fix(ocr): rotate image to the natural orientation before layout prediction

Signed-off-by: Clément Doumouro <clement.doumouro@gmail.com>

docling/models/layout_model.py

@@ -16,6 +16,7 @@ from docling.datamodel.settings import settings
 from docling.models.base_model import BasePageModel
 from docling.utils.accelerator_utils import decide_device
 from docling.utils.layout_postprocessor import LayoutPostprocessor
+from docling.utils.orientation import detect_orientation
 from docling.utils.profiling import TimeRecorder
 from docling.utils.visualization import draw_clusters
 
@@ -152,7 +153,9 @@ class LayoutModel(BasePageModel):
                     assert page.size is not None
                     page_image = page.get_image(scale=1.0)
                     assert page_image is not None
-
+                    page_orientation = detect_orientation(page.cells)
+                    if page_orientation:
+                        page_image = page_image.rotate(-page_orientation, expand=True)
                     clusters = []
                     for ix, pred_item in enumerate(
                         self.layout_predictor.predict(page_image)

docling/models/tesseract_ocr_cli_model.py

@@ -266,7 +266,6 @@ class TesseractOcrCliModel(BaseOcrModel):
                                 orientation=doc_orientation,
                                 rotated_image_size=high_res_image.size,
                             )
-
                             cell = TextCell(
                                 index=ix,
                                 text=str(text),

docling/utils/ocr_utils.py

@@ -24,22 +24,27 @@ def map_tesseract_script(script: str) -> str:
 def reverse_tesseract_preprocessing_rotation(
     box: Box, orientation: int, rotated_im_size: Size
 ) -> tuple[Point, Point, Point, Point]:
+    # The box is left top width height in TOPLEFT coordinates
+    # Bounding rectangle start with r_0 at the bottom left whatever the
+    # coordinate system. Then other corners are found rotating counterclockwise
     l, t, w, h = box
-    rotated_w, rotated_h = rotated_im_size
+    rotated_im_w, rotated_im_h = rotated_im_size
     if orientation == 0:
-        return (l, t), (l + w, t), (l + w, t + h), (l, t + h)
+        r0_x = l
+        r0_y = t + h
+        return (r0_x, r0_y), (r0_x + w, r0_y), (r0_x + w, r0_y - h), (r0_x, r0_y - h)
     if orientation == 90:
-        x0 = rotated_h - t
+        r0_x = rotated_im_h - (t + h)
-        y0 = l
+        r0_y = l
-        return (x0, y0), (x0, y0 + w), (x0 - h, y0 + w), (x0 - h, y0)
+        return (r0_x, r0_y), (r0_x, r0_y + w), (r0_x + h, r0_y + w), (r0_x, r0_y + w)
     if orientation == 180:
-        x0 = rotated_w - l
+        r0_x = rotated_im_w - l
-        y0 = rotated_h - t
+        r0_y = rotated_im_h - (t + h)
-        return (x0, y0), (x0 - w, y0), (x0 - w, y0 - h), (x0, y0 - h)
+        return (r0_x, r0_y), (r0_x - w, r0_y), (r0_x - w, r0_y + h), (r0_x, r0_y + h)
     if orientation == 270:
-        x0 = t
+        r0_x = t + h
-        y0 = rotated_w - l
+        r0_y = rotated_im_w - l
-        return (x0, y0), (x0, y0 - w), (x0 + h, y0 - w), (x0 + h, y0)
+        return (r0_x, r0_y), (r0_x, r0_y - w), (r0_x - h, r0_y - w), (r0_x - h, r0_y)
     msg = (
         f"invalid tesseract document orientation {orientation}, "
         f"expected orientation: {sorted(_TESSERACT_ORIENTATIONS)}"

docling/utils/orientation.py

@@ -0,0 +1,17 @@
+from collections import Counter
+from operator import itemgetter
+
+from docling_core.types.doc.page import TextCell
+
+_ORIENTATIONS = [0, 90, 180, 270]
+
+
+def _clipped_orientation(angle: float) -> int:
+    return min((abs(angle - o) % 360, o) for o in _ORIENTATIONS)[1]
+
+
+def detect_orientation(cells: list[TextCell]) -> int:
+    if not cells:
+        return 0
+    orientation_counter = Counter(_clipped_orientation(c.rect.angle_360) for c in cells)
+    return max(orientation_counter.items(), key=itemgetter(1))[0]

