utf32_word_split() and utf8_word_split() splits a string into words

[disorder] / server / rescan.c

diff --git a/server/rescan.c b/server/rescan.c
index 890e3fe9a6b746369d541697e42470632b78b700..cc31888ab91be9334aed228fba4322a34e68ab88 100644 (file)
--- a/server/rescan.c
+++ b/server/rescan.c
@@ -50,6 +50,7 @@
 #include "trackdb.h"
 #include "trackdb-int.h"
 #include "trackname.h"
+#include "unicode.h"
 
 static DB_TXN *global_tid;
 
@@ -151,6 +152,13 @@ static void rescan_collection(const struct collection *c) {
       error(0, "cannot convert track path to UTF-8: %s", path);
       continue;
     }
+    if(config->dbversion > 1) {
+      /* We use NFC track names */
+      if(!(track = utf8_compose_canon(track, strlen(track), 0))) {
+        error(0, "cannot convert track path to NFC: %s", path);
+        continue;
+      }
+    }
     D(("track %s", track));
     /* only tracks with a known player are admitted */
     for(n = 0; (n < config->player.n