X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=notmuch-index-message.cc;h=bb393641af1da745841f108ec6022f28e4904f9a;hp=dedbff37dec206269e09684f64edcdd12a33ae4e;hb=6363ab32eab64c1618f8d257953f31877d09b2a9;hpb=048b8aec111188385a806c7ebc2930e7b046f8cb

diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc
index dedbff37..bb393641 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);
 }
@@ -135,11 +139,29 @@ gen_terms_address_name (Xapian::TermGenerator term_gen,
 			const char *prefix_name)
 {
     const char *name;
+    int own_name = 0;
 
     name = internet_address_get_name (address);
 
+    /* In the absence of a name, we'll strip the part before the @
+     * from the address. */
+    if (! name) {
+	InternetAddressMailbox *mailbox = INTERNET_ADDRESS_MAILBOX (address);
+	const char *addr = internet_address_mailbox_get_addr (mailbox);
+	const char *at;
+
+	at = strchr (addr, '@');
+	if (at) {
+	    name = strndup (addr, at - addr);
+	    own_name = 1;
+	}
+    }
+
     if (name)
 	gen_terms (term_gen, prefix_name, name);
+
+    if (own_name)
+	free ((void *) name);
 }
 
 static void
@@ -187,86 +209,246 @@ add_terms_address_addrs (Xapian::Document doc,
     }
 }
 
-/* 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 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;
+}
+
 static void
-gen_terms_body (Xapian::TermGenerator term_gen,
-		const char * filename,
-		gint64 body_offset)
+find_messages_by_term (Xapian::Database db,
+		       const char *prefix_name,
+		       const char *value,
+		       Xapian::PostingIterator *begin,
+		       Xapian::PostingIterator *end)
 {
-    GIOChannel *channel;
-    GIOStatus gio_status;
-    GError *error = NULL;
-    char *body_line = NULL;
+    Xapian::PostingIterator i;
+    char *term;
 
-    channel = g_io_channel_new_file (filename, "r", &error);
-    if (channel == NULL) {
-	fprintf (stderr, "Error: %s\n", error->message);
-	exit (1);
+    term = g_strdup_printf ("%s%s", find_prefix (prefix_name), value);
+
+    *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);
+}
+
+/* Return one or more thread_ids, (as a GPtrArray of strings), for the
+ * given message based on looking into the database for any messages
+ * referenced in parents, and also for any messages in the database
+ * referencing message_id.
+ *
+ * Caller should free all strings in the array and the array itself,
+ * (g_ptr_array_free) when done. */
+static GPtrArray *
+find_thread_ids (Xapian::Database db,
+		 GPtrArray *parents,
+		 const char *message_id)
+{
+    Xapian::PostingIterator child, children_end;
+    Xapian::Document doc;
+    GHashTable *thread_ids;
+    GList *keys, *l;
+    unsigned int i;
+    const char *parent_message_id;
+    GPtrArray *result;
+
+    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);
     }
 
-    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);
+    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);
     }
 
-    while (1) {
-	if (body_line)
-	    g_free (body_line);
+    result = g_ptr_array_new ();
+
+    keys = g_hash_table_get_keys (thread_ids);
+    for (l = keys; l; l = l->next) {
+	char *id = (char *) l->data;
+	g_ptr_array_add (result, id);
+    }
+
+    return result;
+}
+
+/* Add a term for each message-id in the References header of the
+ * message. */
+static void
+parse_references (GPtrArray *array,
+		  const char *refs_str)
+{
+    GMimeReferences *refs, *r;
+    const char *message_id;
+
+    if (refs_str == NULL)
+	return;
+
+    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));
+    }
+
+    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;
+
+    if (body == NULL)
+	return;
 
-	gio_status = g_io_channel_read_line (channel, &body_line,
-					     NULL, NULL, &error);
-	if (gio_status == G_IO_STATUS_EOF)
+    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;
+
+	/* Trim whitespace. */
+	while (*next_line && isspace (*next_line))
+	    next_line++;
 
-	if (body_line[0] == '>')
+	/* Skip lines that are quotes. */
+	if (*line == '>')
 	    continue;
 
