1 view

docling-project/docling

Get your documents ready for gen AI

Python★ 67,196+129 that day#6 on trending

Analyzed September 20, 2026 at 890dd42 (main) with jscpd 5.4.0 in 1.7s.

C60.9/100

Project health182,061 lines of code (L)

Duplication66

5.1% in python, latex, apex, bash, perl

Dead code83

Complexity44

60.3%

The score is jscpd's own --health model: markup, data and text left out, tests kept. The numbers below are the trending scan's.

4.08%duplicated code · 3,635 of 89,007 lines

202clones in code

240code files of 1,115 scanned

26.74%duplicated over all files · 4,272 clones

Most complex files

docling/backend/html_backend.py1198 cx
docling/backend/msword_backend.py749 cx
docling/backend/opendocument_backend.py445 cx
tests/test_service_client_sdk_unit.py402 cx
docling/backend/xml/uspto_backend.py361 cx
docling/service_client/client.py329 cx
tests/data/latex/sources/2501.00089/aastex631.cls326 cx
docling/backend/xml/jats_backend.py299 cx
docling/backend/msexcel_backend.py288 cx
tests/test_backend_msword.py245 cx

Dead code 1.99% unused · 163 findings

docling/experimental/pipeline/threaded_layout_vlm_pipeline.pyunused-file
docling/models/stages/ocr/_nemotron_ocr_model.pyunused-file
docling/document_extractor.pyunused-file
docling/models/extraction/nuextract_transformers_model.pyunused-file
docling/pipeline/extraction_vlm_pipeline.pyunused-file
docling/models/extraction/transformers_extraction_model.pyunused-file
docling/models/extraction/prompt_utils.pyunused-file
docling/pipeline/base_extraction_pipeline.pyunused-file
docling/models/stages/ocr/auto_ocr_model.py:EngineTypeunused-import
docling/models/stages/ocr/auto_ocr_model.py:RapidOCRunused-import

The code number leaves out tests, fixtures, docs, examples, data files, styles, translations and vendored code.

Largest clones in the code

266 lines in 2 placespython1,235 tokens

docling/models/stages/ocr/_nemotron_ocr_model.py:175–440docling/models/stages/ocr/nemotron_ocr_model.py:206–471

175raise OcrLanguageNotSupportedError(
176    self._engine_name,
177    language.tag(),
178    supported=self.supported_ocr_languages(),
179    detail=(
180        "nemotron-OCR-v2 ships an English and a multilingual recognizer "
181        "only; write 'multilingual' to run the multilingual one."
182    ),
183)
184

First 10 of 266 lines.

48 lines in 3 placespython270 tokens

docling/backend/msexcel_backend.py:1554–1601docling/backend/mspowerpoint_backend.py:1068–1115docling/backend/msword_backend.py:3665–3712

1554    columns.append((name, values))
1555
1556num_data_rows = max([len(categories)] + [len(values) for _, values in columns])
1557if num_data_rows == 0:
1558    return None
1559
1560num_rows = num_data_rows + 1
1561num_cols = 1 + len(columns)
1562cells: list[TableCell] = []
1563

First 10 of 48 lines.

105 lines in 2 placespython366 tokens

docling/models/stages/ocr/_nemotron_ocr_model.py:57–161docling/models/stages/ocr/nemotron_ocr_model.py:76–181

57# Mappings of nemotron language to the artifacts subdir. Also nemotron's own native vocabulary
58_NEMOTRON_CODE_TO_ARTIFACT = {
59    "english": "v2_english",
60    "multilingual": "v2_multilingual",
61}
62
63
64def nemotron_ocr_model_dir() -> str:
65    return _NEMOTRON_OCR_REPO_ID.replace("/", "--")
66

First 10 of 105 lines.

