swm · commit

Keep truncated status fields valid UTF-8

4c9a1f134acd84479cef0f91f6444fe4d3acc28c

Alexis Sellier committed 1 month ago 1 parent 183d43a8

swm.c +15 -2

void swm_sanitize_field(char *dst, size_t size, const char *src, const char *fallback) {
    size_t i;

    if (!size)
        return;

    if (!src || !*src)
        src = fallback ? fallback : "";

    for (i = 0; src[i] && i + 1 < size; i++) {
        unsigned char ch = (unsigned char)src[i];

        dst[i] = ch < ' ' || ch == 0x7f ? ' ' : src[i];
    }

    /* Wayland strings are UTF-8. Drop a partial final codepoint after truncation. */
    if (src[i] && i) {
        size_t        start = i - 1, len;
        unsigned char ch;

        while (start && ((unsigned char)dst[start] & 0xc0) == 0x80)
            start--;
        ch  = (unsigned char)dst[start];
        len = ch >= 0xc2 && ch <= 0xdf   ? 2
              : ch >= 0xe0 && ch <= 0xef ? 3
              : ch >= 0xf0 && ch <= 0xf4 ? 4
                                         : 1;
        if (i - start < len)
            i = start;
    }
    dst[i] = '\0';
}

/* Parse one persisted window-state record. */
bool swm_parse_window_state(

test/unit.c +2 -0

    assert(swm_rule_matches("*", nullptr, "anything", "title"));
    assert(swm_rule_matches("term", "shell", "my-terminal", "a shell"));
    assert(!swm_rule_matches("browser", nullptr, "terminal", "browser"));
    swm_sanitize_field(small, sizeof(small), "a\tb\x1b", nullptr);
    assert(!strcmp(small, "a b "));
    swm_sanitize_field(small, sizeof(small), "abc\xc3\xa9", nullptr);
    assert(!strcmp(small, "abc"));
    assert(swm_parse_window_state(
        "org.app\tWindow\t1\t-2\t300\t200\n", appid, sizeof(appid), title, sizeof(title), &geometry
    ));
    assert(!strcmp(appid, "org.app"));
    assert(!strcmp(title, "Window"));

1464	1464		void swm_sanitize_field(char dst, size_t size, const char src, const char *fallback) {
1465	1465		size_t i;
1466	1466
1467	1467		if (!size)
1468	1468		return;
1469		-
1470	1469		if (!src \|\| !*src)
1471	1470		src = fallback ? fallback : "";
1472	1471
1473	1472		for (i = 0; src[i] && i + 1 < size; i++) {
1474	1473		unsigned char ch = (unsigned char)src[i];
1475	1474
1476	1475		dst[i] = ch < ' ' \|\| ch == 0x7f ? ' ' : src[i];
1477	1476		}
1478		-
	1477	+	/* Wayland strings are UTF-8. Drop a partial final codepoint after truncation. */
	1478	+	if (src[i] && i) {
	1479	+	size_t start = i - 1, len;
	1480	+	unsigned char ch;
	1481	+
	1482	+	while (start && ((unsigned char)dst[start] & 0xc0) == 0x80)
	1483	+	start--;
	1484	+	ch = (unsigned char)dst[start];
	1485	+	len = ch >= 0xc2 && ch <= 0xdf ? 2
	1486	+	: ch >= 0xe0 && ch <= 0xef ? 3
	1487	+	: ch >= 0xf0 && ch <= 0xf4 ? 4
	1488	+	: 1;
	1489	+	if (i - start < len)
	1490	+	i = start;
	1491	+	}
1479	1492		dst[i] = '\0';
1480	1493		}
1481	1494
1482	1495		/* Parse one persisted window-state record. */
1483	1496		bool swm_parse_window_state(

157	157		assert(swm_rule_matches("*", nullptr, "anything", "title"));
158	158		assert(swm_rule_matches("term", "shell", "my-terminal", "a shell"));
159	159		assert(!swm_rule_matches("browser", nullptr, "terminal", "browser"));
160	160		swm_sanitize_field(small, sizeof(small), "a\tb\x1b", nullptr);
161	161		assert(!strcmp(small, "a b "));
	162	+	swm_sanitize_field(small, sizeof(small), "abc\xc3\xa9", nullptr);
	163	+	assert(!strcmp(small, "abc"));
162	164		assert(swm_parse_window_state(
163	165		"org.app\tWindow\t1\t-2\t300\t200\n", appid, sizeof(appid), title, sizeof(title), &geometry
164	166		));
165	167		assert(!strcmp(appid, "org.app"));
166	168		assert(!strcmp(title, "Window"));