2 * Copyright © 2009 Carl Worth
4 * This program is free software: you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation, either version 3 of the License, or
7 * (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program. If not, see http://www.gnu.org/licenses/ .
17 * Author: Carl Worth <cworth@cworth.org>
28 #include <gmime/gmime.h>
34 #define ARRAY_SIZE(arr) (sizeof (arr) / sizeof (arr[0]))
36 /* Xapian complains if we provide a term longer than this. */
37 #define NOTMUCH_MAX_TERM 245
39 /* These prefix values are specifically chosen to be compatible
40 * with sup, (http://sup.rubyforge.org), written by
41 * William Morgan <wmorgan-sup@masanjin.net>, and released
42 * under the GNU GPL v2.
50 prefix_t NORMAL_PREFIX[] = {
53 { "from_name", "FN" },
59 prefix_t BOOLEAN_PREFIX[] = {
61 { "from_email", "FE" },
67 { "attachment_extension", "O" },
73 /* Similarly, these value numbers are also chosen to be sup
77 NOTMUCH_VALUE_MESSAGE_ID = 0,
78 NOTMUCH_VALUE_THREAD = 1,
79 NOTMUCH_VALUE_DATE = 2
83 find_prefix (const char *name)
87 for (i = 0; i < ARRAY_SIZE (NORMAL_PREFIX); i++)
88 if (strcmp (name, NORMAL_PREFIX[i].name) == 0)
89 return NORMAL_PREFIX[i].prefix;
91 for (i = 0; i < ARRAY_SIZE (BOOLEAN_PREFIX); i++)
92 if (strcmp (name, BOOLEAN_PREFIX[i].name) == 0)
93 return BOOLEAN_PREFIX[i].prefix;
98 int TERM_COMBINED = 0;
101 add_term (Xapian::Document doc,
102 const char *prefix_name,
111 prefix = find_prefix (prefix_name);
113 term = g_strdup_printf ("%s%s", prefix, value);
115 if (strlen (term) <= NOTMUCH_MAX_TERM)
122 gen_terms (Xapian::TermGenerator term_gen,
123 const char *prefix_name,
131 prefix = find_prefix (prefix_name);
133 term_gen.index_text (text, 1, prefix);
137 gen_terms_address_name (Xapian::TermGenerator term_gen,
138 InternetAddress *address,
139 const char *prefix_name)
143 name = internet_address_get_name (address);
146 gen_terms (term_gen, prefix_name, name);
150 gen_terms_address_names (Xapian::TermGenerator term_gen,
151 InternetAddressList *addresses,
152 const char *address_type)
155 InternetAddress *address;
157 for (i = 0; i < internet_address_list_length (addresses); i++) {
158 address = internet_address_list_get_address (addresses, i);
159 gen_terms_address_name (term_gen, address, address_type);
160 gen_terms_address_name (term_gen, address, "name");
161 gen_terms_address_name (term_gen, address, "body");
166 add_term_address_addr (Xapian::Document doc,
167 InternetAddress *address,
168 const char *prefix_name)
170 InternetAddressMailbox *mailbox = INTERNET_ADDRESS_MAILBOX (address);
173 addr = internet_address_mailbox_get_addr (mailbox);
176 add_term (doc, prefix_name, addr);
180 add_terms_address_addrs (Xapian::Document doc,
181 InternetAddressList *addresses,
182 const char *address_type)
185 InternetAddress *address;
187 for (i = 0; i < internet_address_list_length (addresses); i++) {
188 address = internet_address_list_get_address (addresses, i);
189 add_term_address_addr (doc, address, address_type);
190 add_term_address_addr (doc, address, "email");
195 skip_re_in_subject (const char *subject)
197 const char *s = subject;
200 while (*s && isspace (*s))
202 if (strncasecmp (s, "re:", 3) == 0)
212 find_messages_by_term (Xapian::Database db,
213 const char *prefix_name,
215 Xapian::PostingIterator *begin,
216 Xapian::PostingIterator *end)
218 Xapian::PostingIterator i;
221 term = g_strdup_printf ("%s%s", find_prefix (prefix_name), value);
223 *begin = db.postlist_begin (term);
226 *end = db.postlist_end (term);
232 find_message_by_docid (Xapian::Database db, Xapian::docid docid)
234 return db.get_document (docid);
238 find_message_by_message_id (Xapian::Database db, const char *message_id)
240 Xapian::PostingIterator i, end;
242 find_messages_by_term (db, "msgid", message_id, &i, &end);
245 return find_message_by_docid (db, *i);
