From 81e107ecb7b19a2e06a9b435ce24816f61b572ab Mon Sep 17 00:00:00 2001 From: Richard Kunze Date: Sat, 5 May 2012 23:12:05 +0200 Subject: [PATCH] Add builtin charset conversion for systems without sufficient iconv support. Enable with './configure --enable-dvbconv' --- src/dvb/dvb_charset_tables.h | 365 +++++++++++++++++++++++++++++++++++ src/dvb/dvb_support.c | 287 ++++++++++++++++++++++----- support/configure.inc | 3 +- 3 files changed, 605 insertions(+), 50 deletions(-) create mode 100644 src/dvb/dvb_charset_tables.h diff --git a/src/dvb/dvb_charset_tables.h b/src/dvb/dvb_charset_tables.h new file mode 100644 index 00000000..2d781523 --- /dev/null +++ b/src/dvb/dvb_charset_tables.h @@ -0,0 +1,365 @@ +#ifndef _DVB_CHARSET_TABLES_H +#define _DVB_CHARSET_TABLES_H 1 + +// Code tables for the variable part of the iso 8859 family (codes 0xa0 to 0xff) +static uint16_t conv_8859_table[14][96] = { + /* ISO-8859-1 */ { + /* 0xA0 */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, + /* 0xA8 */ 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, + /* 0xB8 */ 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF, + /* 0xC0 */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, + /* 0xC8 */ 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, + /* 0xD8 */ 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, + /* 0xE0 */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, + /* 0xE8 */ 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, + /* 0xF8 */ 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF, + }, + /* ISO-8859-2 */ { + /* 0xA0 */ 0x00A0, 0x0104, 0x02D8, 0x0141, 0x00A4, 0x013D, 0x015A, 0x00A7, + /* 0xA8 */ 0x00A8, 0x0160, 0x015E, 0x0164, 0x0179, 0x00AD, 0x017D, 0x017B, + /* 0xB0 */ 0x00B0, 0x0105, 0x02DB, 0x0142, 0x00B4, 0x013E, 0x015B, 0x02C7, + /* 0xB8 */ 0x00B8, 0x0161, 0x015F, 0x0165, 0x017A, 0x02DD, 0x017E, 0x017C, + /* 0xC0 */ 0x0154, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0139, 0x0106, 0x00C7, + /* 0xC8 */ 0x010C, 0x00C9, 0x0118, 0x00CB, 0x011A, 0x00CD, 0x00CE, 0x010E, + /* 0xD0 */ 0x0110, 0x0143, 0x0147, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x00D7, + /* 0xD8 */ 0x0158, 0x016E, 0x00DA, 0x0170, 0x00DC, 0x00DD, 0x0162, 0x00DF, + /* 0xE0 */ 0x0155, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x013A, 0x0107, 0x00E7, + /* 0xE8 */ 0x010D, 0x00E9, 0x0119, 0x00EB, 0x011B, 0x00ED, 0x00EE, 0x010F, + /* 0xF0 */ 0x0111, 0x0144, 0x0148, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x00F7, + /* 0xF8 */ 0x0159, 0x016F, 0x00FA, 0x0171, 0x00FC, 0x00FD, 0x0163, 0x02D9, + }, + /* ISO-8859-3 */ { + /* 0xA0 */ 0x00A0, 0x0126, 0x02D8, 0x00A3, 0x00A4, 0x0000, 0x0124, 0x00A7, + /* 0xA8 */ 0x00A8, 0x0130, 0x015E, 0x011E, 0x0134, 0x00AD, 0x0000, 0x017B, + /* 0xB0 */ 0x00B0, 0x0127, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x0125, 0x00B7, + /* 0xB8 */ 0x00B8, 0x0131, 0x015F, 0x011F, 0x0135, 0x00BD, 0x0000, 0x017C, + /* 0xC0 */ 0x00C0, 0x00C1, 0x00C2, 0x0000, 0x00C4, 0x010A, 0x0108, 0x00C7, + /* 0xC8 */ 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x0000, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x0120, 0x00D6, 0x00D7, + /* 0xD8 */ 0x011C, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x016C, 0x015C, 0x00DF, + /* 0xE0 */ 0x00E0, 0x00E1, 0x00E2, 0x0000, 0x00E4, 0x010B, 0x0109, 0x00E7, + /* 0xE8 */ 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x0000, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x0121, 