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_180.doctags.txt

@@ -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>

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_180.json

@@ -1 +1 @@
-{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_180.pdf", "filename-prov": null, "document-hash": "a9cbfe0f2a71171face9ee31d2347ca4195649670ad75680520d67d4a863f982", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "baca27070f05dd84cf0903ded39bcf0fc1fa6ef0ac390e79cf8ba90c8c33ba49", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [441.304584329099, 132.09610360960653, 521.9863114205704, 151.67751306395223], "page": 1, "span": [0, 7], "__ref_s3_data": null}], "text": "package", "type": "paragraph", "payload": null, "name": "Text", "font": null}, {"prov": [{"bbox": [89.12133215549848, 77.02339849621205, 523.3501733013318, 124.86176457554109], "page": 1, "span": [0, 86], "__ref_s3_data": null}], "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 841.9216918945312, "page": 1, "width": 595.201171875}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}
+{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_180.pdf", "filename-prov": null, "document-hash": "a9cbfe0f2a71171face9ee31d2347ca4195649670ad75680520d67d4a863f982", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "baca27070f05dd84cf0903ded39bcf0fc1fa6ef0ac390e79cf8ba90c8c33ba49", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [444.6666666666667, 131.58835856119788, 521.6666666666666, 150.25502522786462], "page": 1, "span": [0, 7], "__ref_s3_data": null}], "text": "package", "type": "paragraph", "payload": null, "name": "Text", "font": null}, {"prov": [{"bbox": [92.0, 77.92169189453125, 523.0, 123.25502522786462], "page": 1, "span": [0, 86], "__ref_s3_data": null}], "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 841.9216918945312, "page": 1, "width": 595.201171875}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_270.doctags.txt

@@ -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>

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_270.json

@@ -1 +1 @@
-{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_270.pdf", "filename-prov": null, "document-hash": "52f54e7183bdb73aa3713c7b169baca93e276963a138418c26e7d6a1ea128f14", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "59bc9ddba89e7b008185dd16d384493beb034686e5670546786390c5d237a304", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [691.4680194659409, 442.3948768148814, 709.8255850278712, 523.0765988200898], "page": 1, "span": [0, 7], "__ref_s3_data": null}], "text": "package", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 595.201171875, "page": 1, "width": 841.9216918945312}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}
+{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_270.pdf", "filename-prov": null, "document-hash": "52f54e7183bdb73aa3713c7b169baca93e276963a138418c26e7d6a1ea128f14", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "59bc9ddba89e7b008185dd16d384493beb034686e5670546786390c5d237a304", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [691.6666666666666, 444.53450520833337, 710.3333333333334, 521.5345052083334], "page": 1, "span": [0, 7], "__ref_s3_data": null}], "text": "package", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 595.201171875, "page": 1, "width": 841.9216918945312}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.doctags.txt

@@ -1,3 +1,4 @@
 <document>
-<paragraph><location><page_1><loc_16><loc_12><loc_18><loc_26></location>package</paragraph>
+<paragraph><location><page_1><loc_9><loc_12><loc_11><loc_85></location>Docling bundles PDF document conversion to</paragraph>
+<paragraph><location><page_1><loc_12><loc_12><loc_15><loc_85></location><location><page_1><loc_12><loc_12><loc_15><loc_85></location>JSON and Markdown in an easy self contained package</paragraph>
 </document>

