1 /* thread.cc - Results of thread-based searches from a notmuch database
3 * Copyright © 2009 Carl Worth
5 * This program is free software: you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation, either version 3 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program. If not, see https://www.gnu.org/licenses/ .
18 * Author: Carl Worth <cworth@cworth.org>
21 #include "notmuch-private.h"
22 #include "database-private.h"
24 #include <gmime/gmime.h>
25 #include <glib.h> /* GHashTable */
27 #ifdef DEBUG_THREADING
28 #define THREAD_DEBUG(format, ...) fprintf (stderr, format " (%s).\n", ##__VA_ARGS__, __location__)
30 #define THREAD_DEBUG(format, ...) do {} while (0) /* ignored */
33 struct _notmuch_thread {
34 notmuch_database_t *notmuch;
37 GHashTable *authors_hash;
38 GPtrArray *authors_array;
39 GHashTable *matched_authors_hash;
40 GPtrArray *matched_authors_array;
44 /* All messages, oldest first. */
45 notmuch_message_list_t *message_list;
46 /* Top-level messages, oldest first. */
47 notmuch_message_list_t *toplevel_list;
49 GHashTable *message_hash;
58 _notmuch_thread_destructor (notmuch_thread_t *thread)
60 g_hash_table_unref (thread->authors_hash);
61 g_hash_table_unref (thread->matched_authors_hash);
62 g_hash_table_unref (thread->tags);
63 g_hash_table_unref (thread->message_hash);
65 if (thread->authors_array) {
66 g_ptr_array_free (thread->authors_array, true);
67 thread->authors_array = NULL;
70 if (thread->matched_authors_array) {
71 g_ptr_array_free (thread->matched_authors_array, true);
72 thread->matched_authors_array = NULL;
78 /* Add each author of the thread to the thread's authors_hash and to
79 * the thread's authors_array. */
81 _thread_add_author (notmuch_thread_t *thread,
89 if (g_hash_table_lookup_extended (thread->authors_hash,
93 author_copy = talloc_strdup (thread, author);
95 g_hash_table_insert (thread->authors_hash, author_copy, NULL);
97 g_ptr_array_add (thread->authors_array, author_copy);
100 /* Add each matched author of the thread to the thread's
101 * matched_authors_hash and to the thread's matched_authors_array. */
103 _thread_add_matched_author (notmuch_thread_t *thread,
111 if (g_hash_table_lookup_extended (thread->matched_authors_hash,
115 author_copy = talloc_strdup (thread, author);
117 g_hash_table_insert (thread->matched_authors_hash, author_copy, NULL);
119 g_ptr_array_add (thread->matched_authors_array, author_copy);
122 /* Construct an authors string from matched_authors_array and
123 * authors_array. The string contains matched authors first, then
124 * non-matched authors (with the two groups separated by '|'). Within
125 * each group, authors are listed in date order. */
127 _resolve_thread_authors_string (notmuch_thread_t *thread)
131 int first_non_matched_author = 1;
133 /* First, list all matched authors in date order. */
134 for (i = 0; i < thread->matched_authors_array->len; i++) {
135 author = (char *) g_ptr_array_index (thread->matched_authors_array, i);
137 thread->authors = talloc_asprintf (thread, "%s, %s",
141 thread->authors = author;
144 /* Next, append any non-matched authors that haven't already appeared. */
145 for (i = 0; i < thread->authors_array->len; i++) {
146 author = (char *) g_ptr_array_index (thread->authors_array, i);
147 if (g_hash_table_lookup_extended (thread->matched_authors_hash,
150 if (first_non_matched_author) {
151 thread->authors = talloc_asprintf (thread, "%s| %s",
155 thread->authors = talloc_asprintf (thread, "%s, %s",
160 first_non_matched_author = 0;
163 g_ptr_array_free (thread->authors_array, true);
164 thread->authors_array = NULL;
165 g_ptr_array_free (thread->matched_authors_array, true);
166 thread->matched_authors_array = NULL;
168 if (! thread->authors)
169 thread->authors = talloc_strdup (thread, "");
172 /* clean up the ugly "Lastname, Firstname" format that some mail systems
173 * (most notably, Exchange) are creating to be "Firstname Lastname"
174 * To make sure that we don't change other potential situations where a
175 * comma is in the name, we check that we match one of these patterns
176 * "Last, First" <first.last@company.com>
177 * "Last, First MI" <first.mi.last@company.com>
180 _thread_cleanup_author (notmuch_thread_t *thread,
181 const char *author, const char *from)
183 char *clean_author, *test_author;
190 clean_author = talloc_strdup (thread, author);
191 if (clean_author == NULL)
193 /* check if there's a comma in the name and that there's a
194 * component of the name behind it (so the name doesn't end with
195 * the comma - in which case the string that strchr finds is just
196 * one character long ",\0").