0x00F6, 0x00F7, + /* 0xF8 */ 0x011D, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x016D, 0x015D, 0x02D9, + }, + /* ISO-8859-4 */ { + /* 0xA0 */ 0x00A0, 0x0104, 0x0138, 0x0156, 0x00A4, 0x0128, 0x013B, 0x00A7, + /* 0xA8 */ 0x00A8, 0x0160, 0x0112, 0x0122, 0x0166, 0x00AD, 0x017D, 0x00AF, + /* 0xB0 */ 0x00B0, 0x0105, 0x02DB, 0x0157, 0x00B4, 0x0129, 0x013C, 0x02C7, + /* 0xB8 */ 0x00B8, 0x0161, 0x0113, 0x0123, 0x0167, 0x014A, 0x017E, 0x014B, + /* 0xC0 */ 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, + /* 0xC8 */ 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x012A, + /* 0xD0 */ 0x0110, 0x0145, 0x014C, 0x0136, 0x00D4, 0x00D5, 0x00D6, 0x00D7, + /* 0xD8 */ 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x0168, 0x016A, 0x00DF, + /* 0xE0 */ 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, + /* 0xE8 */ 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x012B, + /* 0xF0 */ 0x0111, 0x0146, 0x014D, 0x0137, 0x00F4, 0x00F5, 0x00F6, 0x00F7, + /* 0xF8 */ 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x0169, 0x016B, 0x02D9, + }, + /* ISO-8859-5 */ { + /* 0xA0 */ 0x00A0, 0x0401, 0x0402, 0x0403, 0x0404, 0x0405, 0x0406, 0x0407, + /* 0xA8 */ 0x0408, 0x0409, 0x040A, 0x040B, 0x040C, 0x00AD, 0x040E, 0x040F, + /* 0xB0 */ 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417, + /* 0xB8 */ 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F, + /* 0xC0 */ 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427, + /* 0xC8 */ 0x0428, 0x0429, 0x042A, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F, + /* 0xD0 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437, + /* 0xD8 */ 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F, + /* 0xE0 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447, + /* 0xE8 */ 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F, + /* 0xF0 */ 0x2116, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457, + /* 0xF8 */ 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x00A7, 0x045E, 0x045F, + }, + /* ISO-8859-6 */ { + /* 0xA0 */ 0x00A0, 0x0000, 0x0000, 0x0000, 0x00A4, 0x0000, 0x0000, 0x0000, + /* 0xA8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x060C, 0x00AD, 0x0000, 0x0000, + /* 0xB0 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xB8 */ 0x0000, 0x0000, 0x0000, 0x061B, 0x0000, 0x0000, 0x0000, 0x061F, + /* 0xC0 */ 0x0000, 0x0621, 0x0622, 0x0623, 0x0624, 0x0625, 0x0626, 0x0627, + /* 0xC8 */ 0x0628, 0x0629, 0x062A, 0x062B, 0x062C, 0x062D, 0x062E, 0x062F, + /* 0xD0 */ 0x0630, 0x0631, 0x0632, 0x0633, 0x0634, 0x0635, 0x0636, 0x0637, + /* 0xD8 */ 0x0638, 0x0639, 0x063A, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xE0 */ 0x0640, 0x0641, 0x0642, 0x0643, 0x0644, 0x0645, 0x0646, 0x0647, + /* 0xE8 */ 0x0648, 0x0649, 0x064A, 0x064B, 0x064C, 0x064D, 0x064E, 0x064F, + /* 0xF0 */ 0x0650, 0x0651, 0x0652, + }, + /* ISO-8859-7 */ { + /* 0xA0 */ 0x00A0, 0x2018, 0x2019, 0x00A3, 0x20AC, 0x20AF, 0x00A6, 0x00A7, + /* 0xA8 */ 0x00A8, 0x00A9, 0x037A, 0x00AB, 0x00AC, 0x00AD, 0x0000, 0x2015, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x0384, 0x0385, 0x0386, 0x00B7, + /* 0xB8 */ 0x0388, 0x0389, 0x038A, 0x00BB, 0x038C, 0x00BD, 0x038E, 0x038F, + /* 0xC0 */ 0x0390, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, + /* 0xC8 */ 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F, + /* 0xD0 */ 0x03A0, 0x03A1, 