SouthpawIN
commited on
Commit
•
e6dfd23
1
Parent(s):
2669095
Upload arxivddl.py with huggingface_hub
Browse files- arxivddl.py +219 -0
arxivddl.py
ADDED
@@ -0,0 +1,219 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import requests
|
3 |
+
import sys
|
4 |
+
import os
|
5 |
+
from datetime import datetime
|
6 |
+
from dateutil.relativedelta import relativedelta
|
7 |
+
import logging
|
8 |
+
from multiprocessing import Pool, cpu_count
|
9 |
+
from tqdm import tqdm
|
10 |
+
import re
|
11 |
+
import feedparser
|
12 |
+
import time
|
13 |
+
|
14 |
+
# Configuration Variables
|
15 |
+
MONTHS_BACK = 99 # Default number of months back to download papers
|
16 |
+
OUTPUT_DIR = 'arxiv_dumps'
|
17 |
+
JSONL_FILE = 'downloaded_papers.jsonl'
|
18 |
+
CACHE_DIR = 'cache'
|
19 |
+
SEARCH_CACHE_FILE = os.path.join(CACHE_DIR, 'search_cache.jsonl')
|
20 |
+
MAX_RESULTS_PER_PAGE = 100
|
21 |
+
MAX_BACKOFF = 120 # Maximum backoff time in seconds
|
22 |
+
CONCURRENT_DOWNLOADS = min(32, cpu_count()) # Number of concurrent downloads
|
23 |
+
|
24 |
+
def setup_logging(enable_logging: bool):
|
25 |
+
if enable_logging:
|
26 |
+
logging.basicConfig(filename='arxivdump.log', filemode='a',
|
27 |
+
format='%(asctime)s - %(levelname)s - %(message)s',
|
28 |
+
level=logging.DEBUG)
|
29 |
+
else:
|
30 |
+
logging.basicConfig(level=logging.CRITICAL)
|
31 |
+
logging.disable(logging.CRITICAL)
|
32 |
+
|
33 |
+
def sanitize_filename(name):
|
34 |
+
return re.sub(r'[^\w\-]', '', name.replace(' ', '_'))
|
35 |
+
|
36 |
+
def load_cache(file_path):
|
37 |
+
if not os.path.exists(file_path):
|
38 |
+
return set()
|
39 |
+
with open(file_path, 'r', encoding='utf-8') as f:
|
40 |
+
return set(json.loads(line).get('id') for line in f)
|
41 |
+
|
42 |
+
def update_cache(file_path, metadata):
|
43 |
+
os.makedirs(os.path.dirname(file_path), exist_ok=True)
|
44 |
+
with open(file_path, 'a', encoding='utf-8') as f:
|
45 |
+
f.write(json.dumps(metadata, ensure_ascii=False) + '\n')
|
46 |
+
|
47 |
+
def fetch_paper_source(entry_id, headers):
|
48 |
+
src_url = entry_id.replace("/abs/", "/src/").replace("https://arxiv.org/", "https://export.arxiv.org/")
|
49 |
+
backoff = 1
|
50 |
+
|
51 |
+
while True:
|
52 |
+
try:
|
53 |
+
response = requests.get(src_url, headers=headers, allow_redirects=True, timeout=60)
|
54 |
+
content_type = response.headers.get('Content-Type', '').lower()
|
55 |
+
content_disposition = response.headers.get('Content-Disposition', '')
|
56 |
+
|
57 |
+
if 'filename=' in content_disposition:
|
58 |
+
filename = content_disposition.split('filename=')[-1].strip('"')
|
59 |
+
else:
|
60 |
+
filename = f"{entry_id.split('/')[-1]}.unlabeled_file_type"
|
61 |
+
|
62 |
+
return (response.content, content_type, dict(response.headers), filename)
|
63 |
+
|
64 |
+
except requests.exceptions.RequestException as e:
|
65 |
+
logging.warning(f"Request failed for {entry_id}: {e}. Retrying in {backoff} seconds.")
