X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=notmuch-index-message.cc;h=1922dd373856a32efb8c0f6f0ac4c857554da4ca;hp=a0faf442dbf9ef1c04910046e3560c90df50f7a8;hb=be72bf30703c5df814d4f572c3d9132d39700b36;hpb=c0da89a8e02190f172e6f53073827bbdf91d60eb

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index a0faf442..1922dd37 100644
--- a/notmuch-index-message.cc
+++ b/notmuch-index-message.cc
@@ -33,6 +33,9 @@ using namespace std;
 
 #define ARRAY_SIZE(arr) (sizeof (arr) / sizeof (arr[0]))
 
+/* Xapian complains if we provide a term longer than this. */
+#define NOTMUCH_MAX_TERM 245
+
 /* These prefix values are specifically chosen to be compatible
  * with sup, (http://sup.rubyforge.org), written by
  * William Morgan <wmorgan-sup@masanjin.net>, and released
@@ -109,7 +112,8 @@ add_term (Xapian::Document doc,
 
     term = g_strdup_printf ("%s%s", prefix, value);
 
-    doc.add_term (term);
+    if (strlen (term) <= NOTMUCH_MAX_TERM)
+	doc.add_term (term);
 
     g_free (term);
 }
@@ -204,122 +208,207 @@ skip_re_in_subject (const char *subject)
     return s;
 }
 
-/* Add a term for each message-id in the References header of the
- * message. */
 static void
-add_terms_references (Xapian::Document doc,
-		      GMimeMessage *message)
+find_messages_by_term (Xapian::Database db,
+		       const char *prefix_name,
+		       const char *value,
+		       Xapian::PostingIterator *begin,
+		       Xapian::PostingIterator *end)
 {
-    const char *refs, *end, *next;
+    Xapian::PostingIterator i;
     char *term;
 
-    refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+    term = g_strdup_printf ("%s%s", find_prefix (prefix_name), value);
 
-    if (refs == NULL)
-	return;
+    *begin = db.postlist_begin (term);
+
+    if (end)
+	*end = db.postlist_end (term);
+
+    free (term);
+}
+
+Xapian::Document
+find_message_by_docid (Xapian::Database db, Xapian::docid docid)
+{
+    return db.get_document (docid);
+}
+
+Xapian::Document
+find_message_by_message_id (Xapian::Database db, const char *message_id)
+{
+    Xapian::PostingIterator i, end;
+
+    find_messages_by_term (db, "msgid", message_id, &i, &end);
+
+    if (i != end)
+	return find_message_by_docid (db, *i);
+    else
+	return Xapian::Document ();
+}
+
+static void
+insert_thread_id (GHashTable *thread_ids, Xapian::Document doc)
+{
+    string value_string;
+    const char *value;
+
+    value_string = doc.get_value (NOTMUCH_VALUE_THREAD);
+    value = value_string.c_str();
+    if (strlen (value))
+	g_hash_table_insert (thread_ids, strdup (value), NULL);
+}
 
-    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) {
-	    term = g_strndup (refs, end - refs + 1);
-	    add_term (doc, "ref", term);
-	    g_free (term);
+static char *
+find_thread_id (Xapian::Database db,
+		GPtrArray *parents,
+		const char *message_id)
+{
+    Xapian::PostingIterator child, children_end;
+    Xapian::Document doc;
+    GHashTable *thread_ids;
+    GList *keys, *l;
+    GString *result = NULL;
+    unsigned int i;
+    const char *parent_message_id;
+
+    thread_ids = g_hash_table_new (g_str_hash, g_str_equal);
+
+    find_messages_by_term (db, "ref", message_id, &child, &children_end);
+    for ( ; child != children_end; child++) {
+	doc = find_message_by_docid (db, *child);
+	insert_thread_id (thread_ids, doc);
+    }
+
+    for (i = 0; i < parents->len; i++) {
+	parent_message_id = (char *) g_ptr_array_index (parents, i);
+	doc = find_message_by_message_id (db, parent_message_id);
+	insert_thread_id (thread_ids, doc);
+    }
+
+    keys = g_hash_table_get_keys (thread_ids);
+    for (l = keys; l; l = l->next) {
+	char *id = (char *) l->data;
+	if (result == NULL) {
+	    result = g_string_new (id);
+	} else {
+	    g_string_append_printf (result, ",%s", id);
 	}
-	refs = next;
+	free (id);
     }
+
+    if (result)
+	return g_string_free (result, FALSE);
+    else
+	return NULL;
 }
 
-/* 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). */
+/* Add a term for each message-id in the References header of the
+ * message. */
 static void
-gen_terms_body (Xapian::TermGenerator term_gen,
-		const char * filename,
-		gint64 body_offset)
+parse_references (GPtrArray *array,
+		  const char *refs_str)
 {
-    GIOChannel *channel;
-    GIOStatus gio_status;
-    GError *error = NULL;
-    char *p, *body_line = NULL, *prev_line = NULL;
+    GMimeReferences *refs, *r;
+    const char *message_id;
 
-    channel = g_io_channel_new_file (filename, "r", &error);
-    if (channel == NULL) {
-	fprintf (stderr, "Error: %s\n", error->message);
-	exit (1);
-    }
+    if (refs_str == NULL)
+	return;
 
-    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);
+    refs = g_mime_references_decode (refs_str);
+
+    for (r = refs; r; r = r->next) {
+	message_id = g_mime_references_get_message_id (r);
+	g_ptr_array_add (array, g_strdup (message_id));
     }
 