0x0000, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7, + /* 0xD8 */ 0x03A8, 0x03A9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF, + /* 0xE0 */ 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, + /* 0xE8 */ 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF, + /* 0xF0 */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7, + /* 0xF8 */ 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, + }, + /* ISO-8859-8 */ { + /* 0xA0 */ 0x00A0, 0x0000, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, + /* 0xA8 */ 0x00A8, 0x00A9, 0x00D7, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, + /* 0xB8 */ 0x00B8, 0x00B9, 0x00F7, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x0000, + /* 0xC0 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xC8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xD0 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xD8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x2017, + /* 0xE0 */ 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7, + /* 0xE8 */ 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF, + /* 0xF0 */ 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7, + /* 0xF8 */ 0x05E8, 0x05E9, 0x05EA, 0x0000, 0x0000, 0x200E, 0x200F, + }, + /* ISO-8859-9 */ { + /* 0xA0 */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, + /* 0xA8 */ 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, + /* 0xB8 */ 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF, + /* 0xC0 */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, + /* 0xC8 */ 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x011E, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, + /* 0xD8 */ 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0130, 0x015E, 0x00DF, + /* 0xE0 */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, + /* 0xE8 */ 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x011F, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, + /* 0xF8 */ 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0131, 0x015F, 0x00FF, + }, + /* ISO-8859-10 */ { + /* 0xA0 */ 0x00A0, 0x0104, 0x0112, 0x0122, 0x012A, 0x0128, 0x0136, 0x00A7, + /* 0xA8 */ 0x013B, 0x0110, 0x0160, 0x0166, 0x017D, 0x00AD, 0x016A, 0x014A, + /* 0xB0 */ 0x00B0, 0x0105, 0x0113, 0x0123, 0x012B, 0x0129, 0x0137, 0x00B7, + /* 0xB8 */ 0x013C, 0x0111, 0x0161, 0x0167, 0x017E, 0x2015, 0x016B, 0x014B, + /* 0xC0 */ 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, + /* 0xC8 */ 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x00D0, 0x0145, 0x014C, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0168, + /* 0xD8 */ 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, + /* 0xE0 */ 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, + /* 0xE8 */ 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x00F0, 0x0146, 0x014D, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0169, + /* 0xF8 */ 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x0138, + }, + /* ISO-8859-11 */ { + /* 0xA0 */ 0x00A0, 0x0E01, 0x0E02, 0x0E03, 0x0E04, 0x0E05, 0x0E06, 0x0E07, + /* 0xA8 */ 0x0E08, 0x0E09, 0x0E0A, 0x0E0B, 0x0E0C, 0x0E0D, 0x0E0E, 0x0E0F, + /* 0xB0 */ 0x0E10, 0x0E11, 0x0E12, 0x0E13, 0x0E14, 0x0E15, 0x0E16, 0x0E17, + /* 0xB8 */ 0x0E18, 0x0E19, 0x0E1A, 0x0E1B, 0x0E1C, 0x0E1D, 0x0E1E, 0x0E1F, + /* 0xC0 */ 0x0E20, 0x0E21, 0x0E22, 0x0E23, 0x0E24, 0x0E25, 0x0E26, 0x0E27, + /* 0xC8 */ 0x0E28, 0x0E29, 0x0E2A, 0x0E2B, 0x0E2C, 0x0E2D, 0x0E2E, 0x0E2F, + /* 0xD0 */ 