Merge pull request #1816 from z2250389/codex-z2250389/fix-windows-unicode-git-root

fix(windows): support Unicode repository paths

src/git/git_context.c

@@ -2,6 +2,7 @@
 
 #include "foundation/compat_fs.h"
 #include "foundation/constants.h"
+#include "foundation/platform.h"
 #include "foundation/str_util.h"
 
 #include <ctype.h>
@@ -9,7 +10,6 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <sys/stat.h>
 
 enum {
     GIT_CMD_MAX = 1024,
@@ -254,8 +254,7 @@ int cbm_git_context_resolve(const char *path, cbm_git_context_t *out) {
         return CBM_NOT_FOUND;
     }
 
-    struct stat st;
-    out->root_exists = (stat(path, &st) == 0);
+    out->root_exists = cbm_is_dir(path);
     if (!out->root_exists) {
         return 0;
     }

src/ui/http_server.c

@@ -715,7 +715,6 @@ static void handle_processes(cbm_http_conn_t *c) {
 
 /* ── Directory browser ────────────────────────────────────────── */
 
-#include <dirent.h>
 
 static void append_roots_json(char *buf, size_t bufsz, int *pos) {
     http_appendf(buf, bufsz, pos, ",\"roots\":[");
@@ -770,7 +769,7 @@ static void handle_browse(cbm_http_conn_t *c, const cbm_http_req_t *req) {
         return;
     }
 
-    DIR *dir = opendir(path);
+    cbm_dir_t *dir = cbm_opendir(path);
     if (!dir) {
         cbm_http_replyf(c, 403, g_cors_json, "{\"error\":\"cannot open directory\"}");
         return;
@@ -781,16 +780,16 @@ static void handle_browse(cbm_http_conn_t *c, const cbm_http_req_t *req) {
     int pos = 0;
     http_appendf(buf, sizeof(buf), &pos, "{\"path\":\"%s\",\"dirs\":[", path);
 
-    struct dirent *ent;
+    cbm_dirent_t *ent;
     int count = 0;
-    while ((ent = readdir(dir)) != NULL) {
+    while ((ent = cbm_readdir(dir)) != NULL) {
         /* Skip hidden dirs and . / .. */
-        if (ent->d_name[0] == '.')
+        if (ent->name[0] == '.')
             continue;
 
         /* Check if it's actually a directory */
         char full[2048];
-        snprintf(full, sizeof(full), "%s/%s", path, ent->d_name);
+        snprintf(full, sizeof(full), "%s/%s", path, ent->name);
         if (!cbm_is_dir(full))
             continue;
 
@@ -799,7 +798,7 @@ static void handle_browse(cbm_http_conn_t *c, const cbm_http_req_t *req) {
         /* Escape directory name to prevent XSS (e.g., names with quotes/angle brackets) */
         {
             char esc[512];
-            cbm_json_escape(esc, (int)sizeof(esc), ent->d_name);
+            cbm_json_escape(esc, (int)sizeof(esc), ent->name);
             http_appendf(buf, sizeof(buf), &pos, "\"%s\"", esc);
         }
         if (pos >= (int)sizeof(buf)) {
@@ -810,7 +809,7 @@ static void handle_browse(cbm_http_conn_t *c, const cbm_http_req_t *req) {
         if (count >= 200)
             break; /* safety limit */
     }
-    closedir(dir);
+    cbm_closedir(dir);
 
     /* Parent path — escape to prevent injection */
     char parent[1024];

tests/test_daemon_application.c

@@ -1859,6 +1859,53 @@ static bool app_env_backup_restore(app_env_backup_t *backup) {
     return status == 0;
 }
 