-	if (strncmp (body_line, "-- ", 3) == 0)
+	/* Also skip lines introducing a quote on the next line. */
+	if (*line_end == ':' && *next_line == '>')
+	    continue;
+
+	/* Finally, bail as soon as we see a signature. */
+	/* XXX: Should only do this if "near" the end of the message. */
+	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;
+    }
 
-int
-main (int argc, char **argv)
+    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
+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, *thread_ids;
 
-    const char *database_path, *filename;
     FILE *file;
 
-    const char *value, *from;
+    const char *subject, *refs, *in_reply_to, *from;
+    const char *message_id;
 
     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) {
@@ -274,93 +456,180 @@ 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");
+    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);
 
-	addresses = g_mime_message_get_all_recipients (message);
-	gen_terms_address_names (term_gen, addresses, "to_name");
+    g_mime_message_foreach (message, gen_terms_part, &term_gen);
 
-	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));
+    refs = g_mime_object_get_header (GMIME_OBJECT (message), "references");
+    parse_references (parents, refs);
 
-	from = g_mime_message_get_sender (message);
-	addresses = internet_address_list_parse_string (from);
+    in_reply_to = g_mime_object_get_header (GMIME_OBJECT (message),
+					    "in-reply-to");
+    parse_references (parents, in_reply_to);
 
-	add_terms_address_addrs (doc, addresses, "from_email");
+    for (i = 0; i < parents->len; i++)
+	add_term (doc, "ref", (char *) g_ptr_array_index (parents, i));
 
-	add_terms_address_addrs (doc,
-				 g_mime_message_get_all_recipients (message),
-				 "to_email");
+    message_id = g_mime_message_get_message_id (message);
 
-	g_mime_message_get_date (message, &time, NULL);
+    thread_ids = find_thread_ids (db, parents, message_id);
 
-	gmtime_r (&time, &gm_time_tm);
+    for (i = 0; i < parents->len; i++)
+	g_free (g_ptr_array_index (parents, i));
+    g_ptr_array_free (parents, TRUE);
 
-	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);
-	}
+    from = g_mime_message_get_sender (message);
+    addresses = internet_address_list_parse_string (from);
 
-	add_term (doc, "date", date_str);
+    add_terms_address_addrs (doc, addresses, "from_email");
 
-	add_term (doc, "label", "inbox");
-	add_term (doc, "label", "unread");
-	add_term (doc, "type", "mail");
+    add_terms_address_addrs (doc,
+			     g_mime_message_get_all_recipients (message),
+			     "to_email");
 
-	value = g_mime_message_get_message_id (message);
-	add_term (doc, "msgid", value);
+    g_mime_message_get_date (message, &time, NULL);
 
-	add_term (doc, "source_id", "1");
+    gmtime_r (&time, &gm_time_tm);
 
-	add_term (doc, "thread", 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);
+    }
 
-	doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, value);
-	doc.add_value (NOTMUCH_VALUE_THREAD, value);
+    add_term (doc, "date", date_str);
 
-	doc.add_value (NOTMUCH_VALUE_DATE, Xapian::sortable_serialise (time));
+    add_term (doc, "label", "inbox");
+    add_term (doc, "label", "unread");
+    add_term (doc, "type", "mail");
+    add_term (doc, "source_id", "1");
 
-	db.add_document (doc);
+    add_term (doc, "msgid", message_id);
+    doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, message_id);
 
-    } catch (const Xapian::Error &error) {
-	cerr << "A Xapian exception occurred: " << error.get_msg () << endl;
-	exit (1);
+    if (thread_ids->len) {
+	unsigned int i;
+	GString *thread_id;
+	char *id;
+
+	for (i = 0; i < thread_ids->len; i++) {
+	    id = (char *) thread_ids->pdata[i];
+
+	    add_term (doc, "thread", id);
+
+	    if (i == 0)
+		thread_id = g_string_new (id);
+	    else
+		g_string_append_printf (thread_id, ",%s", id);
+
+	    free (id);
+	}
+	g_ptr_array_free (thread_ids, TRUE);
+
+	doc.add_value (NOTMUCH_VALUE_THREAD, thread_id->str);
+
+	g_string_free (thread_id, TRUE);
+    } 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));
+
+    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;
 }