0x0E30, 0x0E31, 0x0E32, 0x0E33, 0x0E34, 0x0E35, 0x0E36, 0x0E37, + /* 0xD8 */ 0x0E38, 0x0E39, 0x0E3A, 0x0000, 0x0000, 0x0000, 0x0000, 0x0E3F, + /* 0xE0 */ 0x0E40, 0x0E41, 0x0E42, 0x0E43, 0x0E44, 0x0E45, 0x0E46, 0x0E47, + /* 0xE8 */ 0x0E48, 0x0E49, 0x0E4A, 0x0E4B, 0x0E4C, 0x0E4D, 0x0E4E, 0x0E4F, + /* 0xF0 */ 0x0E50, 0x0E51, 0x0E52, 0x0E53, 0x0E54, 0x0E55, 0x0E56, 0x0E57, + /* 0xF8 */ 0x0E58, 0x0E59, 0x0E5A, 0x0E5B, + }, + /* ISO-8859-13 */ { + /* 0xA0 */ 0x00A0, 0x201D, 0x00A2, 0x00A3, 0x00A4, 0x201E, 0x00A6, 0x00A7, + /* 0xA8 */ 0x00D8, 0x00A9, 0x0156, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00C6, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x201C, 0x00B5, 0x00B6, 0x00B7, + /* 0xB8 */ 0x00F8, 0x00B9, 0x0157, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00E6, + /* 0xC0 */ 0x0104, 0x012E, 0x0100, 0x0106, 0x00C4, 0x00C5, 0x0118, 0x0112, + /* 0xC8 */ 0x010C, 0x00C9, 0x0179, 0x0116, 0x0122, 0x0136, 0x012A, 0x013B, + /* 0xD0 */ 0x0160, 0x0143, 0x0145, 0x00D3, 0x014C, 0x00D5, 0x00D6, 0x00D7, + /* 0xD8 */ 0x0172, 0x0141, 0x015A, 0x016A, 0x00DC, 0x017B, 0x017D, 0x00DF, + /* 0xE0 */ 0x0105, 0x012F, 0x0101, 0x0107, 0x00E4, 0x00E5, 0x0119, 0x0113, + /* 0xE8 */ 0x010D, 0x00E9, 0x017A, 0x0117, 0x0123, 0x0137, 0x012B, 0x013C, + /* 0xF0 */ 0x0161, 0x0144, 0x0146, 0x00F3, 0x014D, 0x00F5, 0x00F6, 0x00F7, + /* 0xF8 */ 0x0173, 0x0142, 0x015B, 0x016B, 0x00FC, 0x017C, 0x017E, 0x2019, + }, + /* ISO-8859-14 */ { + /* 0xA0 */ 0x00A0, 0x1E02, 0x1E03, 0x00A3, 0x010A, 0x010B, 0x1E0A, 0x00A7, + /* 0xA8 */ 0x1E80, 0x00A9, 0x1E82, 0x1E0B, 0x1EF2, 0x00AD, 0x00AE, 0x0178, + /* 0xB0 */ 0x1E1E, 0x1E1F, 0x0120, 0x0121, 0x1E40, 0x1E41, 0x00B6, 0x1E56, + /* 0xB8 */ 0x1E81, 0x1E57, 0x1E83, 0x1E60, 0x1EF3, 0x1E84, 0x1E85, 0x1E61, + /* 0xC0 */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, + /* 0xC8 */ 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x0174, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x1E6A, + /* 0xD8 */ 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x0176, 0x00DF, + /* 0xE0 */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, + /* 0xE8 */ 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x0175, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x1E6B, + /* 0xF8 */ 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x0177, 0x00FF, + }, + /* ISO-8859-15 */ { + /* 0xA0 */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x20AC, 0x00A5, 0x0160, 0x00A7, + /* 0xA8 */ 0x0161, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, + /* 0xB0 */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x017D, 0x00B5, 0x00B6, 0x00B7, + /* 0xB8 */ 0x017E, 0x00B9, 0x00BA, 0x00BB, 0x0152, 0x0153, 0x0178, 0x00BF, + /* 0xC0 */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, + /* 0xC8 */ 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, + /* 0xD0 */ 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, + /* 0xD8 */ 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, + /* 0xE0 */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, + /* 0xE8 */ 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, + /* 0xF0 */ 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, + /* 0xF8 */ 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF, + } +}; + +// Code table for ISO6937 codes from 0xa0 to 0xbf and 0xd0 to 0xff +static const uint16_t iso6937_single_byte[96] = { + /* 0xA0 */ 0x00a0, 0x00a1, 0x00a2, 0x00a3, 0x0000, 