docling 2.56.1__py3-none-any.whl → 2.58.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of docling might be problematic. Click here for more details.

Files changed (30) hide show
  1. docling/backend/abstract_backend.py +24 -3
  2. docling/backend/asciidoc_backend.py +3 -3
  3. docling/backend/docling_parse_v4_backend.py +15 -4
  4. docling/backend/docx/drawingml/utils.py +131 -0
  5. docling/backend/html_backend.py +130 -20
  6. docling/backend/md_backend.py +27 -5
  7. docling/backend/msexcel_backend.py +115 -27
  8. docling/backend/mspowerpoint_backend.py +2 -2
  9. docling/backend/msword_backend.py +104 -29
  10. docling/backend/pdf_backend.py +9 -2
  11. docling/backend/pypdfium2_backend.py +12 -3
  12. docling/cli/main.py +85 -30
  13. docling/datamodel/asr_model_specs.py +408 -6
  14. docling/datamodel/backend_options.py +82 -0
  15. docling/datamodel/base_models.py +17 -2
  16. docling/datamodel/document.py +81 -48
  17. docling/datamodel/pipeline_options_asr_model.py +21 -1
  18. docling/document_converter.py +37 -45
  19. docling/document_extractor.py +12 -11
  20. docling/models/readingorder_model.py +6 -7
  21. docling/pipeline/asr_pipeline.py +139 -3
  22. docling/pipeline/vlm_pipeline.py +53 -33
  23. docling/utils/api_image_request.py +4 -4
  24. docling/utils/layout_postprocessor.py +23 -24
  25. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/METADATA +4 -2
  26. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/RECORD +30 -28
  27. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/WHEEL +0 -0
  28. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/entry_points.txt +0 -0
  29. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/licenses/LICENSE +0 -0
  30. {docling-2.56.1.dist-info → docling-2.58.0.dist-info}/top_level.txt +0 -0
{docling-2.56.1.dist-info → docling-2.58.0.dist-info}/RECORD RENAMED
@@ -1,26 +1,27 @@
1
1
  docling/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
2
- docling/document_converter.py,sha256=gPyBrNegMgeBGxN7iebrjqEDm7zQQOmFNm8hVi-pFEQ,16013
3
- docling/document_extractor.py,sha256=-RbQRvLWLXF15HYqBbV_lJhh08Zl487UEQKhP-_FR8k,11969
2
+ docling/document_converter.py,sha256=_P3f4eZ8Gssv3P3l8xX2RrgzS8WhafY7-x6rWaWOeN4,15511
3
+ docling/document_extractor.py,sha256=Jk1a4hgPxjLkp4UoZR_pdEMid9-jhNiND5_NlPHGy6c,11965
4
4
  docling/exceptions.py,sha256=K1WnCS1leK2JtMB5ewZWKkb0EaijFgl-tRzrO9ntgPM,134
5
5
  docling/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
6
6
  docling/backend/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
7