247 return Xapian::Document ();
251 insert_thread_id (GHashTable *thread_ids, Xapian::Document doc)
256 value_string = doc.get_value (NOTMUCH_VALUE_THREAD);
257 value = value_string.c_str();
259 g_hash_table_insert (thread_ids, strdup (value), NULL);
263 find_thread_id (Xapian::Database db,
265 const char *message_id)
267 Xapian::PostingIterator child, children_end;
268 Xapian::Document doc;
269 GHashTable *thread_ids;
271 GString *result = NULL;
273 const char *parent_message_id;
275 thread_ids = g_hash_table_new (g_str_hash, g_str_equal);
277 find_messages_by_term (db, "ref", message_id, &child, &children_end);
278 for ( ; child != children_end; child++) {
279 doc = find_message_by_docid (db, *child);
280 insert_thread_id (thread_ids, doc);
283 for (i = 0; i < parents->len; i++) {
284 parent_message_id = (char *) g_ptr_array_index (parents, i);
285 doc = find_message_by_message_id (db, parent_message_id);
286 insert_thread_id (thread_ids, doc);
289 keys = g_hash_table_get_keys (thread_ids);
290 for (l = keys; l; l = l->next) {
291 char *id = (char *) l->data;
292 if (result == NULL) {
293 result = g_string_new (id);
295 g_string_append_printf (result, ",%s", id);
301 return g_string_free (result, FALSE);
306 /* Add a term for each message-id in the References header of the
309 parse_references (GPtrArray *array,
310 const char *refs_str)
312 GMimeReferences *refs, *r;
313 const char *message_id;
315 if (refs_str == NULL)
318 refs = g_mime_references_decode (refs_str);
320 for (r = refs; r; r = r->next) {
321 message_id = g_mime_references_get_message_id (r);
322 g_ptr_array_add (array, g_strdup (message_id));
325 g_mime_references_free (refs);
328 /* Given a string representing the body of a message, generate terms
329 * for it, (skipping quoted portions and signatures). */
331 gen_terms_body_str (Xapian::TermGenerator term_gen,
334 char *line, *line_end, *next_line;
346 next_line = strchr (line, '\n');
347 if (next_line == NULL) {
348 next_line = line + strlen (line);
350 line_end = next_line - 1;
352 /* Trim whitespace. */
353 while (*next_line && isspace (*next_line))
356 /* Skip lines that are quotes. */
360 /* Also skip lines introducing a quote on the next line. */
361 while (line_end > line && isspace (*line_end))
364 if (*line_end == ':' && *next_line == '>')
367 /* Finally, bail as soon as we see a signature. */
368 /* XXX: Should only do this if "near" the end of the message. */
369 if (strncmp (line, "-- ", 3) == 0 ||
370 strncmp (line, "----------", 10) == 0 ||
371 strncmp (line, "__________", 10) == 0)
374 *(line_end + 1) = '\0';
375 gen_terms (term_gen, "body", line);
380 /* Callback to generate terms for each mime part of a message. */
382 gen_terms_part (GMimeObject *parent,
386 Xapian::TermGenerator *term_gen = (Xapian::TermGenerator *) user_data;
388 GMimeDataWrapper *wrapper;
389 GByteArray *byte_array;
392 if (! GMIME_IS_PART (part)) {
393 fprintf (stderr, "Warning: Not indexing unknown mime part: %s.\n",
394 g_type_name (G_OBJECT_TYPE (part)));
398 byte_array = g_byte_array_new ();
400 stream = g_mime_stream_mem_new_with_byte_array (byte_array);
401 g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE);
402 wrapper = g_mime_part_get_content_object (GMIME_PART (part));
403 g_mime_data_wrapper_write_to_stream (wrapper, stream);
405 g_object_unref (stream);
407 body = (char *) g_byte_array_free (byte_array, FALSE);
409 gen_terms_body_str (*term_gen, body);
415 index_file (Xapian::WritableDatabase db,
416 Xapian::TermGenerator term_gen,
417 const char *filename)
419 Xapian::Document doc;
423 GMimeMessage *message;
424 InternetAddressList *addresses;
429 const char *subject, *refs, *in_reply_to, *from;
430 const char *message_id, *thread_id;
433 struct tm gm_time_tm;
434 char date_str[16]; /* YYYYMMDDHHMMSS + 1 for Y100k compatibility ;-) */
