Changeset 1db4e2ae in mainline for common/str.c

Timestamp:

2025-04-17T15:10:08Z (5 days ago)

Author:

Jiří Zárevúcky <zarevucky.jiri@…>

Branches:

master

Children:

a65ccc4, af77459, c4cfe4c

Parents:

45adeeb

git-author:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-14 18:44:41)

git-committer:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-17 15:10:08)

Message:

Have str_sanitize also remove C0 and C1 control codes

and use it to sanitize KIO output

File:

• common/str.c (modified) (3 diffs)

common/str.c

@@ -547 +547 @@
 
         for (; n > 0 && b[0]; b++, n--) {
+                if (b[0] < ' ') {
+                        /* C0 control codes */
+                        b[0] = replacement;
+                        count++;
+                        continue;
+                }
+
                 int cont = _continuation_bytes(b[0]);
                 if (__builtin_expect(cont, 0) == 0)
@@ -584 +591 @@
                 }
 
+                bool c1_control = (b[0] == 0b11000010 && b[1] < 0b10100000);
+                if (cont == 1 && c1_control) {
+                        b[0] = replacement;
+                        count++;
+                        continue;
+                }
+
                 /* 0b1110!!!! 0b10!xxxxx 0b10xxxxxx */
                 if (cont == 2 && !(b[0] & 0b00001111) && !(b[1] & 0b00100000)) {
@@ -619 +633 @@
 }
 
+/** Replaces any byte that's not part of a complete valid UTF-8 character
+ * encoding with a replacement byte.
+ * Also replaces C0 and C1 control codes.
+ */
 size_t str_sanitize(char *str, size_t n, uint8_t replacement)
 {