X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fsiconv.c;h=26f36780d449a36e9223abe6b79ed5ce4c4df4a5;hp=8916561feb3255d1f85d17324683df10f71edf43;hb=62fd62f25a6893d042c4c87f8c770929fd8cf95d;hpb=19c0b332044f06ee33eaf66ca92675d9cda8d1b9 diff --git a/src/siconv.c b/src/siconv.c index 8916561..26f3678 100644 --- a/src/siconv.c +++ b/src/siconv.c @@ -2,7 +2,7 @@ * Copyright (C) 1995-2007, Index Data ApS * See the file LICENSE for details. * - * $Id: siconv.c,v 1.34 2007-03-09 08:39:38 adam Exp $ + * $Id: siconv.c,v 1.37 2007-03-20 21:37:32 adam Exp $ */ /** * \file siconv.c @@ -83,8 +83,7 @@ struct yaz_iconv_struct { unsigned long (*read_handle)(yaz_iconv_t cd, unsigned char *inbuf, size_t inbytesleft, size_t *no_read); size_t (*write_handle)(yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last); + char **outbuf, size_t *outbytesleft); size_t (*flush_handle)(yaz_iconv_t cd, char **outbuf, size_t *outbytesleft); int marc8_esc_mode; @@ -356,6 +355,360 @@ static unsigned long yaz_read_wchar_t (yaz_iconv_t cd, unsigned char *inp, } #endif +static unsigned long yaz_read_advancegreek(yaz_iconv_t cd, unsigned char *inp, + size_t inbytesleft, size_t *no_read) +{ + unsigned long x = 0; + int shift = 0; + int tonos = 0; + int dialitika = 0; + + *no_read = 0; + while (inbytesleft > 0) + { + if (*inp == 0x9d) + { + tonos = 1; + } + else if (*inp == 0x9e) + { + dialitika = 1; + } + else if (*inp == 0x9f) + { + shift = 1; + } + else + break; + inp++; + --inbytesleft; + (*no_read)++; + } + if (inbytesleft == 0) + { + cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */ + *no_read = 0; + return 0; + } + switch (*inp) { + case 0x81: + if (shift) + if (tonos) + x = 0x0386; + else + x = 0x0391; + else + if (tonos) + x = 0x03ac; + else + x = 0x03b1; + break; + case 0x82: + if (shift) + x = 0x0392; + else + x = 0x03b2; + + break; + case 0x83: + if (shift) + x = 0x0393; + else + x = 0x03b3; + break; + case 0x84: + if (shift) + x = 0x0394; + else + x = 0x03b4; + break; + case 0x85: + if (shift) + if (tonos) + x = 0x0388; + else + x = 0x0395; + else + if (tonos) + x = 0x03ad; + else + x = 0x03b5; + break; + case 0x86: + if (shift) + x = 0x0396; + else + x = 0x03b6; + break; + case 0x87: + if (shift) + if (tonos) + x = 0x0389; + else + x = 0x0397; + else + if (tonos) + x = 0x03ae; + else + x = 0x03b7; + break; + case 0x88: + if (shift) + x = 0x0398; + else + x = 0x03b8; + break; + case 0x89: + if (shift) + if (tonos) + x = 0x038a; + else + if (dialitika) + x = 0x9e; + else + x = 0x0399; + else + if (tonos) + if (dialitika) + x = 0x0390; + else + x = 0x03af; + + else + if (dialitika) + x = 0x03ca; + else + x = 0x03b9; + break; + case 0x8a: + if (shift) + x = 0x039a; + else + x = 0x03ba; + + break; + case 0x8b: + if (shift) + x = 0x039b; + else + x = 0x03bb; + break; + case 0x8c: + if (shift) + x = 0x039c; + else + x = 0x03bc; + + break; + case 0x8d: + if (shift) + x = 0x039d; + else + x = 0x03bd; + break; + case 0x8e: + if (shift) + x = 0x039e; + else + x = 0x03be; + break; + case 0x8f: + if (shift) + if (tonos) + x = 0x038c; + else + x = 0x039f; + else + if (tonos) + x = 0x03cc; + else + x = 0x03bf; + break; + case 0x90: + if (shift) + x = 0x03a0; + else + x = 0x03c0; + break; + case 0x91: + if (shift) + x = 0x03a1; + else + x = 0x03c1; + break; + case 0x92: + x = 0x03c2; + break; + case 0x93: + if (shift) + x = 0x03a3; + else + x = 0x03c3; + break; + case 0x94: + if (shift) + x = 0x03a4; + else + x = 0x03c4; + break; + case 0x95: + if (shift) + if (tonos) + x = 0x038e; + else + if (dialitika) + x = 0x03ab; + else + x = 0x03a5; + else + if (tonos) + if (dialitika) + x = 0x03b0; + else + x = 0x03cd; + + else + if (dialitika) + x = 0x03cb; + else + x = 0x03c5; + break; + case 0x96: + if (shift) + x = 0x03a6; + else + x = 0x03c6; + break; + case 0x97: + if (shift) + x = 0x03a7; + else + x = 0x03c7; + break; + case 0x98: + if (shift) + x = 0x03a8; + else + x = 0x03c8; + + break; + + case 0x99: + if (shift) + if (tonos) + x = 0x038f; + else + x = 0x03a9; + else + if (tonos) + x = 0x03ce; + else + x = 0x03c9; + break; + default: + x = *inp; + break; + } + (*no_read)++; + + return x; +} + +static size_t yaz_write_advancegreek(yaz_iconv_t cd, unsigned long x, + char **outbuf, size_t *outbytesleft) +{ + size_t k = 0; + unsigned char *out = (unsigned char*) *outbuf; + if (*outbytesleft < 3) + { + cd->my_errno = YAZ_ICONV_E2BIG; /* not room for output */ + return (size_t)(-1); + } + switch (x) + { + case 0x03ac : out[k++]=0x9d; out[k++]=0x81; break; + case 0x03ad : out[k++]=0x9d; out[k++]=0x85; break; + case 0x03ae : out[k++]=0x9d; out[k++]=0x87; break; + case 0x03af : out[k++]=0x9d; out[k++]=0x89; break; + case 0x03cc : out[k++]=0x9d; out[k++]=0x8f; break; + case 0x03cd : out[k++]=0x9d; out[k++]=0x95; break; + case 0x03ce : out[k++]=0x9d; out[k++]=0x99; break; + case 0x0390 : out[k++]=0x9d; out[k++]=0x9e; out[k++]=0x89; break; + case 0x03b0 : out[k++]=0x9d; out[k++]=0x9e; out[k++]=0x95; break; + case 0x0386 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x81; break; + case 0x0388 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x85; break; + case 0x0389 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x87; break; + case 0x038a : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x89; break; + case 0x038c : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x8f; break; + case 0x038e : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x95; break; + case 0x038f : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x99; break; + case 0x03ca : out[k++]=0x9e; out[k++]=0x89; break; + case 0x03cb : out[k++]=0x9e; out[k++]=0x95; break; + case 0x03aa : out[k++]=0x9e; out[k++]=0x9f; out[k++]=0x89; break; + case 0x03ab : out[k++]=0x9e; out[k++]=0x9f; out[k++]=0x95; break; + case 0x0391 : out[k++]=0x9f; out[k++]=0x81; break; + case 0x0392 : out[k++]=0x9f; out[k++]=0x82; break; + case 0x0393 : out[k++]=0x9f; out[k++]=0x83; break; + case 0x0394 : out[k++]=0x9f; out[k++]=0x84; break; + case 0x0395 : out[k++]=0x9f; out[k++]=0x85; break; + case 0x0396 : out[k++]=0x9f; out[k++]=0x86; break; + case 0x0397 : out[k++]=0x9f; out[k++]=0x87; break; + case 0x0398 : out[k++]=0x9f; out[k++]=0x88; break; + case 0x0399 : out[k++]=0x9f; out[k++]=0x89; break; + case 0x039a : out[k++]=0x9f; out[k++]=0x8a; break; + case 0x039b : out[k++]=0x9f; out[k++]=0x8b; break; + case 0x039c : out[k++]=0x9f; out[k++]=0x8c; break; + case 0x039d : out[k++]=0x9f; out[k++]=0x8d; break; + case 0x039e : out[k++]=0x9f; out[k++]=0x8e; break; + case 0x039f : out[k++]=0x9f; out[k++]=0x8f; break; + case 0x03a0 : out[k++]=0x9f; out[k++]=0x90; break; + case 0x03a1 : out[k++]=0x9f; out[k++]=0x91; break; + case 0x03a3 : out[k++]=0x9f; out[k++]=0x93; break; + case 0x03a4 : out[k++]=0x9f; out[k++]=0x94; break; + case 0x03a5 : out[k++]=0x9f; out[k++]=0x95; break; + case 0x03a6 : out[k++]=0x9f; out[k++]=0x96; break; + case 0x03a7 : out[k++]=0x9f; out[k++]=0x97; break; + case 0x03a8 : out[k++]=0x9f; out[k++]=0x98; break; + case 0x03a9 : out[k++]=0x9f; out[k++]=0x99; break; + case 0x03b1 : out[k++]=0x81; break; + case 0x03b2 : out[k++]=0x82; break; + case 0x03b3 : out[k++]=0x83; break; + case 0x03b4 : out[k++]=0x84; break; + case 0x03b5 : out[k++]=0x85; break; + case 