197 * Otherwise just return the copy of the original author name that
199 comma = strchr (author, ',');
200 if (comma && strlen (comma) > 1) {
201 /* let's assemble what we think is the correct name */
202 lname = comma - author;
204 /* Skip all the spaces after the comma */
205 fname = strlen (author) - lname - 1;
207 while (*comma == ' ') {
211 strncpy (clean_author, comma, fname);
213 *(clean_author + fname) = ' ';
214 strncpy (clean_author + fname + 1, author, lname);
215 *(clean_author + fname + 1 + lname) = '\0';
216 /* make a temporary copy and see if it matches the email */
217 test_author = talloc_strdup (thread, clean_author);
219 blank = strchr (test_author, ' ');
220 while (blank != NULL) {
222 blank = strchr (test_author, ' ');
224 if (strcasestr (from, test_author) == NULL)
225 /* we didn't identify this as part of the email address
226 * so let's punt and return the original author */
227 strcpy (clean_author, author);
232 /* Add 'message' as a message that belongs to 'thread'.
234 * The 'thread' will talloc_steal the 'message' and hold onto a
238 _thread_add_message (notmuch_thread_t *thread,
239 notmuch_message_t *message,
240 notmuch_string_list_t *exclude_terms,
241 notmuch_exclude_t omit_exclude)
243 notmuch_tags_t *tags;
245 InternetAddressList *list = NULL;
246 InternetAddress *address;
247 const char *from, *author;
249 bool message_excluded = false;
251 if (omit_exclude != NOTMUCH_EXCLUDE_FALSE) {
252 for (tags = notmuch_message_get_tags (message);
253 notmuch_tags_valid (tags);
254 notmuch_tags_move_to_next (tags)) {
255 tag = notmuch_tags_get (tags);
256 /* Is message excluded? */
257 for (notmuch_string_node_t *term = exclude_terms->head;
260 /* Check for an empty string, and then ignore initial 'K'. */
261 if (*(term->string) && strcmp (tag, (term->string + 1)) == 0) {
262 message_excluded = true;
269 if (message_excluded && omit_exclude == NOTMUCH_EXCLUDE_ALL)
272 _notmuch_message_list_add_message (thread->message_list,
273 talloc_steal (thread, message));
274 thread->total_messages++;
275 thread->total_files += notmuch_message_count_files (message);
277 g_hash_table_insert (thread->message_hash,
278 xstrdup (notmuch_message_get_message_id (message)),
281 from = notmuch_message_get_header (message, "from");
283 list = internet_address_list_parse (NULL, from);
286 address = internet_address_list_get_address (list, 0);
288 author = internet_address_get_name (address);
289 /* We treat quoted empty names as if they were empty. */
290 if (author == NULL || author[0] == '\0') {
291 InternetAddressMailbox *mailbox;
292 mailbox = INTERNET_ADDRESS_MAILBOX (address);
293 author = internet_address_mailbox_get_addr (mailbox);
295 clean_author = _thread_cleanup_author (thread, author, from);
296 _thread_add_author (thread, clean_author);
297 _notmuch_message_set_author (message, clean_author);
299 g_object_unref (G_OBJECT (list));
302 if (! thread->subject) {
304 subject = notmuch_message_get_header (message, "subject");
305 thread->subject = talloc_strdup (thread, subject ? subject : "");
308 for (tags = notmuch_message_get_tags (message);
309 notmuch_tags_valid (tags);
310 notmuch_tags_move_to_next (tags)) {
311 tag = notmuch_tags_get (tags);
312 g_hash_table_insert (thread->tags, xstrdup (tag), NULL);
315 /* Mark excluded messages. */
316 if (message_excluded)
317 notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, true);
321 _thread_set_subject_from_message (notmuch_thread_t *thread,
322 notmuch_message_t *message)
325 const char *cleaned_subject;
327 subject = notmuch_message_get_header (message, "subject");
331 if ((strncasecmp (subject, "Re: ", 4) == 0) ||
332 (strncasecmp (subject, "Aw: ", 4) == 0) ||
333 (strncasecmp (subject, "Vs: ", 4) == 0) ||
334 (strncasecmp (subject, "Sv: ", 4) == 0)) {
336 cleaned_subject = talloc_strndup (thread,
338 strlen (subject) - 4);
340 cleaned_subject = talloc_strdup (thread, subject);
343 if (! EMPTY_STRING (cleaned_subject)) {
345 talloc_free (thread->subject);
