From 475b627e585a1f3132378733e4dfc0076326f6bf Mon Sep 17 00:00:00 2001 From: =?utf8?q?Marian=20=C4=8Eurkovi=C4=8D?= Date: Mon, 14 Sep 2009 09:59:50 +0200 Subject: [PATCH] Provide local routine for ISO_6937 -> UTF-8 conversions --- src/extras/libc.c | 292 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 292 insertions(+) diff --git a/src/extras/libc.c b/src/extras/libc.c index 9ecb4700df..60c7139abf 100644 --- a/src/extras/libc.c +++ b/src/extras/libc.c @@ -210,11 +210,294 @@ char *vlc_gettext( const char *msgid ) #endif } +/***************************************************************************** + * Local conversion routine from ISO_6937 to UTF-8 charset. Support for this + * is still missing in libiconv, hence multiple operating systems lack it. + * The conversion table adds Euro sign (0xA4) as per ETSI EN 300 468 Annex A + *****************************************************************************/ +#ifndef __linux__ +static const uint16_t to_ucs4[128] = +{ + /* 0x80 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087, + /* 0x88 */ 0x0088, 0x0089, 0x008a, 0x008b, 0x008c, 0x008d, 0x008e, 0x008f, + /* 0x90 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097, + /* 0x98 */ 0x0098, 0x0099, 0x009a, 0x009b, 0x009c, 0x009d, 0x009e, 0x009f, + /* 0xa0 */ 0x00a0, 0x00a1, 0x00a2, 0x00a3, 0x20ac, 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, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff, + /* 0xc8 */ 0xffff, 0x0000, 0xffff, 0xffff, 0x0000, 0xffff, 0xffff, 0xffff, + /* 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 +}; + +/* The outer array range runs from 0xc1 to 0xcf, the inner range from 0x40 + to 0x7f. */ +static const uint16_t to_ucs4_comb[15][64] = +{ + /* 0xc1 */ + { + /* 0x40 */ 0x0000, 0x00c0, 0x0000, 0x0000, 0x0000, 0x00c8, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x00cc, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00d2, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00d9, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e0, 0x0000, 0x0000, 0x0000, 0x00e8, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x00ec, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00f2, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00f9, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc2 */ + { + /* 0x40 */ 0x0000, 0x00c1, 0x0000, 0x0106, 0x0000, 0x00c9, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x00cd, 0x0000, 0x0000, 0x0139, 0x0000, 0x0143, 0x00d3, + /* 0x50 */ 0x0000, 0x0000, 0x0154, 0x015a, 0x0000, 0x00da, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x00dd, 0x0179, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e1, 0x0000, 0x0107, 0x0000, 0x00e9, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x00ed, 0x0000, 0x0000, 0x013a, 0x0000, 0x0144, 0x00f3, + /* 0x70 */ 0x0000, 0x0000, 0x0155, 0x015b, 0x0000, 0x00fa, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x00fd, 0x017a, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc3 */ + { + /* 0x40 */ 0x0000, 0x00c2, 0x0000, 0x0108, 0x0000, 0x00ca, 0x0000, 0x011c, + /* 0x48 */ 0x0124, 0x00ce, 0x0134, 0x0000, 0x0000, 0x0000, 0x0000, 0x00d4, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x015c, 0x0000, 0x00db, 0x0000, 0x0174, + /* 0x58 */ 0x0000, 0x0176, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e2, 0x0000, 0x0109, 0x0000, 0x00ea, 0x0000, 0x011d, + /* 0x68 */ 0x0125, 0x00ee, 0x0135, 0x0000, 0x0000, 0x0000, 0x0000, 0x00f4, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x015d, 0x0000, 0x00fb, 0x0000, 0x0175, + /* 0x78 */ 0x0000, 0x0177, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc4 */ + { + /* 0x40 */ 0x0000, 0x00c3, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x0128, 0x0000, 0x0000, 0x0000, 0x0000, 0x00d1, 0x00d5, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0168, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e3, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x0129, 0x0000, 0x0000, 0x0000, 0x0000, 0x00f1, 0x00f5, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0169, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc5 */ + { + /* 0x40 */ 0x0000, 0x0100, 0x0000, 0x0000, 0x0000, 0x0112, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x012a, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x014c, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016a, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0101, 0x0000, 0x0000, 0x0000, 0x0113, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x012b, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x014d, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016b, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc6 */ + { + /* 0x40 */ 0x0000, 0x0102, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x011e, + /* 0x48 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016c, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0103, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x011f, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016d, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc7 */ + { + /* 0x40 */ 0x0000, 0x0000, 0x0000, 0x010a, 0x0000, 0x0116, 0x0000, 0x0120, + /* 0x48 */ 0x0000, 0x0130, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x017b, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0000, 0x0000, 0x010b, 0x0000, 0x0117, 0x0000, 0x0121, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x017c, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc8 */ + { + /* 0x40 */ 0x0000, 0x00c4, 0x0000, 0x0000, 0x0000, 