ConfigFile.py

   1 #
   2 # Stuff for dealing with configuration files.
   3 #
   4 #
   5 # This code is part of the LWN git data miner.
   6 #
   7 # Copyright 2007-11 Eklektix, Inc.
   8 # Copyright 2007-11 Jonathan Corbet <corbet@lwn.net>
   9 #
  10 # This file may be distributed under the terms of the GNU General
  11 # Public License, version 2.
  12 #
  13 import sys, re, datetime, os.path
  14 import database
  15
  16 #
  17 # Read a line and strip out junk.
  18 #
  19 def ReadConfigLine (file):
  20     line = file.readline ()
  21     if not line:
  22         return None
  23     line = line.split('#')[0] # Get rid of any comments
  24     line = line.strip () # and extra white space
  25     if len (line) == 0: # we got rid of everything
  26         return ReadConfigLine (file)
  27     return line
  28
  29 #
  30 # Give up and die.
  31 #
  32 def croak (message):
  33     sys.stderr.write (message + '\n')
  34     sys.exit (1)
  35
  36 #
  37 # Read a list of email aliases.
  38 #
  39 def ReadEmailAliases (name):
  40     try:
  41         file = open (name, 'r')
  42     except IOError:
  43         croak ('Unable to open email alias file %s' % (name))
  44     line = ReadConfigLine (file)
  45     while line:
  46         m = re.match ('^("[^"]+"|\S+)\s+(.+)$', line)
  47         if not m or len (m.groups ()) != 2:
  48             croak ('Funky email alias line "%s"' % (line))
  49         if m and m.group (2).find ('@') <= 0:
  50             croak ('Non-addresses in email alias "%s"' % (line))
  51         database.AddEmailAlias (m.group (1).replace ('"', ''), m.group (2))
  52         line = ReadConfigLine (file)
  53     file.close ()
  54
  55 #
  56 # The Email/Employer map
  57 #
  58 EMMpat = re.compile (r'^([^\s]+)\s+([^<]+)\s*(<\s*(\d+-\d+-\d+)\s*)?$')
  59
  60 def ReadEmailEmployers (name):
  61     try:
  62         file = open (name, 'r')
  63     except IOError:
  64         croak ('Unable to open email/employer file %s' % (name))
  65     line = ReadConfigLine (file)
  66     while line:
  67         m = EMMpat.match (line)
  68         if not m:
  69             croak ('Funky email/employer line "%s"' % (line))
  70         email = m.group (1)
  71         company = m.group (2).strip ()
  72         enddate = ParseDate (m.group (4))
  73         database.AddEmailEmployerMapping (email, company, enddate)
  74         line = ReadConfigLine (file)
  75     file.close ()
  76
  77 def ParseDate (cdate):
  78     if not cdate:
  79         return None
  80     sdate = cdate.split ('-')
  81     return datetime.date (int (sdate[0]), int (sdate[1]), int (sdate[2]))
  82
  83
  84 def ReadGroupMap (fname, employer):
  85     try:
  86         file = open (fname, 'r')
  87     except IOError:
  88         croak ('Unable to open group map file %s' % (fname))
  89     line = ReadConfigLine (file)
  90     while line:
  91         database.AddEmailEmployerMapping (line, employer)
  92         line = ReadConfigLine (file)
  93     file.close ()
  94
  95 #
  96 # Read in a virtual employer description.
  97 #
  98 def ReadVirtual (file, name):
  99     ve = database.VirtualEmployer (name)
 100     line = ReadConfigLine (file)
 101     while line:
 102         sl = line.split (None, 1)
 103         first = sl[0]
 104         if first == 'end':
 105             ve.store ()
 106             return
 107         #
 108         # Zap the "%" syntactic sugar if it's there
 109         #
 110         if first[-1] == '%':
 111             first = first[:-1]
 112         try:
 113             percent = int (first)
 114         except ValueError:
 115             croak ('Bad split value "%s" for virtual empl %s' % (first, name))
 116         if not (0 < percent <= 100):
 117             croak ('Bad split value "%s" for virtual empl %s' % (first, name))
 118         ve.addsplit (' '.join (sl[1:]), percent/100.0)
 119         line = ReadConfigLine (file)
 120     #
 121     # We should never get here
 122     #
 123     croak ('Missing "end" line for virtual employer %s' % (name))
 124
 125 #
 126 # Read file type patterns for more fine graned reports
 127 #
 128 def ReadFileType (filename):
 129     try:
 130         file = open (filename, 'r')
 131     except IOError:
 132         croak ('Unable to open file type mapping file %s' % (filename))
 133     patterns = {}
 134     order = []
 135     regex_order = re.compile ('^order\s+(.*)$')
 136     regex_file_type = re.compile ('^filetype\s+(\S+)\s+(.+)$')
 137     line = ReadConfigLine (file)
 138     while line:
 139         o = regex_order.match (line)
 140         if o:
 141             # Consider only the first definition in the config file
 142             elements = o.group(1).replace (' ', '')
 143             order = order or elements.split(',')
 144             line = ReadConfigLine (file)
 145             continue
 146
 147         m = regex_file_type.match (line)
 148         if not m or len (m.groups ()) != 2:
 149             ConfigFile.croak ('Funky file type line "%s"' % (line))
 150         if not patterns.has_key (m.group (1)):
 151             patterns[m.group (1)] = []
 152         if m.group (1) not in order:
 153             print '%s not found, appended to the last order' % m.group (1)
 154             order.append (m.group (1))
 155
 156         patterns[m.group (1)].append (re.compile (m.group (2), re.IGNORECASE))
 157
 158         line = ReadConfigLine (file)
 159     file.close ()
 160     return patterns, order
 161
 162 #
 163 # Read an overall config file.
 164 #
 165
 166 def ConfigFile (name, confdir):
 167     try:
 168         file = open (name, 'r')
 169     except IOError:
 170         try:
 171             file = open (os.path.join (confdir, name), 'r')
 172         except IOError:
 173             croak ('Unable to open config file %s' % (name))
 174     line = ReadConfigLine (file)
 175     while line:
 176         sline = line.split (None, 2)
 177         if len (sline) < 2:
 178             croak ('Funky config line: "%s"' % (line))
 179         if sline[0] == 'EmailAliases':
 180             ReadEmailAliases (os.path.join (confdir, sline[1]))
 181         elif sline[0] == 'EmailMap':
 182             ReadEmailEmployers (os.path.join (confdir, sline[1]))
 183         elif sline[0] == 'GroupMap':
 184             if len (sline) != 3:
 185                 croak ('Funky group map line "%s"' % (line))
 186             ReadGroupMap (os.path.join (confdir, sline[1]), sline[2])
 187         elif sline[0] == 'VirtualEmployer':
 188             ReadVirtual (file, ' '.join (sline[1:]))
 189         elif sline[0] == 'FileTypeMap':
 190             patterns, order = ReadFileType (os.path.join (confdir, sline[1]))
 191             database.FileTypes = database.FileType (patterns, order)
 192         else:
 193             croak ('Unrecognized config line: "%s"' % (line))
 194         line = ReadConfigLine (file)
 195