|
66 |
+
time.sleep(backoff)
|
67 |
+
backoff = min(backoff * 2, MAX_BACKOFF)
|
68 |
+
|
69 |
+
def save_archive_and_metadata(content, content_type, response_headers, filename, paper_title, index, jsonl_file, paper):
|
70 |
+
sanitized_title = sanitize_filename(paper_title)
|
71 |
+
paper_dir = os.path.join(OUTPUT_DIR, f"{index}_{sanitized_title}")
|
72 |
+
os.makedirs(paper_dir, exist_ok=True)
|
73 |
+
|
74 |
+
if content and filename:
|
75 |
+
archive_path = os.path.join(paper_dir, filename)
|
76 |
+
with open(archive_path, 'wb') as f:
|
77 |
+
f.write(content)
|
78 |
+
logging.info(f"Saved archive to: {archive_path}")
|
79 |
+
else:
|
80 |
+
logging.warning(f"No archive content to save for paper {index}_{sanitized_title}.")
|
81 |
+
|
82 |
+
metadata = {
|
83 |
+
"id": paper['id'],
|
84 |
+
"title": paper['title'],
|
85 |
+
"authors": [author['name'] for author in paper['authors']],
|
86 |
+
"summary": paper['summary'],
|
87 |
+
"categories": [tag['term'] for tag in paper['tags']],
|
88 |
+
"published": paper['published'],
|
89 |
+
"updated": paper['updated'],
|
90 |
+
"links": paper['links'],
|
91 |
+
"source_response_headers": response_headers,
|
92 |
+
"downloaded_at": datetime.utcnow().isoformat() + 'Z'
|
93 |
+
}
|
94 |
+
|
95 |
+
metadata_path = os.path.join(paper_dir, 'metadata.json')
|
96 |
+
with open(metadata_path, 'w', encoding='utf-8') as f:
|
97 |
+
json.dump(metadata, f, ensure_ascii=False, indent=4)
|
98 |
+
|
99 |
+
update_cache(jsonl_file, metadata)
|
100 |
+
|
101 |
+
def cache_search_results(results):
|
102 |
+
os.makedirs(CACHE_DIR, exist_ok=True)
|
103 |
+
with open(SEARCH_CACHE_FILE, 'a', encoding='utf-8') as f:
|
104 |
+
for paper in results:
|
105 |
+
f.write(json.dumps({
|
106 |
+
'id': paper['id'],
|
107 |
+
'title': paper['title'],
|
108 |
+
'published': paper['published'],
|
109 |
+
'updated': paper['updated'],
|
110 |
+
'summary': paper['summary'],
|
111 |
+
'authors': [author['name'] for author in paper['authors']],
|
112 |
+
'categories': [tag['term'] for tag in paper['tags']],
|
113 |
+
'links': paper['links']
|
114 |
+
}, ensure_ascii=False) + '\n')
|
115 |
+
|
116 |
+
def parse_atom_feed(feed_content):
|
117 |
+
parsed_feed = feedparser.parse(feed_content)
|
118 |
+
results = []
|
119 |
+
|
120 |
+
for entry in parsed_feed.entries:
|
121 |
+
links_dict = {link.get('title', link.get('rel', 'unknown')): link.get('href') for link in entry.links}
|
122 |
+
results.append({
|
123 |
+
'id': entry.id,
|
124 |
+
'title': entry.title,
|
125 |
+
'summary': entry.summary,
|
126 |
+
'published': entry.published,
|
127 |
+
'updated': entry.updated,
|
128 |
+
'tags': [{'term': tag.term} for tag in entry.get('tags', [])],
|
129 |
+
'authors': [{'name': author.name} for author in entry.authors],
|
130 |
+
'links': links_dict
|
131 |
+
})
|
132 |
+
|
133 |
+
return results
|
134 |
+
|
135 |
+
def process_paper(args):
|
136 |
+
index, paper, headers, jsonl_file, cached_ids = args
|
137 |
+
entry_id = paper['id']
|
138 |
+
paper_id = entry_id.split('/')[-1]
|
139 |
+
|
140 |
+
if paper_id in cached_ids:
|
141 |
+
return None
|
142 |
+
|
143 |
+
try:
|
144 |
+
content, content_type, response_headers, filename = fetch_paper_source(entry_id, headers)
|
145 |
+
if content and content_type:
