X-Git-Url: https://code.communitydata.science/mediawiki_dump_tools.git/blobdiff_plain/ba886ecf4c1cc5003882be8fa2f8d2e733471c52..4ccde845290a9ae350456107c1171638e81901e0:/wikiq diff --git a/wikiq b/wikiq index 7a2f8e4..7a1b846 100755 --- a/wikiq +++ b/wikiq @@ -13,27 +13,91 @@ from subprocess import Popen, PIPE from collections import deque from hashlib import sha1 -from mw.xml_dump import Iterator -from mw.lib import persistence -from mw.lib import reverts +from mwxml import Dump + +from deltas.tokenizers import wikitext_split +import mwpersistence +import mwreverts from urllib.parse import quote TO_ENCODE = ('title', 'editor') PERSISTENCE_RADIUS=7 +from deltas import SequenceMatcher +from deltas import SegmentMatcher + +class PersistMethod: + none = 0 + sequence = 1 + segment = 2 + legacy = 3 def calculate_persistence(tokens_added): return(sum([(len(x.revisions)-1) for x in tokens_added]), len(tokens_added)) +def matchmake(scanned_content, rev_data, regex, label): + p = re.compile(regex) + + temp_dict = {} + # if there are named capture groups in the regex + if bool(p.groupindex): + capture_groups = list(p.groupindex.keys()) + + # initialize the {capture_group_name:list} for each capture group + for cap_group in capture_groups: + temp_dict["{}_{}".format(label, cap_group)] = [] + + # if there are matches of some sort in this revision content, fill the lists for each cap_group + if p.search(scanned_content) is not None: + m = re.finditer(p,scanned_content) + matchobjects = list(m) + + for cap_group in capture_groups: + temp_list = [] + for match in matchobjects: + # we only want to add the match for the capture group if the match is not None + if match.group(cap_group) != None: + temp_list.append(match.group(cap_group)) + + # if temp_list of matches is empty just make that column None + if len(temp_list)==0: + temp_dict["{}_{}".format(label, cap_group)] = None + # else we put in the list we made in the for-loop above + else: + temp_dict["{}_{}".format(label, cap_group)] = ', '.join(temp_list) + + # there are no matches at all in this revision content, we default values to None + else: + for cap_group in capture_groups: + temp_dict["{}_{}".format(label, cap_group)] = None + + # there are no capture groups, we just search for all the matches of the regex + else: + #given that there are matches to be made + if p.search(scanned_content) is not None: + m = p.findall(scanned_content) + temp_dict[label] = ', '.join(m) + else: + temp_dict[label] = None + # update rev_data with our new columns + rev_data.update(temp_dict) + print(rev_data.keys()) + return rev_data + + class WikiqIterator(): def __init__(self, fh, collapse_user=False): self.fh = fh self.collapse_user = collapse_user - self.mwiterator = Iterator.from_file(self.fh) + self.mwiterator = Dump.from_file(self.fh) + self.namespace_map = { ns.id : ns.name for ns in + self.mwiterator.site_info.namespaces } self.__pages = self.load_pages() def load_pages(self): for page in self.mwiterator: - yield WikiqPage(page, collapse_user=self.collapse_user) + yield WikiqPage(page, + namespace_map = self.namespace_map, + collapse_user=self.collapse_user) def __iter__(self): return self.__pages @@ -46,13 +110,19 @@ class WikiqPage(): 'restrictions', 'mwpage', '__revisions', 'collapse_user') - def __init__(self, page, collapse_user=False): + def __init__(self, page, namespace_map, collapse_user=False): self.id = page.id - self.title = page.title self.namespace = page.namespace - self.redirect = page.redirect + # following mwxml, we assume namespace 0 in cases where + # page.namespace is inconsistent with namespace_map + if page.namespace not in namespace_map: + self.title = page.title + page.namespace = 0 + if page.namespace != 0: + self.title = ':'.join([namespace_map[page.namespace], page.title]) + else: + self.title = page.title self.restrictions = page.restrictions - self.collapse_user = collapse_user self.mwpage = page self.