347 thread->subject = talloc_strdup (thread, cleaned_subject);
351 /* Add a message to this thread which is known to match the original
352 * search specification. The 'sort' parameter controls whether the
353 * oldest or newest matching subject is applied to the thread as a
356 _thread_add_matched_message (notmuch_thread_t *thread,
357 notmuch_message_t *message,
361 notmuch_message_t *hashed_message;
363 date = notmuch_message_get_date (message);
365 if (date < thread->oldest || ! thread->matched_messages) {
366 thread->oldest = date;
367 if (sort == NOTMUCH_SORT_OLDEST_FIRST)
368 _thread_set_subject_from_message (thread, message);
371 if (date > thread->newest || ! thread->matched_messages) {
372 thread->newest = date;
373 const char *cur_subject = notmuch_thread_get_subject (thread);
374 if (sort != NOTMUCH_SORT_OLDEST_FIRST || EMPTY_STRING (cur_subject))
375 _thread_set_subject_from_message (thread, message);
378 if (! notmuch_message_get_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED))
379 thread->matched_messages++;
381 if (g_hash_table_lookup_extended (thread->message_hash,
382 notmuch_message_get_message_id (message), NULL,
383 (void **) &hashed_message)) {
384 notmuch_message_set_flag (hashed_message,
385 NOTMUCH_MESSAGE_FLAG_MATCH, 1);
388 _thread_add_matched_author (thread, _notmuch_message_get_author (hashed_message));
392 _parent_via_in_reply_to (notmuch_thread_t *thread, notmuch_message_t *message)
394 notmuch_message_t *parent;
395 const char *in_reply_to;
397 in_reply_to = _notmuch_message_get_in_reply_to (message);
398 THREAD_DEBUG ("checking message = %s in_reply_to=%s\n",
399 notmuch_message_get_message_id (message), in_reply_to);
401 if (in_reply_to && (! EMPTY_STRING (in_reply_to)) &&
402 g_hash_table_lookup_extended (thread->message_hash,
404 (void **) &parent)) {
405 _notmuch_message_add_reply (parent, message);
413 _parent_or_toplevel (notmuch_thread_t *thread, notmuch_message_t *message)
415 size_t max_depth = 0;
416 notmuch_message_t *new_parent;
417 notmuch_message_t *parent = NULL;
418 const notmuch_string_list_t *references =
419 _notmuch_message_get_references (message);
421 THREAD_DEBUG ("trying to reparent via references: %s\n",
422 notmuch_message_get_message_id (message));
424 for (notmuch_string_node_t *ref_node = references->head;
425 ref_node; ref_node = ref_node->next) {
426 THREAD_DEBUG ("checking reference=%s\n", ref_node->string);
427 if ((g_hash_table_lookup_extended (thread->message_hash,
428 ref_node->string, NULL,
429 (void **) &new_parent))) {
430 size_t new_depth = _notmuch_message_get_thread_depth (new_parent);
431 THREAD_DEBUG ("got depth %lu\n", new_depth);
432 if (new_depth > max_depth || ! parent) {
433 THREAD_DEBUG ("adding at depth %lu parent=%s\n", new_depth, ref_node->string);
434 max_depth = new_depth;
440 THREAD_DEBUG ("adding reply %s to parent=%s\n",
441 notmuch_message_get_message_id (message),
442 notmuch_message_get_message_id (parent));
443 _notmuch_message_add_reply (parent, message);
445 THREAD_DEBUG ("adding as toplevel %s\n",
446 notmuch_message_get_message_id (message));
447 _notmuch_message_list_add_message (thread->toplevel_list, message);
452 _resolve_thread_relationships (notmuch_thread_t *thread)
454 notmuch_message_node_t *node, *first_node;
455 notmuch_message_t *message;
457 notmuch_message_list_t *maybe_toplevel_list;
459 first_node = thread->message_list->head;
463 local = talloc_new (thread);
464 maybe_toplevel_list = _notmuch_message_list_create (local);
466 for (node = first_node->next; node; node = node->next) {
467 message = node->message;
468 if (! _parent_via_in_reply_to (thread, message))
469 _notmuch_message_list_add_message (maybe_toplevel_list, message);
473 * if we reach the end of the list without finding a top-level
474 * message, that means the thread is a cycle (or set of cycles)
475 * and any message can be considered top-level. Choose the oldest
476 * message, which happens to be first in our list.
