X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=notmuch-index-message.cc;h=b8fa205b9cc474d5fc11d795d308f9933ae15007;hp=b7a0d3e7bb37a8c440819def7d40f4ee4ceec8d2;hb=5cbdcbbec58248ef803ba8f3f035e1c49b7ad0a8;hpb=9dbb1facfb1af5f16b3311a75593693ec04fa6a6

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index b7a0d3e7..b8fa205b 100644
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -187,6 +187,53 @@ 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;
+}
+
+/* Add a term for each message-id in the References header of the
+ * message. */
+static void
+parse_references (GPtrArray *array,
+		  const char *refs)
+{
+    const char *end, *next;
+
+    if (refs == NULL)
+	return;
+
+    while (*refs) {
+	while (*refs && isspace (*refs))
+	    refs++;
+	if (*refs == '<')
+	    refs++;
+	end = refs;
+	while (*end && !isspace (*end))
+	    end++;
+	next = end;
+	end--;
+	if (end > refs && *end == '>')
+	    end--;
+	if (end > refs) {
+	    g_ptr_array_add (array, g_strndup (refs, end - refs + 1));
+	}
+	refs = next;
+    }
+}
+
 /* 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). */
@@ -198,7 +245,7 @@ gen_terms_body (Xapian::TermGenerator term_gen,
     GIOChannel *channel;
     GIOStatus gio_status;
     GError *error = NULL;
-    char *body_str;
+    char *p, *body_line = NULL, *prev_line = NULL;
 
     channel = g_io_channel_new_file (filename, "r", &error);
     if (channel == NULL) {
@@ -213,29 +260,78 @@ gen_terms_body (Xapian::TermGenerator term_gen,
 	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);
+    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 ||
+	    strncmp (body_line, "----------", 10) == 0 ||
+	    strncmp (body_line, "__________", 10) == 0)
+	    break;
+
+	gen_terms (term_gen, "body", body_line);
     }
 
-    gen_terms (term_gen, "body", body_str);
+    if (body_line)
+	g_free (body_line);
 
-    g_free (body_str);
     g_io_channel_close (channel);
 }
 
-
-int
-main (int argc, char **argv)
+static void
+index_file (Xapian::WritableDatabase db,
+	    Xapian::TermGenerator term_gen,
+	    const char *filename)
 {
+    Xapian::Document doc;
+
     GMimeStream *stream;
     GMimeParser *parser;
     GMimeMessage *message;
     InternetAddressList *addresses;
+    GPtrArray *parents;
 
-    const char *database_path, *filename;
     FILE *file;
 
     const char *value, *from;
@@ -243,15 +339,7 @@ main (int argc, char **argv)
     time_t time;
     struct tm gm_time_tm;
     char date_str[16]; /* YYYYMMDDHHMMSS + 1 for Y100k compatibility ;-) */
-
-    if (argc < 3) {
-	fprintf (stderr, "Usage: %s <path-to-xapian-database> <mail-message>\n",
-		 argv[0]);
-	exit (1);
-    }
-
-    database_path = argv[1];
-    filename = argv[2];
+    unsigned int i;
 
     file = fopen (filename, "r");
     if (! file) {
@@ -259,93 +347,152 @@ main (int argc, char **argv)
 	exit (1);
     }
 
-    g_mime_init (0);
-
     stream = g_mime_stream_file_new (file);
 
     parser = g_mime_parser_new_with_stream (stream);
 
     message = g_mime_parser_construct_message (parser);
 
-    try {
-	Xapian::WritableDatabase db;
-	Xapian::TermGenerator term_gen;
-	Xapian::Document doc;
+    doc = Xapian::Document ();
 
-	doc = Xapian::Document ();
+    doc.set_data (filename);
 
-	doc.set_data (filename);
+    term_gen.set_stemmer (Xapian::Stem ("english"));
 
-	db = Xapian::WritableDatabase (database_path,
-				       Xapian::DB_CREATE_OR_OPEN);
+    term_gen.set_document (doc);
 
-	term_gen = Xapian::TermGenerator ();
-	term_gen.set_stemmer (Xapian::Stem ("english"));
+    from = g_mime_message_get_sender (message);
+    addresses = internet_address_list_parse_string (from);
 
-	term_gen.set_document (doc);
+    gen_terms_address_names (term_gen, addresses, "from_name");
 
