wikiq_util.py

   1 import sys
   2 import re
   3 from subprocess import Popen, PIPE
   4 from collections import deque
   5 from hashlib import sha1
   6 from deltas.tokenizers import wikitext_split
   7 from mwxml import Dump
   8 import mwpersistence
   9 import mwreverts
  10 from urllib.parse import quote
  11 from urllib.parse import unquote
  12 from deltas import SequenceMatcher
  13
  14 TO_ENCODE = ('title', 'editor')
  15 PERSISTENCE_RADIUS = 7
  16
  17 def try_unquote(obj):
  18     if type(obj) is str:
  19         obj = unquote(obj)
  20         return obj.strip('\"')
  21     else:
  22         return
  23
  24 def calculate_persistence(tokens_added):
  25     return(sum([(len(x.revisions)-1) for x in tokens_added]),
  26            len(tokens_added))
  27
  28 class WikiqIterator():
  29     def __init__(self, fh, collapse_user=False):
  30         self.fh = fh
  31         self.collapse_user = collapse_user
  32         self.mwiterator = Dump.from_file(self.fh)
  33         self.namespace_map = { ns.id : ns.name for ns in
  34                                self.mwiterator.site_info.namespaces }
  35         self.__pages = self.load_pages()
  36
  37     def load_pages(self):
  38         for page in self.mwiterator:
  39             yield WikiqPage(page,
  40                             namespace_map = self.namespace_map,
  41                             collapse_user=self.collapse_user)
  42
  43     def __iter__(self):
  44         return self.__pages
  45
  46     def __next__(self):
  47         return next(self._pages)
  48
  49
  50 class WikiqPage():
  51     __slots__ = ('id', 'title', 'namespace', 'redirect',
  52                  'restrictions', 'mwpage', '__revisions',
  53                  'collapse_user')
  54
  55     def __init__(self, page, namespace_map, collapse_user=False):
  56         self.id = page.id
  57         self.namespace = page.namespace
  58         if page.namespace != 0:
  59             self.title = ':'.join([namespace_map[page.namespace], page.title])
  60         else:
  61             self.title = page.title
  62         self.restrictions = page.restrictions
  63         self.collapse_user = collapse_user
  64         self.mwpage = page
  65         self.__revisions = self.rev_list()
  66
  67     def rev_list(self):
  68         # Outline for how we want to handle collapse_user=True
  69         # iteration   rev.user   prev_rev.user   add prev_rev?
  70         #         0          A            None           Never
  71         #         1          A               A           False
  72         #         2          B               A            True
  73         #         3          A               B            True
  74         #         4          A               A           False
  75         # Post-loop                          A          Always
  76         for i, rev in enumerate(self.mwpage):
  77             # never yield the first time
  78             if i == 0:
  79                 if self.collapse_user:
  80                     collapsed_revs = 1
  81                     rev.collapsed_revs = collapsed_revs
  82
  83             else:
  84                 if self.collapse_user:
  85                     # yield if this is the last edit in a seq by a user and reset
  86                     # also yield if we do know who the user is
  87
  88                     if rev.deleted.user or prev_rev.deleted.user:
  89                         yield prev_rev
  90                         collapsed_revs = 1
  91                         rev.collapsed_revs = collapsed_revs
  92
  93                     elif not rev.user.text == prev_rev.user.text:
  94                         yield prev_rev
  95                         collapsed_revs = 1
  96                         rev.collapsed_revs = collapsed_revs
  97                     # otherwise, add one to the counter
  98                     else:
  99                         collapsed_revs += 1
 100                         rev.collapsed_revs = collapsed_revs
 101                 # if collapse_user is false, we always yield
 102                 else:
 103                     yield prev_rev
 104
 105             prev_rev = rev
 106
 107         # also yield the final time
 108         yield prev_rev
 109
 110     def __iter__(self):
 111         return self.__revisions
 112
 113     def __next__(self):
 114         return next(self.__revisions)
 115
 116 class WikiqParser():
 117
 118     def __init__(self, input_file, output_file, collapse_user=False, persist=False, urlencode=False, persist_legacy=False):
 119
 120         self.input_file = input_file
 121         self.output_file = output_file
 122         self.collapse_user = collapse_user
 123         self.persist = persist
 124         self.persist_legacy = persist_legacy
 125         self.printed_header = False
 126         self.namespaces = []
 127         self.urlencode = urlencode
 128
 129     def __get_namespace_from_title(self, title):
 130         default_ns = None
 131
 132         for ns in self.namespaces:
 133             # skip if the namespace is not defined
 134             if ns == None:
 135                 default_ns = self.namespaces[ns]
