knp distsim call test

knp_distsim.c

/*
# VirtualBoxのディスクがIO負荷で（？）勝手にリードオンリーになった
VBoxManage setextradata "docker-host" "VBoxInternal/Devices/ahci/0/LUN0/Config/FlushInterval" 1000000

# 実行ファイル
gcc -DHAVE_CONFIG_H -I. -I.. -I/knp-4.16/CRF++-0.58 -I/knp-4.16/crf -DUSE_CRF -DKNP_DICT="\"/usr/local/share/knp/dict"\" -DKNP_RULE="\"/usr/local/share/knp/rule"\" -DKNP_RC_DEFAULT="\"/usr/local/etc/knprc"\" -DCF_VERSION="\"CF1.1"\" -DCDB -I/knp-4.16/cdb -I/usr/local//include -I/knp-4.16/distsim -g3 -O0 -MT dsim.o -MD -MP -MF .deps/dsim.Tpo -c -o dsim.o dsim.c
/bin/sh ../libtool --tag=CC   --mode=link g++  -g3 -O0   -o dsim dsim.o tools.o db.o -lm /knp-4.16/distsim/libdistsim.la -lstdc++ -L/knp-4.16/cdb -lcdb  -lz

# 共有ライブラリ
## distsim
/bin/sh ../libtool --tag=CXX   --mode=compile g++ -DHAVE_CONFIG_H -I. -I.. -I../cdb -g -O2 -MT distsim_for_knp.lo -MD -MP -MF .deps/distsim_for_knp.Tpo -c -fPIC -o distsim_for_knp.lo distsim_for_knp.cc
/bin/sh ../libtool --tag=CXX   --mode=compile g++ -DHAVE_CONFIG_H -I. -I.. -I../cdb -g -O2 -MT distsim.lo -MD -MP -MF .deps/distsim.Tpo -c -fPIC -o distsim.lo distsim.cc

## cdb
make shared
cp libcdb.so /usr/local/lib
ldconfig

## system
;コンパイル
gcc -DHAVE_CONFIG_H -I. -I.. -DCDB -I/knp-4.16/cdb -I/usr/local//include -I/knp-4.16/distsim -g3 -O0 -MT tools.o -MD -MP -MF .deps/tools.Tpo -c -fPIC -o tools.o tools.c
gcc -DHAVE_CONFIG_H -I. -I.. -DCDB -I/knp-4.16/cdb -I/usr/local//include -I/knp-4.16/distsim -g3 -O0 -MT db.o    -MD -MP -MF .deps/db.Tpo    -c -fPIC -o db.o db.c
gcc -DHAVE_CONFIG_H -I. -I.. -DCDB -I/knp-4.16/cdb -I/usr/local//include -I/knp-4.16/distsim -g3 -O0 -MT dsim.o  -MD -MP -MF .deps/dsim.Tpo  -c -fPIC -o dsim.o dsim.c

;共有ライブラリの作成
g++ -shared -o libdsim.so dsim.o db.o tools.o ../distsim/distsim.o ../distsim/distsim_for_knp.o -L../cdb/ -lstdc++ -lcdb -lz
cp libdsim.so /usr/local/lib
ldconfig
gcc -I./ dmain.c -o dmain -ldsim

; 共有ライブラリのバージョニングする場合
g++ -shared -Wl,-soname=libdsim.so.1 -o libdsim.so.1.0.0 dsim.o db.o tools.o ../distsim/distsim.o ../distsim/distsim_for_knp.o -L../cdb/ -lstdc++ -lcdb -lz
cp libdsim.so.1.0.0 /usr/local/lib
ldconfig
cd /usr/local/lib
ln -s libdsim.so.1.0.0 libdsim.so

*/

/* dsim.h */
#ifndef DSIM_H
#define DSIM_H
void init_dsim(char *dic_name);
float similarity(char *rep1, char *rep2);
#endif

/* dsim.c */
#include <stdio.h>
#include <sys/stat.h>
#include <const.h>
#include <path.h>
#include "distsim_for_knp.h"
#include "dsim.h"

char static_buffer1[5120];
char static_buffer2[5120];

DBM_FILE mrph2id_db;

char *rep2id2(char *rep, int rep_len, char *buffer)
{
    /* MRPH_MAX * 9(max8桁+"+"の分)以上あるので溢れない */
    buffer[0] = '\0';

    if (rep && rep_len > 0) {
        char *token_start, *token, *value;
        char *copied_rep = (char *)malloc_data(sizeof(char) * rep_len + 1, "rep2id");
        strncpy(copied_rep, rep, rep_len);
        copied_rep[rep_len] = '\0';
        token_start = strtok(copied_rep, "+?");
        token = token_start;
        while (token) {
            value = db_get(mrph2id_db, token);
            if (value) {
                if (buffer[0]) /* 2つ目以降 */
                    strncat(buffer, rep + (token - 1 - token_start), 1);
                strcat(buffer, value);
                free(value);
            }
            token = strtok(NULL, "+?");
        }
        free(copied_rep);
        /* fprintf(stderr, ";; %s -> %s\n", rep, buffer); */
    }

    return buffer;
}

DBM_FILE *open_dict(char *dic_name)
{
    DBM_FILE db;
    db = DB_open(dic_name, O_RDONLY, 0);
    return db;
}

void init_dsim(char *dic_name)
{
    mrph2id_db = open_dict(dic_name);
    init_distsim();
}

float similarity(char *rep1, char *rep2)
{
    char *id1 = rep2id2(rep1, strlen(rep1), &(static_buffer1[0]));
    char *id2 = rep2id2(rep2, strlen(rep2), &(static_buffer2[0]));
    float score = calc_distsim(id1, id2);
    printf("distsim_score(%s:%s, %s:%s)=%f\n",rep1,id1,rep2,id2,score);
    return score; 
}

/* main */
#include <stdio.h>
#include <dsim.h>

int main()
{
    init_dsim("mrph2id.db");
    char rep1[100], rep2[100];
    scanf("%s %s", rep1, rep2);
    float score = similarity(rep1, rep2);
    printf("Score: %f\n",score);
}

/* Python */
import ctypes
dsim = ctypes.cdll.LoadLibrary('libdsim.so.1')
#dsim = ctypes.cdll.LoadLibrary('libdsim.so')
dsim.init_dsim(ctypes.c_char_p('mrph2id.db'))
similarity = dsim.similarity
similarity.restype = ctypes.c_float
similarity.argtypes = [ctypes.c_char_p, ctypes.c_char_p]
similarity('林檎/りんご', '梨/なし')
similarity('林檎/りんご', '引力/いんりょく')