summaryrefslogtreecommitdiffstats
path: root/src/charset.h
diff options
context:
space:
mode:
Diffstat (limited to 'src/charset.h')
-rw-r--r--src/charset.h115
1 files changed, 0 insertions, 115 deletions
diff --git a/src/charset.h b/src/charset.h
deleted file mode 100644
index 8d1e1a2..0000000
--- a/src/charset.h
+++ /dev/null
@@ -1,115 +0,0 @@
-/*
- Copyright (C) 2015 Matthias P. Braendli (http://opendigitalradio.org)
-
- This program is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with this program. If not, see <http://www.gnu.org/licenses/>.
-*/
-/*!
- \file charset.h
- \brief Define the EBU charset according to ETSI TS 101 756v1.8.1 for DLS encoding
-
- \author Matthias P. Braendli <matthias@mpb.li>
- \author Lindsay Cornell
-*/
-
-#ifndef __CHARSET_H_
-#define __CHARSET_H_
-
-#include "utf8.h"
-#include <string>
-#include <vector>
-#include <algorithm>
-
-/**********************************************/
-/************* BIG FAT WARNING ****************/
-/**********************************************/
-/**** Make sure this file is always saved ****/
-/**** encoded in UTF-8, otherwise you will ****/
-/**** mess up the table below ! ****/
-/**********************************************/
-/********* END OF BIG FAT WARNING *************/
-/**********************************************/
-
-
-#define CHARSET_TABLE_OFFSET 1 // NUL at index 0 cannot be represented
-#define CHARSET_TABLE_ENTRIES (256 - CHARSET_TABLE_OFFSET)
-const char* utf8_encoded_EBU_Latin[CHARSET_TABLE_ENTRIES] = {
- "Ę", "Į", "Ų", "Ă", "Ė", "Ď", "Ș", "Ț", "Ċ", "\n","\v","Ġ", "Ĺ", "Ż", "Ń",
-"ą", "ę", "į", "ų", "ă", "ė", "ď", "ș", "ț", "ċ", "Ň", "Ě", "ġ", "ĺ", "ż", "\u0082",
-" ", "!", "\"","#", "ł", "%", "&", "'", "(", ")", "*", "+", ",", "-", ".", "/",
-"0", "1", "2", "3", "4", "5", "6", "7", "8", "9", ":", ";", "<", "=", ">", "?",
-"@", "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O",
-"P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z", "[", "Ů", "]", "Ł", "_",
-"Ą", "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o",
-"p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z", "«", "ů", "»", "Ľ", "Ħ",
-"á", "à", "é", "è", "í", "ì", "ó", "ò", "ú", "ù", "Ñ", "Ç", "Ş", "ß", "¡", "Ÿ",
-"â", "ä", "ê", "ë", "î", "ï", "ô", "ö", "û", "ü", "ñ", "ç", "ş", "ğ", "ı", "ÿ",
-"Ķ", "Ņ", "©", "Ģ", "Ğ", "ě", "ň", "ő", "Ő", "€", "£", "$", "Ā", "Ē", "Ī", "Ū",
-"ķ", "ņ", "Ļ", "ģ", "ļ", "İ", "ń", "ű", "Ű", "¿", "ľ", "°", "ā", "ē", "ī", "ū",
-"Á", "À", "É", "È", "Í", "Ì", "Ó", "Ò", "Ú", "Ù", "Ř", "Č", "Š", "Ž", "Ð", "Ŀ",
-"Â", "Ä", "Ê", "Ë", "Î", "Ï", "Ô", "Ö", "Û", "Ü", "ř", "č", "š", "ž", "đ", "ŀ",
-"Ã", "Å", "Æ", "Œ", "ŷ", "Ý", "Õ", "Ø", "Þ", "Ŋ", "Ŕ", "Ć", "Ś", "Ź", "Ť", "ð",
-"ã", "å", "æ", "œ", "ŵ", "ý", "õ", "ø", "þ", "ŋ", "ŕ", "ć", "ś", "ź", "ť", "ħ"};
-
-class CharsetConverter
-{
- public:
- CharsetConverter() {
- /*! Build the converstion table that contains the known code points,
- * at the indices corresponding to the EBU Latin table
- */
- using namespace std;
- for (size_t i = 0; i < CHARSET_TABLE_ENTRIES; i++) {
- string table_entry(utf8_encoded_EBU_Latin[i]);
- string::iterator it = table_entry.begin();
- uint32_t code_point = utf8::next(it, table_entry.end());
- m_conversion_table.push_back(code_point);
- }
- }
-
- /*! Convert a UTF-8 encoded text line into an EBU Latin encoded byte stream
- */
- std::string convert(std::string line_utf8) {
- using namespace std;
-
- // check for invalid utf-8, we only convert up to the first error
- string::iterator end_it = utf8::find_invalid(line_utf8.begin(), line_utf8.end());
-
- // Convert it to utf-32
- vector<uint32_t> utf32line;
- utf8::utf8to32(line_utf8.begin(), end_it, back_inserter(utf32line));
-
- string encoded_line(utf32line.size(), '0');
-
- // Try to convert each codepoint
- for (size_t i = 0; i < utf32line.size(); i++) {
- vector<uint32_t>::iterator iter = find(m_conversion_table.begin(),
- m_conversion_table.end(), utf32line[i]);
- if (iter != m_conversion_table.end()) {
- size_t index = std::distance(m_conversion_table.begin(), iter);
-
- encoded_line[i] = (char)(index + CHARSET_TABLE_OFFSET);
- }
- else {
- encoded_line[i] = ' ';
- }
- }
- return encoded_line;
- }
-
- private:
-
- std::vector<uint32_t> m_conversion_table;
-};
-
-#endif