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.json

@@ -1 +1 @@
-{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_90.pdf", "filename-prov": null, "document-hash": "4a282813d93824eaa9bc2a0b2a0d6d626ecc8f5f380bd1320e2dd3e8e53c2ba6", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "f8a4dc72d8b159f69d0bc968b97f3fb9e0ac59dcb3113492432755835935d9b3", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [131.21306574279092, 74.12495603322407, 152.19606490864376, 154.19400205373182], "page": 1, "span": [0, 7], "__ref_s3_data": null}], "text": "package", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 595.201171875, "page": 1, "width": 841.9216918945312}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}
+{"_name": "", "type": "pdf-document", "description": {"title": null, "abstract": null, "authors": null, "affiliations": null, "subjects": null, "keywords": null, "publication_date": null, "languages": null, "license": null, "publishers": null, "url_refs": null, "references": null, "publication": null, "reference_count": null, "citation_count": null, "citation_date": null, "advanced": null, "analytics": null, "logs": [], "collection": null, "acquisition": null}, "file-info": {"filename": "ocr_test_rotated_90.pdf", "filename-prov": null, "document-hash": "4a282813d93824eaa9bc2a0b2a0d6d626ecc8f5f380bd1320e2dd3e8e53c2ba6", "#-pages": 1, "collection-name": null, "description": null, "page-hashes": [{"hash": "f8a4dc72d8b159f69d0bc968b97f3fb9e0ac59dcb3113492432755835935d9b3", "model": "default", "page": 1}]}, "main-text": [{"prov": [{"bbox": [78.0, 73.86783854166663, 96.66666666666667, 503.201171875], "page": 1, "span": [0, 42], "__ref_s3_data": null}], "text": "Docling bundles PDF document conversion to", "type": "paragraph", "payload": null, "name": "Text", "font": null}, {"prov": [{"bbox": [104.66666666666667, 72.201171875, 123.33333333333333, 503.201171875], "page": 1, "span": [0, 51], "__ref_s3_data": null}, {"bbox": [104.66666666666667, 72.201171875, 123.33333333333333, 503.201171875], "page": 1, "span": [0, 51], "__ref_s3_data": null}], "text": "JSON and Markdown in an easy self contained package", "type": "paragraph", "payload": null, "name": "Text", "font": null}], "figures": [], "tables": [], "bitmaps": null, "equations": [], "footnotes": [], "page-dimensions": [{"height": 595.201171875, "page": 1, "width": 841.9216918945312}], "page-footers": [], "page-headers": [], "_s3_data": null, "identifiers": null}

tests/data_scanned/groundtruth/docling_v1/ocr_test_rotated_90.md

@@ -1 +1,3 @@
-package
+Docling bundles PDF document conversion to
+
+JSON and Markdown in an easy self contained package

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_180.doctags.txt

@@ -1,3 +1,3 @@
-<doctag><text><loc_371><loc_410><loc_438><loc_422>package</text>
+<doctag><text><loc_374><loc_411><loc_438><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>
+<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>

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_180.json

@@ -1 +1 @@
-{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_180", "origin": {"mimetype": "application/pdf", "binary_hash": 2530576989861832966, "filename": "ocr_test_rotated_180.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 441.304584329099, "t": 151.67751306395223, "r": 521.9863114205704, "b": 132.09610360960653, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 7]}], "orig": "package", "text": "package", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 89.12133215549848, "t": 124.86176457554109, "r": 523.3501733013318, "b": 77.02339849621205, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 86]}], "orig": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 595.201171875, "height": 841.9216918945312}, "image": null, "page_no": 1}}}
+{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_180", "origin": {"mimetype": "application/pdf", "binary_hash": 2530576989861832966, "filename": "ocr_test_rotated_180.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 444.6666666666667, "t": 150.25502522786462, "r": 521.6666666666666, "b": 131.58835856119788, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 7]}], "orig": "package", "text": "package", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 92.0, "t": 123.25502522786462, "r": 523.0, "b": 77.92169189453125, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 86]}], "orig": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 595.201171875, "height": 841.9216918945312}, "image": null, "page_no": 1}}}

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_270.doctags.txt

@@ -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>
+<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_61><loc_422><loc_128>package</text>
+<text><loc_411><loc_62><loc_422><loc_127>package</text>
 </doctag>

