[PATCH 02 of 12] Win32: non-ASCII names support in "include" with wildcards

[Maxim Dounin]

# HG changeset patch
# User Maxim Dounin <mdounin at mdounin.ru>
# Date 1673548899 -10800
#      Thu Jan 12 21:41:39 2023 +0300
# Node ID d05c0adf5890aecc68ce8906ef19ca07502ed071
# Parent  60d845f9505fe1b97c1e04b680523b790e29fdb1
Win32: non-ASCII names support in "include" with wildcards.

Notably, ngx_open_glob() now supports opening directories with non-ASCII
characters, and pathnames returned by ngx_read_glob() are converted to UTF-8.

diff -r 60d845f9505f -r d05c0adf5890 src/os/win32/ngx_files.c
--- a/src/os/win32/ngx_files.c	Thu Jan 12 21:41:30 2023 +0300
+++ b/src/os/win32/ngx_files.c	Thu Jan 12 21:41:39 2023 +0300
@@ -546,14 +546,27 @@ ngx_open_glob(ngx_glob_t *gl)
 {
     u_char     *p;
     size_t      len;
+    u_short    *u;
     ngx_err_t   err;
+    u_short     utf16[NGX_UTF16_BUFLEN];
 
-    gl->dir = FindFirstFile((const char *) gl->pattern, &gl->finddata);
+    len = NGX_UTF16_BUFLEN;
+    u = ngx_utf8_to_utf16(utf16, gl->pattern, &len, 0);
+
+    if (u == NULL) {
+        return NGX_ERROR;
+    }
+
+    gl->dir = FindFirstFileW(u, &gl->finddata);
 
     if (gl->dir == INVALID_HANDLE_VALUE) {
 
         err = ngx_errno;
 
+        if (u != utf16) {
+            ngx_free(u);
+        }
+
         if ((err == ERROR_FILE_NOT_FOUND || err == ERROR_PATH_NOT_FOUND)
              && gl->test)
         {
@@ -561,6 +574,8 @@ ngx_open_glob(ngx_glob_t *gl)
             return NGX_OK;
         }
 
+        ngx_set_errno(err);
+
         return NGX_ERROR;
     }
 
@@ -570,18 +585,10 @@ ngx_open_glob(ngx_glob_t *gl)
         }
     }
 
-    len = ngx_strlen(gl->finddata.cFileName);
-    gl->name.len = gl->last + len;
-
-    gl->name.data = ngx_alloc(gl->name.len + 1, gl->log);
-    if (gl->name.data == NULL) {
-        return NGX_ERROR;
+    if (u != utf16) {
+        ngx_free(u);
     }
 
-    ngx_memcpy(gl->name.data, gl->pattern, gl->last);
-    ngx_cpystrn(gl->name.data + gl->last, (u_char *) gl->finddata.cFileName,
-                len + 1);
-
     gl->ready = 1;
 
     return NGX_OK;
@@ -591,40 +598,25 @@ ngx_open_glob(ngx_glob_t *gl)
 ngx_int_t
 ngx_read_glob(ngx_glob_t *gl, ngx_str_t *name)
 {
-    size_t     len;
-    ngx_err_t  err;
+    u_char     *p;
+    size_t      len;
+    ngx_err_t   err;
+    u_char      utf8[NGX_UTF16_BUFLEN];
 
     if (gl->no_match) {
         return NGX_DONE;
     }
 
     if (gl->ready) {
-        *name = gl->name;
-
         gl->ready = 0;
-        return NGX_OK;
+        goto convert;
     }
 
     ngx_free(gl->name.data);
     gl->name.data = NULL;
 
-    if (FindNextFile(gl->dir, &gl->finddata) != 0) {
-
-        len = ngx_strlen(gl->finddata.cFileName);
-        gl->name.len = gl->last + len;
-
-        gl->name.data = ngx_alloc(gl->name.len + 1, gl->log);
-        if (gl->name.data == NULL) {
-            return NGX_ERROR;
-        }
-
-        ngx_memcpy(gl->name.data, gl->pattern, gl->last);
-        ngx_cpystrn(gl->name.data + gl->last, (u_char *) gl->finddata.cFileName,
-                    len + 1);
-
-        *name = gl->name;
-
-        return NGX_OK;
+    if (FindNextFileW(gl->dir, &gl->finddata) != 0) {
+        goto convert;
     }
 
     err = ngx_errno;
@@ -637,6 +629,43 @@ ngx_read_glob(ngx_glob_t *gl, ngx_str_t 
                   "FindNextFile(%s) failed", gl->pattern);
 
     return NGX_ERROR;
+
+convert:
+
+    len = NGX_UTF16_BUFLEN;
+    p = ngx_utf16_to_utf8(utf8, gl->finddata.cFileName, &len, NULL);
+
+    if (p == NULL) {
+        return NGX_ERROR;
+    }
+
+    gl->name.len = gl->last + len - 1;
+
+    gl->name.data = ngx_alloc(gl->name.len + 1, gl->log);
+    if (gl->name.data == NULL) {
+        goto failed;
+    }
+
+    ngx_memcpy(gl->name.data, gl->pattern, gl->last);
+    ngx_cpystrn(gl->name.data + gl->last, p, len);
+
+    if (p != utf8) {
+        ngx_free(p);
+    }
+
+    *name = gl->name;
+
+    return NGX_OK;
+
+failed:
+
+    if (p != utf8) {
+        err = ngx_errno;
+        ngx_free(p);
+        ngx_set_errno(err);
+    }
+
+    return NGX_ERROR;
 }
 
 
diff -r 60d845f9505f -r d05c0adf5890 src/os/win32/ngx_files.h
--- a/src/os/win32/ngx_files.h	Thu Jan 12 21:41:30 2023 +0300
+++ b/src/os/win32/ngx_files.h	Thu Jan 12 21:41:39 2023 +0300
@@ -44,7 +44,7 @@ typedef struct {
 
 typedef struct {
     HANDLE                          dir;
-    WIN32_FIND_DATA                 finddata;
+    WIN32_FIND_DATAW                finddata;
 
     unsigned                        ready:1;
     unsigned                        test:1;