document.h

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。

#ifndef DOCUMENT_H
#define DOCUMENT_H

#include <time.h>
#include "rfc822.h"

enum document_type {
    PLAIN,	/**< Unformatted text */
    RFC822	/**< Email message */
};

typedef struct document_ document;

document *
document_new (char *source, enum document_type dt);

void
document_free (document *doc);

void
document_add_text (document *doc, char *charset, char *text, int len);

int
document_set_from (document *doc, char *from);

int
document_set_subject (document *doc, char *subject);

int
document_set_language (document *doc, char *lang);

int
document_set_rfc822 (document *doc, char *src);

const char *
document_get_from (document *doc);

const char *
document_get_from_address (document *doc);

const char *
document_get_from_name (document *doc);

const char *
document_get_to (document *doc);

const char *
document_get_subject (document *doc);

time_t
document_get_time (document *doc);

const char *
document_get_language (document *doc);

const char *
document_get_charset (document *doc);

text_part *
document_get_parts (document *doc);

#endif