magic-pdf 1.2.2__py3-none-any.whl → 1.3.0__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.
Files changed (101) hide show
  1. magic_pdf/data/batch_build_dataset.py +156 -0
  2. magic_pdf/data/dataset.py +44 -24
  3. magic_pdf/data/utils.py +108 -9
  4. magic_pdf/dict2md/ocr_mkcontent.py +4 -3
  5. magic_pdf/libs/pdf_image_tools.py +11 -6
  6. magic_pdf/libs/performance_stats.py +12 -1
  7. magic_pdf/libs/version.py +1 -1
  8. magic_pdf/model/batch_analyze.py +175 -201
  9. magic_pdf/model/doc_analyze_by_custom_model.py +137 -92
  10. magic_pdf/model/pdf_extract_kit.py +5 -38
  11. magic_pdf/model/sub_modules/language_detection/utils.py +2 -4
  12. magic_pdf/model/sub_modules/language_detection/yolov11/YOLOv11.py +24 -19
  13. magic_pdf/model/sub_modules/layout/doclayout_yolo/DocLayoutYOLO.py +3 -1
  14. magic_pdf/model/sub_modules/mfd/yolov8/YOLOv8.py +3 -1
  15. magic_pdf/model/sub_modules/mfr/unimernet/Unimernet.py +31 -102
  16. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/__init__.py +13 -0
  17. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/modeling_unimernet.py +189 -0
  18. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_mbart/__init__.py +8 -0
  19. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_mbart/configuration_unimer_mbart.py +163 -0
  20. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_mbart/modeling_unimer_mbart.py +2351 -0
  21. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_swin/__init__.py +9 -0
  22. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_swin/configuration_unimer_swin.py +132 -0
  23. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_swin/image_processing_unimer_swin.py +132 -0
  24. magic_pdf/model/sub_modules/mfr/unimernet/unimernet_hf/unimer_swin/modeling_unimer_swin.py +1084 -0
  25. magic_pdf/model/sub_modules/model_init.py +50 -37
  26. magic_pdf/model/sub_modules/model_utils.py +17 -11
  27. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/__init__.py +1 -0
  28. magic_pdf/model/sub_modules/ocr/{paddleocr → paddleocr2pytorch}/ocr_utils.py +102 -97
  29. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorch_paddle.py +193 -0
  30. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/base_ocr_v20.py +39 -0
  31. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/data/__init__.py +8 -0
  32. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/data/imaug/__init__.py +48 -0
  33. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/data/imaug/operators.py +418 -0
  34. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/architectures/__init__.py +25 -0
  35. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/architectures/base_model.py +105 -0
  36. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/__init__.py +62 -0
  37. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/det_mobilenet_v3.py +269 -0
  38. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/rec_hgnet.py +290 -0
  39. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/rec_lcnetv3.py +516 -0
  40. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/rec_mobilenet_v3.py +136 -0
  41. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/rec_mv1_enhance.py +234 -0
  42. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/backbones/rec_svtrnet.py +638 -0
  43. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/common.py +76 -0
  44. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/heads/__init__.py +43 -0
  45. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/heads/cls_head.py +23 -0
  46. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/heads/det_db_head.py +109 -0
  47. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/heads/rec_ctc_head.py +54 -0
  48. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/heads/rec_multi_head.py +58 -0
  49. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/necks/__init__.py +29 -0
  50. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/necks/db_fpn.py +456 -0
  51. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/necks/intracl.py +117 -0
  52. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/necks/rnn.py +228 -0
  53. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/postprocess/__init__.py +33 -0
  54. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/postprocess/cls_postprocess.py +20 -0
  55. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/postprocess/db_postprocess.py +179 -0
  56. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/postprocess/rec_postprocess.py +690 -0
  57. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/__init__.py +0 -0
  58. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/arch_config.yaml +383 -0
