Nagen is a custom language random name generator. Or word generator, but nagen is more sexy than wogen. Its basic aim is to generate random names using a "language file", which contain a set of generation rules. The language file system permits the generation of similar sounding and pronounceable words. This program can be usefull to fiction writers or rolists.

The second aim is to scan a word list using a symbol file and generate a set of rules which would produce equivalent sounding words.

Simply make install as root or using sudo.

This code has no dependency apart from the libc.

To uninstall you can use make uninstall

nagen [-n numberofname] [-l lengthofname] [-o outputfile] languagefile

• -n numberofname: the number of names (or words) you want to generate. It defaults to 1.

• -l lengthofnames: the lenght of the names (or ... you get the idea) you want to generate in symbols (see next point). It defaults to the Glenschplockfür constant.

• -o outputfile: the file where you to put the generated names (or ... stuff). It defaults to stdout, so your shell basically.

• languagefile: the file containing the language rules (see next point).

You can try it out with the example.lg language file.

The language file is composed of several rules. Each rule use a group of symbols. A symbol is a string of characters, like a voyel or a syllable.

A rule define two things: a group of symbols and a list of connections to other rules, each with a probability weight.

The generation algorithm simply starts by the first rule, picks a random symbol in the group then randomly picks a connection, jump to that rule and start this process again. The algorithm stops after picking a number of symbols equals to lenghtofname.

If the group of symbols is empty then the algorithm simply connects to another rule without picking any symbols. This is useful as a first rule.

If the list of connection is empty then the algorithm stops even if the lenghtofname condition is not met.

The relative probability means that if 'a' has a weight of 1 and 'b' a weight of 3 then 'b' has three time more chance being chosen. So the probability of a symbol (or a connection) being chosen equal its weight over the sum of the weight of all the symbols (or connections) in that rule.

In short, it's just a way of writing bigram quickly.

As a general note, each line can be preceded with any number of newline, tabulation or space character you want.

A rule start with the line [Name of the rule] and end with the start of a new rule or the end of the file. The name of the rule can be anything, it's just an identifier.

Follows the connection list. A connection is defined by a line like this: -Name of the rule 1 Where the name of the rule is the one you're connecting to and the number at the end of the line is the probability weight.

At last comes the symbol group. Each symbol is on a line like this: *symbol 1 Where 'symbol' is the character string and the number at the end of the line is the probability weight.

An example of a dummy language resulting only in strings of 'a' and 'b' atlernating would be:

[start]
-ra 1
-rb 1

[ra]
-rb 1
*a 1

[rb]
-ra 1
*b 1

For a more complex example, see the file example.lg.

My original idea was that you could simplify a real language bigram probability with groups, but this idea is really wrong. I implemented an algorithm that would identify, with a certain tolerance, symbols that could be merged into a group. But whatever was the tolerance equation, the merging happened only at a high value of tolerance and the difference in tolerance beetween one group per symbol and one group for all the symbol was really thin. So I just dropped this algorithm.

The current algorithm just writes one group per symbol resulting in a very inefficient way of writing bigrams. It simply try to transform each word from the word list in a succession of symbols from the symbol file and then count the bigrams. The transformation algorithm is really simple but error prone: it simply moves through the word trying to fit the biggest symbol it can and, if it fails, it just drops the whole word with an error message. No fancy backtracking to find the best solution.

The command to use it is : nagen -s symbolfile [-o outputfile] wordlistfile

• -s symbolfile: a file containing a symbol list, each symbol on one line, which will be analyzed.

• -o outputfile: the file where you to put the generated rules. It defaults to stdout, so your shell basically.

• wordlistfile: the file containing the word list, each word on one line, to analyze.

This file is part of nagen.

Nagen is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

Nagen is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.

Copyright 2012 Charles Hubain haxelion@gmail.com