0x00cb, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x00cf, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00d6, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00dc, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0178, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e4, 0x0000, 0x0000, 0x0000, 0x00eb, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x00ef, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00f6, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x00fc, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x00ff, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xc9 */ + { + 0x0000, + }, + /* 0xca */ + { + /* 0x40 */ 0x0000, 0x00c5, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016e, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x00e5, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x016f, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xcb */ + { + /* 0x40 */ 0x0000, 0x0000, 0x0000, 0x00c7, 0x0000, 0x0000, 0x0000, 0x0122, + /* 0x48 */ 0x0000, 0x0000, 0x0000, 0x0136, 0x013b, 0x0000, 0x0145, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0156, 0x015e, 0x0162, 0x0000, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0000, 0x0000, 0x00e7, 0x0000, 0x0000, 0x0000, 0x0123, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0137, 0x013c, 0x0000, 0x0146, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0157, 0x015f, 0x0163, 0x0000, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xcc */ + { + 0x0000, + }, + /* 0xcd */ + { + /* 0x40 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0150, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0170, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0151, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0171, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xce */ + { + /* 0x40 */ 0x0000, 0x0104, 0x0000, 0x0000, 0x0000, 0x0118, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x012e, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0172, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0105, 0x0000, 0x0000, 0x0000, 0x0119, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x012f, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0173, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + }, + /* 0xcf */ + { + /* 0x40 */ 0x0000, 0x0000, 0x0000, 0x010c, 0x010e, 0x011a, 0x0000, 0x0000, + /* 0x48 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x013d, 0x0000, 0x0147, 0x0000, + /* 0x50 */ 0x0000, 0x0000, 0x0158, 0x0160, 0x0164, 0x0000, 0x0000, 0x0000, + /* 0x58 */ 0x0000, 0x0000, 0x017d, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, + /* 0x60 */ 0x0000, 0x0000, 0x0000, 0x010d, 0x010f, 0x011b, 0x0000, 0x0000, + /* 0x68 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x013e, 0x0000, 0x0148, 0x0000, + /* 0x70 */ 0x0000, 0x0000, 0x0159, 0x0161, 0x0165, 0x0000, 0x0000, 0x0000, + /* 0x78 */ 0x0000, 0x0000, 0x017e, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000 + } +}; + +static size_t ISO6937toUTF8( const unsigned char **inbuf, size_t *inbytesleft, + unsigned char **outbuf, size_t *outbytesleft ) + + +{ + if( !inbuf || !(*inbuf) ) + return (size_t)(0); /* Reset state requested */ + + const unsigned char *iptr = *inbuf; + const unsigned char *iend = iptr + *inbytesleft; + unsigned char *optr = *outbuf; + unsigned char *oend = optr + *outbytesleft; + uint16_t ch; + int err = 0; + + while ( iptr < iend ) + { + if( *iptr < 0x80 ) + { + if( optr >= oend ) + { + err = E2BIG; + break; /* No space in outbuf */ + } + *optr++ = *iptr++; + continue; + } + + + if ( optr + 2 >= oend ) + { + err = E2BIG; + break; /* No space in outbuf for multibyte char */ + } + + ch = to_ucs4[*iptr - 0x80]; + + if( ch == 0xffff ) + { + /* Composed character */ + if ( iptr + 1 >= iend ) + { + err = EINVAL; + break; /* No next character */ + } + if ( iptr[1] < 0x40 || iptr[1] >= 0x80 || + !(ch = to_ucs4_comb[iptr[0] - 0xc1][iptr[1] - 0x40]) ) + { + err = EILSEQ; + break; /* Illegal combination */ + } + iptr += 2; + + } + else + { + if ( !ch ) + { + err = EILSEQ; + break; + } + iptr++; + } + + if ( ch < 0x800 ) + { + optr[1] = 0x80 | (ch & 0x3f); + optr[0] = 0xc0 | (ch >> 6); + optr +=2; + } + else + { + optr[2] = 0x80 | (ch & 0x3f); + ch >>= 6; + optr[1] = 0x80 | (ch & 0x3f); + optr[0] = 0xe0 | (ch >> 6); + optr += 3; + } + + } + *inbuf = iptr; + *outbuf = optr; + *inbytesleft = iend - iptr; + *outbytesleft = oend - optr; + + if( err ) + { + errno = err; + return (size_t)(-1); + } + + return (size_t)(0); + +} +#endif + /***************************************************************************** * iconv wrapper *****************************************************************************/ vlc_iconv_t vlc_iconv_open( const char *tocode, const char *fromcode ) { +#ifndef __linux__ + if( !strcasecmp(tocode, "UTF-8") && !strcasecmp(fromcode, "ISO_6937") ) + return (vlc_iconv_t)(-2); +#endif #if defined(HAVE_ICONV) return iconv_open( tocode, fromcode ); #else @@ -225,6 +508,11 @@ vlc_iconv_t vlc_iconv_open( const char *tocode, const char *fromcode ) size_t vlc_iconv( vlc_iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft ) { +#ifndef __linux__ + if ( cd == (vlc_iconv_t)(-2) ) + return ISO6937toUTF8( (const unsigned char **)inbuf, inbytesleft, + (unsigned char **)outbuf, outbytesleft ); +#endif #if defined(HAVE_ICONV) return iconv( cd, (ICONV_CONST char **)inbuf, inbytesleft, outbuf, outbytesleft ); @@ -235,6 +523,10 @@ size_t vlc_iconv( vlc_iconv_t cd, const char **inbuf, size_t *inbytesleft, int vlc_iconv_close( vlc_iconv_t cd ) { +#ifndef __linux__ + if ( cd == (vlc_iconv_t)(-2) ) + return 0; +#endif #if defined(HAVE_ICONV) return iconv_close( cd ); #else -- 2.39.2