001: #!/usr/bin/env python
002: 
003: import argparse
004: import marshal
005: import random
006: import regex
007: 
008: class Lipsum:
009:     clean = regex.compile ('\\s+$')
010: 
011:     """
012:     Create a new lipsum instance by either training a model from given input
013:     text file, or loading a previously saved model. Behavior actually depends
014:     on whether the 3rd parameter (length) is provided since it's only used when
015:     training a new model (but yeah, that's still an ugly hack).
016:     """
017:     def __init__ (self, path, length = None):
018:         # If length is none we assume input is a trained model
019:         if length is None:
020:             with open (args.model, 'rb') as file:
021:                 (self.model, self.length) = marshal.load (file)
022: 
023:         # Otherwise start training
024:         else:
025:             self.length = length
026:             self.model = {}
027: 
028:             # Tokenize input string into (prefix -> (suffix, count)) tree
029:             tokenize = regex.compile ('((?:[-\\w]+|[^\\n.])\\s*)*(?:[\\n.]\\s*)?')
030:             tree = {}
031: 
032:             with open (path, 'r') as file:
033:                 for match in tokenize.finditer (file.read ()):
034:                     lexems = map (self.clean_lexem, match.captures (1))
035: 
036:                     # Ignore empty sequences
037:                     if len (lexems) < 1:
038:                         continue
039: 
040:                     # Register suffixes, including special "end of line" marker
041:                     prefix = (None, ) * length
042: 
043:                     for lexem in lexems + [None]:
044:                         suffixes = tree.setdefault (prefix, {})
045:                         suffixes[lexem] = suffixes.get (lexem, 0) + 1
046: 
047:                         prefix = prefix[1:] + (lexem, )
048: 
049:             # Convert to (prefix -> (suffix, probability)) model
050:             for (key, suffixes) in tree.iteritems ():
051:                 occurrences = float (sum ((count for (suffix, count) in suffixes.iteritems ())))
052:                 thresholds = []
053:                 total = 0
054: 
055:                 for (lexem, count) in suffixes.iteritems ():
056:                     total += count / occurrences
057: 
058:                     thresholds.append ((lexem, total))
059: 
060:                 self.model[key] = thresholds
061: 
062:     """
063:     Cleanup input lexem by squashing all "dirty" characters (see the "clean"
064:     regular expression above) into a single space character.
065:     """
066:     def clean_lexem (self, lexem):
067:         return self.clean.sub (' ', lexem).lower ()
068: 
069:     """
070:     Find first suffix above given value (used for random suffix selection).
071:     This method could/should be replaced by some functional call like:
072:     bisect (suffixes, lambda suffix: suffix[1], value)
073:     """
074:     def first_above (self, suffixes, value):
075:         i = 0
076: 
077:         while i < len (suffixes) and suffixes[i][1] <= value:
078:             i += 1
079: 
080:         return i < len (suffixes) and suffixes[i][0] or None
081: 
082:     """
083:     Generate a random lexems sequence using currently loaded model.
084:     """
085:     def generate (self):
086:         buffer = ''
087:         prefix = tuple ([None] * self.length)
088: 
089:         while prefix in self.model:
090:             lexem = self.first_above (self.model[prefix], random.random ())
091: 
092:             if lexem is None:
093:                 break
094: 
095:             buffer += lexem
096:             prefix = prefix[1:] + (lexem, )
097: 
098:         return buffer
099: 
100:     """
101:     Save current model to file.
102:     """
103:     def save (self, path):
104:         with open (path, 'wb') as file:
105:             marshal.dump ((self.model, self.length), file)
106: 
107: parser = argparse.ArgumentParser (description = 'Lipsum blabla')
108: parser.add_argument ('-g', '--generate', type = int, default = 1, help = 'Generate N lines', metavar = 'N')
109: parser.add_argument ('-l', '--length', type = int, default = 3, help = 'Set prefix length', metavar = 'LEN')
110: parser.add_argument ('-m', '--model', action = 'store', help = 'Specify path to model', metavar = 'FILE')
111: parser.add_argument ('-t', '--train', action = 'store', help = 'Train from given file (and save if -m is specified)', metavar = 'FILE')
112: 
113: args = parser.parse_args ()
114: 
115: if args.train is not None:
116:     lipsum = Lipsum (args.train, args.length)
117: 
118:     if args.model is not None:
119:         lipsum.save (args.model)
120: 
121: elif args.model is not None:
122:     lipsum = Lipsum (args.model)
123: 
124: else:
125:     raise Exception ('please specify either model or train argument')
126: 
127: for i in range (args.generate):
128:     print lipsum.generate ()