__revisions = self.rev_list() @@ -76,7 +146,14 @@ class WikiqPage(): else: if self.collapse_user: # yield if this is the last edit in a seq by a user and reset - if not rev.contributor.user_text == prev_rev.contributor.user_text: + # also yield if we do know who the user is + + if rev.deleted.user or prev_rev.deleted.user: + yield prev_rev + collapsed_revs = 1 + rev.collapsed_revs = collapsed_revs + + elif not rev.user.text == prev_rev.user.text: yield prev_rev collapsed_revs = 1 rev.collapsed_revs = collapsed_revs @@ -89,6 +166,7 @@ class WikiqPage(): yield prev_rev prev_rev = rev + # also yield the final time yield prev_rev @@ -99,10 +177,12 @@ class WikiqPage(): return next(self.__revisions) class WikiqParser(): + def __init__(self, input_file, output_file, regex_match_revision, regex_match_comment, regex_revision_label, regex_comment_label, collapse_user=False, persist=None, urlencode=False, namespaces = None, revert_radius=15): + """ + Parameters: + persist : what persistence method to use. Takes a PersistMethod value + """ - - def __init__(self, input_file, output_file, collapse_user=False, persist=False, urlencode=False): - self.input_file = input_file self.output_file = output_file self.collapse_user = collapse_user @@ -110,7 +190,17 @@ class WikiqParser(): self.printed_header = False self.namespaces = [] self.urlencode = urlencode - + self.revert_radius = revert_radius + self.regex_match_revision = regex_match_revision + self.regex_revision_label = regex_revision_label + self.regex_match_comment = regex_match_comment + self.regex_comment_label = regex_comment_label + + if namespaces is not None: + self.namespace_filter = set(namespaces) + else: + self.namespace_filter = None + def __get_namespace_from_title(self, title): default_ns = None @@ -126,6 +216,7 @@ class WikiqParser(): # if we've made it this far with no matches, we return the default namespace return default_ns + def process(self): # create a regex that creates the output filename @@ -137,50 +228,117 @@ class WikiqParser(): dump = WikiqIterator(self.input_file, collapse_user=self.collapse_user) # extract list of namspaces - self.namespaces = {ns.name : ns.id for ns in dump.mwiterator.namespaces} + self.namespaces = {ns.name : ns.id for ns in dump.mwiterator.site_info.namespaces} page_count = 0 rev_count = 0 + + # Iterate through pages for page in dump: - if self.persist: - state = persistence.State() + namespace = page.namespace if page.namespace is not None else self.__get_namespace_from_title(page.title) + + # skip namespaces not in the filter + if self.namespace_filter is not None: + if namespace not in self.namespace_filter: + continue + + print(self.revert_radius) + rev_detector = mwreverts.Detector(radius = self.revert_radius) + + + if self.persist != PersistMethod.none: window = deque(maxlen=PERSISTENCE_RADIUS) - rev_detector = reverts.Detector() + if self.persist == PersistMethod.sequence: + state = mwpersistence.DiffState(SequenceMatcher(tokenizer = wikitext_split), + revert_radius=PERSISTENCE_RADIUS) + + elif self.persist == PersistMethod.segment: + state = mwpersistence.DiffState(SegmentMatcher(tokenizer = wikitext_split), + revert_radius=PERSISTENCE_RADIUS) + + # self.persist == PersistMethod.legacy + else: + from mw.lib import persistence + state = persistence.State() # Iterate through a page's revisions for rev in page: + + # initialize rev_data + rev_data = {} - rev_data = {'revid' : rev.id, - 'date_time' : rev.timestamp.strftime('%Y-%m-%d %H:%M:%S'), - 'articleid' : page.id, - 'editor_id' : "" if rev.contributor.id == None else rev.contributor.id, - 'title' : '"' + page.title + '"', - 'namespace' : page.namespace if page.namespace else self.__get_namespace_from_title(page.title), - 'deleted' : "TRUE" if rev.text.deleted else "FALSE" } + # if the command line args only gave a label (and no regular expression is given) + if (self.regex_revision_label != None and self.regex_match_revision == None) or (self.regex_comment_label != None and self.regex_match_comment == None): + sys.exit('The given regex label(s) has no corresponding regex to search for.') + + # if there's anything in the list of regex_match_revision + if self.regex_match_revision is not None: + if (self.regex_revision_label == None) or (len(self.regex_match_revision) != len(self.regex_revision_label)): + sys.exit('Each regular expression *must* come with a corresponding label and vice versa.') + + # initialize and construct the list of regex-label tuples + pairs = [] + for i in