 136                 continue
 137
 138             if title.startswith(ns + ":"):
 139                 return self.namespaces[ns]
 140
 141         # if we've made it this far with no matches, we return the default namespace
 142         return default_ns
 143
 144     def process(self):
 145
 146         # create a regex that creates the output filename
 147         # output_filename = re.sub(r'^.*/(enwiki\-\d+)\-.*p(\d+)p.*$',
 148         #                         r'output/wikiq-\1-\2.tsv',
 149         #                         input_filename)
 150
 151         # Construct dump file iterator
 152         dump = WikiqIterator(self.input_file, collapse_user=self.collapse_user)
 153
 154         # extract list of namspaces
 155         self.namespaces = {ns.name : ns.id for ns in dump.mwiterator.site_info.namespaces}
 156
 157         page_count = 0
 158         rev_count = 0
 159
 160
 161         # Iterate through pages
 162         for page in dump:
 163             rev_detector = mwreverts.Detector()
 164
 165             if self.persist or self.persist_legacy:
 166                 window = deque(maxlen=PERSISTENCE_RADIUS)
 167
 168                 if not self.persist_legacy:
 169                     state = mwpersistence.DiffState(SequenceMatcher(tokenizer = wikitext_split),
 170                                                     revert_radius=PERSISTENCE_RADIUS)
 171
 172                 else:
 173                     from mw.lib import persistence
 174                     state = persistence.State()
 175
 176             # Iterate through a page's revisions
 177             for rev in page:
 178
 179                 rev_data = {'revid' : rev.id,
 180                             'date_time' : rev.timestamp.strftime('%Y-%m-%d %H:%M:%S'),
 181                             'articleid' : page.id,
 182                             'editor_id' : "" if rev.deleted.user == True or rev.user.id is None else rev.user.id,
 183                             'title' : '"' + page.title + '"',
 184                             'namespace' : page.namespace if page.namespace is not None else self.__get_namespace_from_title(page.title),
 185                             'deleted' : "TRUE" if rev.deleted.text else "FALSE" }
 186
 187                 # if revisions are deleted, /many/ things will be missing
 188                 if rev.deleted.text:
 189                     rev_data['text_chars'] = ""
 190                     rev_data['sha1'] = ""
 191                     rev_data['revert'] = ""
 192                     rev_data['reverteds'] = ""
 193
 194                 else:
 195                     # rev.text can be None if the page has no text
 196                     if not rev.text:
 197                         rev.text = ""
 198                     # if text exists, we'll check for a sha1 and generate one otherwise
 199
 200                     if rev.sha1:
 201                         text_sha1 = rev.sha1
 202                     else:
 203
 204                         text_sha1 = sha1(bytes(rev.text, "utf8")).hexdigest()
 205
 206                     rev_data['sha1'] = text_sha1
 207
 208                     # TODO rev.bytes doesn't work.. looks like a bug
 209                     rev_data['text_chars'] = len(rev.text)
 210
 211                     # generate revert data
 212                     revert = rev_detector.process(text_sha1, rev.id)
 213
 214                     if revert:
 215                         rev_data['revert'] = "TRUE"
 216                         rev_data['reverteds'] = '"' + ",".join([str(x) for x in revert.reverteds]) + '"'
 217                     else:
 218                         rev_data['revert'] = "FALSE"
 219                         rev_data['reverteds'] = ""
 220
 221                 # if the fact that the edit was minor can be hidden, this might be an issue
 222                 rev_data['minor'] = "TRUE" if rev.minor else "FALSE"
 223
 224                 if not rev.deleted.user:
 225                     # wrap user-defined editors in quotes for fread
 226                     rev_data['editor'] = '"' + rev.user.text + '"'
 227                     rev_data['anon'] = "TRUE" if rev.user.id == None else "FALSE"
 228
 229                 else:
 230                     rev_data['anon'] = ""
 231                     rev_data['editor'] = ""
 232
 233                 #if re.match(r'^#redirect \[\[.*\]\]', rev.text, re.I):
 234                 #    redirect = True
 235                 #else:
 236                 #    redirect = False
 237
 238                 #TODO missing: additions_size deletions_size
 239
 240                 # if collapse user was on, lets run that
 241                 if self.collapse_user:
 242                     rev_data['collapsed_revs'] = rev.collapsed_revs
 243
 244                 if self.persist or self.persist_legacy:
 245                     if rev.deleted.text:
 246
 247                         for k in ["token_revs", "tokens_added", "tokens_removed", "tokens_window"]:
 248                             old_rev_data[k] = None
 249                     else:
 250
 251                         if not self.persist_legacy:
 252                             _, tokens_added, tokens_removed = state.update(rev.text, rev.id)