  59. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/arabic_dict.txt +162 -0
  60. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/chinese_cht_dict.txt +8421 -0
  61. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/cyrillic_dict.txt +163 -0
  62. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/devanagari_dict.txt +167 -0
  63. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/en_dict.txt +95 -0
  64. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/japan_dict.txt +4399 -0
  65. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/ka_dict.txt +153 -0
  66. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/korean_dict.txt +3688 -0
  67. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/latin_dict.txt +185 -0
  68. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/ppocr_keys_v1.txt +6623 -0
  69. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/ta_dict.txt +128 -0
  70. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/te_dict.txt +151 -0
  71. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/models_config.yml +49 -0
  72. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/__init__.py +1 -0
  73. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/__init__.py +1 -0
  74. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/predict_cls.py +106 -0
  75. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/predict_det.py +217 -0
  76. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/predict_rec.py +440 -0
  77. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/predict_system.py +104 -0
  78. magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/tools/infer/pytorchocr_utility.py +227 -0
  79. magic_pdf/model/sub_modules/table/rapidtable/rapid_table.py +10 -18
  80. magic_pdf/pdf_parse_union_core_v2.py +112 -74
  81. magic_pdf/pre_proc/ocr_dict_merge.py +9 -1
  82. magic_pdf/pre_proc/ocr_span_list_modify.py +51 -0
  83. magic_pdf/resources/model_config/model_configs.yaml +1 -1
  84. magic_pdf/tools/cli.py +30 -12
  85. magic_pdf/tools/common.py +90 -12
  86. {magic_pdf-1.2.2.dist-info → magic_pdf-1.3.0.dist-info}/METADATA +50 -40
  87. magic_pdf-1.3.0.dist-info/RECORD +202 -0
  88. magic_pdf/model/sub_modules/ocr/paddleocr/ppocr_273_mod.py +0 -204
  89. magic_pdf/model/sub_modules/ocr/paddleocr/ppocr_291_mod.py +0 -213
  90. magic_pdf/model/sub_modules/table/structeqtable/struct_eqtable.py +0 -37
  91. magic_pdf/model/sub_modules/table/tablemaster/tablemaster_paddle.py +0 -71
  92. magic_pdf/resources/model_config/UniMERNet/demo.yaml +0 -46
  93. magic_pdf/resources/model_config/layoutlmv3/layoutlmv3_base_inference.yaml +0 -351
  94. magic_pdf-1.2.2.dist-info/RECORD +0 -147
  95. /magic_pdf/model/sub_modules/{ocr/paddleocr/__init__.py → mfr/unimernet/unimernet_hf/unimer_mbart/tokenization_unimer_mbart.py} +0 -0
  96. /magic_pdf/model/sub_modules/{table/structeqtable → ocr/paddleocr2pytorch/pytorchocr}/__init__.py +0 -0
  97. /magic_pdf/model/sub_modules/{table/tablemaster → ocr/paddleocr2pytorch/pytorchocr/modeling}/__init__.py +0 -0
  98. {magic_pdf-1.2.2.dist-info → magic_pdf-1.3.0.dist-info}/LICENSE.md +0 -0
  99. {magic_pdf-1.2.2.dist-info → magic_pdf-1.3.0.dist-info}/WHEEL +0 -0
  100. {magic_pdf-1.2.2.dist-info → magic_pdf-1.3.0.dist-info}/entry_points.txt +0 -0
  101. {magic_pdf-1.2.2.dist-info → magic_pdf-1.3.0.dist-info}/top_level.txt +0 -0
magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/resources/dict/ka_dict.txt ADDED
@@ -0,0 +1,153 @@
1
+ k
2
+ a
3
+ _
4
+ i
5
+ m
6
+ g
7
+ /
8
+ 1
9
+ 2
10
+ I
11
+ L
12
+ S
13
+ V
14
+ R
15
+ C
16
+ 0
17
+ v
18
+ l
19
+ 6
20
+ 4
21
+ 8
22
+ .
23
+ j
24
+ p
25
+ ಗ
26
+ ು
27
+ ಣ
28
+ ಪ
29
+ ಡ
30
+ ಿ
31
+ ಸ
32
+ ಲ
33
+ ಾ
34
+ ದ
35
+ ್
36
+ 7
37
+ 5
38
+ 3
39
+ ವ
40
+ ಷ
41
+ ಬ
42
+ ಹ
43
+ ೆ
44
+ 9
45
+ ಅ
46
+ ಳ
47
+ ನ
48
+ ರ
49
+ ಉ
50
+ ಕ
51
+ ಎ
52
+ ೇ
53
+ ಂ
54
+ ೈ
55
+ ೊ
56
+ ೀ
57
+ ಯ
58
+ ೋ
59
+ ತ
60
+ ಶ
61
+ ಭ
62
+ ಧ
63
+ ಚ
64
+ ಜ
65
+ ೂ
66
+ ಮ
67
+ ಒ
68
+ ೃ
69
+ ಥ
70
+ ಇ
71
+ ಟ
72
+ ಖ
73
+ ಆ
74
+ ಞ
75
+ ಫ
76
+ -
77
+ ಢ
78
+ ಊ
79
+ ಓ
80
+ ಐ
81
+ ಃ
82
+ ಘ
83
+ ಝ
84
+ ೌ
85
+ ಠ
86
+ ಛ
87
+ ಔ
88
+ ಏ
89
+ ಈ
90
+ ಋ
91
+ ೨
92
+ ೦
93
+ ೧
94
+ ೮
95
+ ೯
96
+ ೪
97
+ ,
98
+ ೫
99
+ ೭
100
+ ೩
101
+ ೬
102
+ ಙ
103
+ s
104
+ c
105
+ e
106
+ n
107
+ w
108
+ o
109
+ u
110
+ t
111
+ d
112
+ E
113
+ A
114
+ T
115
+ B
116
+ Z
117
+ N
118
+ G
119
+ O
120
+ q
121
+ z
122
+ r
123
+ x
124
+ P
125
+ K
126
+ M
127
+ J
128
+ U
129
+ D
130
+ f
131
+ F
132
+ h
133
+ b
134
+ W
135
+ Y
136
+ y
137
+ H
138
+ X
139
+ Q
140
+ '
141
+ #
142
+ &
143
+ !
144
+ @
145
+ $
146
+ :
147
+ %
148
+ é
149
+ É
150
+ (
151
+ ?
152
+ +
153
+