- docling/backend/abstract_backend.py,sha256=1lNxzwDTn303aXduPDVmTyXn-5ZIoWMLYqNxANGWmQQ,1658
8
- docling/backend/asciidoc_backend.py,sha256=ARpMBzrNCV-x6g2I0KytDM3pGLac3z4ql3hDKi3FI04,14403
7
+ docling/backend/abstract_backend.py,sha256=_xKSjLpR-ia93Kz0dto0yyVsaeIqEepUhVEGo18MuWw,2169
8
+ docling/backend/asciidoc_backend.py,sha256=DR8AUTNvy_SCHkieMpqZXg_NLRTy4roEqa0V8sILPWk,14400
9
9
  docling/backend/csv_backend.py,sha256=2g9famYG2W-ID9jEdZPxc6O8QGv1vWQfjN8pL-QMBE0,4536
10
10
  docling/backend/docling_parse_backend.py,sha256=9rUo1vPxX6QLzGqF-2B2iEYglZg6YQ3Uea00XrLluTg,7918
11
11
  docling/backend/docling_parse_v2_backend.py,sha256=3ckTfke8IICjaImlIzc3TRhG7KDuxDDba0AuCEcjA-M,9500
12
- docling/backend/docling_parse_v4_backend.py,sha256=xCBbaaXjNNrOaod9tmBuCbe5mL_ipmTNG2XOxVbGG3w,7891
13
- docling/backend/html_backend.py,sha256=iuRyYztUduyP214X0SyDvl1dP_h0eccp5RkuM72rV8o,48664
14
- docling/backend/md_backend.py,sha256=TWboEPHl93pqI_Go1a3XpP-KpzI3d17xo5ZW42Ul0kY,22764
12
+ docling/backend/docling_parse_v4_backend.py,sha256=tBJR0BbKFOIDKSngjVDu0BrzTj7qUZAhFdRT8GvAJ18,8232
13
+ docling/backend/html_backend.py,sha256=m91kRxMhQ1w-7G6MHA9l01dgF8-YQNn8ZNx9lwG467M,52935
14
+ docling/backend/md_backend.py,sha256=_0ToiecsGwU4H4BBso4ar9TGJi8OTwSXjgmi66vSJVQ,23513
15
15
  docling/backend/mets_gbs_backend.py,sha256=EA8sY6tbmGiysKGYPPZiNlK-i7Adn8bLTo-7Ym15hTU,12774
16
- docling/backend/msexcel_backend.py,sha256=GOuA-MlShpzFmCmJq3-Z28iquwWUg4k8v-AT4O-aAQI,19305
17
- docling/backend/mspowerpoint_backend.py,sha256=wJgB2JStEPfD7MPpWQlpPN7bffPxaHFUnKD4wj8SLxU,15114
18
- docling/backend/msword_backend.py,sha256=Jfd57hzG8iFVAzqsOAHe5jG8LCHAIBXJhQCW0tESnMM,54405
16
+ docling/backend/msexcel_backend.py,sha256=-iWLdIonMZl2FCfPAXFQKIQzFOJn5InpH6KDAJ_L64o,22760
17
+ docling/backend/mspowerpoint_backend.py,sha256=71W_iV31Rggqn9UcMzXmsZ3QKMRpsBT8fCwdjsIIKAs,15109
18
+ docling/backend/msword_backend.py,sha256=zNJy-KM3Ia-L8IQ4sjYxATW4owFxbg2CK0rzke8y-7w,57451
19
19
  docling/backend/noop_backend.py,sha256=EOPbD86FzZPX-K_DpNrJh0_lC0bZz--4DpG-OagDNGY,1688
20
- docling/backend/pdf_backend.py,sha256=Wcd1NSrAMjXK8VicTki5p-j-JLofklt07eF0kIG17_0,3361
21
- docling/backend/pypdfium2_backend.py,sha256=AYhWs9S8W_TkAK0-OkRmUNf4HUZl26FP7-XYjwU5zDk,14209
20
+ docling/backend/pdf_backend.py,sha256=UovGV3RJG6qllzMPYzhDB6GID7buGV6w1uxl5dOAEw4,3563
21
+ docling/backend/pypdfium2_backend.py,sha256=tx0FnUW87zPsyafCvOuLcls2k5QdpPKWweyjNTfclNc,14509
22
22
  docling/backend/webvtt_backend.py,sha256=9xPcfWVLuqhEAFrkv8aU36qHnSgjeINZAXT_C9C6XJA,19165
23
23
