Re: [BUG] Unicode filenames handling in `git log --stat`

From: Torsten Bögershausen <tboegi@web.de>
Date: Aug 10, 2022, 09:51 UTC
Message-ID: <20220810095157.wo4jaumtu47qplsb@tb-raspi4>
In-Reply-To: <CA+VDVVUKf48Q9A0hWPnBE+qG_7tBDuXKkdo+wWDU7iC3Wg=oEg@mail.gmail.com>

On Wed, Aug 10, 2022 at 12:56:11PM +0400, Alexander Meshcheryakov wrote:

Thanks for digging.

(And please, try to avoid top-posting here in this list)

Show 7 quoted lines

> I believe I have found exact place where strlen is used incorrectly
> This is at diff.c:show_stats
>
> https://github.com/git/git/blob/c50926e1f48891e2671e1830dbcd2912a4563450/diff.c#L2623
>
> It probably should be replaced with one of utf8_width, utf8_strnwidth
> or utf8_strwidth from utf8.c

That did not help here. If I understand it right, this function is not at all involved in our `git log --stat` ?

I tried this patch (not 100% git-style) and didn't see any print.

--- a/diff.c
+++ b/diff.c
@@ -2620,7 +2620,14 @@ static void show_stats(struct diffstat_t *data, struct diff_options *options)
                        continue;
			                }
					                fill_print_name(file);
							-               len = strlen(file->print_name);
							+               {
							+                       const char *cp = file->print_name;
							+                       size_t l = strlen(file->print_name);
							+                       len = utf8_width(&cp, &l);
							+                       fprintf(stderr, "%s/%s:%d file->print_name='%s' len=%lu\n",
							+                               __FILE__, __FUNCTION__, __LINE__,
							+                               file->print_name, (unsigned long)len);
							+               }
							                if (max_len < len)
									                        max_len = len;


And looking here, it seems as we are calculating max_len here.
Still more digging needed (but I don't promise anything today)