7 more

15 lines in 6 placespython72 tokens
docling/datamodel/asr_model_specs.py:55–69 docling/datamodel/asr_model_specs.py:97–111 docling/datamodel/asr_model_specs.py:139–153 docling/datamodel/asr_model_specs.py:181–195 docling/datamodel/asr_model_specs.py:223–237and 1 more
52 lines in 2 placespython276 tokens
docling/backend/xml/uspto_backend.py:580–631 docling/backend/xml/uspto_backend.py:1155–1206
17 lines in 3 placespython127 tokens
docling/models/extraction/nuextract_transformers_model.py:190–206 docling/models/extraction/transformers_extraction_model.py:125–141 docling/models/vlm_pipeline_models/hf_transformers_model.py:264–280
48 lines in 2 placespython251 tokens
docling/backend/xml/uspto_backend.py:233–280 docling/backend/xml/uspto_backend.py:584–631
23 lines in 3 placespython122 tokens
docling/backend/xml/uspto_backend.py:343–365 docling/backend/xml/uspto_backend.py:692–714 docling/backend/xml/uspto_backend.py:1267–1289
42 lines in 2 placespython238 tokens
docling/models/inference_engines/image_classification/api_kserve_v2_engine.py:95–136 docling/models/inference_engines/object_detection/api_kserve_v2_engine.py:109–150
42 lines in 2 placespython220 tokens
docling/backend/xml/uspto_backend.py:392–433 docling/backend/xml/uspto_backend.py:1313–1354

Largest clones over all files, tests and data included

5168 lines in 2 placesjsontests17,552 tokens

tests/data/xlsx/groundtruth/xlsx_01.xlsx.json:12–5179tests/data/xlsx/groundtruth/xlsx_04_inflated.xlsx.json:12–5179

12  "children": [],
13  "content_layer": "furniture",
14  "name": "_root_",
15  "label": "unspecified"
16},
17"body": {
18  "self_ref": "#/body",
19  "children": [
20    {
21      "$ref": "#/groups/0"

First 10 of 5168 lines.

276 lines in 3 placesapextests3,151 tokens

tests/data/latex/sources/2501.00089/aastex631.cls:3817–4092tests/data/latex/sources/2501.00089/aastex631.cls:4124–4399tests/data/latex/sources/2501.00089/aastex631.cls:4440–4715

3817\ifcase\totalcolumncount%
3818\or%
3819\csname tab1\endcsname%
3820\or%
3821\csname tab1\endcsname&\csname tab2\endcsname%
3822\or%
3823\csname tab1\endcsname&\csname tab2\endcsname&%
3824\csname tab3\endcsname%
3825\or%
3826\csname tab1\endcsname&\csname tab2\endcsname&%

First 10 of 276 lines.

466 lines in 4 placesjsontests1,630 tokens
tests/data/csv/groundtruth/csv-comma.csv.json:885–1350 tests/data/csv/groundtruth/csv-pipe.csv.json:885–1350 tests/data/csv/groundtruth/csv-semicolon.csv.json:885–1350 tests/data/csv/groundtruth/csv-tab.csv.json:885–1350
```
885  "text": "Martin, Lang and Andrade",
886  "column_header": false,
887  "row_header": false,
888  "row_section": false,
889  "fillable": false
890},
891{
892  "row_span": 1,
893  "col_span": 1,
894  "start_row_offset_idx": 5,
```
First 10 of 466 lines.
1309 lines in 2 placesjsontests4,563 tokens
tests/data/ppt/groundtruth/legacy_sample.ppt.json:1021–2329 tests/data/pptx/groundtruth/powerpoint_sample.pptx.json:1021–2329
365 lines in 4 placesjsontests1,291 tokens
tests/data/csv/groundtruth/csv-comma.csv.json:40–404 tests/data/csv/groundtruth/csv-pipe.csv.json:40–404 tests/data/csv/groundtruth/csv-semicolon.csv.json:40–404 tests/data/csv/groundtruth/csv-tab.csv.json:40–404

By format, code only

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	227	87,680	202	3,635	4.15%
bash	12	1,153	0	0	0%
perl	1	174	0	0	0%

By format, all files

Format	Files	Lines	Clones	Duplicated lines	Duplication
json	208	344,213	3,448	138,638	40.28%
python	480	150,481	597	8,854	5.88%
apex	4	9,540	78	1,188	12.45%
markdown	267	31,456	77	1,643	5.22%
markup	40	4,498	37	1,277	28.39%
yaml	25	3,281	10	163	4.97%
latex	24	19,200	9	59	0.31%
txt	17	2,384	8	66	2.77%
toml	5	1,736	6	187	10.77%
bash	27	1,476	1	6	0.41%

Trending appearances

Day	Rank	Stars	Files	Lines	Clones	All files	Code	Health	Commit
Sep 20, 2026	#6	67,196 (+129)	1,115	568,797	4,272	26.74%	4.08%	C 60.9	`890dd42`

Want this for your own project? Install jscpd and run jscpd ., or measure it the way the baselines were measured.