Mercurial > hg > cc > azure

annotate master/wecu/generate_file_list.py @ 57:ac1a20e627a9

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted, sac not quite working yet
author Henry S. Thompson <ht@markup.co.uk>
date Wed, 27 May 2020 20:54:34 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
57
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
1 #!/bin/python
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
2
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
3 from urllib.request import Request, urlopen, URLopener
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
4 import sys
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
5 import json
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
6 import subprocess
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
7 import os
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
8
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
9
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
10 def get_list_of_crawls():
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
11 req = Request('https://index.commoncrawl.org/collinfo.json')
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
12 resp = urlopen(req).read().decode('utf-8')
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
13 crawls_obj = json.loads(resp)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
14
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
15 crawls = []
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
16 for c in crawls_obj:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
17 crawls.append((c['name'], c['id']))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
18
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
19 return crawls
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
20
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
21 def print_crawls(crawls):
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
22 i = 0
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
23 for c in crawls:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
24 if 'ARC' not in c[0]:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
25 print('{}. {}'.format(i, c[0].replace('Index', '')))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
26 i += 1
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
27
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
28 def choose_crawl_and_download_paths():
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
29 crawls = get_list_of_crawls()
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
30 print("Select a crawl [0-{}]:".format(len(crawls)))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
31 print_crawls(crawls)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
32 try:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
33 crawl_no = int(input("Crawl number [0-{}]:".format(len(crawls))))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
34 except:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
35 print('Error: Enter a valid crawl number')
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
36 sys.exit(1)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
37
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
38 file_type = input("File Type [wat/wet/warc]:").lower()
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
39
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
40 if file_type not in ['warc', 'wat', 'wet']:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
41 print("Error: Enter a valid file type")
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
42 sys.exit(1)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
43
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
44 url_to_fetch = "https://commoncrawl.s3.amazonaws.com/crawl-data/{}/{}.paths.gz".format(crawls[crawl_no][1], file_type)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
45 path_file_opener = URLopener()
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
46 path_file_opener.retrieve(url_to_fetch, "paths.gz")
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
47
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
48 subprocess.check_output(['gunzip', '--force', 'paths.gz'])
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
49
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
50 return crawls[crawl_no][0]
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
51
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
52 ########
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
53 # MAIN #
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
54 ########
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
55
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
56
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
57 def main():
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
58 name_of_crawl = choose_crawl_and_download_paths()
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
59
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
60 full_size_str = input("Full crawl or sample? [f/s]: ")
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
61 if full_size_str == 'f':
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
62 os.system('mv paths input_paths')
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
63 else:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
64 #Sample selection
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
65 subset_size = 0
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
66 try:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
67 subset_size = int(input("Choose subset size: "))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
68 except:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
69 print("Enter a valid integer")
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
70 sys.exit(1)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
71
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
72 random_str = input("Random sample? [y/n]: ")
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
73 if random_str == 'y':
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
74 os.system('shuf -n {} paths > input_paths'.format(subset_size))
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
75 else:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
76 subprocess.call("head -{} paths > input_paths".format(subset_size), shell=True)
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
77
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
78 with open("crawl_name.txt", "w+") as cn_f:
ac1a20e627a9 from lukasz git repo 2020-05-26 (see ~/src/wecu), then editted,
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
79 cn_f.write(name_of_crawl.replace("Index", ""))