3 # original wikiq headers are: title articleid revid date_time anon
4 # editor editor_id minor text_size text_entropy text_md5 reversion
5 # additions_size deletions_size
12 from subprocess import Popen, PIPE
13 from collections import deque
14 from hashlib import sha1
16 from mwxml import Dump
18 from deltas.tokenizers import wikitext_split
21 from urllib.parse import quote
22 TO_ENCODE = ('title', 'editor')
24 from deltas import SequenceMatcher
25 from deltas import SegmentMatcher
33 def calculate_persistence(tokens_added):
34 return(sum([(len(x.revisions)-1) for x in tokens_added]),
37 class WikiqIterator():
38 def __init__(self, fh, collapse_user=False):
40 self.collapse_user = collapse_user
41 self.mwiterator = Dump.from_file(self.fh)
42 self.namespace_map = { ns.id : ns.name for ns in
43 self.mwiterator.site_info.namespaces }
44 self.__pages = self.load_pages()
47 for page in self.mwiterator:
49 namespace_map = self.namespace_map,
50 collapse_user=self.collapse_user)
56 return next(self._pages)
59 __slots__ = ('id', 'title', 'namespace', 'redirect',
60 'restrictions', 'mwpage', '__revisions',
63 def __init__(self, page, namespace_map, collapse_user=False):
65 self.namespace = page.namespace
66 # following mwxml, we assume namespace 0 in cases where
67 # page.namespace is inconsistent with namespace_map
68 if page.namespace not in namespace_map:
69 self.title = page.title
71 if page.namespace != 0:
72 self.title = ':'.join([namespace_map[page.namespace], page.title])
74 self.title = page.title
75 self.restrictions = page.restrictions
76 self.collapse_user = collapse_user
78 self.__revisions = self.rev_list()
81 # Outline for how we want to handle collapse_user=True
82 # iteration rev.user prev_rev.user add prev_rev?
89 for i, rev in enumerate(self.mwpage):
90 # never yield the first time
92 if self.collapse_user:
94 rev.collapsed_revs = collapsed_revs
97 if self.collapse_user:
98 # yield if this is the last edit in a seq by a user and reset
99 # also yield if we do know who the user is
101 if rev.deleted.user or prev_rev.deleted.user:
104 rev.collapsed_revs = collapsed_revs
106 elif not rev.user.text == prev_rev.user.text:
109 rev.collapsed_revs = collapsed_revs
110 # otherwise, add one to the counter
113 rev.collapsed_revs = collapsed_revs
114 # if collapse_user is false, we always yield
120 # also yield the final time
124 return self.__revisions
127 return next(self.__revisions)
131 def __init__(self, input_file, output_file, collapse_user=False, persist=None, urlencode=False):
134 persist : what persistence method to use. Takes a PersistMethod value
137 self.input_file = input_file
138 self.output_file = output_file
139 self.collapse_user = collapse_user
140 self.persist = persist
141 self.printed_header = False
143 self.urlencode = urlencode
145 def __get_namespace_from_title(self, title):
148 for ns in self.namespaces:
149 # skip if the namespace is not defined
151 default_ns = self.namespaces[ns]
154 if title.startswith(ns + ":"):
155 return self.namespaces[ns]
157 # if we've made it this far with no matches, we return the default namespace
162 # create a regex that creates the output filename
163 # output_filename = re.sub(r'^.*/(enwiki\-\d+)\-.*p(\d+)p.*$',
164 # r'output/wikiq-\1-\2.tsv',
167 # Construct dump file iterator
168 dump = WikiqIterator(self.input_file, collapse_user=self.collapse_user)
170 # extract list of namspaces
171 self.namespaces = {ns.name : ns.id for ns in dump.mwiterator.site_info.namespaces}
177 # Iterate through pages
179 rev_detector = mwreverts.Detector()
181 if self.persist != PersistMethod.none:
182 window = deque(maxlen=PERSISTENCE_RADIUS)
184 if self.persist == PersistMethod.sequence:
185 state = mwpersistence.DiffState(SequenceMatcher(tokenizer = wikitext_split),
186 revert_radius=PERSISTENCE_RADIUS)
188 elif self.persist == PersistMethod.segment:
189 state = mwpersistence.DiffState(SegmentMatcher(tokenizer = wikitext_split),
190 revert_radius=PERSISTENCE_RADIUS)
192 # self.persist == PersistMethod.legacy
194 from mw.lib import persistence
195 state = persistence.State()
197 # Iterate through a page's revisions
200 rev_data = {'revid' : rev.id,
201 'date_time' : rev.timestamp.strftime('%Y-%m-%d %H:%M:%S'),
202 'articleid' : page.id,
203 'editor_id' : "" if rev.deleted.user == True or rev.user.id is None else rev.user.id,
204 'title' : '"' + page.title + '"',
