Merge pull request #2611 from myhloli/dev

Dev

Merge pull request #2611 from myhloli/dev
Dev
0c7a0882 · Xiaomeng Zhao · GitHub · 3bd0ecf1 · a392f445 · 3bd0ecf1
Unverified Commit 0c7a0882 authored Jun 12, 2025 by Xiaomeng Zhao Committed by GitHub Jun 12, 2025
20 changed files
--- a/magic_pdf/model/sub_modules/layout/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3.py
+++ b/magic_pdf/model/sub_modules/layout/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3.py
--- a/magic_pdf/model/sub_modules/layout/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3_fast.py
+++ b/magic_pdf/model/sub_modules/layout/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3_fast.py
--- a/magic_pdf/model/sub_modules/layout/layoutlmv3/model_init.py
+++ b/magic_pdf/model/sub_modules/layout/layoutlmv3/model_init.py
--- a/magic_pdf/model/sub_modules/layout/layoutlmv3/rcnn_vl.py
+++ b/magic_pdf/model/sub_modules/layout/layoutlmv3/rcnn_vl.py
--- a/magic_pdf/model/sub_modules/layout/layoutlmv3/visualizer.py
+++ b/magic_pdf/model/sub_modules/layout/layoutlmv3/visualizer.py
--- a/magic_pdf/model/sub_modules/mfd/__init__.py
+++ b/magic_pdf/model/sub_modules/mfd/__init__.py
--- a/magic_pdf/model/sub_modules/mfd/yolov8/__init__.py
+++ b/magic_pdf/model/sub_modules/mfd/yolov8/__init__.py
--- a/magic_pdf/model/sub_modules/mfr/__init__.py
+++ b/magic_pdf/model/sub_modules/mfr/__init__.py
--- a/magic_pdf/model/sub_modules/mfr/unimernet/__init__.py
+++ b/magic_pdf/model/sub_modules/mfr/unimernet/__init__.py
--- a/magic_pdf/model/sub_modules/ocr/__init__.py
+++ b/magic_pdf/model/sub_modules/ocr/__init__.py
--- a/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/__init__.py
+++ b/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/__init__.py
--- a/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/__init__.py
+++ b/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/modeling/__init__.py
--- a/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/__init__.py
+++ b/magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorchocr/utils/__init__.py
--- a/magic_pdf/model/sub_modules/reading_oreder/__init__.py
+++ b/magic_pdf/model/sub_modules/reading_oreder/__init__.py
--- a/magic_pdf/model/sub_modules/reading_oreder/layoutreader/__init__.py
+++ b/magic_pdf/model/sub_modules/reading_oreder/layoutreader/__init__.py
--- a/magic_pdf/model/sub_modules/table/__init__.py
+++ b/magic_pdf/model/sub_modules/table/__init__.py
--- a/magic_pdf/model/sub_modules/table/rapidtable/__init__.py
+++ b/magic_pdf/model/sub_modules/table/rapidtable/__init__.py
--- a/magic_pdf/model/sub_modules/table/table_utils.py
+++ b/magic_pdf/model/sub_modules/table/table_utils.py
-import re
-
-
-def minify_html(html):
-    # 移除多余的空白字符
-    html = re.sub(r'\s+', ' ', html)
-    # 移除行尾的空白字符
-    html = re.sub(r'\s*>\s*', '>', html)
-    # 移除标签前的空白字符
-    html = re.sub(r'\s*<\s*', '<', html)
-    return html.strip()
\ No newline at end of file
--- a/magic_pdf/operators/__init__.py
+++ b/magic_pdf/operators/__init__.py
--- a/magic_pdf/operators/models.py
+++ b/magic_pdf/operators/models.py