Replace g_utf8_make_valid() with own function

The function ws_utf8_make_valid() is all-around better and
also does maximal substitution of subparts.

4 files changed, 26 insertions, 13 deletions

diff --git a/wsutil/unicode-utils.c b/wsutil/unicode-utils.c
index 8af7910554..10c13a62d9 100644
--- a/wsutil/unicode-utils.c
+++ b/wsutil/unicode-utils.c
@@ -189,8 +189,8 @@ utf_8_validate(const guint8 *start, ssize_t length, const guint8 **end)
  * indication of whether there was an invalid character (i.e.
  * REPLACEMENT CHARACTER was used.)
  */
-guint8 *
-ws_utf8_make_valid(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length)
+wmem_strbuf_t *
+ws_utf8_make_valid_strbuf(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length)
 {
     wmem_strbuf_t *str;
 
@@ -215,7 +215,14 @@ ws_utf8_make_valid(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length)
         }
     }
 
-    return (guint8 *) wmem_strbuf_finalize(str);
+    return str;
+}
+
+guint8 *
+ws_utf8_make_valid(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length)
+{
+    wmem_strbuf_t *str = ws_utf8_make_valid_strbuf(scope, ptr, length);
+    return wmem_strbuf_finalize(str);
 }
 
 #ifdef _WIN32
diff --git a/wsutil/unicode-utils.h b/wsutil/unicode-utils.h
index 7303f607e0..7c85614bbe 100644
--- a/wsutil/unicode-utils.h
+++ b/wsutil/unicode-utils.h
@@ -70,6 +70,12 @@ int ws_utf8_seqlen[256];
 WS_DLL_PUBLIC guint8 *
 ws_utf8_make_valid(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length);
 
+/*
+ * Same as ws_utf8_make_valid() but returns a wmem_strbuf_t.
+ */
+WS_DLL_PUBLIC wmem_strbuf_t *
+ws_utf8_make_valid_strbuf(wmem_allocator_t *scope, const guint8 *ptr, ssize_t length);
+
 #ifdef _WIN32
 
 /** Given a UTF-8 string, convert it to UTF-16.  This is meant to be used
diff --git a/wsutil/wmem/wmem_strbuf.c b/wsutil/wmem/wmem_strbuf.c
index 3b67ab7b38..11c9548d67 100644
--- a/wsutil/wmem/wmem_strbuf.c
+++ b/wsutil/wmem/wmem_strbuf.c
@@ -18,6 +18,8 @@
 #include "wmem-int.h"
 #include "wmem_strutl.h"
 
+#include <wsutil/unicode-utils.h>
+
 #define DEFAULT_MINIMUM_SIZE 16
 
 /* _ROOM accounts for the null-terminator, _RAW_ROOM does not.
@@ -444,16 +446,14 @@ wmem_strbuf_utf8_validate(wmem_strbuf_t *strbuf, const char **endpptr)
 void
 wmem_strbuf_utf8_make_valid(wmem_strbuf_t *strbuf)
 {
-    /* Sanitize the contents to a temporary string. */
-    char *tmp = g_utf8_make_valid(strbuf->str, strbuf->len);
+    wmem_strbuf_t *tmp = ws_utf8_make_valid_strbuf(strbuf->allocator, strbuf->str, strbuf->len);
 
-    /* Reset the strbuf, keeping the backing memory allocation */
-    *strbuf->str = '\0';
-    strbuf->len = 0;
+    wmem_free(strbuf->allocator, strbuf->str);
+    strbuf->str = tmp->str;
+    strbuf->len = tmp->len;
+    strbuf->alloc_size = tmp->alloc_size;
 
-    /* Copy the temporary string to the strbuf. */
-    wmem_strbuf_append(strbuf, tmp);
-    g_free(tmp);
+    wmem_free(strbuf->allocator, tmp);
 }
 
 /*
diff --git a/wsutil/wmem/wmem_test.c b/wsutil/wmem/wmem_test.c
index c887db9e3d..6ac917fb57 100644
--- a/wsutil/wmem/wmem_test.c
+++ b/wsutil/wmem/wmem_test.c
@@ -1157,7 +1157,7 @@ wmem_test_strbuf_validate(void)
 
     strbuf = wmem_strbuf_new(NULL, "TEST\xEF ABC");
     g_assert_false(wmem_strbuf_utf8_validate(strbuf, &endptr));
-    g_assert(endptr == &strbuf->str[4]);
+    g_assert_true(endptr == &strbuf->str[4]);
     wmem_strbuf_destroy(strbuf);
 
     strbuf = wmem_strbuf_new(NULL, NULL);
@@ -1168,7 +1168,7 @@ wmem_test_strbuf_validate(void)
     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]);
+    g_assert_true(endptr == &strbuf->str[5]);
     wmem_strbuf_destroy(strbuf);
 
     strbuf = wmem_strbuf_new(NULL, NULL);