Downloads: wiki2corpus-1.0.1.py

File wiki2corpus-1.0.1.py, 11.2 KB (added by admin, 6 years ago)

Line
1	#!/usr/bin/python
2	#coding=utf-8
3
4	"""
5	MIT License
6
7	Copyright (c) 2016 Vit Baisa, Vit Suchomel
8
9	Permission is hereby granted, free of charge, to any person obtaining a copy
10	of this software and associated documentation files (the "Software"), to deal
11	in the Software without restriction, including without limitation the rights
12	to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
13	copies of the Software, and to permit persons to whom the Software is
14	furnished to do so, subject to the following conditions:
15
16	The above copyright notice and this permission notice shall be included in all
17	copies or substantial portions of the Software.
18
19	THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
20	IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21	FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
22	AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
23	LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
24	OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
25	SOFTWARE.
26	"""
27
28	VERSION = '1.0.1'
29
30	import re
31	import os
32	import sys
33	import argparse
34	import urllib2
35	import urllib
36	import socket #socket.error
37	import datetime
38	import json
39	import time
40	import codecs
41	import datetime
42	import gzip
43	from justext import core as justext
44	from lxml.etree import XMLSyntaxError, ParserError
45	remove_links_re = re.compile(u'</?a[^>]*>')
46
47	from unicodedata import category as unicode_char_category
48
49	class MissingPage(Exception):
50	pass
51
52	class EmptyHTML(Exception):
53	pass
54
55	class EmptyJusText(Exception):
56	pass
57
58	LATEST = 'https://dumps.wikimedia.org/%swiki/latest/%swiki-latest-all-titles-in-ns0.gz'
59	WIKI_URL = 'https://%s.wikipedia.org/wiki/'
60	API_HTML = 'https://%s.wikipedia.org/w/api.php?action=parse&page=%s&format=json'
61	API_JSON = 'https://%s.wikipedia.org/w/api.php?action=query&prop=revisions&titles=%s&format=json'
62
63	# TODO: look at https://en.wikipedia.org/w/api.php?action=query&prop=extracts&format=json&titles=xxx
64
65	def html2prevert(s, justext_wordlist):
66	# TODO: preclean HTML (remove tables, infoboxes, References, TOC, ...)
67	try:
68	html_root = justext.preprocess(html_text=s, encoding='utf-8')
69	paragraphs = justext.make_paragraphs(html_root)
70	except (ParserError, XMLSyntaxError):
71	return ('', 0, 0)
72	#use Justext to classify paragraphs
73	justext.classify_paragraphs(
74	paragraphs=paragraphs,
75	stoplist=justext_wordlist,
76	length_low=70, #character count < length_low => bad or short
77	length_high=200, #character count > length_high => good
78	stopwords_low=0.2, #number of words frequent in the language >= stopwords_low => neargood
79	stopwords_high=0.3, #number of words frequent in the language >= stopwords_high => good or neargood
80	max_link_density=0.4 #density of link words (words inside the <a> tag) > max_link_density => bad
81	)
82	justext.revise_paragraph_classification(
83	paragraphs=paragraphs,
84	max_heading_distance=200, #Short/near-good heads in the distance [chars] before a good par => good
85	)
86	#extract good paragraphs
87	prevert_paragraphs, paragraph_count, plaintext_len = [], 0, 0
88	for p in paragraphs:
89	#if p['class'] == 'good': # TODO find why this does not produce a good result
90	if p['cfclass'] in ('good', 'neargood'): #'good', 'neargood', 'short', 'bad'
91	p_text = justext.html_escape(p['text']).strip()
92	if p_text:
93	paragraph_count += 1
94	plaintext_len += len(p_text)
95	heading = u' heading="1"' if p['heading'] else u''
96	prevert_paragraphs.append(u'<p%s>\n%s\n</p>' % (heading, p_text))
97	return (u'\n'.join(prevert_paragraphs), paragraph_count, plaintext_len)
98
99	def api_wait(last, wait_interval):
100	n = datetime.datetime.now()
101	interval = (n-last).seconds + ((n-last).microseconds / 1.0e6)
102	if interval < wait_interval:
103	time.sleep(wait_interval - interval)
104
105	def process_article(langcode, title, linksf, raw_response_fp, justext_wordlist, logf, last_api_parse, wait_interval):
106	api_wait(last_api_parse, wait_interval)
107	api_url = API_HTML % (langcode, title)
108	try:
109	response_data = urllib2.urlopen(api_url).read()
110	parse_time = datetime.datetime.now()
111	except urllib2.HTTPError:
112	raise MissingPage()
113	try:
114	data = json.loads(response_data)
115	except socket.error:
116	raise MissingPage()
117	if not data or 'error' in data:
118	raise MissingPage()
119	#store the API response to allow re-processing without downloading in the future
120	raw_response_fp.write('%s\t%d\n' % (api_url, len(response_data) + 1))
121	raw_response_fp.write(response_data)
122	raw_response_fp.write('\n')
123	#parse the API response
124	p = data['parse']
125	html = p['text']['*'].strip()
126	if html:
127	#remove <a/> tags (Justext makes extra spaces there) # TODO: correct Justext and remove
128	html = remove_links_re.sub('', html)
129	prevert, paragraph_count, plaintext_len = html2prevert(
130	html.encode('utf-8'), justext_wordlist) # justext decodes!
131	else:
132	print >>logf, '\tempty HTML parse returned by API'
133	raise EmptyHTML()
134	if not prevert:
