wmem: Make strbuf_utf8_validate() accept embedded NUL bytes

wsutil/wmem/wmem_core.h

@@ -18,6 +18,7 @@
 #include <glib.h>
 #include <ws_symbol_export.h>
 #include <ws_attributes.h>
+#include <ws_posix_compat.h>
 
 #ifdef __cplusplus
 extern "C" {

wsutil/wmem/wmem_strbuf.c

@@ -415,10 +415,35 @@ wmem_strbuf_destroy(wmem_strbuf_t *strbuf)
     wmem_free(strbuf->allocator, strbuf);
 }
 
-bool
-wmem_strbuf_utf8_validate(wmem_strbuf_t *strbuf, const char **endptr)
+static bool
+string_utf8_validate(const char *str, ssize_t max_len, const char **endpptr)
 {
-    return g_utf8_validate(strbuf->str, strbuf->len, endptr);
+    bool valid;
+
+    if (max_len <= 0)
+        return true;
+
+    valid = g_utf8_validate(str, max_len, endpptr);
+    if (valid || **endpptr != '\0')
+        return valid;
+
+    /* Invalid because of a nul byte. Skip nuls and continue. */
+    max_len -= *endpptr - str;
+    str = *endpptr;
+    while (max_len > 0 && *str == '\0') {
+        str++;
+        max_len--;
+    }
+    return string_utf8_validate(str, max_len, endpptr);
+}
+
+/* g_utf8_validate() returns FALSE in the string contains embedded NUL
+ * bytes. We accept \x00 as valid and work around that to validate the
+ * entire len bytes. */
+bool
+wmem_strbuf_utf8_validate(wmem_strbuf_t *strbuf, const char **endpptr)
+{
+    return string_utf8_validate(strbuf->str, strbuf->len, endpptr);
 }
 
 void

wsutil/wmem/wmem_test.c

@@ -1175,6 +1175,34 @@ wmem_test_strbuf(void)
     wmem_destroy_allocator(allocator);
 }
 
+static void
+wmem_test_strbuf_validate(void)
+{
+    wmem_strbuf_t *strbuf;
+    const char *endptr;
+
+    strbuf = wmem_strbuf_new(NULL, "TEST\xEF ABC");
+    g_assert_false(wmem_strbuf_utf8_validate(strbuf, &endptr));
+    g_assert(endptr == &strbuf->str[4]);
+    wmem_strbuf_destroy(strbuf);
+
+    strbuf = wmem_strbuf_new(NULL, NULL);
+    wmem_strbuf_append_len(strbuf, "TEST\x00\x00 ABC", 10);
+    g_assert_true(wmem_strbuf_utf8_validate(strbuf, &endptr));
+    wmem_strbuf_destroy(strbuf);
+
+    strbuf = wmem_strbuf_new(NULL, NULL);
+    wmem_strbuf_append_len(strbuf, "TEST\x00\xEF ABC", 10);
+    g_assert_false(wmem_strbuf_utf8_validate(strbuf, &endptr));
+    g_assert(endptr == &strbuf->str[5]);
+    wmem_strbuf_destroy(strbuf);
+
+    strbuf = wmem_strbuf_new(NULL, NULL);
+    wmem_strbuf_append_len(strbuf, "TEST\x00 ABC \x00 DEF \x00", 17);
+    g_assert_true(wmem_strbuf_utf8_validate(strbuf, &endptr));
+    wmem_strbuf_destroy(strbuf);
+}
+
 static void
 wmem_test_tree(void)
 {
@@ -1466,6 +1494,7 @@ main(int argc, char **argv)
     g_test_add_func("/wmem/datastruct/queue",  wmem_test_queue);
     g_test_add_func("/wmem/datastruct/stack",  wmem_test_stack);
     g_test_add_func("/wmem/datastruct/strbuf", wmem_test_strbuf);
+    g_test_add_func("/wmem/datastruct/strbuf/validate", wmem_test_strbuf_validate);
     g_test_add_func("/wmem/datastruct/tree",   wmem_test_tree);
     g_test_add_func("/wmem/datastruct/itree",  wmem_test_itree);