languages.c

使用具有增量学习的监控式学习方法。包括几个不同的分类算法。

/* Copyright (C) 2002  Mikael Ylikoski
 * See the accompanying file "README" for the full copyright notice */
/**
 * @file
 * Handle multiple languages.
 *
 * @author  Mikael Ylikoski
 * @date    2002
 */

#include <stdlib.h>
#include <string.h>
#include "languages.h"
#include "stemmer.h"
#include "stopword.h"
#include "utility.h"


/**
 * Multiple language handler.
 */
struct languages_ {
    language *langs;	/**< languages */
    language *unknown;	/**< default language */
    int size;		/**< number of languages */
};


/**
 * Create multi language handler.
 */
languages *
languages_create (void) {
    languages *langs;

    langs = my_malloc (sizeof(languages));
    langs->size = 0;
    langs->langs = NULL;
    langs->unknown = NULL;

    return langs;
}


/**
 * Free memory of multi language handler.
 */
void
languages_free (languages *langs) {
    // FIXME free rest too
    free (langs->langs);
    free (langs);
}


/**
 * Add language to multi language handler.
 */
int
languages_add (languages *langs, const char *lang, const char *locale,
	       word_stopper *stop, stemmer_functions *stem) {
    language *newlang;

    newlang = my_realloc (langs->langs, sizeof(language) * (langs->size + 1));
    langs->langs = newlang;
    newlang = &langs->langs[langs->size];
    newlang->language = lang;
    newlang->locale = locale;
    newlang->stopper = stop;
    newlang->stemmer = stem;
    langs->size++;

    return 0;
}


/**
 * Set default language.
 */
int
languages_set_unknown (languages *langs, const char *lang) {
    int i;

    for (i = 0; i < langs->size; i++)
	if (!strcmp (langs->langs[i].language, lang)) {
	    langs->unknown = &langs->langs[i];
	    return 0;
	}
    return -1;
}


/**
 * Find language struct for language.
 */
language *
languages_find_language (languages *langs, const char *lang) {
    int i;

    if (!lang)
	return NULL;
    if (!strcmp (LANGUAGE_UNKNOWN, lang))
	return langs->unknown;
    for (i = 0; i < langs->size; i++)
	if (!strcmp (langs->langs[i].language, lang))
	    return &langs->langs[i];
    return NULL;
}