目录重构

pdf_tools/libs/nlp_utils.py → magic_pdf/libs/nlp_utils.py

@@ -10,7 +10,7 @@ import spacy
 import en_core_web_sm
 import zh_core_web_sm
 
-from pdf_tools.libs.language import detect_lang
+from magic_pdf.libs.language import detect_lang
 
 
 class NLPModels:

pdf_tools/libs/pdf_image_tools.py → magic_pdf/libs/pdf_image_tools.py

@@ -4,9 +4,9 @@ from typing import Tuple
 import io
 
 # from app.common.s3 import get_s3_client
-from pdf_tools.libs.commons import fitz
+from magic_pdf.libs.commons import fitz
 from loguru import logger
-from pdf_tools.libs.commons import parse_bucket_key, join_path
+from magic_pdf.libs.commons import parse_bucket_key, join_path
 
 
 def cut_image(bbox: Tuple, page_num: int, page: fitz.Page, save_parent_path: str, s3_return_path=None, img_s3_client=None, upload_switch=True):

pdf_tools/libs/vis_utils.py → magic_pdf/libs/vis_utils.py

-from pdf_tools.libs.commons import fitz
+from magic_pdf.libs.commons import fitz
 import os
 
 

pdf_tools/para/block_continuation_processor.py → magic_pdf/para/block_continuation_processor.py

 import os
 import unicodedata
 
-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 
 if sys.version_info[0] >= 3:

pdf_tools/para/block_termination_processor.py → magic_pdf/para/block_termination_processor.py

-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 
 if sys.version_info[0] >= 3:

pdf_tools/para/commons.py → magic_pdf/para/commons.py

 import sys
 
-from pdf_tools.libs.commons import fitz
+from magic_pdf.libs.commons import fitz
 from termcolor import cprint
 
 

pdf_tools/para/denoise.py → magic_pdf/para/denoise.py

 import math
 
 from collections import defaultdict
-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 if sys.version_info[0] >= 3:
     sys.stdout.reconfigure(encoding="utf-8")  # type: ignore

pdf_tools/para/draw.py → magic_pdf/para/draw.py

-from pdf_tools.libs.commons import fitz
+from magic_pdf.libs.commons import fitz
 
-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 
 if sys.version_info[0] >= 3:

pdf_tools/para/layout_match_processor.py → magic_pdf/para/layout_match_processor.py

 import math
-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 
 if sys.version_info[0] >= 3:

pdf_tools/para/para_pipeline.py → magic_pdf/para/para_pipeline.py

 import os
 import json
 
-from pdf_tools.para.commons import *
-
-from pdf_tools.para.raw_processor import RawBlockProcessor
-from pdf_tools.para.layout_match_processor import LayoutFilterProcessor
-from pdf_tools.para.stats import BlockStatisticsCalculator
-from pdf_tools.para.stats import DocStatisticsCalculator
-from pdf_tools.para.title_processor import TitleProcessor
-from pdf_tools.para.block_termination_processor import BlockTerminationProcessor
-from pdf_tools.para.block_continuation_processor import BlockContinuationProcessor
-from pdf_tools.para.draw import DrawAnnos
-from pdf_tools.para.exceptions import (
+from magic_pdf.para.commons import *
+
+from magic_pdf.para.raw_processor import RawBlockProcessor
+from magic_pdf.para.layout_match_processor import LayoutFilterProcessor
+from magic_pdf.para.stats import BlockStatisticsCalculator
+from magic_pdf.para.stats import DocStatisticsCalculator
+from magic_pdf.para.title_processor import TitleProcessor
+from magic_pdf.para.block_termination_processor import BlockTerminationProcessor
+from magic_pdf.para.block_continuation_processor import BlockContinuationProcessor
+from magic_pdf.para.draw import DrawAnnos
+from magic_pdf.para.exceptions import (
     DenseSingleLineBlockException,
     TitleDetectionException,
     TitleLevelException,

pdf_tools/para/stats.py → magic_pdf/para/stats.py

 from collections import Counter
 import numpy as np
 
-from pdf_tools.para.commons import *
+from magic_pdf.para.commons import *
 
 
 if sys.version_info[0] >= 3: