wikifirc

commit 991c20fe310edcac62574ad84ce05ebf0bdc5383
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sat,  7 Apr 2012 17:08:10 +0200

initial commit

Diffstat:
README
 | 
15
+++++++++++++++
wikifirc
 | 
93
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 108 insertions(+), 0 deletions(-)
diff --git a/README b/README
@@ -0,0 +1,15 @@
+wikifirc -- filter irc.wikimedia.org on specific pages and users
+Copyright (C) 2012 by Antoine Amarilli
+
+Run as wikifirc DUMP ADMIN. DUMP should be a file to dump the currently followed
+pages, it will be read at startup if it exists and will be written on exit.
+ADMIN should be the name of an administrator who will send commands to the
+program. Events from irc.wikimedia.org should be provided in the irctk format on
+standard input, the data of matching lines will be sent on standard output.
+
+The list of pages and users to follow is initially empty. ADMIN can send the
+name of new users to follow as the body of a IRC message to the program. The
+user and user talk pages of an user are followed. Whenever a followed user
+modifies a page, the page is followed. The program will display modifications
+that are performed on a followed page or performed by a followed user.
+
diff --git a/wikifirc b/wikifirc
@@ -0,0 +1,93 @@
+#!/usr/bin/python3
+
+import sys
+import time
+
+class Line:
+  def __init__(self, project, data):
+    self.project = project
+    self.data = data
+    self.time = time.time()
+    page, sep, rest = data.partition(']]')
+    self.page = page[8:-3]
+    fields = rest.split(' ')
+    fields.pop(0) # trailing characters of title
+    self.flags = fields.pop(0)[:-2]
+    self.diff = fields.pop(0)[3:-1]
+    self.flags2 = fields.pop(0)[2:-1]
+    username, sep, rest = ' '.join(fields).partition('*')
+    self.username = username[3:-4]
+    fields = rest.split(' ')
+    fields.pop(0) # rest of flags3
+    self.diff = fields.pop(0)[1:-1]
+    self.message = ' '.join(fields)[3:-1]
+
+  def __str__(self):
+    return self.data
+
+# other languages should be added here
+user_namespaces = [
+    'User:', 'User talk:',
+    'Utilisateur:', 'Discussion utilisateur:',
+  ]
+
+if __name__ == "__main__":
+
+  pages = set()
+  exception = None
+
+  try:
+    save = sys.argv[1]
+    admin = sys.argv[2]
+  except IndexError:
+    print ("Usage: %s DUMP ADMIN" % sys.argv[0])
+    sys.exit(1)
+
+  try:
+    f = open(save, 'r')
+    while True:
+      l = f.readline()
+      if not l:
+        break
+      pages.add(l.rstrip())
+    f.close()
+  except IOError:
+    pass
+
+  try:
+    while True:
+      data = sys.stdin.readline()
+      if not data:
+        break
+      fields = data.strip().split()
+      project = fields.pop(0)[1:-1]
+      if not project.startswith('#'):
+        user = fields.pop(0)[1:-1]
+        if user != admin:
+          continue
+        new_user = fields.pop(0)
+        for namespace in user_namespaces:
+          pages.add(namespace + new_user)
+        continue
+      fields.pop(0) # bot username
+      data = ' '.join(fields)
+      line = Line(project, data)
+      # a user is followed if its user page is followed
+      if user_namespaces[0] + line.username in pages:
+        pages.add(line.page)
+      if line.page in pages:
+        print(line)
+  except Exception as e:
+    exception = e
+    raise e
+
+  f = open(save, 'w')
+  for x in pages:
+    print(x, file=f)
+  f.close()
+
+  if exception:
+    raise exception
+
+  sys.exit(0)
+