#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;
}
+Xapian::Document
+find_message (Xapian::Database db, const char *message_id)
+{
+ Xapian::PostingIterator i;
+ char *term;
+
+ term = g_strdup_printf ("%s%s", find_prefix ("msgid"), message_id);
+ i = db.postlist_begin (term);
+ if (i != db.postlist_end (term))
+ return db.get_document (*i);
+ else
+ return Xapian::Document ();
+}
+
+static char *
+find_thread_id (Xapian::Database db, GPtrArray *parents)
+{
+ Xapian::Document doc;
+ GHashTable *thread_ids;
+ GList *keys, *l;
+ GString *result = NULL;
+ unsigned int i;
+ string value_string;
+ const char *value;
+
+ thread_ids = g_hash_table_new (g_str_hash, g_str_equal);
+
+ for (i = 0; i < parents->len; i++) {
+ doc = find_message (db, (char *) g_ptr_array_index (parents, i));
+ 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);
+ }
+
+ 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
-add_terms_references (Xapian::Document doc,
- GMimeMessage *message)
+parse_references (GPtrArray *array,
+ const char *refs)
{
- const char *refs, *end, *next;
- char *term;
+ const char *end, *next;
- refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+ if (refs == NULL)
+ return;
while (*refs) {
while (*refs && isspace (*refs))
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 *value, *from, *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) {
gen_terms_body (term_gen, filename,
g_mime_parser_get_headers_end (parser));
- add_terms_references (doc, message);
+ parents = g_ptr_array_new ();
+
+ value = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+ parse_references (parents, value);
+
+ value = g_mime_object_get_header (GMIME_OBJECT (message), "in-reply-to");
+ parse_references (parents, value);
+
+ for (i = 0; i < parents->len; i++)
+ add_term (doc, "ref", (char *) g_ptr_array_index (parents, i));
+
+ thread_id = find_thread_id (db, parents);
+
+ 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);
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);
+
+ 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", value);
+ doc.add_value (NOTMUCH_VALUE_THREAD, value);
+ }
doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
g_object_unref (stream);
}
+static void
+usage (const char *argv0)
+{
+ fprintf (stderr, "Usage: %s <path-to-xapian-database>\n", argv0);
+ fprintf (stderr, "\n");
+ fprintf (stderr, "Messages to be indexed are read from stdnin as absolute filenames\n");
+ fprintf (stderr, "one file per line.");
+}
+
int
main (int argc, char **argv)
{
- const char *database_path, *filename;
+ const char *database_path;
+ char *filename;
+ GIOChannel *channel;
+ GIOStatus gio_status;
+ GError *error = NULL;
- if (argc < 3) {
- fprintf (stderr, "Usage: %s <path-to-xapian-database> <mail-message>\n",
- argv[0]);
+ if (argc < 2) {
+ usage (argv[0]);
exit (1);
}
database_path = argv[1];
- filename = argv[2];
g_mime_init (0);
term_gen = Xapian::TermGenerator ();
- index_file (db, term_gen, filename);
+ channel = g_io_channel_unix_new (fileno (stdin));
+
+ while (1) {
+ gio_status = g_io_channel_read_line (channel, &filename,
+ NULL, NULL, &error);
+ if (gio_status == G_IO_STATUS_EOF)
+ break;
+ if (gio_status != G_IO_STATUS_NORMAL) {
+ fprintf (stderr, "An error occurred reading from stdin: %s\n",
+ error->message);
+ exit (1);
+ }
+
+ g_strchomp (filename);
+ index_file (db, term_gen, filename);
+
+ g_free (filename);
+ }
} catch (const Xapian::Error &error) {
cerr << "A Xapian exception occurred: " << error.get_msg () << endl;
projects / notmuch / blobdiff