namegen.py: first working version of the script functions.

1 files changed, 118 insertions, 0 deletions

diff --git a/namegen.py b/namegen.py
new file mode 100755
index 0000000..194208f
--- /dev/null
+++ b/namegen.py
@@ -0,0 +1,118 @@
+#!/usr/bin/env python2
+
+"""
+namegen.py
+==========
+
+---------------------
+random name generator
+---------------------
+
+:Authors: Elena ``of Valhalla'' Grandi, Diego Roversi
+:Date: 2011-09-06
+:Copyright: 2011 Elena Grandi, Diego Roversi
+:Version: 0.20110906
+:Manual Section: 1
+
+SYNOPSIS
+--------
+
+   namegen.py (-s <stat_file>)|(-l <list_file>) [-S <dest_file>] [-n <n>]
+
+DESCRIPTION
+-----------
+
+OPTIONS
+-------
+
+-s <stat_file>, --stats=<stat_file>
+   Load rules from <stat_file>.
+-l <list_file>, --list=<list_file>
+   Generate rules from the files in <list_file>.
+-S <dest_file>, --save_stats=<dest_file>
+   Save stats in <dest_file>.
+-n <n>, --number=<n>
+   Generate <n> names (0 to generate no name).
+-c <n>, --context=<n>
+   Use <n> characters of context in the stats.
+-h, --help
+   Show an help message.
+
+LICENSE
+-------
+
+"""
+
+import sys
+import random
+
+try:
+    import yaml
+    skip_yaml = False
+except ImportError:
+    sys.stderr.write("Could not find module YAML: will not be able to save / load statistics\n")
+    skip_yaml = True
+
+class Stats:
+
+    def __init__(self,context):
+        """"""
+        self.stats = {}
+        self.context = context
+
+    def _increment_stats(self,cont,key):
+        if cont in self.stats:
+            if key in self.stats[cont]:
+                self.stats[cont][key] += 1
+            else:
+                self.stats[cont][key] = 1
+        else:
+            self.stats[cont] = {key: 1}
+
+    def read_names(self,fp):
+        """Calculate stats from a list of names in a file-like object."""
+
+        for name in fp.readlines():
+            for i in xrange(min(self.context,len(name))):
+                self._increment_stats(name[0:i],name[i])
+            for i in xrange(self.context,len(name)):
+                self._increment_stats(name[i-self.context:i],name[i])
+
+    def load_stats(self,fp):
+        """Load stats from a yaml in a string or file-like object"""
+        if skip_yaml:
+            return False
+        self.stats = yaml.safe_load(fp)
+
+    def save_stats(self,fp):
+        """Save stats in yaml to a file-like object"""
+        if skip_yaml:
+            return False
+        yaml.safe_dump(self.stats,fp)
+        return True
+
+
+    def generate_name(self):
+        """Generate a name following the stats"""
+        name = ''
+        c = ''
+        while c != '\n':
+            if len(name) <= self.context:
+                cont = name
+            else:
+                cont = name[-self.context:]
+            
+            chars = ''.join([k*self.stats[cont][k] for k in self.stats[cont]])
+            c = random.choice(chars)
+            name += c
+
+        return name
+
+def main():
+    mystats = Stats(3)
+    fp = open('data/nomif2.txt')
+    mystats.read_names(fp)
+    fp.close()
+    print mystats.generate_name(),
+
+if __name__ == '__main__': main()