Doc/tools/findacks

   1 #!/usr/bin/env python
   2 """Script to locate email addresses in the CVS logs."""
   3 __version__ = '$Revision$'
   4
   5 import os
   6 import re
   7 import sys
   8 import UserDict
   9
  10 import cvsinfo
  11
  12
  13 class Acknowledgements(UserDict.UserDict):
  14     def add(self, email, name, path):
  15         d = self.data
  16         d.setdefault(email, {})[path] = name
  17
  18
  19 def open_cvs_log(info, paths=None):
  20     cvsroot = info.get_cvsroot()
  21     cmd = "cvs -q -d%s log " % cvsroot
  22     if paths:
  23         cmd += " ".join(paths)
  24     return os.popen(cmd, "r")
  25
  26
  27 email_rx = re.compile("<([a-z][-a-z0-9._]*@[-a-z0-9.]+)>", re.IGNORECASE)
  28
  29 def find_acks(f, acks):
  30     prev = ''
  31     filename = None
  32     MAGIC_WORDS = ('van', 'von')
  33     while 1:
  34         line = f.readline()
  35         if not line:
  36             break
  37         if line.startswith("Working file: "):
  38             filename = line.split(None, 2)[2].strip()
  39             prev = line
  40             continue
  41         m = email_rx.search(line)
  42         if m:
  43             words = prev.split() + line[:m.start()].split()
  44             L = []
  45             while words \
  46                   and (words[-1][0].isupper() or words[-1] in MAGIC_WORDS):
  47                 L.insert(0, words.pop())
  48             name = " ".join(L)
  49             email = m.group(1).lower()
  50             acks.add(email, name, filename)
  51         prev = line
  52
  53
  54 def load_cvs_log_acks(acks, args):
  55     repolist = cvsinfo.get_repository_list(args or [""])
  56     for info, paths in repolist:
  57         print >>sys.stderr, "Repository:", info.get_cvsroot()
  58         f = open_cvs_log(info, paths)
  59         find_acks(f, acks)
  60         f.close()
  61
  62
  63 def load_tex_source_acks(acks, args):
  64     for path in args:
  65         path = path or os.curdir
  66         if os.path.isfile(path):
  67             read_acks_from_tex_file(acks, path)
  68         else:
  69             read_acks_from_tex_dir(acks, path)
  70
  71
  72 def read_acks_from_tex_file(acks, path):
  73     f = open(path)
  74     while 1:
  75         line = f.readline()
  76         if not line:
  77             break
  78         if line.startswith(r"\sectionauthor{"):
  79             line = line[len(r"\sectionauthor"):]
  80             name, line = extract_tex_group(line)
  81             email, line = extract_tex_group(line)
  82             acks.add(email, name, path)
  83
  84
  85 def read_acks_from_tex_dir(acks, path):
  86     stack = [path]
  87     while stack:
  88         p = stack.pop()
  89         for n in os.listdir(p):
  90             n = os.path.join(p, n)
  91             if os.path.isdir(n):
  92                 stack.insert(0, n)
  93             elif os.path.normpath(n).endswith(".tex"):
  94                 read_acks_from_tex_file(acks, n)
  95
  96
  97 def extract_tex_group(s):
  98     c = 0
  99     for i in range(len(s)):
 100         if s[i] == '{':
 101             c += 1
 102         elif s[i] == '}':
 103             c -= 1
 104             if c == 0:
 105                 return s[1:i], s[i+1:]
 106
 107
 108 def print_acks(acks):
 109     first = 1
 110     for email, D in acks.items():
 111         if first:
 112             first = 0
 113         else:
 114             print
 115         L = D.items()
 116         L.sort()
 117         prefname = L[0][1]
 118         for file, name in L[1:]:
 119             if name != prefname:
 120                 prefname = ""
 121                 break
 122         if prefname:
 123             print prefname, "<%s>:" % email
 124         else:
 125             print email + ":"
 126         for file, name in L:
 127             if name == prefname:
 128                 print "    " + file
 129             else:
 130                 print "    %s (as %s)" % (file, name)
 131
 132
 133 def print_ack_names(acks):
 134     names = []
 135     for email, D in acks.items():
 136         L = D.items()
 137         L.sort()
 138         prefname = L[0][1]
 139         for file, name in L[1:]:
 140             prefname = prefname or name
 141         names.append(prefname or email)
 142     def f(s1, s2):
 143         s1 = s1.lower()
 144         s2 = s2.lower()
 145         return cmp((s1.split()[-1], s1),
 146                    (s2.split()[-1], s2))
 147     names.sort(f)
 148     for name in names:
 149         print name
 150
 151
 152 def main():
 153     args = sys.argv[1:]
 154     acks = Acknowledgements()
 155     load_cvs_log_acks(acks, args)
 156     load_tex_source_acks(acks, args)
 157     print_ack_names(acks)
 158
 159
 160 if __name__ == "__main__":
 161     main()