling_v2/2305.03393v1.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4965
tests/data/groundtruth/docling_v2/amt_handbook_sample.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4518
tests/data/groundtruth/docling_v2/code_and_formula.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

9633
tests/data/groundtruth/docling_v2/multi_page.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2570
tests/data/groundtruth/docling_v2/picture_classification.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

46512
tests/data/groundtruth/docling_v2/redp5110_sampled.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

1309
tests/data/groundtruth/docling_v2/right_to_left_01.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

8015
tests/data/groundtruth/docling_v2/right_to_left_02.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2809
tests/data/groundtruth/docling_v2/right_to_left_03.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

8
tests/data_scanned/groundtruth/docling_v1/ocr_test.json vendored
View File

@ -44,10 +44,10 @@
"prov": [
{
"bbox": [
69.6796630536824,
689.0124221922704,
504.8720051760782,
764.9216921155637
72.33333333333333,
691.5883585611979,
503.3333333333333,
763.9216918945312
],
"page": 1,
"span": [

1611
tests/data_scanned/groundtruth/docling_v1/ocr_test.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_180.doctags.txt vendored
View File

@ -1,4 +1,4 @@
<document>
<paragraph><location><page_1><loc_74><loc_16><loc_88><loc_18></location>package</paragraph>
<paragraph><location><page_1><loc_75><loc_16><loc_88><loc_18></location>package</paragraph>
<paragraph><location><page_1><loc_15><loc_9><loc_88><loc_15></location>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</paragraph>
</document>

16
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_180.json vendored
View File

@ -44,10 +44,10 @@
"prov": [
{
"bbox": [
441.2561096985719,
131.89488404865142,
522.0347860494834,
151.87873262042876
444.6666666666667,
131.58835856119788,
521.6666666666666,
150.25502522786462
],
"page": 1,
"span": [
@ -67,10 +67,10 @@
"prov": [
{
"bbox": [
89.23887497045128,
77.02339852098021,
523.208764293368,
124.75312428291147
92.0,
77.92169189453125,
523.0,
123.25502522786462
],
"page": 1,
"span": [

1663
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_180.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_270.doctags.txt vendored
View File

@ -1,3 +1,3 @@
<document>
<paragraph><location><page_1><loc_82><loc_74><loc_84><loc_88></location>package</paragraph>
<paragraph><location><page_1><loc_82><loc_75><loc_84><loc_88></location>package</paragraph>
</document>

8
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_270.json vendored
View File

@ -44,10 +44,10 @@
"prov": [
{
"bbox": [
690.2441821046808,
442.39487414368364,
709.8255852011977,
523.076601235155
691.6666666666666,
444.53450520833337,
710.3333333333334,
521.5345052083334
],
"page": 1,
"span": [

1667
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_270.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.doctags.txt vendored
View File

@ -1,3 +1,3 @@
<document>
<paragraph><location><page_1><loc_16><loc_12><loc_18><loc_26></location>package</paragraph>
<paragraph><location><page_1><loc_16><loc_12><loc_18><loc_25></location>package</paragraph>
</document>

8
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.json vendored
View File

@ -44,10 +44,10 @@
"prov": [
{
"bbox": [
131.21306574279092,
74.12495603322407,
152.19606490864376,
154.19400205373182
131.66666666666666,
73.53450520833337,
150.33333333333334,
150.53450520833331
],
"page": 1,
"span": [

1663
tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

2
tests/data_scanned/groundtruth/docling_v2/ocr_test.doctags.txt vendored
View File

@ -1,2 +1,2 @@
<doctag><text><loc_59><loc_46><loc_424><loc_91>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained package</text>
<doctag><text><loc_61><loc_46><loc_423><loc_89>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained package</text>
</doctag>

