[PATCH v2 0/1] MR7473: winetest: Avoid underflow when computing filtered output size.

[Rémi Bernon (＠rbernon)]

--
  v2: winetest: Avoid underflow when computing filtered output size.

https://gitlab.winehq.org/wine/wine/-/merge_requests/7473

[PATCH v2 1/1] winetest: Avoid underflow when computing filtered output size.

[Rémi Bernon]

From: Rémi Bernon

---
 programs/winetest/main.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/programs/winetest/main.c b/programs/winetest/main.c
index f505b85323a..755e4eb6e2c 100644
--- a/programs/winetest/main.c
+++ b/programs/winetest/main.c
@@ -909,11 +909,12 @@ static void *filter_data( const char *data, DWORD size, DWORD *output_size )
     char *ret;

if (!(ret = malloc( size + 1 ))) return NULL;
-    for (i = j = eol = 0; i < size; i++)
+    for (i = j = 0, eol = -1; i < size; i++)
     {
         if (data[i] == '\x1b' && data[i + 1] == '[')
         {
             while (data[i] && data[i] != 'm') i++;
+            eol = i;
         }
         else if (data[i]) ret[j++] = data[i];
         if (!strncmp( data + i, " Test succeeded", 15 )) ignore += i + 15 - eol;

[Rémi Bernon (＠rbernon)]

Happens because j is the length without ANSI codes, and `ignore` striped entire "\n.*Test succeeded" lines, which may also include ANSI codes, removing them twice from the total.

v2: Fix it properly instead of hacking it.