KaraKaraWitch commited on
Commit
bf76393
1 Parent(s): 643c37e

Upload script/extract-text.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. script/extract-text.py +147 -0
script/extract-text.py ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import json
3
+ from bs4 import BeautifulSoup
4
+ import ebooklib
5
+ from ebooklib import epub
6
+ import re
7
+ import xml.etree.ElementTree as ET
8
+
9
+ folder_path = './books' # Replace with your folder path
10
+ output_file_pattern = './output/output_part_{}.jsonl' # Pattern for output files
11
+ part_size = 1 # Number of files per part
12
+ part_counter = 0
13
+ file_counter = 0
14
+
15
+ def correct_french_punctuation(text):
16
+ # Correct spaces before punctuation in French
17
+ text = re.sub(r'\s+([?!:;])', r'\1', text) # Remove space before punctuation
18
+ text = re.sub(r'([?!:;])\s*', r'\1 ', text) # Add space after punctuation if not already present
19
+ text = re.sub(r'\s*-\s*', '-', text)
20
+ text = re.sub(r'\s*–\s*', '-', text)
21
+ return text
22
+
23
+ def find_navpoint_2_in_toc(book):
24
+ toc_item = book.get_item_with_id('ncx')
25
+ if toc_item is None:
26
+ return None
27
+ toc_content = toc_item.get_content()
28
+ namespaces = {'ncx': 'http://www.daisy.org/z3986/2005/ncx/'}
29
+ toc_root = ET.fromstring(toc_content)
30
+
31
+ nav_points = toc_root.findall('.//ncx:navPoint', namespaces)
32
+ for nav_point in nav_points:
33
+ if nav_point.attrib.get('id') == 'navpoint-2':
34
+ return nav_point.text if nav_point.text else None
35
+ return None
36
+
37
+ def find_section_href_in_toc(book, section_title):
38
+ toc_item = book.get_item_with_id('ncx')
39
+ if toc_item is None:
40
+ return None
41
+ toc_content = toc_item.get_content()
42
+ namespaces = {'ncx': 'http://www.daisy.org/z3986/2005/ncx/'}
43
+ toc_root = ET.fromstring(toc_content)
44
+ nav_points = toc_root.findall('.//ncx:navPoint', namespaces)
45
+ for nav_point in nav_points:
46
+ text_elements = nav_point.findall('.//ncx:navLabel/ncx:text', namespaces)
47
+ for text_element in text_elements:
48
+ if text_element.text == section_title:
49
+ content_element = nav_point.find('.//ncx:content', namespaces)
50
+ if content_element is not None:
51
+ return content_element.attrib['src']
52
+ return None
53
+
54
+ def extract_content_from_epub(book):
55
+ text = ''
56
+ start_section = find_section_href_in_toc(book, "Avant propos") or find_section_href_in_toc(book, "Premier Chapitre")
57
+ end_section_1 = find_section_href_in_toc(book, "À propos de cette édition électronique")
58
+ end_section_2 = find_section_href_in_toc(book, "Bibliographie – Œuvres complètes")
59
+
60
+ # Determine the final end section
61
+ if end_section_1 is not None and end_section_2 is not None:
62
+ end_section = end_section_1 if end_section_1 < end_section_2 else end_section_2
63
+ elif end_section_1 is not None:
64
+ end_section = end_section_1
65
+ else:
66
+ end_section = end_section_2
67
+
68
+ extracting = start_section is None # Start extracting if no specific start section
69
+
70
+ for item in book.get_items():
71
+ if item.get_type() == ebooklib.ITEM_DOCUMENT:
72
+ item_id = item.get_name()
73
+ if start_section and start_section in item_id:
74
+ extracting = True
75
+ if end_section and end_section in item_id:
76
+ break
77
+ if extracting or not start_section:
78
+ try:
79
+ soup = BeautifulSoup(item.get_content(), 'html.parser')
80
+ for p in soup.find_all('p'): # Process paragraph by paragraph
81
+ paragraph = p.get_text(separator='\n')
82
+ paragraph = paragraph.replace(u'\xa0', ' ')
83
+ paragraph = correct_french_punctuation(paragraph)
84
+ text += paragraph + '\n'
85
+ # Check for end phrases after each paragraph
86
+ if "FIN" in paragraph:
87
+ text = text.split("FIN", 1)[0]
88
+ print("End of book reached")
89
+ return text
90
+ elif "la Bibliothèque électronique du Québec" in paragraph:
91
+ text = text.split("la Bibliothèque électronique du Québec", 1)[0]
92
+ print("End of book reached")
93
+ return text
94
+ elif "ouvrage est le" in paragraph:
95
+ text = text.split("ouvrage est le", 1)[0]
96
+ print("End of book reached")
97
+ return text
98
+ except Exception as e:
99
+ print(f"Error processing content: {e}")
100
+
101
+ if not text:
102
+ print("Fallback: Adding all text as no specific sections were found.")
103
+ for item in book.get_items():
104
+ if item.get_type() == ebooklib.ITEM_DOCUMENT:
105
+ try:
106
+ soup = BeautifulSoup(item.get_content(), 'html.parser')
107
+ text += soup.get_text(separator='\n').replace(u'\xa0', ' ') + '\n'
108
+ except Exception as e:
109
+ print(f"Error in fallback processing: {e}")
110
+
111
+ return text
112
+
113
+
114
+
115
+
116
+ def extract_metadata_from_epub(book):
117
+ metadata = {}
118
+ try:
119
+ metadata['title'] = book.get_metadata('DC', 'title')
120
+ metadata['author'] = book.get_metadata('DC', 'creator')
121
+ metadata['publisher'] = book.get_metadata('DC', 'publisher')
122
+ # Add more metadata fields if needed
123
+ except Exception as e:
124
+ print(f"Error extracting metadata: {e}")
125
+ return metadata
126
+
127
+ for file in os.listdir(folder_path):
128
+ if file.endswith('.epub'):
129
+ if file_counter % part_size == 0:
130
+ if 'jsonl_file' in locals():
131
+ jsonl_file.close()
132
+ part_counter += 1
133
+ jsonl_file = open(output_file_pattern.format(part_counter), 'w', encoding='utf-8')
134
+
135
+ full_path = os.path.join(folder_path, file)
136
+ try:
137
+ book = epub.read_epub(full_path)
138
+ text = extract_content_from_epub(book)
139
+ meta = extract_metadata_from_epub(book)
140
+ jsonl_file.write(json.dumps({"text": text, "meta": meta}, ensure_ascii=False) + '\n')
141
+ file_counter += 1
142
+ print(f"reading file {file}")
143
+ except Exception as e:
144
+ print(f"Error reading file {file}: {e}")
145
+
146
+ if 'jsonl_file' in locals():
147
+ jsonl_file.close()