205 'namespace' : page.namespace if page.namespace is not None else self.__get_namespace_from_title(page.title),
206 'deleted' : "TRUE" if rev.deleted.text else "FALSE" }
208 # if revisions are deleted, /many/ things will be missing
210 rev_data['text_chars'] = ""
211 rev_data['sha1'] = ""
212 rev_data['revert'] = ""
213 rev_data['reverteds'] = ""
216 # rev.text can be None if the page has no text
219 # if text exists, we'll check for a sha1 and generate one otherwise
225 text_sha1 = sha1(bytes(rev.text, "utf8")).hexdigest()
227 rev_data['sha1'] = text_sha1
229 # TODO rev.bytes doesn't work.. looks like a bug
230 rev_data['text_chars'] = len(rev.text)
232 # generate revert data
233 revert = rev_detector.process(text_sha1, rev.id)
236 rev_data['revert'] = "TRUE"
237 rev_data['reverteds'] = '"' + ",".join([str(x) for x in revert.reverteds]) + '"'
239 rev_data['revert'] = "FALSE"
240 rev_data['reverteds'] = ""
242 # if the fact that the edit was minor can be hidden, this might be an issue
243 rev_data['minor'] = "TRUE" if rev.minor else "FALSE"
245 if not rev.deleted.user:
246 # wrap user-defined editors in quotes for fread
247 rev_data['editor'] = '"' + rev.user.text + '"'
248 rev_data['anon'] = "TRUE" if rev.user.id == None else "FALSE"
251 rev_data['anon'] = ""
252 rev_data['editor'] = ""
254 #if re.match(r'^#redirect \[\[.*\]\]', rev.text, re.I):
259 #TODO missing: additions_size deletions_size
261 # if collapse user was on, lets run that
262 if self.collapse_user:
263 rev_data['collapsed_revs'] = rev.collapsed_revs
265 if self.persist != PersistMethod.none:
267 for k in ["token_revs", "tokens_added", "tokens_removed", "tokens_window"]:
268 old_rev_data[k] = None
271 if self.persist != PersistMethod.legacy:
272 _, tokens_added, tokens_removed = state.update(rev.text, rev.id)
275 _, tokens_added, tokens_removed = state.process(rev.text, rev.id, text_sha1)
277 window.append((rev.id, rev_data, tokens_added, tokens_removed))
279 if len(window) == PERSISTENCE_RADIUS:
280 old_rev_id, old_rev_data, old_tokens_added, old_tokens_removed = window[0]
282 num_token_revs, num_tokens = calculate_persistence(old_tokens_added)
284 old_rev_data["token_revs"] = num_token_revs
285 old_rev_data["tokens_added"] = num_tokens
286 old_rev_data["tokens_removed"] = len(old_tokens_removed)
287 old_rev_data["tokens_window"] = PERSISTENCE_RADIUS-1
289 self.print_rev_data(old_rev_data)
292 self.print_rev_data(rev_data)
296 if self.persist != PersistMethod.none:
297 # print out metadata for the last RADIUS revisions
298 for i, item in enumerate(window):
299 # if the window was full, we've already printed item 0
300 if len(window) == PERSISTENCE_RADIUS and i == 0:
303 rev_id, rev_data, tokens_added, tokens_removed = item
304 num_token_revs, num_tokens = calculate_persistence(tokens_added)
306 rev_data["token_revs"] = num_token_revs
307 rev_data["tokens_added"] = num_tokens
308 rev_data["tokens_removed"] = len(tokens_removed)
309 rev_data["tokens_window"] = len(window)-(i+1)
311 self.print_rev_data(rev_data)
315 print("Done: %s revisions and %s pages." % (rev_count, page_count),
318 def print_rev_data(self, rev_data):
319 # if it's the first time through, print the header
321 for field in TO_ENCODE:
322 rev_data[field] = quote(str(rev_data[field]))
324 if not self.printed_header:
325 print("\t".join([str(k) for k in sorted(rev_data.keys())]), file=self.output_file)
326 self.printed_header = True
328 print("\t".join([str(v) for k, v in sorted(rev_data.items())]), file=self.output_file)
331 def open_input_file(input_filename):
332 if re.match(r'.*\.7z$', input_filename):
333 cmd = ["7za", "x", "-so", input_filename, '*']
334 elif re.match(r'.*\.gz$', input_filename):
335 cmd = ["zcat", input_filename]
336 elif re.match(r'.*\.bz2$', input_filename):
337 cmd = ["bzcat", "-dk", input_filename]
340 input_file = Popen(cmd, stdout=PIPE).stdout
342 input_file = open(input_filename, 'r')
346 def open_output_file(input_filename):
347 # create a regex that creates the output filename
348 output_filename = re.sub(r'\.(7z|gz|bz2)?$', '', input_filename)
349 output_filename = re.sub(r'\.xml', '', output_filename)
350 output_filename = output_filename + ".tsv"
351 output_file = open(output_filename, "w")
355 parser = argparse.ArgumentParser(description='Parse MediaWiki XML database dumps into tab delimitted data.')