8
tests/data_scanned/groundtruth/docling_v2/ocr_test.json vendored
View File

@ -42,10 +42,10 @@
{
"page_no": 1,
"bbox": {
"l": 69.6796630536824,
"t": 764.9216921155637,
"r": 504.8720051760782,
"b": 689.0124221922704,
"l": 72.33333333333333,
"t": 763.9216918945312,
"r": 503.3333333333333,
"b": 691.5883585611979,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [

1611
tests/data_scanned/groundtruth/docling_v2/ocr_test.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_180.doctags.txt vendored
View File

@ -1,3 +1,3 @@
<doctag><text><loc_371><loc_410><loc_439><loc_422>package</text>
<text><loc_75><loc_426><loc_440><loc_454>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</text>
<doctag><text><loc_374><loc_411><loc_438><loc_422>package</text>
<text><loc_77><loc_427><loc_439><loc_454>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</text>
</doctag>

16
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_180.json vendored
View File

@ -45,10 +45,10 @@
{
"page_no": 1,
"bbox": {
"l": 441.2561096985719,
"t": 151.87873262042876,
"r": 522.0347860494834,
"b": 131.89488404865142,
"l": 444.6666666666667,
"t": 150.25502522786462,
"r": 521.6666666666666,
"b": 131.58835856119788,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [
@ -74,10 +74,10 @@
{
"page_no": 1,
"bbox": {
"l": 89.23887497045128,
"t": 124.75312428291147,
"r": 523.208764293368,
"b": 77.02339852098021,
"l": 92.0,
"t": 123.25502522786462,
"r": 523.0,
"b": 77.92169189453125,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [

1663
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_180.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_270.doctags.txt vendored
View File

@ -1,3 +1,3 @@
<doctag><page_header><loc_426><loc_60><loc_454><loc_424>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</page_header>
<text><loc_410><loc_61><loc_422><loc_128>package</text>
<doctag><page_header><loc_427><loc_61><loc_454><loc_423>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</page_header>
<text><loc_411><loc_62><loc_422><loc_127>package</text>
</doctag>

16
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_270.json vendored
View File

@ -45,10 +45,10 @@
{
"page_no": 1,
"bbox": {
"l": 717.168585936602,
"t": 524.2990550512769,
"r": 764.8982839673505,
"b": 90.3291657283603,
"l": 718.6666666666666,
"t": 522.8678385416666,
"r": 764.0,
"b": 91.86783854166669,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [
@ -74,10 +74,10 @@
{
"page_no": 1,
"bbox": {
"l": 690.2441821046808,
"t": 523.076601235155,
"r": 709.8255852011977,
"b": 442.39487414368364,
"l": 691.6666666666666,
"t": 521.5345052083334,
"r": 710.3333333333334,
"b": 444.53450520833337,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [

1667
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_270.pages.json vendored
View File

File diff suppressed because it is too large Load Diff

4
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.doctags.txt vendored
View File

@ -1,3 +1,3 @@
<doctag><page_header><loc_46><loc_75><loc_75><loc_440>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</page_header>
<text><loc_78><loc_370><loc_90><loc_438>package</text>
<doctag><page_header><loc_46><loc_77><loc_73><loc_439>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</page_header>
<text><loc_78><loc_374><loc_89><loc_438>package</text>
</doctag>

16
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.json vendored
View File

@ -45,10 +45,10 @@
{
"page_no": 1,
"bbox": {
"l": 77.10171545548258,
"t": 506.0744964609271,
"r": 126.08064862014129,
"b": 71.87755635676046,
"l": 78.0,
"t": 503.201171875,
"r": 123.33333333333333,
"b": 72.201171875,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [
@ -74,10 +74,10 @@
{
"page_no": 1,
"bbox": {
"l": 131.21306574279092,
"t": 154.19400205373182,
"r": 152.19606490864376,
"b": 74.12495603322407,
"l": 131.66666666666666,
"t": 150.53450520833331,
"r": 150.33333333333334,
"b": 73.53450520833337,
"coord_origin": "BOTTOMLEFT"
},
"charspan": [

1663
tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.pages.json vendored
View File

File diff suppressed because it is too large Load Diff