gout

commit 6d1e33e1549fcdcdbb03c956cd566e35aaaef2bd
parent c4080dcdb82c89b83170f36df9ada4d604f19407
Author: Chris Bracken <chris@bracken.jp>
Date:   Thu, 19 Feb 2026 13:39:05 +0900

format: filter out control chars in print_xml_encoded

Don't print characters in the control char range. Since we're comparing
char to a range (>= 0x20) we need to cast to unsigned so that chars with
the high bit set and're treated as < 0 and missed.

Adds some tests.

Diffstat:
M
src/format.c
 | 
10
++++++++--
M
src/format_tests.c
 | 
49
+++++++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 57 insertions(+), 2 deletions(-)
diff --git a/src/format.c b/src/format.c
@@ -90,7 +90,7 @@ void print_xml_encoded_len(FILE* out,
                            bool output_crlf) {
   size_t len = (str_len >= 0) ? (size_t)str_len : strlen(str);
   for (size_t i = 0; i < len && str[i] != '\0'; i++) {
-    char c = str[i];
+    unsigned char c = (unsigned char)str[i];
     switch (c) {
       case '<':
         fprintf(out, "&lt;");
@@ -107,6 +107,9 @@ void print_xml_encoded_len(FILE* out,
       case '"':
         fprintf(out, "&quot;");
         break;
+      case '\t':
+        fprintf(out, "\t");
+        break;
       case '\r':
       case '\n':
         if (output_crlf) {
@@ -114,7 +117,10 @@ void print_xml_encoded_len(FILE* out,
         }
         break;
       default:
-        fprintf(out, "%c", c);
+        if (c >= 0x20) {
+          fprintf(out, "%c", c);
+        }
+        break;
     }
   }
 }
diff --git a/src/format_tests.c b/src/format_tests.c
@@ -256,6 +256,55 @@ UTEST(print_xml_encoded_len, ZeroLength) {
   free(buf);
 }
 
+UTEST(print_xml_encoded_len, ControlCharacters) {
+  char* buf = NULL;
+  size_t size = 0;
+  FILE* out = open_memstream(&buf, &size);
+  ASSERT_NE(NULL, out);
+
+  /* Tab (\t), LF (\n), CR (\r) should be preserved.
+   * Other control characters like \x01 (SOH) should be filtered. */
+  const char* test_str = "a\t\n\r\x01" "b";
+  print_xml_encoded_len(out, test_str, -1, true);
+  fclose(out);
+
+  EXPECT_STREQ("a\t\n\rb", buf);
+
+  free(buf);
+}
+
+UTEST(print_xml_encoded_len, MultiByteUtf8) {
+  char* buf = NULL;
+  size_t size = 0;
+  FILE* out = open_memstream(&buf, &size);
+  ASSERT_NE(NULL, out);
+
+  /* UTF-8 'é' is 0xC3 0xA9. These should be preserved as-is. */
+  const char* test_str = "caf\xC3\xA9";
+  print_xml_encoded_len(out, test_str, -1, true);
+  fclose(out);
+
+  EXPECT_STREQ("caf\xC3\xA9", buf);
+
+  free(buf);
+}
+
+UTEST(print_xml_encoded_len, EmbeddedNull) {
+  char* buf = NULL;
+  size_t size = 0;
+  FILE* out = open_memstream(&buf, &size);
+  ASSERT_NE(NULL, out);
+
+  /* The function should stop at \0 even if str_len is larger. */
+  const char* test_str = "abc\0def";
+  print_xml_encoded_len(out, test_str, 10, true);
+  fclose(out);
+
+  EXPECT_STREQ("abc", buf);
+
+  free(buf);
+}
+
 UTEST(print_gopher_text_len, BasicWithLf) {
   char* buf = NULL;
   size_t size = 0;