135	print >>logf, '\tempty prevert returned by jusText'
136	raise EmptyJusText()
137	revid = p['revid']
138	langlinks_len = len(p['langlinks'])
139	categories = '\|'.join([d['*'].replace('"', '') for d in p['categories']])
140	if linksf and p['externallinks']:
141	linksf.write('### %s\n' % title)
142	for line in p['externallinks']:
143	linksf.write(line.encode('utf-8') + '\n')
144	linksf.write('\n')
145	print >>logf, '\t%d chars' % plaintext_len
146	article_attrs = 'url="%s" title="%s" categories="%s" translations="%d" paragraphs="%d" chars="%d" downloaded="%s"' % \
147	((WIKI_URL % langcode) + title, title, categories.encode('utf-8'),
148	langlinks_len, paragraph_count, plaintext_len,
149	parse_time.strftime('%Y-%m-%d %H:%M'))
150	article = '<doc %s>\n%s\n</doc>\n' % (article_attrs, prevert.encode('utf-8'))
151	return (article, paragraph_count, revid, parse_time)
152
153	def main(langcode, cachefn, raw_response_path, logfn, newest, links, justext_wordlist, logf, wait_interval, nicetitles):
154	last_api_request = datetime.datetime.now()
155	last_api_parse = datetime.datetime.now()
156
157	linksf = open(links, 'a') if links else None
158	cache = {}
159	if os.path.exists(cachefn):
160	print >>logf, 'Cache: %s' % cachefn
161	with open(cachefn) as cf:
162	for line in cf:
163	try:
164	title, revid = line.split('\t')
165	cache[title.strip()] = revid.strip()
166	except ValueError:
167	continue
168	cf = open(cachefn, 'a') # cache file
169	raw_response_fp = open(raw_response_path, 'a') #raw API responses
170	print >>logf, 'Getting all titles from latest Wikipedia dump'
171	processed_articles = 0
172	skipped_articles = 0
173	empty_articles = 0
174	filename, _ = urllib.urlretrieve(LATEST % (langcode, langcode))
175	with gzip.open(filename) as df:
176	for line in df:
177	# TODO: download also talk pages
178	title = line.strip().replace('"', "'")
179	# TODO: filter titles, use RE as parameter
180	print >>logf, '%s' % title
181	if nicetitles and not unicode_char_category(unicode(title, 'utf-8')[0])[0] == 'L':
182	print >>logf, '\tskipping (not a nice title)'
183	continue
184	if title in cache:
185	if newest: # download the newest revision
186	previous_revid = cache[title]
187	api_wait(last_api_request, wait_interval)
188	last_api_request = datetime.datetime.now()
189	resp = urllib2.urlopen(API_JSON % (langcode, title))
190	data = json.load(response)
191	try:
192	dqp = data['query']['pages']
193	for key in dqp.keys():
194	current_revid = dqp[key]['revisions']['revid']
195	if previous_revid != current_revid:
196	article, parlen, revid, last_api_parse =\
197	process_article(langcode, title, linksf,
198	raw_response_fp, justext_wordlist, logf,
199	last_api_parse, wait_interval)
200	cache[title] = revid
201	cf.write(title + '\t' + revid + '\n')
202	else:
203	print >>logf, '\tskipping cached'
204	except (MissingPage, EmptyHTML, EmptyJusText):
205	article = ''
206	empty_articles += 1
207	else:
208	# do not download
209	print >>logf, '\tskip already downloaded'
210	skipped_articles += 1
211	article = ''
212	else:
213	try:
214	article, parlen, revid, last_api_parse =\
215	process_article(langcode, title, linksf,
216	raw_response_fp, justext_wordlist, logf,
217	last_api_parse, wait_interval)
218	cache[title] = revid
219	cf.write('%s\t%d\n' % (title, revid))
220	print >>logf, '\t%d paragraphs' % parlen
221	processed_articles += 1
222	except (MissingPage, EmptyHTML, EmptyJusText):
223	article = ''
224	empty_articles += 1
225	if article:
226	sys.stdout.write(article)
227	print >>logf, 'Updated cache database stored in %s' % cachefn
228	linksf.close()
229	cf.close()
230	print >>logf, 'Processed: %d' % processed_articles
231	print >>logf, 'Empty: %d' % empty_articles
232	print >>logf, 'Skipped: %d' % skipped_articles
233	logf.close()
234
235	if __name__ == '__main__':
236	parser = argparse.ArgumentParser(description='Wikipedia downloader')
237	parser.add_argument('langcode', help='Wikipedia language prefix', type=str)
238	parser.add_argument('wordlist', help='Path to a list of ~2000 most frequent words in the language (UTF-8, one per line)', type=str)
239	parser.add_argument('--cache', help='Directory with previously downloaded pages and data', type=str, default='')
240	parser.add_argument('--wait', help='Time interval between GET requests', type=float, default=1.0)
241	parser.add_argument('--newest', help='Download the newest versions of articles', action='store_true')
242	parser.add_argument('--links', help='Gather external links from Wikipedia', type=str, default='')
243	parser.add_argument('--nicetitles', help='Download only titles starting with alphabetical character', action='store_true')
244	args = parser.parse_args()
245	logfn = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S') + '.log'
246	logfile = open(logfn, 'w')
247	cachefile = args.cache or args.langcode + 'wiki.cache'
248	raw_response_file = (args.cache or args.langcode) + '_raw_data'
249	with open(args.wordlist) as fp:
250	justext_wordlist = set([line.decode('utf-8').rstrip() for line in fp])
251	main(args.langcode, cachefile, raw_response_file, logfile, args.newest, args.links, justext_wordlist, logfile, args.wait, args.nicetitles)

Download in other formats:

Original Format