Factor out parsing of reference-header values and pickup In-reply-to.

[notmuch] / notmuch-index-message.cc

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index 79b3bf3845c14ca1a1de487db68312daaad2f43e..b8fa205b9cc474d5fc11d795d308f9933ae15007 100644 (file)
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -207,13 +207,13 @@ skip_re_in_subject (const char *subject)
 /* 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))
@@ -228,9 +228,7 @@ add_terms_references (Xapian::Document doc,
        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;
     }
@@ -307,7 +305,9 @@ gen_terms_body (Xapian::TermGenerator term_gen,
 
        /* 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);
@@ -330,6 +330,7 @@ index_file (Xapian::WritableDatabase db,
     GMimeParser *parser;
     GMimeMessage *message;
     InternetAddressList *addresses;
+    GPtrArray *parents;
 
     FILE *file;
 
@@ -338,6 +339,7 @@ index_file (Xapian::WritableDatabase db,
     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) {
@@ -375,7 +377,20 @@ index_file (Xapian::WritableDatabase db,
     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", (const char *) g_ptr_array_index (parents, i));
+
+    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);
@@ -419,19 +434,30 @@ index_file (Xapian::WritableDatabase db,
     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);
 
@@ -444,7 +470,24 @@ main (int argc, char **argv)
 
        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;