range(0,len(self.regex_match_revision)): + pairs.append((self.regex_match_revision[i], self.regex_revision_label[i])) + + # for each regex/label pair, we now run matchmake to check and output columns + for pair in pairs: + # pair[0] corresponds to the regex, pair[1] to the label + rev_data = matchmake(rev.text, rev_data, pair[0], pair[1]) + + # if there's anything in the list of regex_match_comment + if self.regex_match_comment is not None: + if (self.regex_comment_label == None) or (len(self.regex_match_comment) != len(self.regex_comment_label)): + sys.exit('Each regular expression *must* come with a corresponding label and vice versa.') + + # initialize and construct the list of regex-label tuples + pairs = [] + for i in range(0,len(self.regex_match_comment)): + pairs.append((self.regex_match_comment[i], self.regex_comment_label[i])) + + # for each regex/label pair, we now run matchmake to check and output columns + for pair in pairs: + # pair[0] corresponds to the regex, pair[1] to the label + rev_data = matchmake(rev.comment, rev_data, pair[0], pair[1]) + + # we fill out the rest of the data structure now + rev_data['revid'] = rev.id + rev_data['date_time'] = rev.timestamp.strftime('%Y-%m-%d %H:%M:%S') + rev_data['articleid'] = page.id + rev_data['editor_id'] = "" if rev.deleted.user == True or rev.user.id is None else rev.user.id + rev_data['title'] = '"' + page.title + '"' + rev_data['namespace'] = namespace + rev_data['deleted'] = "TRUE" if rev.deleted.text else "FALSE" # if revisions are deleted, /many/ things will be missing - if rev.text.deleted: + if rev.deleted.text: rev_data['text_chars'] = "" rev_data['sha1'] = "" rev_data['revert'] = "" rev_data['reverteds'] = "" else: + # rev.text can be None if the page has no text + if not rev.text: + rev.text = "" # if text exists, we'll check for a sha1 and generate one otherwise + if rev.sha1: text_sha1 = rev.sha1 else: + text_sha1 = sha1(bytes(rev.text, "utf8")).hexdigest() rev_data['sha1'] = text_sha1 # TODO rev.bytes doesn't work.. looks like a bug rev_data['text_chars'] = len(rev.text) - + # generate revert data revert = rev_detector.process(text_sha1, rev.id) + if revert: rev_data['revert'] = "TRUE" rev_data['reverteds'] = '"' + ",".join([str(x) for x in revert.reverteds]) + '"' @@ -191,10 +349,10 @@ class WikiqParser(): # if the fact that the edit was minor can be hidden, this might be an issue rev_data['minor'] = "TRUE" if rev.minor else "FALSE" - if rev.contributor.user_text: + if not rev.deleted.user: # wrap user-defined editors in quotes for fread - rev_data['editor'] = '"' + rev.contributor.user_text + '"' - rev_data['anon'] = "TRUE" if rev.contributor.id == None else "FALSE" + rev_data['editor'] = '"' + rev.user.text + '"' + rev_data['anon'] = "TRUE" if rev.user.id == None else "FALSE" else: rev_data['anon'] = "" @@ -211,12 +369,18 @@ class WikiqParser(): if self.collapse_user: rev_data['collapsed_revs'] = rev.collapsed_revs - if self.persist: - if rev.text.deleted: + if self.persist != PersistMethod.none: + if rev.deleted.text: for k in ["token_revs", "tokens_added", "tokens_removed", "tokens_window"]: old_rev_data[k] = None else: - _, tokens_added, tokens_removed = state.process(rev.text, rev.id, text_sha1) + + if self.persist != PersistMethod.legacy: + _, tokens_added, tokens_removed = state.update(rev.text, rev.id) + + else: + _, tokens_added, tokens_removed = state.process(rev.text, rev.id, text_sha1) + window.append((rev.id, rev_data, tokens_added, tokens_removed)) if len(window) == PERSISTENCE_RADIUS: @@ -236,7 +400,7 @@ class WikiqParser(): rev_count += 1 - if self.persist: + if self.persist != PersistMethod.none: # print out metadata for the last RADIUS revisions for i, item in enumerate(window): # if the window was full, we've already printed item 0 @@ -263,7 +427,7 @@ class WikiqParser(): if self.urlencode: for field in TO_ENCODE: rev_data[field] = quote(str(rev_data[field])) - + if not self.printed_header: print("\t".join([str(k) for k in sorted(rev_data.keys())]), file=self.output_file) self.printed_header = True @@ -277,7 +441,7 @@ def open_input_file(input_filename): elif re.match(r'.*\.gz$', input_filename): cmd = ["zcat", input_filename] elif re.match(r'.