|
146 |
+
save_archive_and_metadata(content, content_type, response_headers, filename, paper['title'], index, jsonl_file, paper)
|
147 |
+
return paper['published']
|
148 |
+
except Exception as e:
|
149 |
+
logging.error(f"Exception occurred while processing paper {paper_id}: {e}")
|
150 |
+
|
151 |
+
return None
|
152 |
+
|
153 |
+
def main():
|
154 |
+
setup_logging(True) # Always enable logging for now
|
155 |
+
|
156 |
+
print(f"Starting arXiv paper download for the past {MONTHS_BACK} months")
|
157 |
+
logging.info(f"Starting arXiv paper download for the past {MONTHS_BACK} months")
|
158 |
+
|
159 |
+
query = "cat:cs.*"
|
160 |
+
sort_by = "submittedDate"
|
161 |
+
sort_order = "descending"
|
162 |
+
print(f"Search query: {query} | sortBy={sort_by} | sortOrder={sort_order}")
|
163 |
+
logging.info(f"Search query: {query} | sortBy={sort_by} | sortOrder={sort_order}")
|
164 |
+
|
165 |
+
start = 0
|
166 |
+
papers = []
|
167 |
+
cutoff_date = datetime.now() - relativedelta(months=MONTHS_BACK)
|
168 |
+
reached_cutoff = False
|
169 |
+
|
170 |
+
while not reached_cutoff:
|
171 |
+
search_url = f"http://export.arxiv.org/api/query?search_query={query}&sortBy={sort_by}&sortOrder={sort_order}&start={start}&max_results={MAX_RESULTS_PER_PAGE}"
|
172 |
+
print(f"Fetching page starting from {start}: {search_url}")
|
173 |
+
logging.info(f"Fetching page starting from {start}: {search_url}")
|
174 |
+
|
175 |
+
response = requests.get(search_url)
|
176 |
+
parsed_results = parse_atom_feed(response.content)
|
177 |
+
if not parsed_results:
|
178 |
+
break
|
179 |
+
|
180 |
+
for paper in parsed_results:
|
181 |
+
published_date = datetime.strptime(paper['published'], "%Y-%m-%dT%H:%M:%SZ")
|
182 |
+
if published_date < cutoff_date:
|
183 |
+
reached_cutoff = True
|
184 |
+
break
|
185 |
+
papers.append(paper)
|
186 |
+
|
187 |
+
cache_search_results(parsed_results)
|
188 |
+
start += len(parsed_results)
|
189 |
+
|
190 |
+
if len(parsed_results) < MAX_RESULTS_PER_PAGE:
|
191 |
+
break
|
192 |
+
|
193 |
+
time.sleep(3) # Respectful delay between API calls
|
194 |
+
|
195 |
+
headers = {
|
196 |
+
"User-Agent": "Mozilla/5.0 (compatible; ArXivDownloader/1.0; +https://github.com/yourusername/arxiv-downloader)",
|
197 |
+
"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8",
|
198 |
+
"Accept-Language": "en-US,en;q=0.5",
|
199 |
+
"Referer": "https://arxiv.org/",
|
200 |
+
"Connection": "keep-alive",
|
201 |
+
"Upgrade-Insecure-Requests": "1"
|
202 |
+
}
|
203 |
+
|
204 |
+
cached_ids = load_cache(JSONL_FILE)
|
205 |
+
|
206 |
+
print(f"Starting to process {len(papers)} papers with multiprocessing...")
|
207 |
+
logging.info(f"Starting to process {len(papers)} papers with multiprocessing.")
|
208 |
+
os.makedirs(OUTPUT_DIR, exist_ok=True)
|
209 |
+
|
210 |
+
args_list = [(index, paper, headers, JSONL_FILE, cached_ids) for index, paper in enumerate(papers, start=1)]
|
211 |
+
|
212 |
+
with Pool(processes=CONCURRENT_DOWNLOADS) as pool:
|
213 |
+
list(tqdm(pool.imap_unordered(process_paper, args_list), total=len(papers), desc="Downloading papers"))
|
214 |
+
|
215 |
+
print("Processing complete.")
|
216 |
+
logging.info("Processing complete.")
|
217 |
+
|
218 |
+
if __name__ == "__main__":
|
219 |
+
main()
|