FIX: POSIX: query not handling correctly file names with non ASCII …

…chars resolves: Oldes/Rebol-issues#2442
Oldes · Jan 27, 2021 · 040c9c6 · 040c9c6
1 parent 2f20ab8
commit 040c9c6
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 0 deletions.
diff --git a/src/core/p-file.c b/src/core/p-file.c
@@ -108,6 +108,13 @@
 **
 ***********************************************************************/
 {
+#if !defined(TO_WINDOWS)
+ // on Posix the file.path is in UTF8 format, so must be decoded bellow
+ REBSER *ser = BUF_UTF8;
+ REBINT len;
+ const REBYTE *path;
+#endif
+
  switch (mode) {
  case SYM_SIZE:
  if(file->file.size == MIN_I64) {
@@ -123,7 +130,15 @@
  Set_File_Date(file, ret);
  break;
  case SYM_NAME:
+#ifdef TO_WINDOWS
  Set_Series(REB_FILE, ret, To_REBOL_Path(file->file.path, 0, OS_WIDE, 0));
+#else
+ path = cb_cast(file->file.path);
+ len = (REBINT)LEN_BYTES(path);
+ len = Decode_UTF8(UNI_HEAD(ser), path, len, 0);
+ if (len < 0) len = -len; // negative len means ASCII chars only
+ Set_Series(REB_FILE, ret, To_REBOL_Path(UNI_HEAD(ser), len, TRUE, 0));
+#endif
  break;
  default:
  return FALSE;

diff --git a/src/tests/units/files/čeština.txt b/src/tests/units/files/čeština.txt
@@ -0,0 +1 @@
+Čeština je flektivní jazyk vyznačující se komplikovaným systémem skloňování a časování a velmi volným slovosledem. K písemným záznamům používá latinku doplněnou o znaky s diakritikou. 
diff --git a/src/tests/units/port-test.r3 b/src/tests/units/port-test.r3
@@ -202,6 +202,11 @@ if system/platform = 'Windows [
  --assert [file 51732] = query/mode file [type size]
  --assert [type: file size: 51732] = query/mode file [type: size:]
 
+ --test-- "query file name"
+ ;@@ https://github.com/Oldes/Rebol-issues/issues/2442
+ file: %units/files/čeština.txt
+ --assert not none? find (query/mode file 'name) file
+
  --test-- "query file info (port)"
  file: open %units/files/alice29.txt.gz
  --assert [name size date type] = query/mode file none