0x00a5, 0x0000, 0x00a7, + /* 0xA8 */ 0x00a4, 0x2018, 0x201c, 0x00ab, 0x2190, 0x2191, 0x2192, 0x2193, + /* 0xB0 */ 0x00b0, 0x00b1, 0x00b2, 0x00b3, 0x00d7, 0x00b5, 0x00b6, 0x00b7, + /* 0xB8 */ 0x00f7, 0x2019, 0x201d, 0x00bb, 0x00bc, 0x00bd, 0x00be, 0x00bf, + /* 0xC0 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xC8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0xD0 */ 0x2014, 0x00b9, 0x00ae, 0x00a9, 0x2122, 0x266a, 0x00ac, 0x00a6, + /* 0xD8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x215b, 0x215c, 0x215d, 0x215e, + /* 0xE0 */ 0x2126, 0x00c6, 0x00d0, 0x00aa, 0x0126, 0x0000, 0x0132, 0x013f, + /* 0xE8 */ 0x0141, 0x00d8, 0x0152, 0x00ba, 0x00de, 0x0166, 0x014a, 0x0149, + /* 0xF0 */ 0x0138, 0x00e6, 0x0111, 0x00f0, 0x0127, 0x0131, 0x0133, 0x0140, + /* 0xF8 */ 0x0142, 0x00f8, 0x0153, 0x00df, 0x00fe, 0x0167, 0x014b, 0x00ad, +}; + +// Code table for ISO6937 two-byte sequences combining an accent byte. +// in the range 0xc0 to 0xcf with a space character (0x20) +static const uint16_t iso6937_lone_accents[16] = { + /* 0xC0 */ 0x0000, 0x00B4, 0x0000, 0x0000, 0x0000, 0x00Af, 0x02D8, 0x02D9, + /* 0xC8 */ 0x00A8, 0x0000, 0x02DA, 0x00B8, 0x0000, 0x02DD, 0x02DB, 0x02C7, +}; + +// Code tables for ISO6937 two-byte sequences combining an accent byte. +// in the range 0xc1 to 0xcf with a letter in one of the ranges +// 0x41 to 0x5a (A-Z) or 0x61 to 0x7a (a-z) +static const uint16_t iso6937_multi_byte[16][52] = { + /* 0xC0 */ { + 0x00, + }, + /* 0xC1 */ { + /* A-Z */ + 0xc0, 0x00, 0x00, 0x00, 0xc8, 0x00, 0x00, 0x00, 0xcc, 0x00, 0x00, 0x00, 0x00, + 0x00, 0xd2, 0x00, 0x00, 0x00, 0x00, 0x00, 0xd9, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0xe0, 0x00, 0x00, 0x00, 0xe8, 0x00, 0x00, 0x00, 0xec, 0x00, 0x00, 0x00, 0x00, + 0x00, 0xf2, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xC2 */ + { + /* A-Z */ + 0xc1, 0x00, 0x0106, 0x00, 0xc9, 0x00, 0x00, 0x00, 0xcd, 0x00, 0x00, 0x0139, + 0x00, 0x0143, 0xd3, 0x00, 0x00, 0x0154, 0x015a, 0x00, 0xda, 0x00, 0x00, 0x00, + 0xdd, 0x0179, + /* a-z */ + 0xe1, 0x00, 0x0107, 0x00, 0xe9, 0x00, 0x00, 0x00, 0xed, 0x00, 0x00, 0x013a, + 0x00, 0x0144, 0xf3, 0x00, 0x00, 0x0155, 0x015b, 0x00, 0xfa, 0x00, 0x00, 0x00, + 0xfd, 0x017a + }, + /* 0xC3 */ + { + /* A-Z */ + 0xc2, 0x00, 0x0108, 0x00, 0xca, 0x00, 0x011c, 0x0124, 0xce, 0x0134, 0x00, + 0x00, 0x00, 0x00, 0xd4, 0x00, 0x00, 0x00, 0x015c, 0x00, 0xdb, 0x00, 0x0174, + 0x00, 0x0176, 0x00, + /* a-z */ + 0xe2, 0x00, 0x0109, 0x00, 0xea, 0x00, 0x011d, 0x0125, 0xee, 0x0135, 0x00, + 0x00, 0x00, 0x00, 0xf4, 0x00, 0x00, 0x00, 0x015d, 0x00, 0xfb, 0x00, 0x0175, + 0x00, 0x0177, 0x00 + }, + /* 0xC4 */ + { + /* A-Z */ + 0xc3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0128, 0x00, 0x00, 0x00, 0x00, + 0xd1, 0xd5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0168, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0xe3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0129, 0x00, 0x00, 0x00, 0x00, + 0xf1, 0xf5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0169, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xC5 */ + { + /* A-Z */ + 0x0100, 0x00, 0x00, 0x00, 0x0112, 0x00, 0x00, 0x00, 0x012a, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x014c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016a, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0x0101, 0x00, 0x00, 0x00, 0x0113, 0x00, 0x00, 0x00, 