  docling/backend/docx/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
24
+ docling/backend/docx/drawingml/utils.py,sha256=E9Iq8_052eEV5L1IN3ZqFX9eBidH56DKNlh6Tk7Do0I,3640
24
25
  docling/backend/docx/latex/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
25
26
  docling/backend/docx/latex/latex_dict.py,sha256=tFJp4ScT_AkY2ON7nLEa560p601Jq2glcZvMKxxjn7w,6593
26
27
  docling/backend/docx/latex/omml.py,sha256=4vh9FCbXh-Tb6KJGqNwzlMUMYEnnJgBtBI24dwy6t2U,12416
@@ -31,18 +32,19 @@ docling/backend/xml/jats_backend.py,sha256=_BWpQQg3SlsHAOOj0v2qRJoVqaQzL91GqN1tK
31
32
  docling/backend/xml/uspto_backend.py,sha256=Tv4CE7V5_QwxTNJPl90CAd_mAbwaLGy8S6s6evh1Xow,70910
32
33
  docling/chunking/__init__.py,sha256=h83TDs0AuOV6oEPLAPrn9dpGKiU-2Vg6IRNo4cv6GDA,346
33
34
  docling/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
34
- docling/cli/main.py,sha256=cvDS6CTME2B2Mrm4l9yNynOUDVsZ9ZTlA6mM_jsa5jU,34258
35
+ docling/cli/main.py,sha256=x_mPS3g3Zw60_9bL_oo9OfPBmuSd-aJV7oKTPD0GjS4,36772
35
36
  docling/cli/models.py,sha256=zZBFQJAD7C5sespnYy5M__4qC_GyqAZ-QpfWtgPRDB0,6343
36
37
  docling/cli/tools.py,sha256=QhtRxQG0TVrfsMqdv5i7J0_qQy1ZZyWYnHPwJl7b5oY,322
37
38
  docling/datamodel/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
38
39
  docling/datamodel/accelerator_options.py,sha256=wv6dOFTVAwr9onkE-0pfUqX_fDb6gX53iPPE6o8nKjI,2511
39
- docling/datamodel/asr_model_specs.py,sha256=Wg7z3zm_wXIWu122iPVy0RMECsA_JCFHrlFF-xxHoVQ,2187
40
- docling/datamodel/base_models.py,sha256=CQ6eThPzVeVD2Gq7BNz9Q5RDLwhe4NgMzk7tdLtk1c8,12382
41
- docling/datamodel/document.py,sha256=HyO3kdJcXIJ3wL95sPoL3zvsO4Rww3-qHH6IkL4I0q4,17483
40
+ docling/datamodel/asr_model_specs.py,sha256=gQJkW7DaSPiOuW_0QoI5OzR1_DQGRkw7yQlrVJ4hyo0,14473
41
+ docling/datamodel/backend_options.py,sha256=2zSbJRtBmJ6Twywj8pLOKaHhklY85XaGXUmSLX_SfgQ,2473
42
+ docling/datamodel/base_models.py,sha256=pC9CvVxMzcujKAG0TTObkYznKp8gIFdzDMDmgk5FjMQ,12697
43
+ docling/datamodel/document.py,sha256=T9OogC1kIm0VDSC2ZFcFgWdcOjXzw5JvGr2y2hMlx3s,18795
42
44
  docling/datamodel/extraction.py,sha256=7dgvtK5SuvgfB8LHAwS1FwrW1kcMQJuJG0ol8uAQgoQ,1323
43
45
  docling/datamodel/layout_model_specs.py,sha256=GSkJ-Z_0PVgwWGi7C7TsxbzRjlrWS9ZrHJjHumv-Z5U,2339
44
46
  docling/datamodel/pipeline_options.py,sha256=dklSaA7P6VkjbBB-Pz2OyzO2SQuV9y0I8VVr9XHJusw,11692
45
- docling/datamodel/pipeline_options_asr_model.py,sha256=7X068xl-qpbyPxC7-TwX7Q6tLyZXGT5h1osZ_xLNLM0,1454
47
+ docling/datamodel/pipeline_options_asr_model.py,sha256=cLqtRHBr2kbTNXRJ1ZhFGiXIK7Nl9RFmz2Wd7tJF2Jg,2172
46
48
