How to bind Lua with Mecab?

Question

I want to use Mecab in Lua,but I really can't figure out the procedure of bindings,i am work no windows 7,is bingdings meaning that to create a shared library? if so? how to?i see some binding files about Java,the files in package org.chasen.mecab shows that it create by swig,it make me confused.so where does it derive from ? or just write by ourself?after bindings,what else should i do if i want to use in Lua,by the way I use mingw.can someone give some simple steps to me that I can keep trying to work on it.

to greatwolf: i use followings command to do it

swig -lua  -c++ MeCab.i
g++ -c MeCab_wrap.cxx -I C:\Lua\5.1\include -I ..\src
g++ -LC:\Lua\5.1\lib -shared MeCab_wrap.o -llua51  -o MeCab.dll

and i got errors below.

 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2960): undefined reference to `_imp___ZN5MeCab12createTaggerEPKc'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2981): undefined reference to `_imp___ZN5MeCab12getLastErrorEv'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x29c9): undefined reference to `_imp___ZN5MeCab12createTaggerEPKc'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x29d9): undefined reference to `_imp___ZN5MeCab12getLastErrorEv'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2a8d): undefined reference to `_imp___ZN5MeCab11createModelEPKc'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2aae): undefined reference to `_imp___ZN5MeCab12getLastErrorEv'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2af6): undefined reference to `_imp___ZN5MeCab11createModelEPKc'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2b06): undefined reference to `_imp___ZN5MeCab12getLastErrorEv'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x2b6d): undefined reference to `_imp___ZN5MeCab13createLatticeEv'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x11050): undefined reference to `MeCab::Model::version()'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x11457): undefined reference to `MeCab::Model::create(int, char**)'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x11732): undefined reference to `MeCab::Model::create(char const*)'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x122bf): undefined reference to `MeCab::Tagger::parse(MeCab::Model const&, MeCab::Lattice*)'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x17737): undefined reference to `MeCab::Tagger::create(int, char**)'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x17a12): undefined reference to `MeCab::Tagger::create(char const*)'
 MeCab_wrap.o:MeCab_wrap.cxx:(.text+0x17d83): undefined reference to `MeCab::Tagger::version()'
 collect2: ld returned 1 exit status

it seems that these function declared in mecab.h，but i don't know how to handle it.

Answer 1

I don't know of any Lua bindings for mecab. Try googling for them first.

To create the bindings yourself you must be proficient both with Lua C API and C (or C++). A deep understanding of Lua itself (the language, I mean) is advisable.

Search the Lua WIKI for some more pointers. In particular BindingCodeToLua page.

Answer 2

It looks like mecab provides a swig file for automatically generated bindings. As luck would have it, lua is one of swig's supported targets.

A reasonable starting point would be to check out the corresponding makefile to see how bindings get created for other languages. From mecab's swig makefile:

SWIG = swig
PREFIX = MeCab

all: perl ruby python java csharp

# ...

perl:   
    $(SWIG) -perl -shadow -c++ $(PREFIX).i
    mv -f $(PREFIX)_wrap.cxx ../perl
    mv -f $(PREFIX).pm ../perl

ruby:   
    $(SWIG) -ruby -c++ $(PREFIX).i
    mv -f $(PREFIX)_wrap.cxx ../ruby/$(PREFIX)_wrap.cpp

python: 
    $(SWIG) -python -shadow -c++ $(PREFIX).i
    mv -f $(PREFIX)_wrap.cxx ../python
    mv -f $(PREFIX).py ../python

# ...

Extrapolating from the above, you can try generating swig bindings something like the following:

swig -lua -shadow -c++ MeCab.i

This will substantially reduce the effort you would otherwise need from manually creating the bindings yourself.

Answer 3

I recently needed this and since I didn't find anything I wrote a module so you can use Mecab with Lua.

It works like this:

mecab = require "mecab" 
parser = mecab:new("") -- you can pass mecab config options here, like "-Owakati"

print(parser:parse("吾輩は猫である"))

You can also install it via LuaRocks as mecab.

It just provides access to the parse method of the Tagger class, but when working with Mecab that's all I've ever needed. If you'd like support for other Mecab features please feel free to file an issue on Github.