X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fsiconv.c;h=36dc4b263f0caea687922b6dc616eb9b25960ebd;hp=b6bcf0e466d31a5b09926e85173bdcfd4228110a;hb=96c6e58f286787106e4a7b3bb3900a36051968d6;hpb=ee6ab2ee3a9ee1a8c65d7272ec7fba1d886f5af0 diff --git a/src/siconv.c b/src/siconv.c index b6bcf0e..36dc4b2 100644 --- a/src/siconv.c +++ b/src/siconv.c @@ -21,9 +21,6 @@ #include #include #include -#if HAVE_WCHAR_H -#include -#endif #if HAVE_ICONV_H #include @@ -33,262 +30,28 @@ #include #include "iconv-p.h" -yaz_conv_func_t yaz_marc8_42_conv; -yaz_conv_func_t yaz_marc8_45_conv; -yaz_conv_func_t yaz_marc8_67_conv; -yaz_conv_func_t yaz_marc8_62_conv; -yaz_conv_func_t yaz_marc8_70_conv; -yaz_conv_func_t yaz_marc8_32_conv; -yaz_conv_func_t yaz_marc8_4E_conv; -yaz_conv_func_t yaz_marc8_51_conv; -yaz_conv_func_t yaz_marc8_33_conv; -yaz_conv_func_t yaz_marc8_34_conv; -yaz_conv_func_t yaz_marc8_53_conv; -yaz_conv_func_t yaz_marc8_31_conv; - struct yaz_iconv_struct { int my_errno; int init_flag; +#if 0 size_t (*init_handle)(yaz_iconv_t cd, unsigned char *inbuf, size_t inbytesleft, size_t *no_read); unsigned long (*read_handle)(yaz_iconv_t cd, unsigned char *inbuf, size_t inbytesleft, size_t *no_read); - int g0_mode; - int g1_mode; - - int comb_offset; - int comb_size; - unsigned long comb_x[8]; - size_t comb_no_read[8]; +#endif size_t no_read_x; unsigned long unget_x; #if HAVE_ICONV_H iconv_t iconv_cd; #endif struct yaz_iconv_encoder_s encoder; + struct yaz_iconv_decoder_s decoder; }; -static unsigned long yaz_read_ISO8859_1(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read) -{ - unsigned long x = inp[0]; - *no_read = 1; - return x; -} - -#if HAVE_WCHAR_H -static unsigned long yaz_read_wchar_t(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read) -{ - unsigned long x = 0; - - if (inbytesleft < sizeof(wchar_t)) - { - cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */ - *no_read = 0; - } - else - { - wchar_t wch; - memcpy(&wch, inp, sizeof(wch)); - x = wch; - *no_read = sizeof(wch); - } - return x; -} -#endif - - -static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read, - int *comb); - -static unsigned long yaz_read_marc8(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read) -{ - unsigned long x; - if (cd->comb_offset < cd->comb_size) - { - *no_read = cd->comb_no_read[cd->comb_offset]; - x = cd->comb_x[cd->comb_offset]; - - /* special case for double-diacritic combining characters, - INVERTED BREVE and DOUBLE TILDE. - We'll increment the no_read counter by 1, since we want to skip over - the processing of the closing ligature character - */ - /* this code is no longer necessary.. our handlers code in - yaz_marc8_?_conv (generated by charconv.tcl) now returns - 0 and no_read=1 when a sequence does not match the input. - The SECOND HALFs in codetables.xml produces a non-existant - entry in the conversion trie.. Hence when met, the input byte is - skipped as it should (in yaz_iconv) - */ -#if 0 - if (x == 0x0361 || x == 0x0360) - *no_read += 1; -#endif - cd->comb_offset++; - return x; - } - - cd->comb_offset = 0; - for (cd->comb_size = 0; cd->comb_size < 8; cd->comb_size++) - { - int comb = 0; - - if (inbytesleft == 0 && cd->comb_size) - { - cd->my_errno = YAZ_ICONV_EINVAL; - x = 0; - *no_read = 0; - break; - } - x = yaz_read_marc8_comb(cd, inp, inbytesleft, no_read, &comb); - if (!comb || !x) - break; - cd->comb_x[cd->comb_size] = x; - cd->comb_no_read[cd->comb_size] = *no_read; - inp += *no_read; - inbytesleft = inbytesleft - *no_read; - } - return x; -} - -static unsigned long yaz_read_marc8s(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read) -{ - unsigned long x = yaz_read_marc8(cd, inp, inbytesleft, no_read); - if (x && cd->comb_size == 1) - { - if (yaz_iso_8859_1_lookup_x12(x, cd->comb_x[0], &x)) - { - *no_read += cd->comb_no_read[0]; - cd->comb_size = 0; - } - } - return x; -} - -static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp, - size_t inbytesleft, size_t *no_read, - int *comb) -{ - *no_read = 0; - while (inbytesleft > 0 && *inp == 27) - { - int *modep = &cd->g0_mode; - size_t inbytesleft0 = inbytesleft; - - inbytesleft--; - inp++; - if (inbytesleft == 0) - goto incomplete; - if (*inp == '$') /* set with multiple bytes */ - { - inbytesleft--; - inp++; - } - if (inbytesleft == 0) - goto incomplete; - if (*inp == '(' || *inp == ',') /* G0 */ - { - inbytesleft--; - inp++; - } - else if (*inp == ')' || *inp == '-') /* G1 */ - { - inbytesleft--; - inp++; - modep = &cd->g1_mode; - } - if (inbytesleft == 0) - goto incomplete; - if (*inp == '!') /* ANSEL is a special case */ - { - inbytesleft--; - inp++; - } - if (inbytesleft == 0) - goto incomplete; - *modep = *inp++; /* Final character */ - inbytesleft--; - - (*no_read) += inbytesleft0 - inbytesleft; - } - if (inbytesleft == 0) - return 0; - else if (*inp == ' ') - { - *no_read += 1; - return ' '; - } - else - { - unsigned long x; - size_t no_read_sub = 0; - int mode = *inp < 128 ? cd->g0_mode : cd->g1_mode; - *comb = 0; - - switch(mode) - { - case 'B': /* Basic ASCII */ - case 's': /* ASCII */ - x = yaz_marc8_42_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'E': /* ANSEL */ - x = yaz_marc8_45_conv(inp, inbytesleft, &no_read_sub, comb, 127, 128); - break; - case 'g': /* Greek */ - x = yaz_marc8_67_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'b': /* Subscripts */ - x = yaz_marc8_62_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'p': /* Superscripts */ - x = yaz_marc8_70_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case '2': /* Basic Hebrew */ - x = yaz_marc8_32_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'N': /* Basic Cyrillic */ - x = yaz_marc8_4E_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'Q': /* Extended Cyrillic */ - x = yaz_marc8_51_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case '3': /* Basic Arabic */ - x = yaz_marc8_33_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case '4': /* Extended Arabic */ - x = yaz_marc8_34_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case 'S': /* Greek */ - x = yaz_marc8_53_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - case '1': /* Chinese, Japanese, Korean (EACC) */ - x = yaz_marc8_31_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0); - break; - default: - *no_read = 0; - cd->my_errno = YAZ_ICONV_EILSEQ; - return 0; - } - *no_read += no_read_sub; - return x; - } -incomplete: - *no_read = 0; - cd->my_errno = YAZ_ICONV_EINVAL; - return 0; -} - - - int yaz_iconv_isbuiltin(yaz_iconv_t cd) { - return cd->read_handle && cd->encoder.write_handle; + return cd->decoder.read_handle && cd->encoder.write_handle; } @@ -311,6 +74,25 @@ static int prepare_encoders(yaz_iconv_t cd, const char *tocode) return 0; } +static int prepare_decoders(yaz_iconv_t cd, const char *tocode) +{ + if (yaz_marc8_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_utf8_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_ucs4_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_iso_8859_1_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_iso_5428_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_advancegreek_decoder(tocode, &cd->decoder)) + return 1; + if (yaz_wchar_decoder(tocode, &cd->decoder)) + return 1; + return 0; +} + yaz_iconv_t yaz_iconv_open(const char *tocode, const char *fromcode) { yaz_iconv_t cd = (yaz_iconv_t) xmalloc (sizeof(*cd)); @@ -321,8 +103,11 @@ yaz_iconv_t yaz_iconv_open(const char *tocode, const char *fromcode) cd->encoder.init_handle = 0; cd->encoder.destroy_handle = 0; - cd->read_handle = 0; - cd->init_handle = 0; + cd->decoder.data = 0; + cd->decoder.read_handle = 0; + cd->decoder.init_handle = 0; + cd->decoder.destroy_handle = 0; + cd->my_errno = YAZ_ICONV_UNKNOWN; /* a useful hack: if fromcode has leading @, @@ -331,37 +116,13 @@ yaz_iconv_t yaz_iconv_open(const char *tocode, const char *fromcode) fromcode++; else { - if (!yaz_matchstr(fromcode, "UTF8")) - { - cd->read_handle = yaz_read_UTF8; - cd->init_handle = yaz_init_UTF8; - } - else if (!yaz_matchstr(fromcode, "ISO88591")) - cd->read_handle = yaz_read_ISO8859_1; - else if (!yaz_matchstr(fromcode, "UCS4")) - cd->read_handle = yaz_read_UCS4; - else if (!yaz_matchstr(fromcode, "UCS4LE")) - cd->read_handle = yaz_read_UCS4LE; - else if (!yaz_matchstr(fromcode, "MARC8")) - cd->read_handle = yaz_read_marc8; - else if (!yaz_matchstr(fromcode, "MARC8s")) - cd->read_handle = yaz_read_marc8s; - else if (!yaz_matchstr(fromcode, "advancegreek")) - cd->read_handle = yaz_read_advancegreek; - else if (!yaz_matchstr(fromcode, "iso54281984")) - cd->read_handle = yaz_read_iso5428_1984; - else if (!yaz_matchstr(fromcode, "iso5428:1984")) - cd->read_handle = yaz_read_iso5428_1984; -#if HAVE_WCHAR_H - else if (!yaz_matchstr(fromcode, "WCHAR_T")) - cd->read_handle = yaz_read_wchar_t; -#endif prepare_encoders(cd, tocode); + prepare_decoders(cd, fromcode); } - if (cd->read_handle && cd->encoder.write_handle) + if (cd->decoder.read_handle && cd->encoder.write_handle) { #if HAVE_ICONV_H - cd->iconv_cd = 0; + cd->iconv_cd = (iconv_t) (-1); #endif ; } @@ -390,7 +151,7 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, size_t r = 0; #if HAVE_ICONV_H - if (cd->iconv_cd) + if (cd->iconv_cd != (iconv_t) (-1)) { size_t r = iconv(cd->iconv_cd, inbuf, inbytesleft, outbuf, outbytesleft); @@ -421,22 +182,21 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, if (cd->init_flag) { cd->my_errno = YAZ_ICONV_UNKNOWN; - cd->g0_mode = 'B'; - cd->g1_mode = 'E'; - cd->comb_offset = cd->comb_size = 0; - if (cd->encoder.init_handle) (*cd->encoder.init_handle)(&cd->encoder); cd->unget_x = 0; cd->no_read_x = 0; - if (cd->init_handle && inbuf && *inbuf) + if (cd->decoder.init_handle) { size_t no_read = 0; - size_t r = (cd->init_handle)(cd, (unsigned char *) *inbuf, - *inbytesleft, &no_read); + size_t r = (cd->decoder.init_handle)( + cd, &cd->decoder, + inbuf ? (unsigned char *) *inbuf : 0, + inbytesleft ? *inbytesleft : 0, + &no_read); if (r) { if (cd->my_errno == YAZ_ICONV_EINVAL) @@ -444,8 +204,10 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, cd->init_flag = 0; return r; } - *inbytesleft -= no_read; - *inbuf += no_read; + if (inbytesleft) + *inbytesleft -= no_read; + if (inbuf) + *inbuf += no_read; } } cd->init_flag = 0; @@ -483,8 +245,9 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, r = *inbuf - inbuf0; break; } - x = (*cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft, - &no_read); + x = (*cd->decoder.read_handle)( + cd, &cd->decoder, + (unsigned char *) *inbuf, *inbytesleft, &no_read); if (no_read == 0) { r = (size_t)(-1); @@ -522,11 +285,13 @@ int yaz_iconv_error(yaz_iconv_t cd) int yaz_iconv_close(yaz_iconv_t cd) { #if HAVE_ICONV_H - if (cd->iconv_cd) + if (cd->iconv_cd != (iconv_t) (-1)) iconv_close(cd->iconv_cd); #endif if (cd->encoder.destroy_handle) (*cd->encoder.destroy_handle)(&cd->encoder); + if (cd->decoder.destroy_handle) + (*cd->decoder.destroy_handle)(&cd->decoder); xfree(cd); return 0; }