  docling/datamodel/pipeline_options_vlm_model.py,sha256=Szdq5_MhqQ8xBCvOUkdn_LLV29ZMQJcF4xnItYlkmXQ,3090
47
49
  docling/datamodel/settings.py,sha256=c0MTw6pO5be_BKxHKYl4SaBJAw_qL-aapxp-g5HHj1A,2084
48
50
  docling/datamodel/vlm_model_specs.py,sha256=9TTmihDEFcI-TY1jJ2GTnTcrGa3bLg0e6anN4gPtFgU,10035
@@ -62,7 +64,7 @@ docling/models/picture_description_api_model.py,sha256=o3EkV5aHW_6WzE_fdj_VRnNCr
62
64
  docling/models/picture_description_base_model.py,sha256=kLthLhdlgwhootQ4_xhhcAk6A-vso5-qcsFJ3TcYfO0,2991
63
65
  docling/models/picture_description_vlm_model.py,sha256=Uja_BQSk7F-U1J2hm4yeLguirUzKYv1K8zRyw1IYomY,4150
64
66
  docling/models/rapid_ocr_model.py,sha256=JGeed1aNO64SYFgxlOifdut4fynUJyBuyyQrfuSno-4,13182
65
- docling/models/readingorder_model.py,sha256=-j-UuvnsYWqZvY0gByKz0bjcBwOhWQTHerCopig_jVs,17266
67
+ docling/models/readingorder_model.py,sha256=gnRFfJAXH-zKtQJws5Zb1_KCVvu_dAq9pgaDYQKCt9s,17236
66
68
  docling/models/table_structure_model.py,sha256=7g_mFf1YzfF8PXQfefNu6XYZu7TzJAn86zKb6IEUdCg,12518
67
69
  docling/models/tesseract_ocr_cli_model.py,sha256=KuO4rXc-88C2-cAymvcr41TqFi3hNg4gerEzoI3Z6m4,13039
68
70
  docling/models/tesseract_ocr_model.py,sha256=W_476USwExjSfhelXG8B9eNIVXXlm_dNFA60TZ5rq7E,11216
@@ -81,20 +83,20 @@ docling/models/vlm_models_inline/mlx_model.py,sha256=ae7hDMgBsMLkqulmbKDamGSSrLJ
81
83
  docling/models/vlm_models_inline/nuextract_transformers_model.py,sha256=jLNtlkMDheUyWot7Oqq-GHQIYzJ0fZrbReq5xCnYb9E,10506
82
84
  docling/models/vlm_models_inline/vllm_model.py,sha256=vXClayYxPGX1jzQ1Rvf3vvwtW9khgApGvcRz4Qbyu7I,10293
83
85
  docling/pipeline/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
84
- docling/pipeline/asr_pipeline.py,sha256=oRluG28no3ezjbtL7nJLpDcxxxJuuULNXheq1W-qklM,10629
86
+ docling/pipeline/asr_pipeline.py,sha256=44lweVOCkFe8KikgXJjqDtfHewIotYvc242Xvgl9fV0,15744
85
87
  docling/pipeline/base_extraction_pipeline.py,sha256=GYrEz83IXv-tdIHjtNWxMBNczFwL8SZyf9vnPJ3STaI,2627
86
88
  docling/pipeline/base_pipeline.py,sha256=NPMQDTyis-LgQ4SybY2f5AESZl5PxogF-FRQuCDckXg,12748
87
89
  docling/pipeline/extraction_vlm_pipeline.py,sha256=veUOTe8nGdnduZKaGn1RRb-NfU1H6t_EN4QAsb022Zg,8260
88
90
  docling/pipeline/simple_pipeline.py,sha256=FSL_ucDd9k0D9DjNKMUkyCULIU8a057dvWfLEPmAc2A,2287
89
91
  docling/pipeline/standard_pdf_pipeline.py,sha256=xOge0zP5wli51n_6QLrFHQlwwvsivI7OMt00tht3my4,10479
90
92
  docling/pipeline/threaded_standard_pdf_pipeline.py,sha256=i67G5AOW7PIFCe5JS2sdBmPAKvAH6ScxIBhjwOGZcrI,28183
91
- docling/pipeline/vlm_pipeline.py,sha256=oMcdgzym_UQbVN3bajux_hENY40XGOnb6NU6Kwje2Os,15376