479 message = first_node->message;
480 THREAD_DEBUG ("checking first message %s\n",
481 notmuch_message_get_message_id (message));
483 if (_notmuch_message_list_empty (maybe_toplevel_list) ||
484 ! _parent_via_in_reply_to (thread, message)) {
486 THREAD_DEBUG ("adding first message as toplevel = %s\n",
487 notmuch_message_get_message_id (message));
488 _notmuch_message_list_add_message (maybe_toplevel_list, message);
492 for (notmuch_messages_t *messages = _notmuch_messages_create (maybe_toplevel_list);
493 notmuch_messages_valid (messages);
494 notmuch_messages_move_to_next (messages)) {
495 notmuch_message_t *message = notmuch_messages_get (messages);
496 _notmuch_message_label_depths (message, 0);
499 for (notmuch_messages_t *roots = _notmuch_messages_create (maybe_toplevel_list);
500 notmuch_messages_valid (roots);
501 notmuch_messages_move_to_next (roots)) {
502 notmuch_message_t *message = notmuch_messages_get (roots);
503 if (_notmuch_messages_has_next (roots) || ! _notmuch_message_list_empty (thread->toplevel_list))
504 _parent_or_toplevel (thread, message);
506 _notmuch_message_list_add_message (thread->toplevel_list, message);
509 /* XXX this could be made conditional on messages being inserted
510 * (out of order) in later passes
512 thread->toplevel_list = _notmuch_message_sort_subtrees (thread, thread->toplevel_list);
517 /* Create a new notmuch_thread_t object by finding the thread
518 * containing the message with the given doc ID, treating any messages
519 * contained in match_set as "matched". Remove all messages in the
520 * thread from match_set.
522 * Creating the thread will perform a database search to get all
523 * messages belonging to the thread and will get the first subject
524 * line, the total count of messages, and all authors in the thread.
525 * Each message in the thread is checked against match_set to allow
526 * for a separate count of matched messages, and to allow a viewer to
527 * display these messages differently.
529 * Here, 'ctx' is talloc context for the resulting thread object.
531 * This function returns NULL in the case of any error.
