select_ifile.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
 * A command-line interface to Select.
 * This was inspired by Jason Rennie's ifile.
 *
 * @author  Mikael Ylikoski
 * @date    2002
 */

#include <getopt.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "protocol_c.h"

static int text;		/**< Treat input as plain text */
static int numbers;		/**< Use folder numbers instead of names */

static char *buf;		/**< Input buffer */
static int blen;		/**< Input buffer length */
static int len;			/**< Current input length */

static protocol_c_data *pdata;	/**< Protocol data */

static struct option longopts[] = {
    /* ifile options */
    { "help", no_argument, NULL, '?' },
    { "insert", required_argument, NULL, 'i' },
    { "query", no_argument, NULL, 'q' },
    { "query-insert", no_argument, NULL, 'Q' },
    { "update", required_argument, NULL, 'u' },
    /* other options */
    { "address", required_argument, NULL, 'a' },
    { "numbers", no_argument, NULL, 'n' },
    { "text", no_argument, NULL, 't' },
    { 0, 0, 0, 0 }
};


static int
input_file (FILE *fd) {
    char ch, *str;

    len = 0;
    while ((ch = fgetc (fd)) != EOF) {
	buf[len] = ch;
	len++;
	if (len == blen) {
	    str = realloc (buf, blen + 10000);
	    if (!str) {
		fprintf (stderr, "Error: Out of memory\n");
		return -1;
	    }
	    buf = str;
	    blen += 10000;
	}
    }
    return 0;
}


static int
read_file (char *file) {
    FILE *fd;
    int i;

    fd = fopen (file, "r");
    if (!fd)
	return -1;
    i = input_file (fd);
    fclose (fd);
    return i;
}


static int
do_insert (char *folder) {
    if (protocol_c_open (pdata)) {
	fprintf (stderr, "Error: Cannot connect to selectd\n");
	return -1;
    }
    protocol_c_part (pdata, "text", NULL, buf, len);

    if (protocol_c_learn_folder (pdata, folder)) {
	fprintf (stderr, "Error: Cannot teach selectd\n");
	return -1;
    }
    protocol_c_close (pdata);

    return 0;
}


static int
do_query (void) {
    double val, oval;
    int i, j, pos;
    double_array *dl;
    str_array *sa;

    if (protocol_c_open (pdata)) {
	fprintf (stderr, "Error: Cannot connect to selectd\n");
	return -1;
    }

    protocol_c_part (pdata, "text", NULL, buf, len);

    sa = NULL;
    if (!numbers)
	sa = protocol_c_get_table (pdata, "folders");

    dl = protocol_c_classify_score (pdata, "0");
    protocol_c_close (pdata);

    if (!dl)
	return -1;

    /* find min */
    oval = 0;
    for (j = 0; j < dl->len; j++)
	if (dl->array[j] != 0 && oval >= dl->array[j])
	    oval = dl->array[j] - 10;

    /* set zeros to min */
    for (j = 0; j < dl->len; j++)
	if (dl->array[j] == 0)
	    dl->array[j] = oval;

    /* print list */
    val = 0;
    for (i = 0; i < dl->len; i++) {
	pos = -1;
	for (j = 0; j < dl->len; j++)
	    if (dl->array[j] >= oval && (pos == -1 || val < dl->array[j])) {
		val = dl->array[j];
		pos = j;
	    }

	dl->array[pos] = oval - 1;

	if (numbers)
	    printf ("%d %.8f\n", pos, val);
	else
	    printf ("%s %.8f\n", sa->array[pos], val);
    }
    printf ("---------\n");

    free (dl);
    if (sa) {
	free (sa->array);
	free (sa);
    }

    return -1;
}


int
main (int argc, char **argv) {
    int i;
    int insert, query, query_insert, update;
    char *address;
    char *folder;
    str_array *sa;

    insert = 0;
    query = 0;
    query_insert = 0;
    text = 0;
    update = 0;
    address = NULL;
    folder = NULL;

    buf = malloc (10000);
    if (!buf) {
	printf ("Error: Out of memory!\n");
	return 1;
    }
    blen = 10000;
    len = 0;

    while ((i = getopt_long (argc, argv, "a:i:nqu:", longopts, NULL)) != EOF) {
        switch (i) {
        case 'a':
	    address = optarg;
            break;
        case 'i':
            folder = optarg;
	    insert = 1;
            break;
        case 'n':
	    numbers = 1;
            break;
        case 'q':
	    query = 1;
            break;
        case 'Q':
	    query_insert = 1;
            break;
        case 't':
	    text = 1;
            break;
        case 'u':
            folder = optarg;
	    insert = 1;
	    update = 1;
            break;
        default:
	    printf ("Usage: ...\n");
	    return -1;
        }
    }

    if (insert + query != 1) {
	fprintf (stderr, "Error: One (and only one) of the options -i, -u and "
		 "-q must be given.\n");
	return 1;
    }

    pdata = protocol_c_new (15000, address);

    /*
    protocol_c_open (pdata);
    protocol_c_open_send (pdata);
    sa = protocol_c_get_table (pdata, "classifiers");
    protocol_c_close (pdata);
    */

    if (update) {
	protocol_c_open (pdata);
	sa = protocol_c_get_table (pdata, "folders");
	protocol_c_close (pdata);
	for (i = 0; i < sa->len; i++)
	    if (strcmp (sa->array[i], folder))
		break;
	if (i == sa->len)
	    return 1;
	free (sa->array);
	free (sa);
    }

    if (insert) {
	if (optind == argc) {
	    i = input_file (stdin);
	    i = do_insert (folder);
	} else {
	    for (; optind < argc; optind++) {
		i = read_file (argv[optind]);
		i = do_insert (folder);
	    }
	}
    } else if (query) {
	if (optind == argc) {
	    i = input_file (stdin);
	    i = do_query ();
	} else {
	    for (; optind < argc; optind++) {
		if (read_file (argv[optind])) {
		    fprintf (stderr, "Error: Cannot read file\n");
		    continue;
		}
		i = do_query ();
	    }
	}
    }

    return 0;
}