Add test case and fix for string with broken escape sequences

src/unicode.c

@@ -165,7 +165,7 @@ uint32_t *advance_next32(const uint32_t *s, size_t *invis)
             /* Found '[' char after ESC. A CSI sequence has started. */
             (*invis)++;
             ansipos++;
-        } else if (ansipos == 1 && c >= 0x40 && c <= 0x5f) {
+        } else if (ansipos == 1 && c >= 0x40 && c <= 0x5f) { /* between '@' and '_' (mostly uppercase letters) */
             /* Found a byte designating the end of a two-byte escape sequence */
             (*invis)++;
             ansipos = 0;
@@ -183,6 +183,9 @@ uint32_t *advance_next32(const uint32_t *s, size_t *invis)
             break;
         }
     }
+    if (rest == NULL) {
+        rest = s + u32_strlen(s);
+    }
     return (uint32_t *) rest;
 }
 

src/unicode.h

@@ -137,7 +137,7 @@ ucs4_t to_utf32(char ascii);
  * @param invis Will contain the number of invisible characters skipped in order to get to the new position.
  *      This will be 0 unless `s` pointed to an ESC char, in which case it contains the length in characters of that
  *      escape sequence.
- * @return The next position, or 0 if the end of the string was reached
+ * @return The next position, or a pointer to the terminating 0 if the end of the string was reached
  */
 uint32_t *advance_next32(const uint32_t *s, size_t *invis);
 

utest/bxstring_test.c

@@ -303,6 +303,44 @@ void test_ansi_unicode_tabs(void **state)
 
 
 
+void test_ansi_unicode_broken_escapes(void **state)
+{
+    UNUSED(state);
+
+    /* "\x1b[38;5;203 X" is the first escape sequence, followed by
+     * "X " (visible) and
+     * "\x1b[0m" (regular escape sequence) and
+     * "__" (visible)
+     * "\x1b[38;5;203m" (regular, invisible)
+     * "b" (visible)
+     * "\x1b[" (broken, counts as invisible, and belonging to the previous 'b')
+     */
+    uint32_t *ustr32 = u32_strconv_from_arg("\x1b[38;5;203 XX \x1b[0m__\x1b[38;5;203mb\x1b[", "ASCII");
+    fprintf(stderr, "BOO\n");
+    assert_non_null(ustr32);
+    bxstr_t *actual = bxs_from_unicode(ustr32);
+    fprintf(stderr, "BAA\n");
+
+    assert_non_null(actual);
+    assert_non_null(actual->memory);
+    assert_string_equal("X __b", actual->ascii);
+    assert_int_equal(0, (int) actual->indent);
+    assert_int_equal(5, (int) actual->num_columns);
+    assert_int_equal(34, (int) actual->num_chars);
+    assert_int_equal(5, (int) actual->num_chars_visible);
+    assert_int_equal(29, (int) actual->num_chars_invisible);
+    assert_int_equal(0, (int) actual->trailing);
+    int expected_firstchar_idx[] = {0, 13, 18, 19, 20, 34};
+    assert_array_equal(expected_firstchar_idx, actual->first_char, 6);
+    int expected_vischar_idx[] = {12, 13, 18, 19, 31, 34};
+    assert_array_equal(expected_vischar_idx, actual->visible_char, 6);
+
+    BFREE(ustr32);
+    bxs_free(actual);
+}
+
+
+
 void test_ansi_unicode_null(void **state)
 {
     UNUSED(state);
@@ -935,8 +973,6 @@ void test_bxs_valid_in_filename_error(void **state)
 }
 
 
-// TODO test case for incomplete/broken escape sequences
-
 
 void test_bxs_free_null(void **state)
 {

utest/bxstring_test.h

@@ -33,6 +33,7 @@ void test_ansi_unicode_blanks(void **state);
 void test_ansi_unicode_invisible_only(void **state);
 void test_ansi_unicode_illegalchar(void **state);
 void test_ansi_unicode_tabs(void **state);
+void test_ansi_unicode_broken_escapes(void **state);
 void test_ansi_unicode_null(void **state);
 
 void test_bxs_strdup(void **state);

utest/main.c

@@ -128,6 +128,7 @@ int main(void)
         cmocka_unit_test_setup(test_ansi_unicode_invisible_only, beforeTest),
         cmocka_unit_test_setup(test_ansi_unicode_illegalchar, beforeTest),
         cmocka_unit_test_setup(test_ansi_unicode_tabs, beforeTest),
+        cmocka_unit_test_setup(test_ansi_unicode_broken_escapes, beforeTest),
         cmocka_unit_test_setup(test_ansi_unicode_null, beforeTest),
         cmocka_unit_test_setup(test_bxs_strdup, beforeTest),
         cmocka_unit_test_setup(test_bxs_trimdup_null, beforeTest),