[FIX] escape special characters ( < > & ) to HTML context

docs/CHANGES.TXT

@@ -41,6 +41,7 @@
 - Fix: Resolve compile-time error about implicit declarations (#1646)
 - Fix: fatal out of memory error extracting from a VOB PS
 - Fix: Unit Test Rust failing due to changes in Rust Version 1.86.0
+- Fix: escape special characters ( < > & ) to HTML context
 
 0.94 (2021-12-14)
 -----------------

src/lib_ccx/ccx_decoders_708_output.c

@@ -156,7 +156,10 @@ void dtvcc_write_row(dtvcc_writer_ctx *writer, dtvcc_service_decoder *decoder, i
 		}
 		else
 		{
-			size_t size = write_utf16_char(tv->chars[row_index][i].sym, buf + buf_len);
+			unsigned char *buffer = buf + buf_len;
+			size_t size = write_utf16_char(tv->chars[row_index][i].sym, buffer);
+			if (size == 1)
+				size = ascii_to_html(*buffer, buffer);
 			buf_len += size;
 		}
 	}

src/lib_ccx/ccx_encoders_helpers.c

@@ -3,6 +3,7 @@
 #include "ccx_common_constants.h"
 #include "ccx_common_structs.h"
 #include "ccx_decoders_common.h"
+#include "utility.h"
 
 #include <assert.h>
 
@@ -350,14 +351,22 @@ unsigned get_decoder_line_encoded(struct encoder_ctx *ctx, unsigned char *buffer
 		{
 			case CCX_ENC_UTF_8:
 				bytes = get_char_in_utf_8(buffer, line[i]);
+				if (bytes == 1)
+					bytes = ascii_to_html(*buffer, buffer);
 				break;
 			case CCX_ENC_LATIN_1:
 				get_char_in_latin_1(buffer, line[i]);
-				bytes = 1;
+				bytes = ascii_to_html(*buffer, buffer);
 				break;
 			case CCX_ENC_UNICODE:
 				get_char_in_unicode(buffer, line[i]);
-				bytes = 2;
+				if (buffer[1] == 0)
+				{
+					bytes = ascii_to_html(*buffer, buffer);
+					bytes = ascii_to_unicode(buffer, bytes, buffer);
+				}
+				else
+					bytes = 2;
 				break;
 			case CCX_ENC_ASCII: // Consider to remove ASCII encoding or write get_char_in_ascii(...)
 				break;

src/lib_ccx/utility.c

@@ -662,6 +662,38 @@ size_t utf16_to_utf8(unsigned short utf16_char, unsigned char *out)
 	return 0;
 }
 
+size_t ascii_to_html(unsigned char ascii_char, unsigned char *out)
+{
+	unsigned char *write = NULL;
+	if (ascii_char == '&')
+		write = "&";
+	else if (ascii_char == '<')
+		write = "&lt;";
+	else if (ascii_char == '>')
+		write = "&gt;";
+	if (write != NULL)
+	{
+		strcpy(out, write);
+		return strlen(write);
+	}
+	else
+	{
+		out[0] = ascii_char;
+		return 1;
+	}
+	return 0;
+}
+
+size_t ascii_to_unicode(unsigned char *ascii_char_buf, int ascii_char_len, unsigned char *out)
+{
+	for (int i = ascii_char_len - 1; i >= 0; i--)
+	{
+		out[i * 2] = ascii_char_buf[i];
+		out[i * 2 + 1] = 0;
+	}
+	return ascii_char_len * 2;
+}
+
 LLONG change_timebase(LLONG val, struct ccx_rational cur_tb, struct ccx_rational dest_tb)
 {
 	/* val = (value * current timebase) / destination timebase */

src/lib_ccx/utility.h

@@ -45,6 +45,8 @@ const char *get_file_extension(const enum ccx_output_format write_format);
 char *create_outfilename(const char *basename, const char *suffix, const char *extension);
 int verify_crc32(uint8_t *buf, int len);
 size_t utf16_to_utf8(unsigned short utf16_char, unsigned char *out);
+size_t ascii_to_html(unsigned char ascii_char, unsigned char *out);
+size_t ascii_to_unicode(unsigned char *ascii_char_buf, int ascii_char_len, unsigned char *out);
 LLONG change_timebase(LLONG val, struct ccx_rational cur_tb, struct ccx_rational dest_tb);
 char *str_reallocncat(char *dst, char *src);