Add some sort of progress indicator.

[notmuch] / notmuch-index-message.cc

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index 6795430b8c72aa8d8e8e34c6027c561851366a92..60370b62b99621f6d8ec1eea491d0b5b96c003ad 100644 (file)
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -17,11 +17,44 @@
  * Author: Carl Worth <cworth@cworth.org>
  */
 
+/* This indexer creates a Xapian mail index that is remarkably similar
+ * to that created by sup. The big difference, (and the thing that
+ * will keep a notmuch index from being used by sup directly), is that
+ * sup expects a serialized ruby data structure in the document's data
+ * field, but notmuch just puts the mail's filename there (trusting
+ * that the email client can get the data in needs from the filename).
+ *
+ * Note: One bug here is that sup actually merges together fields such
+ * as To, CC, Bcc etc. when finding multiple emails with the same
+ * message ID. To support something similar, notmuch should list
+ * multiple files in the data field.
+ *
+ * Other differences between sup and notmuch-index identified so far:
+ *
+ *   o sup supports encrypted mime parts by prompting for a passphrase
+ *     to decrypt the message. So far, notmuch doesn't support this,
+ *     both because I'm lazy to code it, and I also think doing so
+ *     would present a security leak.
+ *
+ *   o sup and notmuch have different heuristics for identifying (and
+ *     thus ignoring) signatures. For example, sup considers a line
+ *     consisting of two hypens as a signature separator, while
+ *     notmuch expects those two hyphens to be followed by a space
+ *     character.
+ *
+ *   o sup as been seen to split some numbers before indexing
+ *     them. For example, the number 1754 in an email message was
+ *     indexed by sup as separate terms 17 and 54. I couldn't find any
+ *     explanation for this behavior and did not try to replicate it
+ *     in notmuch.
+ */
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <errno.h>
 #include <time.h>
+#include <sys/time.h>
 
 #include <iostream>
 
@@ -172,6 +205,9 @@ gen_terms_address_names (Xapian::TermGenerator term_gen,
     int i;
     InternetAddress *address;
 
+    if (addresses == NULL)
+       return;
+
     for (i = 0; i < internet_address_list_length (addresses); i++) {
        address = internet_address_list_get_address (addresses, i);
        gen_terms_address_name (term_gen, address, address_type);
@@ -202,6 +238,9 @@ add_terms_address_addrs (Xapian::Document doc,
     int i;
     InternetAddress *address;
 
+    if (addresses == NULL)
+       return;
+
     for (i = 0; i < internet_address_list_length (addresses); i++) {
        address = internet_address_list_get_address (addresses, i);
        add_term_address_addr (doc, address, address_type);
@@ -379,8 +418,8 @@ gen_terms_body_str (Xapian::TermGenerator term_gen,
        }
        line_end = next_line - 1;
 
-       /* Trim whitespace. */
-       while (*next_line && isspace (*next_line))
+       /* Get to the next non-blank line. */
+       while (*next_line == '\n')
            next_line++;
 
        /* Skip lines that are quotes. */
@@ -439,11 +478,24 @@ gen_terms_part (Xapian::TermGenerator term_gen,
        return;
     }
 
-    disposition = g_mime_object_get_content_disposition (GMIME_OBJECT (part));
+    disposition = g_mime_object_get_content_disposition (part);
     if (disposition &&
        strcmp (disposition->disposition, GMIME_DISPOSITION_ATTACHMENT) == 0)
     {
+       const char *filename = g_mime_part_get_filename (GMIME_PART (part));
+       const char *extension;
+
        add_term (term_gen.get_document (), "label", "attachment");
+       gen_terms (term_gen, "attachment", filename);
+
+       if (filename) {
+           extension = strchr (filename, '.');
+           if (extension) {
+               add_term (term_gen.get_document (), "attachment_extension",
+                         extension + 1);
+           }
+       }
+
        return;
     }
 
@@ -452,7 +504,8 @@ gen_terms_part (Xapian::TermGenerator term_gen,
     stream = g_mime_stream_mem_new_with_byte_array (byte_array);
     g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE);
     wrapper = g_mime_part_get_content_object (GMIME_PART (part));
-    g_mime_data_wrapper_write_to_stream (wrapper, stream);
+    if (wrapper)
+       g_mime_data_wrapper_write_to_stream (wrapper, stream);
 
     g_object_unref (stream);
 
@@ -625,6 +678,8 @@ main (int argc, char **argv)
     GIOChannel *channel;
     GIOStatus gio_status;
     GError *error = NULL;
+    int count;
+    struct timeval tv_start, tv_now;
 
     if (argc < 2) {
        usage (argv[0]);
@@ -646,6 +701,10 @@ main (int argc, char **argv)
 
        channel = g_io_channel_unix_new (fileno (stdin));
 
+       count = 0;
+
+       gettimeofday (&tv_start, NULL);
+
        while (1) {
            gio_status = g_io_channel_read_line (channel, &filename,
                                                 NULL, NULL, &error);
@@ -661,6 +720,14 @@ main (int argc, char **argv)
            index_file (db, term_gen, filename);
 
            g_free (filename);
+
+           count++;
+           if (count % 250 == 0) {
+               gettimeofday (&tv_now, NULL);
+               printf ("Indexed %d messages (%g messages/second)\n",
+                       count, count / ((tv_now.tv_sec - tv_start.tv_sec) +
+                                       (tv_now.tv_usec - tv_start.tv_usec) / 1e6));
+           }
        }
 
     } catch (const Xapian::Error &error) {