0x012b, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x014d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016b, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xC6 */ + { + /* A-Z */ + 0x0102, 0x00, 0x00, 0x00, 0x00, 0x00, 0x011e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016c, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0x0103, 0x00, 0x00, 0x00, 0x00, 0x00, 0x011f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016d, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xC7 */ + { + /* A-Z */ + 0x00, 0x010a, 0x00, 0x0116, 0x00, 0x0120, 0x00, 0x0130, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x017b, + /* a-z */ + 0x00, 0x010b, 0x00, 0x0117, 0x00, 0x0121, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x017c + }, + /* 0xC8 */ + { + /* A-Z */ + 0xc4, 0x00, 0x00, 0x00, 0xcb, 0x00, 0x00, 0x00, 0xcf, 0x00, 0x00, 0x00, 0x00, + 0x00, 0xd6, 0x00, 0x00, 0x00, 0x00, 0x00, 0xdc, 0x00, 0x00, 0x00, 0x0178, 0x00, + /* a-z */ + 0xe4, 0x00, 0x00, 0x00, 0xeb, 0x00, 0x00, 0x00, 0xef, 0x00, 0x00, 0x00, 0x00, + 0x00, 0xf6, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfc, 0x00, 0x00, 0x00, 0xff, 0x00 + }, + /* 0xC9 */ + { + 0x00, + }, + /* 0xCA */ + { + /* A-Z */ + 0xc5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016e, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0xe5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x016f, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xCB */ + { + /* A-Z */ + 0x00, 0x00, 0xc7, 0x00, 0x00, 0x00, 0x0122, 0x00, 0x00, 0x00, 0x0136, 0x013b, 0x00, + 0x0145, 0x00, 0x00, 0x00, 0x0156, 0x015e, 0x0162, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0x00, 0x00, 0xe7, 0x00, 0x00, 0x00, 0x0123, 0x00, 0x00, 0x00, 0x0137, 0x013c, 0x00, + 0x0146, 0x00, 0x00, 0x00, 0x0157, 0x015f, 0x0163, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xCC */ + { + 0x00, + }, + /* 0xCD */ + { + /* A-Z */ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x0150, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0170, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x0151, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0171, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xCE */ + { + /* A-Z */ + 0x0104, 0x00, 0x00, 0x00, 0x0118, 0x00, 0x00, 0x00, 0x012e, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0172, 0x00, 0x00, 0x00, 0x00, 0x00, + /* a-z */ + 0x0105, 0x00, 0x00, 0x00, 0x0119, 0x00, 0x00, 0x00, 0x012f, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0173, 0x00, 0x00, 0x00, 0x00, 0x00 + }, + /* 0xCF */ + { + /* A-Z */ + 0x00, 0x00, 0x010c, 0x010e, 0x011a, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x013d, + 0x00, 0x0147, 0x00, 0x00, 0x00, 0x0158, 0x0160, 0x0164, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x017d, + /* a-z */ + 0x00, 0x00, 0x010d, 0x010f, 0x011b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x013e, + 0x00, 0x0148, 0x00, 0x00, 0x00, 0x0159, 0x0161, 0x0165, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x017e + } +}; + +#endif diff --git a/src/dvb/dvb_support.c b/src/dvb/dvb_support.c index 94c6be2a..2a6f31fe 100644 --- a/src/dvb/dvb_support.c +++ b/src/dvb/dvb_support.c @@ -22,7 +22,6 @@ #include #include #include -#include #include #include @@ -35,18 +34,199 @@ #include "dvb_support.h" #include "dvb.h" -/** - * - */ -static iconv_t convert_iso_8859[16]; -static iconv_t convert_utf8; -static iconv_t convert_latin1; +#ifdef CONFIG_DVBCONV // Use builtin