]> code.communitydata.science - mediawiki_dump_tools.git/blobdiff - wikiq
Merge branch 'master' into regex_scanner
[mediawiki_dump_tools.git] / wikiq
diff --git a/wikiq b/wikiq
index b11f3a480e5c3d7c74b48aaabf1c40b822b79094..b982eaa5f0a8bc32dd7f5dcae77941339ad38304 100755 (executable)
--- a/wikiq
+++ b/wikiq
@@ -13,26 +13,64 @@ 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 matchmaker(rev_data, regular_expression, scanner, rev): #rev_data,self.regex,self.scanner, rev
+    for location in scanner: #presumably 'comment' 'text' 'comment text' made into a list by args
+        if location == "comment":
+            matching_string = rev.comment
+        elif location == "text":
+            matching_string = rev.text
+        else:
+            sys.exit("regex scanner location must be 'comment' or 'text'.")
+
+        if (re.search(regular_expression, matching_string) is not None): # we know that there is a match somewhere
+            m = re.finditer(regular_expression, matching_string) # all our matchObjects in a list
+            blob=""
+            for result in m:
+                blob = blob + "," + result.group(0)
+            # columns we want
+            rev_data['matches'] = blob #### the list of matchObjects. gleaned in post-processing       
+        else:
+            rev_data['matches'] = None
+       
+    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
@@ -44,14 +82,20 @@ class WikiqPage():
     __slots__ = ('id', 'title', 'namespace', 'redirect',
                  '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()
@@ -75,7 +119,14 @@ 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
@@ -88,6 +139,7 @@ class WikiqPage():
                     yield prev_rev
 
             prev_rev = rev
+
         # also yield the final time
         yield prev_rev
 
@@ -98,13 +150,27 @@ class WikiqPage():
         return next(self.__revisions)
 
 class WikiqParser():
-    def __init__(self, input_file, output_file, collapse_user=False, persist=False):
+    
+    def __init__(self, input_file, output_file, scanner, match_regex, collapse_user=False, persist=None, urlencode=False, namespaces = None):
+        """ 
+        Parameters:
+           persist : what persistence method to use. Takes a PersistMethod value
+        """
+
         self.input_file = input_file
         self.output_file = output_file
         self.collapse_user = collapse_user
         self.persist = persist
         self.printed_header = False
         self.namespaces = []
+        self.urlencode = urlencode
+        self.scanner = scanner
+        self.match_regex = match_regex
+
+        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
@@ -121,8 +187,8 @@ class WikiqParser():
         # if we've made it this far with no matches, we return the default namespace
         return default_ns
 
+
     def process(self):
-        print("Processing file: %s" % self.input_file.name, file=sys.stderr)
 
         # create a regex that creates the output filename
         # output_filename = re.sub(r'^.*/(enwiki\-\d+)\-.*p(\d+)p.*$',
@@ -133,50 +199,93 @@ 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
+
+            rev_detector = mwreverts.Detector()
+
+            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:
-
-                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" } 
+                ## m = re.finditer() #so we can find all instances
+                ## m.groupdict() #so we can look at them all with their names
+
+                # initialize rev_dat
+                rev_data = {}
+
+                if self.scanner is not None: # we know we want to do a regex search 
+                    ## comment = want to look in comment attached to revision
+                    ## text = want to look in revision text
+
+                    ### call the scanner function
+                    rev_data = matchmaker(rev_data, self.match_regex, self.scanner, rev)
+       
+                if self.scanner is not None and rev_data['matches'] is None:
+                    next
+
+                # 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]) + '"'
@@ -187,10 +296,10 @@ 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'] = ""
@@ -207,12 +316,18 @@ 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:
@@ -232,7 +347,7 @@ 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
@@ -246,7 +361,7 @@ class WikiqParser():
                     rev_data["tokens_added"] = num_tokens
                     rev_data["tokens_removed"] = len(tokens_removed)
                     rev_data["tokens_window"] = len(window)-(i+1)
-                                           
+                    
                     self.print_rev_data(rev_data)
 
             page_count += 1
@@ -256,6 +371,10 @@ class WikiqParser():
 
     def print_rev_data(self, rev_data):
         # if it's the first time through, print the header
+        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
@@ -264,12 +383,12 @@ class WikiqParser():
 
 
 def open_input_file(input_filename):
-    if re.match(r'.*\.7z', input_filename):
-        cmd = ["7za", "x", "-so", input_filename] 
-    elif re.match(r'.*\.gz', input_filename):
-        cmd = ["zcat", input_filename] 
-    elif re.match(r'.*\.bz2', input_filename):
+    if re.match(r'.*\.7z$', input_filename):
+        cmd = ["7za", "x", "-so", input_filename, '*'] 
+    elif re.match(r'.*\.gz$', input_filename):
         cmd = ["zcat", input_filename] 
+    elif re.match(r'.*\.bz2$', input_filename):
+        cmd = ["bzcat", "-dk", input_filename] 
 
     try:
         input_file = Popen(cmd, stdout=PIPE).stdout
@@ -280,7 +399,8 @@ def open_input_file(input_filename):
 
 def open_output_file(input_filename):
     # create a regex that creates the output filename
-    output_filename = re.sub(r'\.xml(\.(7z|gz|bz2))?$', '', input_filename)
+    output_filename = re.sub(r'\.(7z|gz|bz2)?$', '', input_filename)
+    output_filename = re.sub(r'\.xml', '', output_filename)
     output_filename = output_filename + ".tsv"
     output_file = open(output_filename, "w")
 
@@ -301,30 +421,65 @@ 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('-rs', '--regex-scanner', dest="scanner",type=str, action='append',
+                    help="Find the regex match specified by -R/--match searching in: (1) comment (2) text.")
+
+parser.add_argument('-R', '--match', dest="match_regex", type=str, 
+                    help="The regular expression you would like to find in the string and put in capture group")
 
 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)
 
-        # open file for output
+        # open directory for output
+        if args.output_dir:
+            output_dir = args.output_dir[0]
+        else:
+            output_dir = "."
+
+        print("Processing file: %s" % filename, file=sys.stderr)
+
         if args.stdout:
             output_file = sys.stdout
         else:
-            if args.output_dir:
-                output_dir = args.output_dir[0]
-            else:
-                output_dir = "."
-
             filename = os.path.join(output_dir, os.path.basename(filename))
             output_file = open_output_file(filename)
 
         wikiq = WikiqParser(input_file, output_file, 
-                           collapse_user=args.collapse_user,
-                           persist=args.persist)
+                            collapse_user=args.collapse_user,
+                            persist=persist,
+                            urlencode=args.urlencode,
+                            namespaces = namespaces,
+                            match_regex=args.match_regex, # adding in the new 2 args for regex searching
+                            scanner=args.scanner)
+
         wikiq.process()
 
         # close things 
@@ -332,8 +487,13 @@ if len(args.dumpfiles) > 0:
         output_file.close()
 else:
     wikiq = WikiqParser(sys.stdin, sys.stdout,
-                       collapse_user=args.collapse_user,
-                       persist=args.persist)
+                        collapse_user=args.collapse_user,
+                        persist=persist,
+                        #persist_legacy=args.persist_legacy,
+                        urlencode=args.urlencode,
+                        namespaces = namespaces,
+                        match_regex=args.match_regex, # adding in the new 2 args for regex searching
+                        scanner=args.scanner)
     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"

Community Data Science Collective || Want to submit a patch?