notmuch-index-message: Don't index the "re:" prefix in subjects.

[notmuch] / notmuch-index-message.cc

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index 39be64eca5156241b59c2733009b8be374ab32f5..3e81ecde9e24ada28ce5f4883818b1106b1947c3 100644 (file)
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -187,6 +187,107 @@ add_terms_address_addrs (Xapian::Document doc,
     }
 }
 
     }
 }
 

+static const char *
+skip_re_in_subject (const char *subject)
+{
+    const char *s = subject;
+
+    while (*s) {
+       while (*s && isspace (*s))
+           s++;
+       if (strncasecmp (s, "re:", 3) == 0)
+           s += 3;
+       else
+           break;
+    }
+
+    return s;
+}
+
+/* Generate terms for the body of a message, given the filename of the
+ * message and the offset at which the headers of the message end,
+ * (and hence the body begins). */
+static void
+gen_terms_body (Xapian::TermGenerator term_gen,
+               const char * filename,
+               gint64 body_offset)
+{
+    GIOChannel *channel;
+    GIOStatus gio_status;
+    GError *error = NULL;
+    char *p, *body_line = NULL, *prev_line = NULL;
+
+    channel = g_io_channel_new_file (filename, "r", &error);
+    if (channel == NULL) {
+       fprintf (stderr, "Error: %s\n", error->message);
+       exit (1);
+    }
+
+    gio_status = g_io_channel_seek_position (channel, body_offset,
+                                            G_SEEK_SET, &error);
+    if (gio_status != G_IO_STATUS_NORMAL) {
+       fprintf (stderr, "Error: %s\n", error->message);
+       exit (1);
+    }
+
+    while (1) {
+       if (body_line)
+           g_free (body_line);
+
+       gio_status = g_io_channel_read_line (channel, &body_line,
+                                            NULL, NULL, &error);
+       if (gio_status == G_IO_STATUS_EOF)
+           break;
+       if (gio_status != G_IO_STATUS_NORMAL) {
+           fprintf (stderr, "Error: %s\n", error->message);
+           exit (1);
+       }
+
+       if (strlen (body_line) == 0)
+           continue;
+
+       /* If the line looks like it might be introducing a quote,
+        * save it until we see if the next line begins a quote. */
+       p = body_line + strlen (body_line) - 1;
+       while (p > body_line and isspace (*p))
+           p--;
+       if (*p == ':') {
+           prev_line = body_line;
+           body_line = NULL;
+           continue;
+       }
+
+       /* Skip quoted lines, (and previous lines that introduced them) */
+       if (body_line[0] == '>') {
+           if (prev_line) {
+               g_free (prev_line);
+               prev_line = NULL;
+           }
+           continue;
+       }
+
+       /* Now that we're not looking at a quote we can add the prev_line */
+       if (prev_line) {
+           gen_terms (term_gen, "body", prev_line);
+           g_free (prev_line);
+           prev_line = NULL;
+       }
+
+       /* Skip signatures */
+       /* XXX: Should only do this if "near" the end of the message. */
+       if (strncmp (body_line, "-- ", 3) == 0)
+           break;
+
+       gen_terms (term_gen, "body", body_line);
+    }
+
+    if (body_line)
+       g_free (body_line);
+
+    g_io_channel_close (channel);
+}
+
+

 int
 main (int argc, char **argv)
 {
 int
 main (int argc, char **argv)
 {


@@ -194,14 +295,9 @@ main (int argc, char **argv)
     GMimeParser *parser;
     GMimeMessage *message;
     InternetAddressList *addresses;
     GMimeParser *parser;
     GMimeMessage *message;
     InternetAddressList *addresses;

-    GIOChannel *channel;
-    GIOStatus gio_status;
-    GError *error = NULL;

 
     const char *database_path, *filename;
     FILE *file;
 
     const char *database_path, *filename;
     FILE *file;

-    gint64 body_offset;
-    char *body_str;

 
     const char *value, *from;
 
 
     const char *value, *from;
 


@@ -258,34 +354,12 @@ main (int argc, char **argv)
        gen_terms_address_names (term_gen, addresses, "to_name");
 
        value = g_mime_message_get_subject (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);
 
        gen_terms (term_gen, "subject", value);
        gen_terms (term_gen, "body", value);
 

-       body_offset = g_mime_parser_get_headers_end (parser);
-       channel = g_io_channel_new_file (filename, "r", &error);
-       if (channel == NULL) {
-           fprintf (stderr, "Error: %s\n", error->message);
-           exit (1);
-       }
-
-       gio_status = g_io_channel_seek_position (channel, body_offset,
-                                                G_SEEK_SET, &error);
-       if (gio_status != G_IO_STATUS_NORMAL) {
-           fprintf (stderr, "Error: %s\n", error->message);
-           exit (1);
-       }
-
-       gio_status = g_io_channel_read_to_end (channel, &body_str,
-                                              NULL, &error);
-       if (gio_status != G_IO_STATUS_NORMAL) {
-           fprintf (stderr, "Error: %s\n", error->message);
-           exit (1);
-       }
-
-       gen_terms (term_gen, "body", body_str);
-
-       g_free (body_str);
-       g_io_channel_close (channel);
+       gen_terms_body (term_gen, filename,
+                       g_mime_parser_get_headers_end (parser));

 
        from = g_mime_message_get_sender (message);
        addresses = internet_address_list_parse_string (from);
 
        from = g_mime_message_get_sender (message);
        addresses = internet_address_list_parse_string (from);