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_270.json

@@ -1 +1 @@
-{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_270", "origin": {"mimetype": "application/pdf", "binary_hash": 10890858393843077593, "filename": "ocr_test_rotated_270.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "furniture", "label": "page_header", "prov": [{"page_no": 1, "bbox": {"l": 717.1685859527342, "t": 524.2990548540179, "r": 764.8982839673505, "b": 90.32916553110118, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 86]}], "orig": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 691.4680194659409, "t": 523.0765988200898, "r": 709.8255850278712, "b": 442.3948768148814, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 7]}], "orig": "package", "text": "package", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 841.9216918945312, "height": 595.201171875}, "image": null, "page_no": 1}}}
+{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_270", "origin": {"mimetype": "application/pdf", "binary_hash": 10890858393843077593, "filename": "ocr_test_rotated_270.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "furniture", "label": "page_header", "prov": [{"page_no": 1, "bbox": {"l": 718.6666666666666, "t": 522.8678385416666, "r": 764.0, "b": 91.86783854166669, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 86]}], "orig": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 691.6666666666666, "t": 521.5345052083334, "r": 710.3333333333334, "b": 444.53450520833337, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 7]}], "orig": "package", "text": "package", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 841.9216918945312, "height": 595.201171875}, "image": null, "page_no": 1}}}

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.doctags.txt

@@ -1,3 +1,3 @@
-<doctag><page_header><loc_46><loc_75><loc_74><loc_440>Docling bundles PDF document conversion to JSON and Markdown in an easy self contained</page_header>
+<doctag><text><loc_46><loc_77><loc_57><loc_438>Docling bundles PDF document conversion to</text>
-<text><loc_78><loc_370><loc_90><loc_438>package</text>
+<text><loc_62><loc_77><loc_73><loc_439><loc_62><loc_77><loc_73><loc_439>JSON and Markdown in an easy self contained package</text>
 </doctag>

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.json

@@ -1 +1 @@
-{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_90", "origin": {"mimetype": "application/pdf", "binary_hash": 6989291015361162334, "filename": "ocr_test_rotated_90.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "furniture", "label": "page_header", "prov": [{"page_no": 1, "bbox": {"l": 77.10171546422428, "t": 506.07735421856773, "r": 124.91101654503161, "b": 71.88562244773436, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 86]}], "orig": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "text": "Docling bundles PDF document conversion to JSON and Markdown in an easy self contained", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 131.21306574279092, "t": 154.19400205373182, "r": 152.19606490864376, "b": 74.12495603322407, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 7]}], "orig": "package", "text": "package", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 841.9216918945312, "height": 595.201171875}, "image": null, "page_no": 1}}}
+{"schema_name": "DoclingDocument", "version": "1.3.0", "name": "ocr_test_rotated_90", "origin": {"mimetype": "application/pdf", "binary_hash": 6989291015361162334, "filename": "ocr_test_rotated_90.pdf", "uri": null}, "furniture": {"self_ref": "#/furniture", "parent": null, "children": [], "content_layer": "furniture", "name": "_root_", "label": "unspecified"}, "body": {"self_ref": "#/body", "parent": null, "children": [{"cref": "#/texts/0"}, {"cref": "#/texts/1"}], "content_layer": "body", "name": "_root_", "label": "unspecified"}, "groups": [], "texts": [{"self_ref": "#/texts/0", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 78.0, "t": 503.201171875, "r": 96.66666666666667, "b": 73.86783854166663, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 42]}], "orig": "Docling bundles PDF document conversion to", "text": "Docling bundles PDF document conversion to", "formatting": null, "hyperlink": null}, {"self_ref": "#/texts/1", "parent": {"cref": "#/body"}, "children": [], "content_layer": "body", "label": "text", "prov": [{"page_no": 1, "bbox": {"l": 104.66666666666667, "t": 503.201171875, "r": 123.33333333333333, "b": 72.201171875, "coord_origin": "BOTTOMLEFT"}, "charspan": [0, 43]}, {"page_no": 1, "bbox": {"l": 104.66666666666667, "t": 503.201171875, "r": 123.33333333333333, "b": 72.201171875, "coord_origin": "BOTTOMLEFT"}, "charspan": [44, 51]}], "orig": "JSON and Markdown in an easy self contained package", "text": "JSON and Markdown in an easy self contained package", "formatting": null, "hyperlink": null}], "pictures": [], "tables": [], "key_value_items": [], "form_items": [], "pages": {"1": {"size": {"width": 841.9216918945312, "height": 595.201171875}, "image": null, "page_no": 1}}}

tests/data_scanned/groundtruth/docling_v2/ocr_test_rotated_90.md

@@ -1 +1,3 @@
-package
+Docling bundles PDF document conversion to
+
+JSON and Markdown in an easy self contained package