437 file = fopen (filename, "r");
439 fprintf (stderr, "Error opening %s: %s\n", filename, strerror (errno));
443 stream = g_mime_stream_file_new (file);
445 parser = g_mime_parser_new_with_stream (stream);
447 message = g_mime_parser_construct_message (parser);
449 doc = Xapian::Document ();
451 doc.set_data (filename);
453 term_gen.set_stemmer (Xapian::Stem ("english"));
455 term_gen.set_document (doc);
457 from = g_mime_message_get_sender (message);
458 addresses = internet_address_list_parse_string (from);
460 gen_terms_address_names (term_gen, addresses, "from_name");
462 addresses = g_mime_message_get_all_recipients (message);
463 gen_terms_address_names (term_gen, addresses, "to_name");
465 subject = g_mime_message_get_subject (message);
466 subject = skip_re_in_subject (subject);
467 gen_terms (term_gen, "subject", subject);
468 gen_terms (term_gen, "body", subject);
470 g_mime_message_foreach (message, gen_terms_part, &term_gen);
472 parents = g_ptr_array_new ();
474 refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
475 parse_references (parents, refs);
477 in_reply_to = g_mime_object_get_header (GMIME_OBJECT (message),
479 parse_references (parents, in_reply_to);
481 for (i = 0; i < parents->len; i++)
482 add_term (doc, "ref", (char *) g_ptr_array_index (parents, i));
484 message_id = g_mime_message_get_message_id (message);
486 thread_id = find_thread_id (db, parents, message_id);
488 for (i = 0; i < parents->len; i++)
489 g_free (g_ptr_array_index (parents, i));
490 g_ptr_array_free (parents, TRUE);
492 from = g_mime_message_get_sender (message);
493 addresses = internet_address_list_parse_string (from);
495 add_terms_address_addrs (doc, addresses, "from_email");
497 add_terms_address_addrs (doc,
498 g_mime_message_get_all_recipients (message),
501 g_mime_message_get_date (message, &time, NULL);
503 gmtime_r (&time, &gm_time_tm);
505 if (strftime (date_str, sizeof (date_str),
506 "%Y%m%d%H%M%S", &gm_time_tm) == 0) {
507 fprintf (stderr, "Internal error formatting time\n");
511 add_term (doc, "date", date_str);
513 add_term (doc, "label", "inbox");
514 add_term (doc, "label", "unread");
515 add_term (doc, "type", "mail");
516 add_term (doc, "source_id", "1");
518 add_term (doc, "msgid", message_id);
519 doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, message_id);
522 add_term (doc, "thread", thread_id);
523 doc.add_value (NOTMUCH_VALUE_THREAD, thread_id);
524 free ((void *) thread_id);
526 /* If not referenced thread, use the message ID */
527 add_term (doc, "thread", message_id);
528 doc.add_value (NOTMUCH_VALUE_THREAD, message_id);
531 doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
533 db.add_document (doc);
535 g_object_unref (message);
536 g_object_unref (parser);
537 g_object_unref (stream);
541 usage (const char *argv0)
543 fprintf (stderr, "Usage: %s <path-to-xapian-database>\n", argv0);
544 fprintf (stderr, "\n");
545 fprintf (stderr, "Messages to be indexed are read from stdnin as absolute filenames\n");
546 fprintf (stderr, "one file per line.");
550 main (int argc, char **argv)
552 const char *database_path;
555 GIOStatus gio_status;
556 GError *error = NULL;
563 database_path = argv[1];
568 Xapian::WritableDatabase db;
569 Xapian::TermGenerator term_gen;
571 db = Xapian::WritableDatabase (database_path,
572 Xapian::DB_CREATE_OR_OPEN);
574 term_gen = Xapian::TermGenerator ();
576 channel = g_io_channel_unix_new (fileno (stdin));
579 gio_status = g_io_channel_read_line (channel, &filename,
581 if (gio_status == G_IO_STATUS_EOF)
583 if (gio_status != G_IO_STATUS_NORMAL) {
584 fprintf (stderr, "An error occurred reading from stdin: %s\n",
589 g_strchomp (filename);
590 index_file (db, term_gen, filename);
595 } catch (const Xapian::Error &error) {
596 cerr << "A Xapian exception occurred: " << error.get_msg () << endl;