X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=notmuch-index-message.cc;h=efc7eb14cda4f2f6c386365443909857110149a2;hp=636757a165ac68e47a56af69817c25496afa1123;hb=48d2e2dc440d49d7305324c25460e98d5fa40bda;hpb=29974af08f2eb3ebdff3ae1052a07b6d7361d511

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index 636757a1..efc7eb14 100644
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -17,6 +17,38 @@
  * 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>
@@ -172,6 +204,9 @@ 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 +237,9 @@ 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 +417,8 @@ 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. */
@@ -443,9 +481,20 @@ gen_terms_part (Xapian::TermGenerator term_gen,
     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",
-		  g_mime_part_get_filename (GMIME_PART (part)));
+	gen_terms (term_gen, "attachment", filename);
+
+	if (filename) {
+	    extension = strchr (filename, '.');
+	    if (extension) {
+		add_term (term_gen.get_document (), "attachment_extension",
+			  extension + 1);
+	    }
+	}
+
 	return;
     }