stopword.c

来自「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。」· C语言 代码 · 共 97 行

/**
 * @file
 * Handle stopword removal.
 *
 * @author  Mikael Ylikoski
 * @date    2001-2002
 */

#include <glib.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "stopword.h"
#include "utility.h"


/**
 * Word stopper.
 */
struct word_stopper_ {
    GHashTable *hash;	/**< Hash table of stopwords */
};


/**
 * Create a new word stopper.
 *
 * @return The word stopper.
 */
word_stopper *
stopword_new (void) {
    word_stopper *ws;

    ws = my_malloc (sizeof(word_stopper));
    ws->hash = g_hash_table_new (g_str_hash, g_str_equal);
    if (!ws->hash) {
	free (ws);
	return NULL;
    }

    return ws;
}


/**
 * Free a word stopper.
 *
 * @param ws  word stopper to free
 */
void
stopword_free (word_stopper *ws) {
    g_hash_table_destroy (ws->hash);
    free (ws);
}


/**
 * Load stopwords from a file and add to a word stopper.
 *
 * @param ws    the word stopper
 * @param file  file of stop words
 * @return Zero if ok, nonzero otherwise.
 */
int
stopword_load (word_stopper *ws, const char *file) {
    char buf[30], *s;
    FILE *fp;

    fp = fopen (file, "r");
    if (!fp)
	return -1;

    while (get_line_nows (fp, buf, 30)) {
	s = g_hash_table_lookup (ws->hash, buf);
	if (!s) {
	    s = my_strdup (buf);
	    g_hash_table_insert (ws->hash, s, (gpointer)1);
	}
    }

    fclose (fp);
    return 0;
}


/**
 * Test whether a word is a stop word.
 *
 * @param ws    word stopper
 * @param word  word to test
 * @return Nonzero if the word is a stop word, zero otherwise.
 */
int
stopword_is (word_stopper *ws, const char *word) {
    return (int)g_hash_table_lookup (ws->hash, word);
}