horrorscope

   1 #!/usr/bin/python
   2
   3 # TODO:
   4 #  - Filter packages by tags (for example, keep only role::program packages)
   5 #  - Go from packages to tags
   6 #  - Build a tag cloud or a word cloud
   7 #  - Extract significant phrases from significant package descriptions
   8
   9 # wget http://popcon.debian.org/by_inst.gz
  10
  11 import gzip
  12 import math
  13 import os.path
  14 import apt
  15 import sys
  16 from debian import debtags
  17 from optparse import OptionParser
  18 from collections import Counter
  19
  20 if not os.path.exists('/var/lib/debtags/package-tags'):
  21     sys.stderr.write("Sorry, /var/lib/debtags/package-tags doesn't exist and is required")
  22     sys.exit(2)
  23
  24 class Parser(OptionParser):
  25     def __init__(self, *args, **kwargs):
  26         OptionParser.__init__(self, *args, **kwargs)
  27
  28     def error(self, msg):
  29         sys.stderr.write("%s: error: %s\n\n" % (self.get_prog_name(), msg))
  30         self.print_help(sys.stderr)
  31         sys.exit(2)
  32
  33 parser = Parser(usage="usage: %prog [options] filename",
  34         version="%prog 0.1",
  35         description="Count of packages grouped by implementation language")
  36
  37 (opts, args) = parser.parse_args()
  38
  39 horror_urls = {
  40     "implemented-in::c-sharp": "http://www.fsf.org/news/dont-depend-on-mono",
  41     "implemented-in::ecmascript": "http://wiki.theory.org/YourLanguageSucks#JavaScript_sucks_because:",
  42     "implemented-in::java": "http://wiki.theory.org/YourLanguageSucks#Java_sucks_because:",
  43     "implemented-in::objc": "http://wiki.theory.org/YourLanguageSucks#Objective-C_sucks_because:",
  44     "implemented-in::php": "http://me.veekun.com/blog/2012/04/09/php-a-fractal-of-bad-design/",
  45     "implemented-in::ruby": "http://wiki.theory.org/YourLanguageSucks#Ruby_sucks_because:",
  46     "implemented-in::tcl": "http://wiki.tcl.tk/640",
  47 }
  48
  49
  50 def parse_vote(file):
  51     """
  52     Parse a popcon vote file, generating the names of the valid packages in
  53     the vote
  54     """
  55     for line in file:
  56         if line.startswith("POPULARITY"):
  57             continue
  58         elif line.startswith("END-POPULARITY"):
  59             continue
  60         else:
  61             data = line[:-1].split(" ")
  62             if len(data) < 4:
  63                 continue
  64             if data[3] == '<NOFILES>':
  65                 # Empty/virtual packages
  66                 yield data[2], 0.1
  67             elif len(data) == 4:
  68                 # Used packages
  69                 yield data[2], 1.
  70             elif data[4] == '<OLD>':
  71                 # Unused packages
  72                 yield data[2], 0.3
  73             elif data[4] == '<RECENT-CTIME>':
  74                 # Recently installed packages
  75                 yield data[2], 0.8
  76
  77
  78 apt_cache = apt.Cache()
  79 by_lang = Counter()
  80
  81 with open("/var/log/popularity-contest") as infd:
  82     db = debtags.DB()
  83     db.read(open('/var/lib/debtags/package-tags'))
  84     for t in db.iter_tags():
  85         if t.startswith("implemented-in::"):
  86             by_lang[t] = 0
  87
  88     for pkg, tf in parse_vote(infd):
  89         if not pkg in apt_cache: continue
  90         aptpkg = apt_cache[pkg]
  91         if not aptpkg.installed: continue
  92         tags = db.tags_of_package(pkg)
  93
  94         if args:
  95             for t in tags:
  96                 if t == args[0]:
  97                     print pkg
  98                     break
  99         else:
 100             for t in tags:
 101                 if t.startswith("implemented-in::"):
 102                     by_lang[t] += 1
 103
 104 if not args:
 105     for k, v in sorted(by_lang.iteritems()):
 106         if v == 0:
 107             comment = "WIN!"
 108         else:
 109             comment = horror_urls.get(k, "")
 110         print "%4d %-30s %s" % (v, k, comment)
 111