93
+ docling/pipeline/vlm_pipeline.py,sha256=HSbSoGZyy4eIK8eOL2g_NymrHg8r-DrB2buggJQAqHU,16189
92
94
  docling/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
93
95
  docling/utils/accelerator_utils.py,sha256=DSajLxVx1JEVT0zt5de26llciLNlVfIDfSa2zYCFJzQ,2909
94
- docling/utils/api_image_request.py,sha256=kQDmTvQT6M2IgXnGYeoNflI6sLUG6WTCcEft94CRwWg,5379
96
+ docling/utils/api_image_request.py,sha256=xrn4O8ax8wdQPkLgbRhX22qWUangCXwaOzIXy_86LCs,5367
95
97
  docling/utils/export.py,sha256=VwVUnYDk3mhGmISDbVm306fwpGNnoojouStBD4UajXI,4673
96
98
  docling/utils/glm_utils.py,sha256=TKOWQqWAHsX_w4fvoAA7_2xCi_urhnp1DsmjY8_sk5w,12274
97
- docling/utils/layout_postprocessor.py,sha256=sE9UR3Nv4iOk26uoIsN3bFioE7ScfAjj0orDBDneLXg,25166
99
+ docling/utils/layout_postprocessor.py,sha256=bwDIhgUg5rKianzccGPTotTjqjkWtIQSoZwgKio8YC4,25124
98
100
  docling/utils/locks.py,sha256=RzqQtD5UispgV71pGN_nU6GYfeN11BN0Sh_Dq9ycqGo,52
99
101
  docling/utils/model_downloader.py,sha256=qrkL5NTpwk6yF4bcipcUtLRxl0Tqh7zoSa_WtLsMySA,5325
100
102
  docling/utils/ocr_utils.py,sha256=nmresYyfin0raanpQc_GGeU3WoLsfExf6SEXNIQ7Djg,2325
@@ -102,9 +104,9 @@ docling/utils/orientation.py,sha256=jTyLxyT31FlOodZoBMlADHNQK2lAWKYVs5z7pXd_6Cg,
102
104
  docling/utils/profiling.py,sha256=YaMGoB9MMZpagF9mb5ndoHj8Lpb9aIdb7El-Pl7IcFs,1753
103
105
  docling/utils/utils.py,sha256=kJtIYuzXeOyJHYlxmLAo7dGM5rEsDa1i84qEsUj1nio,1908
104
106
  docling/utils/visualization.py,sha256=tY2ylE2aiQKkmzlSLnFW-HTfFyqUUMguW18ldd1PLfo,2868
105
- docling-2.56.1.dist-info/licenses/LICENSE,sha256=mBb7ErEcM8VS9OhiGHnQ2kk75HwPhr54W1Oiz3965MY,1088
106
- docling-2.56.1.dist-info/METADATA,sha256=LgkCRBKjI20tarzBBeFpRi0vo_fZ4Sh6rD6K8yQkrY8,11364
107
- docling-2.56.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
108
- docling-2.56.1.dist-info/entry_points.txt,sha256=hzVlbeE0aMSTQ9S0-NTYN0Hmgsn6qL_EA2qX4UbkAuY,149
109
- docling-2.56.1.dist-info/top_level.txt,sha256=vkIywP-USjFyYo1AIRQbWQQaL3xB5jf8vkCYdTIfNic,8
110
- docling-2.56.1.dist-info/RECORD,,
107
+ docling-2.58.0.dist-info/licenses/LICENSE,sha256=mBb7ErEcM8VS9OhiGHnQ2kk75HwPhr54W1Oiz3965MY,1088
108
+ docling-2.58.0.dist-info/METADATA,sha256=py9js2V38fIWft1SmMe_iD_trav0WEwojgwxlHMsNv4,11642
109
+ docling-2.58.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
110
+ docling-2.58.0.dist-info/entry_points.txt,sha256=hzVlbeE0aMSTQ9S0-NTYN0Hmgsn6qL_EA2qX4UbkAuY,149
111
+ docling-2.58.0.dist-info/top_level.txt,sha256=vkIywP-USjFyYo1AIRQbWQQaL3xB5jf8vkCYdTIfNic,8
112
+ docling-2.58.0.dist-info/RECORD,,