tools/dev/mlpatch.py

   1 #!/usr/bin/env python
   2
   3 # mlpatch.py: Run with no arguments for usage
   4
   5 import sys, os
   6 import sgmllib
   7 from htmlentitydefs import entitydefs
   8 import fileinput
   9 from urllib2 import urlopen
  10
  11 CHUNKSIZE = 8 * 1024
  12
  13 class MyParser(sgmllib.SGMLParser):
  14   def __init__(self):
  15     self.baseclass = sgmllib.SGMLParser
  16     self.baseclass.__init__(self)
  17     self.entitydefs = entitydefs
  18     self.entitydefs["nbsp"] = " "
  19     self.inbody = False
  20     self.complete_line = False
  21     self.discard_gathered()
  22
  23   def discard_gathered(self):
  24     self.gather_data = False
  25     self.gathered_data = ""
  26
  27   def noop(self):
  28     pass
  29
  30   def out(self, data):
  31     sys.stdout.write(data)
  32
  33   def handle_starttag(self, tag, method, attrs):
  34     if not self.inbody: return
  35     self.baseclass.handle_starttag(self, tag, method, attrs)
  36
  37   def handle_endtag(self, tag, method):
  38     if not self.inbody: return
  39     self.baseclass.handle_endtag(self, tag, method)
  40
  41   def handle_data(self, data):
  42     if not self.inbody: return
  43     data = data.replace('\n','')
  44     if len(data) == 0: return
  45     if self.gather_data:
  46       self.gathered_data += data
  47     else:
  48       if self.complete_line:
  49         if data[0] in ('+', '-', ' ', '#') \
  50             or data.startswith("Index:") \
  51             or data.startswith("@@ ") \
  52             or data.startswith("======"):
  53           # Real new line
  54           self.out('\n')
  55         else:
  56           # Presume that we are wrapped
  57           self.out(' ')
  58       self.complete_line = False
  59       self.out(data)
  60
  61   def handle_charref(self, ref):
  62     if not self.inbody: return
  63     self.baseclass.handle_charref(self, ref)
  64
  65   def handle_entityref(self, ref):
  66     if not self.inbody: return
  67     self.baseclass.handle_entityref(self, ref)
  68
  69   def handle_comment(self, comment):
  70     if comment == ' body="start" ':
  71       self.inbody = True
  72     elif comment == ' body="end" ':
  73       self.inbody = False
  74
  75   def handle_decl(self, data):
  76     if not self.inbody: return
  77     print "DECL: " + data
  78
  79   def unknown_starttag(self, tag, attrs):
  80     if not self.inbody: return
  81     print "UNKTAG: %s %s" % (tag, attrs)
  82
  83   def unknown_endtag(self, tag):
  84     if not self.inbody: return
  85     print "UNKTAG: /%s" % (tag)
  86
  87   def do_br(self, attrs):
  88     self.complete_line = True
  89
  90   def do_p(self, attrs):
  91     if self.complete_line:
  92       self.out('\n')
  93     self.out(' ')
  94     self.complete_line = True
  95
  96   def start_a(self, attrs):
  97     self.gather_data = True
  98
  99   def end_a(self):
 100     self.out(self.gathered_data.replace('_at_', '@'))
 101     self.discard_gathered()
 102
 103   def close(self):
 104     if self.complete_line:
 105       self.out('\n')
 106     self.baseclass.close(self)
 107
 108
 109 def main():
 110   if len(sys.argv) == 1:
 111     sys.stderr.write(
 112     "usage:   mlpatch.py dev|users year month msgno > foobar.patch\n" +
 113     "example: mlpatch.py dev 2005 01 0001 > issue-XXXX.patch\n" +
 114     """
 115     Very annoyingly, the http://svn.haxx.se/ subversion mailing list archives
 116     mangle inline patches, and provide no raw message download facility
 117     (other than for an entire month's email as an mbox).
 118
 119     So, I wrote this script, to demangle them. It's not perfect, as it has to
 120     guess about whitespace, but it does an acceptable job.\n""")
 121     sys.exit(0)
 122   elif len(sys.argv) != 5:
 123     sys.stderr.write("error: mlpatch.py: Bad parameters - run with no "
 124     + "parameters for usage\n")
 125     sys.exit(1)
 126   else:
 127     list, year, month, msgno = sys.argv[1:]
 128     url = "http://svn.haxx.se/" \
 129         + "%(list)s/archive-%(year)s-%(month)s/%(msgno)s.shtml" % locals()
 130     print "MsgUrl: " + url
 131     msgfile = urlopen(url)
 132     p = MyParser()
 133     buffer = msgfile.read(CHUNKSIZE)
 134     while buffer:
 135       p.feed(buffer)
 136       buffer = msgfile.read(CHUNKSIZE)
 137     p.close()
 138     msgfile.close()
 139
 140 if __name__ == '__main__':
 141   main()