*\.bz2$', input_filename): - cmd = ["zcat", input_filename] + cmd = ["bzcat", "-dk", input_filename] try: input_file = Popen(cmd, stdout=PIPE).stdout @@ -310,14 +474,53 @@ parser.add_argument('-s', '--stdout', dest="stdout", action="store_true", parser.add_argument('--collapse-user', dest="collapse_user", action="store_true", 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.") -parser.add_argument('-p', '--persistence', dest="persist", action="store_true", - 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.") +parser.add_argument('-p', '--persistence', dest="persist", default=None, const='', type=str, choices = ['','segment','sequence','legacy'], nargs='?', + 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. The defualt is -p=sequence, which uses the same algorithm as in the past, but with improvements to wikitext parsing. Use -p=legacy for old behavior used in older research projects. Use -p=segment for advanced persistence calculation method that is robust to content moves, but prone to bugs, and slower.") parser.add_argument('-u', '--url-encode', dest="urlencode", action="store_true", 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.") +parser.add_argument('-n', '--namespace-include', dest="namespace_filter", type=int, action='append', + help="Id number of namspace to include. Can be specified more than once.") + +parser.add_argument('-rr', + '--revert-radius', + dest="revert_radius", + type=int, + action='store', + default=15, + help="Number of edits to check when looking for reverts (default: 15)") + +parser.add_argument('-RP', '--revision-pattern', dest="regex_match_revision", default=None, type=str, action='append', + help="The regular expression to search for in revision text. The regex must be surrounded by quotes.") + +parser.add_argument('-RPl', '--revision-pattern-label', dest="regex_revision_label", default=None, type=str, action='append', + help="The label for the outputted column based on matching the regex in revision text.") + +parser.add_argument('-CP', '--comment-pattern', dest="regex_match_comment", default=None, type=str, action='append', + help="The regular expression to search for in comments of revisions.") + +parser.add_argument('-CPl', '--comment-pattern-label', dest="regex_comment_label", default=None, type=str, action='append', + help="The label for the outputted column based on matching the regex in comments.") + args = parser.parse_args() +# set persistence method + +if args.persist is None: + persist = PersistMethod.none +elif args.persist == "segment": + persist = PersistMethod.segment +elif args.persist == "legacy": + persist = PersistMethod.legacy +else: + persist = PersistMethod.sequence + +if args.namespace_filter is not None: + namespaces = args.namespace_filter +else: + namespaces = None + if len(args.dumpfiles) > 0: for filename in args.dumpfiles: input_file = open_input_file(filename) @@ -336,11 +539,17 @@ if len(args.dumpfiles) > 0: filename = os.path.join(output_dir, os.path.basename(filename)) output_file = open_output_file(filename) - wikiq = WikiqParser(input_file, output_file, + wikiq = WikiqParser(input_file, + output_file, collapse_user=args.collapse_user, - persist=args.persist, - urlencode=args.urlencode) - + persist=persist, + urlencode=args.urlencode, + namespaces=namespaces, + revert_radius=args.revert_radius, + regex_match_revision = args.regex_match_revision, + regex_revision_label = args.regex_revision_label, + regex_match_comment = args.regex_match_comment, + regex_comment_label = args.regex_comment_label) wikiq.process() @@ -348,11 +557,20 @@ if len(args.dumpfiles) > 0: input_file.close() output_file.close() else: - wikiq = WikiqParser(sys.stdin, sys.stdout, + wikiq = WikiqParser(sys.stdin, + sys.stdout, collapse_user=args.collapse_user, - persist=args.persist, - urlencode=args.urlencode) - wikiq.process() + persist=persist, + #persist_legacy=args.persist_legacy, + urlencode=args.urlencode, + namespaces=namespaces, + revert_radius=args.revert_radius, + regex_match_revision = args.regex_match_revision, + regex_revision_label = args.regex_revision_label, + regex_match_comment = args.regex_match_comment, + regex_comment_label = args.regex_comment_label) + + wikiq.process() # 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" # stop_words = stop_words.split(",")