+/* The Graph UI enters through the daemon's background coordinator rather than
+ * the MCP tool path. Keep that path UTF-8-safe as well. */
+TEST(daemon_application_ui_index_accepts_non_ascii_directory) {
+    app_env_backup_t cache_environment;
+    bool cache_saved = app_env_backup_capture(&cache_environment, "CBM_CACHE_DIR");
+    char root[APP_TEST_PATH_CAP];
+    char cache[APP_TEST_PATH_CAP];
+    (void)snprintf(root, sizeof(root),
+                   "%s/cbm-app-ui-index-\xE6\x97\xA5\xE6\x9C\xAC\xE8\xAA\x9E-XXXXXX", cbm_tmpdir());
+    (void)snprintf(cache, sizeof(cache), "%s/cbm-app-ui-index-cache-XXXXXX", cbm_tmpdir());
+    bool dirs_created = cbm_mkdtemp(root) != NULL && cbm_mkdtemp(cache) != NULL;
+    bool cache_set = cache_saved && dirs_created && cbm_setenv("CBM_CACHE_DIR", cache, 1) == 0;
+
+    app_fake_worker_context_t fake;
+    app_fake_worker_context_init(&fake);
+    atomic_store(&fake.allow_completion, true);
+    cbm_daemon_application_worker_ops_t worker_ops = {
+        .context = &fake,
+        .start = app_fake_worker_start,
+        .poll = app_fake_worker_poll,
+        .cancel = app_fake_worker_cancel,
+        .log_path = app_fake_worker_log_path,
+        .destroy = app_fake_worker_destroy,
+    };
+    cbm_daemon_application_config_t config = {.worker_ops = &worker_ops};
+    cbm_daemon_application_t *application = cache_set ? cbm_daemon_application_new(&config) : NULL;
+    int result = application ? cbm_daemon_application_index(application, "", root) : -1;
+    bool stopped = application && cbm_daemon_application_shutdown(application, APP_TEST_TIMEOUT_MS);
+    cbm_daemon_application_free(application);
+    bool root_cleaned = th_rmtree(root) == 0;
+    bool cache_cleaned = th_rmtree(cache) == 0;
+    bool cache_restored = app_env_backup_restore(&cache_environment);
+
+    ASSERT_TRUE(cache_saved);
+    ASSERT_TRUE(dirs_created);
+    ASSERT_TRUE(cache_set);
+    ASSERT_NOT_NULL(application);
+    ASSERT_EQ(result, 0);
+    ASSERT_EQ(atomic_load(&fake.starts), 1);
+    ASSERT_EQ(atomic_load(&fake.destroys), 1);
+    ASSERT_TRUE(stopped);
+    ASSERT_TRUE(root_cleaned);
+    ASSERT_TRUE(cache_cleaned);
+    ASSERT_TRUE(cache_restored);
+    PASS();
+}
+
 typedef struct {
     atomic_int starts;
     atomic_int start_failures_remaining;
@@ -5652,6 +5699,7 @@ SUITE(daemon_application) {
     RUN_TEST(daemon_application_initialize_coalesces_auto_index_for_full_sessions);
     RUN_TEST(daemon_application_sensitive_root_blocks_auto_index_but_preserves_controls);
     RUN_TEST(daemon_application_sensitive_root_blocks_watch_but_preserves_controls);
+    RUN_TEST(daemon_application_ui_index_accepts_non_ascii_directory);
     RUN_TEST(daemon_application_index_args_compare_repo_path_separator_equivalently);
     RUN_TEST(daemon_application_programmatic_index_injects_resource_policy);
     RUN_TEST(daemon_application_auto_index_honors_tracked_file_limit);

tests/test_git_context.c

@@ -8,8 +8,9 @@
  * (input_path), not to worktree_root. Joining it with worktree_root and then
  * string-stripping "/.git" left unresolved ".." components in the result.
  *
- * These tests shell out to `git`, so they SKIP_PLATFORM on Windows (the CI
- * shell there cannot init a repo via system()).
+ * The canonical_root tests shell out to `git`, so they SKIP_PLATFORM on
+ * Windows (the CI shell there cannot init a repo via system()). The root-path
+ * existence test runs on every platform.
  *
  * Reproduce-first guard: canonical_root_subdir is the genuine RED-without-the-fix
  * guard — a repo indexed from a subdirectory yields a relative --git-common-dir
@@ -34,9 +35,29 @@
 #include <limits.h>
 #endif
 
-/* These helpers shell out to git and are only used by the non-Windows test
- * bodies below; on Windows every test SKIP_PLATFORMs, so guard them here too or
- * they'd be unused-static functions and fail the -Werror build. */
+/* The Windows narrow stat() API interprets UTF-8 paths through the active ANSI
+ * code page. That made an existing Unicode repository root appear absent and
+ * stopped Git discovery before git was invoked (#1240). */
+TEST(root_exists_non_ascii_directory) {
+    /* UTF-8 for "Japanese language", written as bytes to keep the source ASCII. */
+    char *tmp = th_mktempdir("cbm_gitctx_\xE6\x97\xA5\xE6\x9C\xAC\xE8\xAA\x9E");
+    if (!tmp)
+        FAIL("th_mktempdir returned NULL for non-ASCII path");
+
+    cbm_git_context_t ctx = {0};
+    int rc = cbm_git_context_resolve(tmp, &ctx);
+    bool root_exists = ctx.root_exists;
+    cbm_git_context_free(&ctx);
+    th_rmtree(tmp);
+
+    ASSERT_EQ(rc, 0);
+    ASSERT_TRUE(root_exists);
+    PASS();
+}
+
+/* These helpers shell out to git and are only used by the non-Windows
+ * canonical_root test bodies below; guard them here too or they'd be
+ * unused-static functions and fail the -Werror build on Windows. */
 #ifndef _WIN32
 /* Run a git command inside dir, return 0 on success. */
 static int git_run(const char *dir, const char *args) {
@@ -235,6 +256,7 @@ TEST(canonical_root_linked_worktree) {
 /* ── Suite ──────────────────────────────────────────────────────── */
 