 253
 254                         else:
 255                             _, tokens_added, tokens_removed = state.process(rev.text, rev.id, text_sha1)
 256
 257                         window.append((rev.id, rev_data, tokens_added, tokens_removed))
 258
 259                         if len(window) == PERSISTENCE_RADIUS:
 260                             old_rev_id, old_rev_data, old_tokens_added, old_tokens_removed = window[0]
 261
 262                             num_token_revs, num_tokens = calculate_persistence(old_tokens_added)
 263
 264                             old_rev_data["token_revs"] = num_token_revs
 265                             old_rev_data["tokens_added"] = num_tokens
 266                             old_rev_data["tokens_removed"] = len(old_tokens_removed)
 267                             old_rev_data["tokens_window"] = PERSISTENCE_RADIUS-1
 268
 269                             self.print_rev_data(old_rev_data)
 270
 271                 else:
 272                     self.print_rev_data(rev_data)
 273
 274                 rev_count += 1
 275
 276             if self.persist or self.persist_legacy:
 277                 # print out metadata for the last RADIUS revisions
 278                 for i, item in enumerate(window):
 279                     # if the window was full, we've already printed item 0
 280                     if len(window) == PERSISTENCE_RADIUS and i == 0:
 281                         continue
 282
 283                     rev_id, rev_data, tokens_added, tokens_removed = item
 284                     num_token_revs, num_tokens = calculate_persistence(tokens_added)
 285
 286                     rev_data["token_revs"] = num_token_revs
 287                     rev_data["tokens_added"] = num_tokens
 288                     rev_data["tokens_removed"] = len(tokens_removed)
 289                     rev_data["tokens_window"] = len(window)-(i+1)
 290
 291                     self.print_rev_data(rev_data)
 292
 293             page_count += 1
 294
 295         print("Done: %s revisions and %s pages." % (rev_count, page_count),
 296               file=sys.stderr)
 297
 298     def print_rev_data(self, rev_data):
 299         # if it's the first time through, print the header
 300         if self.urlencode:
 301             for field in TO_ENCODE:
 302                 rev_data[field] = quote(str(rev_data[field]))
 303
 304         if not self.printed_header:
 305             print("\t".join([str(k) for k in sorted(rev_data.keys())]), file=self.output_file)
 306             self.printed_header = True
 307
 308         print("\t".join([str(v) for k, v in sorted(rev_data.items())]), file=self.output_file)
 309
 310
 311 def open_input_file(input_filename):
 312     if re.match(r'.*\.7z$', input_filename):
 313         cmd = ["7za", "x", "-so", input_filename, '*']
 314     elif re.match(r'.*\.gz$', input_filename):
 315         cmd = ["zcat", input_filename]
 316     elif re.match(r'.*\.bz2$', input_filename):
 317         cmd = ["bzcat", "-dk", input_filename]
 318
 319     try:
 320         input_file = Popen(cmd, stdout=PIPE).stdout
 321     except NameError:
 322         input_file = open(input_filename, 'r')
 323
 324     return input_file
 325
 326 def open_output_file(input_filename):
 327     # create a regex that creates the output filename
 328     output_filename = re.sub(r'\.(7z|gz|bz2)?$', '', input_filename)
 329     output_filename = re.sub(r'\.xml', '', output_filename)
 330     output_filename = output_filename + ".tsv"
 331     output_file = open(output_filename, "w")
 332
 333     return output_file
 334
 335
 336 class IPCheck(object):
 337
 338     # IP address regexes taken from https://gist.github.com/mnordhoff/2213179
 339     ipv4_address = re.compile('^(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])$')
 340
 341     ipv6_address_or_addrz = re.compile('^(?:(?:[0-9A-Fa-f]{1,4}:){6}(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|::(?:[0-9A-Fa-f]{1,4}:){5}(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){4}(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){3}(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:(?:[0-9A-Fa-f]{1,4}:){,2}[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){2}(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:(?:[0-9A-Fa-f]{1,4}:){,3}[0-9A-Fa-f]{1,4})?::[0-9A-Fa-f]{1,4}:(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:(?:[0-9A-Fa-f]{1,4}:){,4}[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\\.){3}(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))|(?:(?:[0-9A-Fa-f]{1,4}:){,5}[0-9A-Fa-f]{1,4})?::[0-9A-Fa-f]{1,4}|(?:(?:[0-9A-Fa-f]{1,4}:){,6}[0-9A-Fa-f]{1,4})?::)(?:%25(?:[A-Za-z0-9\\-._~]|%[0-9A-Fa-f]{2})+)?$')
 342
 343     @staticmethod
 344     def is_ip(username):
 345         if not type(username) is str:
 346             return False
 347
 348         '''Check if a username is an ip (v4 or v6) address. We use this as
 349         a marker of whether the user is anonymous.'''
 350         if IPCheck.ipv4_address.match(username) or  IPCheck.ipv6_address_or_addrz.match(username):
 351             return True
 352         else:
 353             return False