#define ARRAY_SIZE(arr) (sizeof (arr) / sizeof (arr[0]))
+/* Xapian complains if we provide a term longer than this. */
+#define NOTMUCH_MAX_TERM 245
+
/* These prefix values are specifically chosen to be compatible
* with sup, (http://sup.rubyforge.org), written by
* William Morgan <wmorgan-sup@masanjin.net>, and released
term = g_strdup_printf ("%s%s", prefix, value);
- doc.add_term (term);
+ if (strlen (term) <= NOTMUCH_MAX_TERM)
+ doc.add_term (term);
g_free (term);
}
return s;
}
-/* Add a term for each message-id in the References header of the
- * message. */
static void
-add_terms_references (Xapian::Document doc,
- GMimeMessage *message)
+find_messages_by_term (Xapian::Database db,
+ const char *prefix_name,
+ const char *value,
+ Xapian::PostingIterator *begin,
+ Xapian::PostingIterator *end)
{
- const char *refs, *end, *next;
+ Xapian::PostingIterator i;
char *term;
- refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+ term = g_strdup_printf ("%s%s", find_prefix (prefix_name), value);
+
+ *begin = db.postlist_begin (term);
+
+ if (end)
+ *end = db.postlist_end (term);
+
+ free (term);
+}
+
+Xapian::Document
+find_message_by_docid (Xapian::Database db, Xapian::docid docid)
+{
+ return db.get_document (docid);
+}
+
+Xapian::Document
+find_message_by_message_id (Xapian::Database db, const char *message_id)
+{
+ Xapian::PostingIterator i, end;
+
+ find_messages_by_term (db, "msgid", message_id, &i, &end);
+
+ if (i != end)
+ return find_message_by_docid (db, *i);
+ else
+ return Xapian::Document ();
+}
+
+static void
+insert_thread_id (GHashTable *thread_ids, Xapian::Document doc)
+{
+ string value_string;
+ const char *value;
+
+ value_string = doc.get_value (NOTMUCH_VALUE_THREAD);
+ value = value_string.c_str();
+ if (strlen (value))
+ g_hash_table_insert (thread_ids, strdup (value), NULL);
+}
+
+static char *
+find_thread_id (Xapian::Database db,
+ GPtrArray *parents,
+ const char *message_id)
+{
+ Xapian::PostingIterator child, children_end;
+ Xapian::Document doc;
+ GHashTable *thread_ids;
+ GList *keys, *l;
+ GString *result = NULL;
+ unsigned int i;
+ const char *parent_message_id;
+
+ thread_ids = g_hash_table_new (g_str_hash, g_str_equal);
+
+ find_messages_by_term (db, "ref", message_id, &child, &children_end);
+ for ( ; child != children_end; child++) {
+ doc = find_message_by_docid (db, *child);
+ insert_thread_id (thread_ids, doc);
+ }
+
+ for (i = 0; i < parents->len; i++) {
+ parent_message_id = (char *) g_ptr_array_index (parents, i);
+ doc = find_message_by_message_id (db, parent_message_id);
+ insert_thread_id (thread_ids, doc);
+ }
+
+ keys = g_hash_table_get_keys (thread_ids);
+ for (l = keys; l; l = l->next) {
+ char *id = (char *) l->data;
+ if (result == NULL) {
+ result = g_string_new (id);
+ } else {
+ g_string_append_printf (result, ",%s", id);
+ }
+ free (id);
+ }
+
+ if (result)
+ return g_string_free (result, FALSE);
+ else
+ return NULL;
+}
+
+/* Add a term for each message-id in the References header of the
+ * message. */
+static void
+parse_references (GPtrArray *array,
+ const char *refs)
+{
+ const char *end, *next;
if (refs == NULL)
return;
if (end > refs && *end == '>')
end--;
if (end > refs) {
- term = g_strndup (refs, end - refs + 1);
- add_term (doc, "ref", term);
- g_free (term);
+ g_ptr_array_add (array, g_strndup (refs, end - refs + 1));
}
refs = next;
}
/* Skip signatures */
/* XXX: Should only do this if "near" the end of the message. */
- if (strncmp (body_line, "-- ", 3) == 0)
+ if (strncmp (body_line, "-- ", 3) == 0 ||
+ strncmp (body_line, "----------", 10) == 0 ||
+ strncmp (body_line, "__________", 10) == 0)
break;
gen_terms (term_gen, "body", body_line);
GMimeParser *parser;
GMimeMessage *message;
InternetAddressList *addresses;
+ GPtrArray *parents;
FILE *file;
- const char *value, *from;
+ const char *subject, *refs, *in_reply_to, *from;
+ const char *message_id, *thread_id;
time_t time;
struct tm gm_time_tm;
char date_str[16]; /* YYYYMMDDHHMMSS + 1 for Y100k compatibility ;-) */
+ unsigned int i;
file = fopen (filename, "r");
if (! file) {
addresses = g_mime_message_get_all_recipients (message);
gen_terms_address_names (term_gen, addresses, "to_name");
- value = g_mime_message_get_subject (message);
- value = skip_re_in_subject (value);
- gen_terms (term_gen, "subject", value);
- gen_terms (term_gen, "body", value);
+ subject = g_mime_message_get_subject (message);
+ subject = skip_re_in_subject (subject);
+ gen_terms (term_gen, "subject", subject);
+ gen_terms (term_gen, "body", subject);
gen_terms_body (term_gen, filename,
g_mime_parser_get_headers_end (parser));
- add_terms_references (doc, message);
+ parents = g_ptr_array_new ();
+
+ refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+ parse_references (parents, refs);
+
+ in_reply_to = g_mime_object_get_header (GMIME_OBJECT (message),
+ "in-reply-to");
+ parse_references (parents, in_reply_to);
+
+ for (i = 0; i < parents->len; i++)
+ add_term (doc, "ref", (char *) g_ptr_array_index (parents, i));
+
+ message_id = g_mime_message_get_message_id (message);
+
+ thread_id = find_thread_id (db, parents, message_id);
+
+ for (i = 0; i < parents->len; i++)
+ g_free (g_ptr_array_index (parents, i));
+ g_ptr_array_free (parents, TRUE);
from = g_mime_message_get_sender (message);
addresses = internet_address_list_parse_string (from);
add_term (doc, "type", "mail");
add_term (doc, "source_id", "1");
- value = g_mime_message_get_message_id (message);
- add_term (doc, "msgid", value);
- add_term (doc, "thread", value);
-
- doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, value);
- doc.add_value (NOTMUCH_VALUE_THREAD, value);
+ add_term (doc, "msgid", message_id);
+ doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, message_id);
+
+ if (thread_id) {
+ add_term (doc, "thread", thread_id);
+ doc.add_value (NOTMUCH_VALUE_THREAD, thread_id);
+ free ((void *) thread_id);
+ } else {
+ /* If not referenced thread, use the message ID */
+ add_term (doc, "thread", message_id);
+ doc.add_value (NOTMUCH_VALUE_THREAD, message_id);
+ }
doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
projects / notmuch / blobdiff