-	from = g_mime_message_get_sender (message);
-	addresses = internet_address_list_parse_string (from);
+    addresses = g_mime_message_get_all_recipients (message);
+    gen_terms_address_names (term_gen, addresses, "to_name");
 
-	gen_terms_address_names (term_gen, addresses, "from_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);
 
-	addresses = g_mime_message_get_all_recipients (message);
-	gen_terms_address_names (term_gen, addresses, "to_name");
+    gen_terms_body (term_gen, filename,
+		    g_mime_parser_get_headers_end (parser));
 
-	value = g_mime_message_get_subject (message);
-	gen_terms (term_gen, "subject", value);
-	gen_terms (term_gen, "body", value);
+    parents = g_ptr_array_new ();
 
-	gen_terms_body (term_gen, filename,
-			g_mime_parser_get_headers_end (parser));
+    value = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+    parse_references (parents, value);
 
-	from = g_mime_message_get_sender (message);
-	addresses = internet_address_list_parse_string (from);
+    value = g_mime_object_get_header (GMIME_OBJECT (message), "in-reply-to");
+    parse_references (parents, value);
 
-	add_terms_address_addrs (doc, addresses, "from_email");
+    for (i = 0; i < parents->len; i++)
+	add_term (doc, "ref", (const char *) g_ptr_array_index (parents, i));
 
-	add_terms_address_addrs (doc,
-				 g_mime_message_get_all_recipients (message),
-				 "to_email");
+    for (i = 0; i < parents->len; i++)
+	g_free (g_ptr_array_index (parents, i));
+    g_ptr_array_free (parents, TRUE);
 
-	g_mime_message_get_date (message, &time, NULL);
+    from = g_mime_message_get_sender (message);
+    addresses = internet_address_list_parse_string (from);
 
-	gmtime_r (&time, &gm_time_tm);
+    add_terms_address_addrs (doc, addresses, "from_email");
 
-	if (strftime (date_str, sizeof (date_str),
-		      "%Y%m%d%H%M%S", &gm_time_tm) == 0) {
-	    fprintf (stderr, "Internal error formatting time\n");
-	    exit (1);
-	}
+    add_terms_address_addrs (doc,
+			     g_mime_message_get_all_recipients (message),
+			     "to_email");
 
-	add_term (doc, "date", date_str);
+    g_mime_message_get_date (message, &time, NULL);
 
-	add_term (doc, "label", "inbox");
-	add_term (doc, "label", "unread");
-	add_term (doc, "type", "mail");
+    gmtime_r (&time, &gm_time_tm);
 
-	value = g_mime_message_get_message_id (message);
-	add_term (doc, "msgid", value);
+    if (strftime (date_str, sizeof (date_str),
+		  "%Y%m%d%H%M%S", &gm_time_tm) == 0) {
+	fprintf (stderr, "Internal error formatting time\n");
+	exit (1);
+    }
 
-	add_term (doc, "source_id", "1");
+    add_term (doc, "date", date_str);
 
-	add_term (doc, "thread", value);
+    add_term (doc, "label", "inbox");
+    add_term (doc, "label", "unread");
+    add_term (doc, "type", "mail");
+    add_term (doc, "source_id", "1");
 
-	doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, value);
-	doc.add_value (NOTMUCH_VALUE_THREAD, value);
+    value = g_mime_message_get_message_id (message);
+    add_term (doc, "msgid", value);
+    add_term (doc, "thread", value);
 
-	doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
+    doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, value);
+    doc.add_value (NOTMUCH_VALUE_THREAD, value);
 
-	db.add_document (doc);
+    doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
 
-    } catch (const Xapian::Error &error) {
-	cerr << "A Xapian exception occurred: " << error.get_msg () << endl;
-	exit (1);
-    }
+    db.add_document (doc);
 
     g_object_unref (message);
     g_object_unref (parser);
     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;
+    char *filename;
+    GIOChannel *channel;
+    GIOStatus gio_status;
+    GError *error = NULL;
+
+    if (argc < 2) {
+	usage (argv[0]);
+	exit (1);
+    }
+
+    database_path = argv[1];
+
+    g_mime_init (0);
+
+    try {
+	Xapian::WritableDatabase db;
+	Xapian::TermGenerator term_gen;
+
+	db = Xapian::WritableDatabase (database_path,
+				       Xapian::DB_CREATE_OR_OPEN);
+
+	term_gen = Xapian::TermGenerator ();
+
+	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;
+	exit (1);
+    }
 
     return 0;
 }