static gboolean
is_spam (GMimeMessage *mime_message)
{
const char *spam;
g_return_val_if_fail(GMIME_IS_MESSAGE(mime_message), FALSE);
/* SpamAssassin */
spam = g_mime_message_get_header(mime_message, "X-Spam-Status");
if (spam && mn_ascii_str_case_has_prefix(spam, "yes"))
return TRUE;
/* bogofilter */
spam = g_mime_message_get_header(mime_message, "X-Bogosity");
if (spam && mn_ascii_str_case_has_prefix(spam, "yes"))
return TRUE;
return FALSE;
}
parsed_article *parse_file(const char *file_name) {
static parsed_article pa;
GMimeStream *stream;
GMimeMessage *msg = NULL;
int offset;
int file;
printf("%s\n", file_name);
if ((file = open(file_name, O_RDONLY|O_STREAMING)) == -1) {
fprintf(stderr, "Can't open %s\n", file_name);
return NULL;
}
stream = g_mime_stream_fs_new(file);
msg = g_mime_parser_construct_message(stream);
g_mime_stream_unref(stream);
if (msg != 0) {
hstrcpy(pa.from, g_mime_message_get_header(msg, "From"));
hstrcpy(pa.subject, g_mime_message_get_subject(msg));
hstrcpy(pa.message_id, g_mime_message_get_message_id(msg));
hstrcpy(pa.references, g_mime_message_get_header(msg, "references"));
hstrcpy(pa.xref, g_mime_message_get_header(msg, "xref"));
hstrcpy(pa.original_message_id,
g_mime_message_get_header(msg, "original-message-id"));
g_mime_message_get_date(msg, &pa.date, &offset);
if (pa.xref != NULL &&
strstr(pa.xref, "gmane.spam.detected") != NULL)
pa.spamp = 1;
else
pa.spamp = 0;
g_mime_message_foreach_part(msg, count_part, (gpointer) &pa);
g_mime_object_unref(GMIME_OBJECT(msg));
}
close(file);
return &pa;
}
document* parse_file(const char *file_name) {
int tallied_length = 0;
GMimeStream *stream;
GMimeMessage *msg = 0;
// struct stat stat_buf;
const char *author, *subject, *xref, *xref_end;
time_t date;
int offset;
int num_words = 0;
int file;
InternetAddress *iaddr;
InternetAddressList *iaddr_list;
char *address;
#if DEBUG
printf("%s\n", file_name);
#endif
/*
if ((file = stat(file_name, &stat_buf)) == -1) {
perror("tokenizer");
return NULL;
}
*/
// |O_STREAMING
if ((file = open(file_name, O_RDONLY|O_STREAMING)) == -1) {
perror("tokenizer");
return NULL;
}
#ifdef POSIX_FADV_NOREUSE
no_reuse(file);
#endif
stream = g_mime_stream_fs_new(file);
msg = g_mime_parser_construct_message(stream);
g_mime_stream_unref(stream);
if (msg != 0) {
table = g_hash_table_new(g_str_hash, g_str_equal);
bufp = buffer;
dword_table[0].word = NULL;
bzero(saved_body, MAX_SAVED_BODY_LENGTH);
saved_body_length = 0;
author = g_mime_message_get_sender(msg);
subject = g_mime_message_get_subject(msg);
xref = g_mime_message_get_header(msg, "Xref");
g_mime_message_get_date(msg, &date, &offset);
if (author != NULL && subject != NULL && xref != NULL) {
tallied_length = tally_string(author, tallied_length);
strncpy(doc.author, author, MAX_HEADER_LENGTH-1);
/* Get the address from the From header. */
if ((iaddr_list = internet_address_parse_string(author)) != NULL) {
iaddr = iaddr_list->address;
internet_address_set_name(iaddr, NULL);
address = internet_address_to_string(iaddr, FALSE);
strncpy(doc.address, address, MAX_HEADER_LENGTH-1);
downcase_string(doc.address);
free(address);
internet_address_list_destroy(iaddr_list);
} else {
*doc.address = 0;
}
tallied_length = tally_string(subject, tallied_length);
strncpy(doc.subject, subject, MAX_HEADER_LENGTH-1);
doc.time = date;
if ((xref = strchr(xref, ' ')) != NULL) {
xref++;
xref_end = strchr(xref, ':');
*doc.group = 0;
strncat(doc.group, xref, min(xref_end-xref, MAX_HEADER_LENGTH-1));
xref_end++;
sscanf(xref_end, "%d", &doc.article);
}
g_mime_message_foreach_part(msg, partFound, (gpointer) &tallied_length);
strncpy(doc.body, saved_body, MAX_SAVED_BODY_LENGTH);
g_hash_table_foreach(table, add_word_to_table, (gpointer) &num_words);
dword_table[num_words].word = NULL;
g_hash_table_destroy(table);
g_mime_object_unref(GMIME_OBJECT(msg));
} else {
close(file);
return NULL;
}
}
close(file);
doc.words = dword_table;
doc.num_words = num_words;
return &doc;
}
MNMessage *
mn_message_new_from_mime_message_full (GType type,
MNMailbox *mailbox,
GMimeMessage *mime_message,
const char *mid,
const char *uri,
MNMessageFlags flags,
gboolean handle_status)
{
MNMessage *message;
const char *id;
time_t sent_time;
const char *from;
const char *subject;
char *decoded_from;
char *decoded_subject;
g_return_val_if_fail(type != 0, NULL);
g_return_val_if_fail(MN_IS_MAILBOX(mailbox), NULL);
g_return_val_if_fail(GMIME_IS_MESSAGE(mime_message), NULL);
if (is_spam(mime_message))
return NULL;
if (handle_status)
{
const char *status;
status = g_mime_message_get_header(mime_message, "Status");
if (status && strchr(status, 'R'))
return NULL; /* the message was read */
else if (status && strchr(status, 'O'))
flags &= ~MN_MESSAGE_NEW;
else
flags |= MN_MESSAGE_NEW;
}
id = g_mime_message_get_message_id(mime_message);
g_mime_message_get_date(mime_message, &sent_time, NULL);
from = g_mime_message_get_sender(mime_message);
subject = g_mime_message_get_subject(mime_message);
decoded_from = from ? g_mime_utils_header_decode_text(from) : NULL;
decoded_subject = subject ? g_mime_utils_header_decode_text(subject) : NULL;
message = g_object_new(type,
MN_MESSAGE_PROP_MAILBOX(mailbox),
MN_MESSAGE_PROP_SENT_TIME(sent_time),
MN_MESSAGE_PROP_ID((char *) id),
MN_MESSAGE_PROP_MID((char *) mid),
MN_MESSAGE_PROP_FROM(decoded_from),
MN_MESSAGE_PROP_SUBJECT(decoded_subject),
MN_MESSAGE_PROP_URI((char *) uri),
MN_MESSAGE_PROP_FLAGS(flags),
NULL);
g_free(decoded_from);
g_free(decoded_subject);
return message;
}