357 # arguments for the input direction
358 parser.add_argument('dumpfiles', metavar="DUMPFILE", nargs="*", type=str,
359 help="Filename of the compressed or uncompressed XML database dump. If absent, we'll look for content on stdin and output on stdout.")
361 parser.add_argument('-o', '--output-dir', metavar='DIR', dest='output_dir', type=str, nargs=1,
362 help="Directory for output files.")
364 parser.add_argument('-s', '--stdout', dest="stdout", action="store_true",
365 help="Write output to standard out (do not create dump file)")
367 parser.add_argument('--collapse-user', dest="collapse_user", action="store_true",
368 help="Operate only on the final revision made by user a user within all sequences of consecutive edits made by a user. This can be useful for addressing issues with text persistence measures.")
370 parser.add_argument('-p', '--persistence', dest="persist", default=None, const='', type=str, choices = ['','segment','sequence','legacy'], nargs='?',
371 help="Compute and report measures of content persistent: (1) persistent token revisions, (2) tokens added, and (3) number of revision used in computing the first measure. This may by slow. Use -p=segment for advanced persistence calculation method that is robust to content moves. This might be very slow. Use -p=legacy for legacy behavior.")
373 parser.add_argument('-u', '--url-encode', dest="urlencode", action="store_true",
374 help="Output url encoded text strings. This works around some data issues like newlines in editor names. In the future it may be used to output other text data.")
376 parser.add_argument('-n', '--namespace-include', dest="namespace_filter", type=int, action='append',
377 help="Id number of namspace to include.")
380 args = parser.parse_args()
382 # set persistence method
384 if args.persist is None:
385 persist = PersistMethod.none
386 elif args.persist == "segment":
387 persist = PersistMethod.segment
388 elif args.persist == "legacy":
389 persist = PersistMethod.legacy
391 persist = PersistMethod.sequence
393 if len(args.dumpfiles) > 0:
394 for filename in args.dumpfiles:
395 input_file = open_input_file(filename)
397 # open directory for output
399 output_dir = args.output_dir[0]
403 print("Processing file: %s" % filename, file=sys.stderr)
406 output_file = sys.stdout
408 filename = os.path.join(output_dir, os.path.basename(filename))
409 output_file = open_output_file(filename)
411 wikiq = WikiqParser(input_file, output_file,
412 collapse_user=args.collapse_user,
414 urlencode=args.urlencode)
422 wikiq = WikiqParser(sys.stdin, sys.stdout,
423 collapse_user=args.collapse_user,
425 persist_legacy=args.persist_legacy,
426 urlencode=args.urlencode)
429 # stop_words = "a,able,about,across,after,all,almost,also,am,among,an,and,any,are,as,at,be,because,been,but,by,can,cannot,could,dear,did,do,does,either,else,ever,every,for,from,get,got,had,has,have,he,her,hers,him,his,how,however,i,if,in,into,is,it,its,just,least,let,like,likely,may,me,might,most,must,my,neither,no,nor,not,of,off,often,on,only,or,other,our,own,rather,said,say,says,she,should,since,so,some,than,that,the,their,them,then,there,these,they,this,tis,to,too,twas,us,wants,was,we,were,what,when,where,which,while,who,whom,why,will,with,would,yet,you,your"
430 # stop_words = stop_words.split(",")