charset conversion +#define dvbconv_t int +static dvbconv_t convert_iso_8859[16]; +static dvbconv_t convert_utf8; +static dvbconv_t convert_latin1; +#include "dvb_charset_tables.h" -static iconv_t +#define CONV_UTF8 14 +#define CONV_ISO6937 15 +void +dvb_conversion_init(void) +{ + int i; + convert_utf8 = CONV_UTF8; + convert_iso_8859[0] = -1; + for (i=1; i<=11; i++) { + convert_iso_8859[i] = i-1; + } + convert_iso_8859[12] = -1; // There is no ISO-8859-12 + for (i=13; i<=15; i++) { + convert_iso_8859[i] = i-2; + } + convert_latin1 = CONV_ISO6937; +} + +static inline int encode_utf8(uint16_t c, char *outb, int outleft) +{ + if (c <= 0x7F && outleft >= 1) { + *outb = c; + return 1; + } else if (c <= 0x7FF && outleft >=2) { + *outb++ = ((c >> 6) & 0x1F) | 0xC0; + *outb++ = ( c & 0x3F) | 0x80; + return 2; + } else if (c <= 0xFFFF && outleft >= 3) { + *outb++ = ((c >> 12) & 0x0F) | 0xE0; + *outb++ = ((c >> 6) & 0x3F) | 0x80; + *outb++ = ( c & 0x3F) | 0x80; + return 3; + } else if (c <= 0x10FFFF && outleft >= 4) { + *outb++ = ((c >> 18) & 0x07) | 0xF0; + *outb++ = ((c >> 12) & 0x3F) | 0x80; + *outb++ = ((c >> 6) & 0x3F) | 0x80; + *outb++ = ( c & 0x3F) | 0x80; + return 4; + } else { + return -1; + } +} + +static inline size_t conv_utf8(const uint8_t *src, size_t srclen, + char *dst, size_t *dstlen) +{ + while (srclen>0 && (*dstlen)>0) { + *dst = (char) *src; + srclen--; (*dstlen)--; + src++; dst++; + } + if (srclen>0) { + errno = E2BIG; + return -1; + } + return 0; +} + +static inline size_t conv_8859(dvbconv_t conv, + const uint8_t *src, size_t srclen, + char *dst, size_t *dstlen) +{ + uint16_t *table = conv_8859_table[conv]; + + while (srclen>0 && (*dstlen)>0) { + uint8_t c = *src; + if (c <= 0x7f) { + // lower half of iso-8859-* is identical to utf-8 + *dst = (char) *src; + (*dstlen)--; + dst++; + } else if (c <= 0x9f) { + // codes 0x80 - 0x9f (control codes) are mapped to ' ' + *dst = ' '; + (*dstlen)--; + dst++; + } else { + // map according to character table, skipping + // unmapped chars (value 0 in the table) + uint16_t uc = table[c-0xa0]; + if (uc != 0) { + int len = encode_utf8(uc, dst, *dstlen); + if (len == -1) { + errno = E2BIG; + return -1; + } else { + (*dstlen) -= len; + dst += len; + } + } + } + srclen--; + src++; + } + if (srclen>0) { + errno = E2BIG; + return -1; + } + return 0; +} + +static inline size_t conv_6937(const uint8_t *src, size_t srclen, + char *dst, size_t *dstlen) +{ + while (srclen>0 && (*dstlen)>0) { + uint8_t c = *src; + if (c <= 0x7f) { + // lower half of iso6937 is identical to utf-8 + *dst = (char) *src; + (*dstlen)--; + dst++; + } else if (c <= 0x9f) { + // codes 0x80 - 0x9f (control codes) are mapped to ' ' + *dst = ' '; + (*dstlen)--; + dst++; + } else { + uint16_t uc; + if (c <= 0x9f) { + // map two-byte sequence, skipping illegal combinations. + if (srclen<2) { + errno = EINVAL; + return -1; + } + srclen--; + src++; + uint8_t c2 = *src; + if (c2 == 0x20) { + uc = iso6937_lone_accents[c-0xc0]; + } else if (c2 >= 0x41 && c2 <= 0x5a) { + uc = iso6937_multi_byte[c-0xc0][c2-0x41]; + } else if (c2 >= 0x61 && c2 <= 0x7a) { + uc = iso6937_multi_byte[c-0xc0][c2-0x61]; + } else { + uc = 0; + } + } else { + // map according to single character table, skipping + // unmapped chars (value 0 in the table) + uc = iso6937_single_byte[c-0xa0]; + } + if (uc != 0) { + int len = encode_utf8(uc, dst, *dstlen); + if (len == -1) { + errno = E2BIG; + return -1; + } else { + (*dstlen) -= len; + dst += len; + } + } + } + srclen--; + src++; + } + if (srclen>0) { + errno = E2BIG; + return -1; + } + return 0; +} + +static size_t dvb_convert(dvbconv_t conv, + const uint8_t *src, size_t srclen, + char *dst, size_t *dstlen) +{ + switch (conv) { + case CONV_UTF8: return conv_utf8(src, srclen, dst, dstlen); + case CONV_ISO6937: return conv_6937(src, srclen, dst, dstlen); + default: return conv_8859(conv, src, srclen, dst, dstlen); + } +} + +#else // use iconv for charset conversion + +#include +#define dvbconv_t iconv_t +static dvbconv_t convert_iso_8859[16]; +static dvbconv_t convert_utf8; +static dvbconv_t convert_latin1; + +static dvbconv_t dvb_iconv_open(const char *srcencoding) { - iconv_t ic; + dvbconv_t ic; ic = iconv_open("UTF-8", srcencoding); return ic; } @@ -64,11 +244,54 @@ dvb_conversion_init(void) convert_utf8 = dvb_iconv_open("UTF-8"); convert_latin1 = dvb_iconv_open("ISO6937"); - if(convert_latin1 == (iconv_t)(-1)) { + if(convert_latin1 == (dvbconv_t)(-1)) { convert_latin1 = dvb_iconv_open("ISO_8859-1"); } } +static size_t dvb_convert(dvbconv_t ic, + const uint8_t *src, size_t srclen, + char *dst, size_t *outlen) { + char *in, *out; + size_t inlen; + unsigned char *tmp; + int i; + int r; + tmp = alloca(srclen + 1); + memcpy(tmp, src, srclen); + tmp[srclen] = 0; + + /* Escape control codes */ + if(ic != convert_utf8) { + for(i = 0; i < srclen; i++) { + if(tmp[i] >= 0x80 && tmp[i] <= 0x9f) + tmp[i] = ' '; + } + } + + out = dst; + in = (char *)tmp; + inlen = srclen; + + while(inlen > 0) { + r = iconv(ic, &in, &inlen, &out, outlen); + + if(r == (size_t) -1) { + if(errno == EILSEQ) { + in++; + inlen--; + continue; + } else { + return -1; + } + } + } + return 0; +} +#endif +/** + * + */ /* @@ -79,14 +302,9 @@ dvb_conversion_init(void) int dvb_get_string(char *dst, size_t dstlen, const uint8_t *src, size_t srclen, char *dvb_default_charset) { - iconv_t ic; - int len; - char *in, *out; - size_t inlen, outlen; - int utf8 = 0; + dvbconv_t ic; + size_t len, outlen; int i; - unsigned char *tmp; - int r; if(srclen < 1) { *dst = 0; @@ -118,7 +336,6 @@ dvb_get_string(char *dst, size_t dstlen, const uint8_t *src, size_t srclen, char case 0x15: ic = convert_utf8; - utf8 = 1; break; case 0x16 ... 0x1f: return -1; @@ -141,41 +358,13 @@ dvb_get_string(char *dst, size_t dstlen, const uint8_t *src, size_t srclen, char return 0; } - tmp = alloca(srclen + 1); - memcpy(tmp, src, srclen); - tmp[srclen] = 0; - - - /* Escape control codes */ - - if(!utf8) { - for(i = 0; i < srclen; i++) { - if(tmp[i] >= 0x80 && tmp[i] <= 0x9f) - tmp[i] = ' '; - } - } - - if(ic == (iconv_t) -1) + if(ic == (dvbconv_t) -1) return -1; - inlen = srclen; outlen = dstlen - 1; - out = dst; - in = (char *)tmp; - - while(inlen > 0) { - r = iconv(ic, &in, &inlen, &out, &outlen); - - if(r == (size_t) -1) { - if(errno == EILSEQ) { - in++; - inlen--; - continue; - } else { - return -1; - } - } + if (dvb_convert(ic, src, srclen, dst, &outlen) == -1) { + return -1; } len = dstlen - outlen - 1; diff --git a/support/configure.inc b/support/configure.inc index 048e74bd..588daada 100644 --- a/support/configure.inc +++ b/support/configure.inc @@ -25,6 +25,7 @@ CONFIG_LIST=" linuxdvb v4l execinfo + dvbconv " die() { @@ -133,4 +134,4 @@ EOF echo "Final binary is: build.${PLATFORM}/tvheadend" die -} \ No newline at end of file +}