 SUITE(git_context) {
+    RUN_TEST(root_exists_non_ascii_directory);
     RUN_TEST(canonical_root_repo_root);
     RUN_TEST(canonical_root_subdir);
     RUN_TEST(canonical_root_linked_worktree);

tests/test_httpd.c

@@ -2150,6 +2150,59 @@ TEST(ui_server_rejects_non_loopback_host) {
     PASS();
 }
 
+/* The browser sends UTF-8 repository paths in a percent-encoded query. On
+ * Windows, the handler's narrow opendir/readdir calls accepted the ASCII parent
+ * but could neither list nor open a non-ASCII child, even though cbm_is_dir's
+ * wide-path check had already accepted it. Exercise both user-visible steps:
+ * discover the Unicode directory from its parent, then browse into it. */
+TEST(ui_server_browse_non_ascii_directory) {
+    static const char utf8_name[] = "\xE9\x81\x93\xE5\x85\xB7\xE7\xAE\xB1"; /* toolbox */
+    char *created = th_mktempdir("cbm_browse_utf8");
+    if (!created)
+        FAIL("mktempdir");
+
+    char base[512];
+    snprintf(base, sizeof(base), "%s", created);
+    cbm_normalize_path_sep(base);
+
+    char unicode_dir[768];
+    char child[1024];
+    int unicode_len = snprintf(unicode_dir, sizeof(unicode_dir), "%s/%s", base, utf8_name);
+    int child_len = snprintf(child, sizeof(child), "%s/Percy", unicode_dir);
+    if (unicode_len <= 0 || (size_t)unicode_len >= sizeof(unicode_dir) || child_len <= 0 ||
+        (size_t)child_len >= sizeof(child) || th_mkdir_p(child) != 0) {
+        th_cleanup(base);
+        FAIL("failed to create non-ASCII browse fixture");
+    }
+
+    th_server_t ts;
+    if (th_server_start(&ts) != 0) {
+        th_cleanup(base);
+        FAIL("server start");
+    }
+
+    int port = cbm_http_server_port(ts.srv);
+    char request[1536];
+    char response[8192];
+    snprintf(request, sizeof(request),
+             "GET /api/browse?path=%s/%%E9%%81%%93%%E5%%85%%B7%%E7%%AE%%B1 HTTP/1.1\r\n\r\n", base);
+    int n = th_http(port, request, response, sizeof(response));
+    bool child_browse_ok =
+        n > 0 && th_status(response) == 200 && strstr(response, "\"Percy\"") != NULL;
+
+    snprintf(request, sizeof(request), "GET /api/browse?path=%s HTTP/1.1\r\n\r\n", base);
+    n = th_http(port, request, response, sizeof(response));
+    bool parent_lists_unicode =
+        n > 0 && th_status(response) == 200 && strstr(response, utf8_name) != NULL;
+
+    th_server_stop(&ts);
+    th_cleanup(base);
+
+    ASSERT_TRUE(parent_lists_unicode);
+    ASSERT_TRUE(child_browse_ok);
+    PASS();
+}
+
 /* The directory browser formats readdir() entries into a fixed 32 KB response
  * buffer. The per-entry loop is clamped, but the trailing "parent"/"roots"
  * appends were not — once the entries filled the buffer, pos ran past the end
@@ -2367,6 +2420,7 @@ TEST(ui_server_index_status_long_paths_no_overflow) {
 /* ── Suite ────────────────────────────────────────────────────── */
 
 SUITE(httpd) {
+    RUN_TEST(ui_server_browse_non_ascii_directory);
     RUN_TEST(ui_server_browse_wide_dir_no_overflow);
     RUN_TEST(ui_server_logs_escape_dense_no_overflow);
     RUN_TEST(ui_server_index_status_long_paths_no_overflow);