-    while (1) {
-	if (body_line)
-	    g_free (body_line);
+    g_mime_references_free (refs);
+}
+
+/* Given a string representing the body of a message, generate terms
+ * for it, (skipping quoted portions and signatures). */
+static void
+gen_terms_body_str (Xapian::TermGenerator term_gen,
+		    char *body)
+{
+    char *line, *line_end, *next_line;
 
-	gio_status = g_io_channel_read_line (channel, &body_line,
-					     NULL, NULL, &error);
-	if (gio_status == G_IO_STATUS_EOF)
+    if (body == NULL)
+	return;
+
+    next_line = body;
+
+    while (1) {
+	line = next_line;
+	if (*line == '\0')
 	    break;
-	if (gio_status != G_IO_STATUS_NORMAL) {
-	    fprintf (stderr, "Error: %s\n", error->message);
-	    exit (1);
+
+	next_line = strchr (line, '\n');
+	if (next_line == NULL) {
+	    next_line = line + strlen (line);
 	}
+	line_end = next_line - 1;
 
-	if (strlen (body_line) == 0)
-	    continue;
+	/* Trim whitespace. */
+	while (*next_line && isspace (*next_line))
+	    next_line++;
 
-	/* 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;
+	/* Skip lines that are quotes. */
+	if (*line == '>')
 	    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;
-	}
+	/* Also skip lines introducing a quote on the next line. */
+	while (line_end > line && isspace (*line_end))
+	    line_end--;
 
-	/* 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;
-	}
+	if (*line_end == ':' && *next_line == '>')
+	    continue;
 
-	/* Skip signatures */
+	/* Finally, bail as soon as we see a signature. */
 	/* XXX: Should only do this if "near" the end of the message. */
-	if (strncmp (body_line, "-- ", 3) == 0)
+	if (strncmp (line, "-- ", 3) == 0 ||
+	    strncmp (line, "----------", 10) == 0 ||
+	    strncmp (line, "__________", 10) == 0)
 	    break;
 
-	gen_terms (term_gen, "body", body_line);
+	*(line_end + 1) = '\0';
+	gen_terms (term_gen, "body", line);
     }
+}
 
-    if (body_line)
-	g_free (body_line);
 
-    g_io_channel_close (channel);
+/* Callback to generate terms for each mime part of a message. */
+static void
+gen_terms_part (GMimeObject *parent,
+		GMimeObject *part,
+		gpointer user_data)
+{
+    Xapian::TermGenerator *term_gen = (Xapian::TermGenerator *) user_data;
+    GMimeStream *stream;
+    GMimeDataWrapper *wrapper;
+    GByteArray *byte_array;
+    char *body;
+
+    if (! GMIME_IS_PART (part)) {
+	fprintf (stderr, "Warning: Not indexing unknown mime part: %s.\n",
+		 g_type_name (G_OBJECT_TYPE (part)));
+	return;
+    }
+
+    byte_array = g_byte_array_new ();
+
+    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);
+
+    g_object_unref (stream);
+
+    body = (char *) g_byte_array_free (byte_array, FALSE);
+
+    gen_terms_body_str (*term_gen, body);
+
+    free (body);
 }
 
 static void
@@ -333,14 +422,17 @@ index_file (Xapian::WritableDatabase db,
     GMimeParser *parser;
     GMimeMessage *message;
     InternetAddressList *addresses;
+    GPtrArray *parents;
 
     FILE *file;
 
-    const char *value, *from;
+    const char *subject, *refs, *in_reply_to, *from;
+    const char *message_id, *thread_id;
 
     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) {
@@ -370,15 +462,32 @@ index_file (Xapian::WritableDatabase db,
     addresses = g_mime_message_get_all_recipients (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);
+    subject = g_mime_message_get_subject (message);
+    subject = skip_re_in_subject (subject);
+    gen_terms (term_gen, "subject", subject);
+    gen_terms (term_gen, "body", subject);
+
+    g_mime_message_foreach (message, gen_terms_part, &term_gen);
+
+    parents = g_ptr_array_new ();
+
+    refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+    parse_references (parents, refs);
+
+    in_reply_to = g_mime_object_get_header (GMIME_OBJECT (message),
+					    "in-reply-to");
+    parse_references (parents, in_reply_to);
+
+    for (i = 0; i < parents->len; i++)
+	add_term (doc, "ref", (char *) g_ptr_array_index (parents, i));
+
+    message_id = g_mime_message_get_message_id (message);
 
-    gen_terms_body (term_gen, filename,
-		    g_mime_parser_get_headers_end (parser));
+    thread_id = find_thread_id (db, parents, message_id);
 
-    add_terms_references (doc, message);
+    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);
@@ -406,12 +515,18 @@ index_file (Xapian::WritableDatabase db,
     add_term (doc, "type", "mail");
     add_term (doc, "source_id", "1");
 
-    value = g_mime_message_get_message_id (message);
-    add_term (doc, "msgid", value);
-    add_term (doc, "thread", value);
-
-    doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, value);
-    doc.add_value (NOTMUCH_VALUE_THREAD, value);
+    add_term (doc, "msgid", message_id);
+    doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, message_id);
+
+    if (thread_id) {
+	add_term (doc, "thread", thread_id);
+	doc.add_value (NOTMUCH_VALUE_THREAD, thread_id);
+	free ((void *) thread_id);
+    } else {
+	/* If not referenced thread, use the message ID */
+	add_term (doc, "thread", message_id);
+	doc.add_value (NOTMUCH_VALUE_THREAD, message_id);
+    }
 
     doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));