534 _notmuch_thread_create (void *ctx,
535 notmuch_database_t *notmuch,
536 unsigned int seed_doc_id,
537 notmuch_doc_id_set_t *match_set,
538 notmuch_string_list_t *exclude_terms,
539 notmuch_exclude_t omit_excluded,
542 void *local = talloc_new (ctx);
543 notmuch_thread_t *thread = NULL;
544 notmuch_message_t *seed_message;
545 const char *thread_id;
546 char *thread_id_query_string;
547 notmuch_query_t *thread_id_query;
549 notmuch_messages_t *messages;
550 notmuch_message_t *message;
551 notmuch_status_t status;
553 seed_message = _notmuch_message_create (local, notmuch, seed_doc_id, NULL);
555 INTERNAL_ERROR ("Thread seed message %u does not exist", seed_doc_id);
557 thread_id = notmuch_message_get_thread_id (seed_message);
558 thread_id_query_string = talloc_asprintf (local, "thread:%s", thread_id);
559 if (unlikely (thread_id_query_string == NULL))
562 thread_id_query = talloc_steal (
563 local, notmuch_query_create (notmuch, thread_id_query_string));
564 if (unlikely (thread_id_query == NULL))
567 thread = talloc (local, notmuch_thread_t);
568 if (unlikely (thread == NULL))
571 talloc_set_destructor (thread, _notmuch_thread_destructor);
573 thread->notmuch = notmuch;
574 thread->thread_id = talloc_strdup (thread, thread_id);
575 thread->subject = NULL;
576 thread->authors_hash = g_hash_table_new_full (g_str_hash, g_str_equal,
578 thread->authors_array = g_ptr_array_new ();
579 thread->matched_authors_hash = g_hash_table_new_full (g_str_hash,
582 thread->matched_authors_array = g_ptr_array_new ();
583 thread->authors = NULL;
584 thread->tags = g_hash_table_new_full (g_str_hash, g_str_equal,
587 thread->message_list = _notmuch_message_list_create (thread);
588 thread->toplevel_list = _notmuch_message_list_create (thread);
589 if (unlikely (thread->message_list == NULL ||
590 thread->toplevel_list == NULL)) {
595 thread->message_hash = g_hash_table_new_full (g_str_hash, g_str_equal,
598 thread->total_messages = 0;
599 thread->total_files = 0;
600 thread->matched_messages = 0;
604 /* We use oldest-first order unconditionally here to obtain the
605 * proper author ordering for the thread. The 'sort' parameter
606 * passed to this function is used only to indicate whether the
607 * oldest or newest subject is desired. */
608 notmuch_query_set_sort (thread_id_query, NOTMUCH_SORT_OLDEST_FIRST);
610 status = notmuch_query_search_messages (thread_id_query, &messages);
615 notmuch_messages_valid (messages);
616 notmuch_messages_move_to_next (messages)) {
619 message = notmuch_messages_get (messages);
620 doc_id = _notmuch_message_get_doc_id (message);
621 if (doc_id == seed_doc_id)
622 message = seed_message;
624 _thread_add_message (thread, message, exclude_terms, omit_excluded);
626 if ( _notmuch_doc_id_set_contains (match_set, doc_id)) {
627 _notmuch_doc_id_set_remove (match_set, doc_id);
628 _thread_add_matched_message (thread, message, sort);
631 _notmuch_message_close (message);
634 _resolve_thread_authors_string (thread);
636 _resolve_thread_relationships (thread);
638 /* Commit to returning thread. */
639 (void) talloc_steal (ctx, thread);
647 notmuch_thread_get_toplevel_messages (notmuch_thread_t *thread)
649 return _notmuch_messages_create (thread->toplevel_list);
653 notmuch_thread_get_messages (notmuch_thread_t *thread)
655 return _notmuch_messages_create (thread->message_list);
659 notmuch_thread_get_thread_id (notmuch_thread_t *thread)
661 return thread->thread_id;
665 notmuch_thread_get_total_messages (notmuch_thread_t *thread)
667 return thread->total_messages;
671 notmuch_thread_get_total_files (notmuch_thread_t *thread)
673 return thread->total_files;
677 notmuch_thread_get_matched_messages (notmuch_thread_t *thread)
679 return thread->matched_messages;
683 notmuch_thread_get_authors (notmuch_thread_t *thread)
685 return thread->authors;
689 notmuch_thread_get_subject (notmuch_thread_t *thread)
691 return thread->subject;
695 notmuch_thread_get_oldest_date (notmuch_thread_t *thread)
697 return thread->oldest;
701 notmuch_thread_get_newest_date (notmuch_thread_t *thread)
703 return thread->newest;
707 notmuch_thread_get_tags (notmuch_thread_t *thread)
709 notmuch_string_list_t *tags;
712 tags = _notmuch_string_list_create (thread);
713 if (unlikely (tags == NULL))
716 keys = g_hash_table_get_keys (thread->tags);
718 for (l = keys; l; l = l->next)
719 _notmuch_string_list_append (tags, (char *) l->data);
723 _notmuch_string_list_sort (tags);
725 return _notmuch_tags_create (thread, tags);
729 notmuch_thread_destroy (notmuch_thread_t *thread)
731 talloc_free (thread);