0x03b6 : out[k++]=0x86; break; + case 0x03b7 : out[k++]=0x87; break; + case 0x03b8 : out[k++]=0x88; break; + case 0x03b9 : out[k++]=0x89; break; + case 0x03ba : out[k++]=0x8a; break; + case 0x03bb : out[k++]=0x8b; break; + case 0x03bc : out[k++]=0x8c; break; + case 0x03bd : out[k++]=0x8d; break; + case 0x03be : out[k++]=0x8e; break; + case 0x03bf : out[k++]=0x8f; break; + case 0x03c0 : out[k++]=0x90; break; + case 0x03c1 : out[k++]=0x91; break; + case 0x03c2 : out[k++]=0x92; break; + case 0x03c3 : out[k++]=0x93; break; + case 0x03c4 : out[k++]=0x94; break; + case 0x03c5 : out[k++]=0x95; break; + case 0x03c6 : out[k++]=0x96; break; + case 0x03c7 : out[k++]=0x96; break; + case 0x03c8 : out[k++]=0x98; break; + case 0x03c9 : out[k++]=0x99; break; + default: + if (x > 255) + { + cd->my_errno = YAZ_ICONV_EILSEQ; + return (size_t) -1; + } + out[k++] = x; + break; + } + *outbytesleft -= k; + (*outbuf) += k; + return 0; +} + static unsigned long yaz_read_marc8_comb (yaz_iconv_t cd, unsigned char *inp, size_t inbytesleft, size_t *no_read, @@ -394,6 +747,7 @@ static unsigned long yaz_read_marc8 (yaz_iconv_t cd, unsigned char *inp, 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; @@ -509,8 +863,7 @@ static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp, } static size_t yaz_write_UTF8(yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { return yaz_write_UTF8_char(x, outbuf, outbytesleft, &cd->my_errno); } @@ -575,10 +928,8 @@ size_t yaz_write_UTF8_char(unsigned long x, return 0; } - static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { /* list of two char unicode sequence that, when combined, are equivalent to single unicode chars that can be represented in @@ -613,7 +964,7 @@ static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x, cd->compose_char = 0; } - if (!last && x > 32 && x < 127 && cd->compose_char == 0) + if (x > 32 && x < 127 && cd->compose_char == 0) { cd->compose_char = x; return 0; @@ -634,10 +985,27 @@ static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x, return 0; } +static size_t yaz_flush_ISO8859_1(yaz_iconv_t cd, + char **outbuf, size_t *outbytesleft) +{ + if (cd->compose_char) + { + unsigned char *outp = (unsigned char *) *outbuf; + if (*outbytesleft < 1) + { + cd->my_errno = YAZ_ICONV_E2BIG; + return (size_t)(-1); + } + *outp++ = (unsigned char) cd->compose_char; + (*outbytesleft)--; + *outbuf = (char *) outp; + cd->compose_char = 0; + } + return 0; +} static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { unsigned char *outp = (unsigned char *) *outbuf; if (*outbytesleft >= 4) @@ -658,8 +1026,7 @@ static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x, } static size_t yaz_write_UCS4LE (yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { unsigned char *outp = (unsigned char *) *outbuf; if (*outbytesleft >= 4) @@ -687,7 +1054,7 @@ static unsigned long lookup_marc8(yaz_iconv_t cd, char *utf8_outbuf = utf8_buf; size_t utf8_outbytesleft = sizeof(utf8_buf)-1, r; - r = yaz_write_UTF8(cd, x, &utf8_outbuf, &utf8_outbytesleft, 0); + r = yaz_write_UTF8(cd, x, &utf8_outbuf, &utf8_outbytesleft); if (r == (size_t)(-1)) { cd->my_errno = YAZ_ICONV_EILSEQ; @@ -811,9 +1178,51 @@ static size_t flush_combos(yaz_iconv_t cd, return 0; } +static size_t yaz_write_marc8_page_chr(yaz_iconv_t cd, + char **outbuf, size_t *outbytesleft, + const char *page_chr) +{ + const char *old_page_chr = cd->write_marc8_page_chr; + if (strcmp(page_chr, old_page_chr)) + { + size_t plen = 0; + const char *page_out = page_chr; + + if (*outbytesleft < 8) + { + cd->my_errno = YAZ_ICONV_E2BIG; + + return (size_t) (-1); + } + cd->write_marc8_page_chr = page_chr; + + if (!strcmp(old_page_chr, "\033p") + || !strcmp(old_page_chr, "\033g") + || !strcmp(old_page_chr, "\033b")) + { + /* Technique 1 leave */ + page_out = "\033s"; + if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */ + { + /* Must leave script + enter new page */ + plen = strlen(page_out); + memcpy(*outbuf, page_out, plen); + (*outbuf) += plen; + (*outbytesleft) -= plen; + page_out = page_chr; + } + } + plen = strlen(page_out); + memcpy(*outbuf, page_out, plen); + (*outbuf) += plen; + (*outbytesleft) -= plen; + } + return 0; +} + + static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { int comb = 0; const char *page_chr = 0; @@ -835,56 +1244,13 @@ static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x, else { size_t r = flush_combos(cd, outbuf, outbytesleft); - const char *old_page_chr = cd->write_marc8_page_chr; if (r) return r; - if (strcmp(page_chr, old_page_chr)) - { - size_t plen = 0; - const char *page_out = page_chr; - - if (*outbytesleft < 8) - { - cd->my_errno = YAZ_ICONV_E2BIG; - - return (size_t) (-1); - } - cd->write_marc8_page_chr = page_chr; - if (!strcmp(old_page_chr, "\033p") - || !strcmp(old_page_chr, "\033g") - || !strcmp(old_page_chr, "\033b")) - { - /* Technique 1 leave */ - page_out = "\033s"; - if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */ - { - /* Must leave script + enter new page */ - plen = strlen(page_out); - memcpy(*outbuf, page_out, plen); - (*outbuf) += plen; - (*outbytesleft) -= plen; - page_out = page_chr; - } - } - plen = strlen(page_out); - memcpy(*outbuf, page_out, plen); - (*outbuf) += plen; - (*outbytesleft) -= plen; - } - cd->write_marc8_last = y; - } - if (last) - { - size_t r = flush_combos(cd, outbuf, outbytesleft); + r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, page_chr); if (r) - { - if (comb) - cd->write_marc8_comb_no--; - else - cd->write_marc8_last = 0; return r; - } + cd->write_marc8_last = y; } return 0; } @@ -892,23 +1258,14 @@ static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x, static size_t yaz_flush_marc8(yaz_iconv_t cd, char **outbuf, size_t *outbytesleft) { - if (strcmp(cd->write_marc8_page_chr, "\033(B")) - { - if (*outbytesleft < 3) - { - cd->my_errno = YAZ_ICONV_E2BIG; - return (size_t) (-1); - } - memcpy(*outbuf, "\033(B", 3); - (*outbuf) += 3; - *outbytesleft -= 3; - } - return 0; + size_t r = flush_combos(cd, outbuf, outbytesleft); + if (r) + return r; + return yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, "\033(B"); } static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) + char **outbuf, size_t *outbytesleft) { int i; for (i = 0; latin1_comb[i].x1; i++) @@ -922,11 +1279,11 @@ static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x, int last_ch = cd->write_marc8_last; r = yaz_write_marc8_2(cd, latin1_comb[i].x1, - outbuf, outbytesleft, 0); + outbuf, outbytesleft); if (r) return r; r = yaz_write_marc8_2(cd, latin1_comb[i].x2, - outbuf, outbytesleft, last); + outbuf, outbytesleft); if (r && cd->my_errno == YAZ_ICONV_E2BIG) { /* not enough room. reset output to original values */ @@ -937,14 +1294,13 @@ static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x, return r; } } - return yaz_write_marc8_2(cd, x, outbuf, outbytesleft, last); + return yaz_write_marc8_2(cd, x, outbuf, outbytesleft); } #if HAVE_WCHAR_H -static size_t yaz_write_wchar_t (yaz_iconv_t cd, unsigned long x, - char **outbuf, size_t *outbytesleft, - int last) +static size_t yaz_write_wchar_t(yaz_iconv_t cd, unsigned long x, + char **outbuf, size_t *outbytesleft) { unsigned char *outp = (unsigned char *) *outbuf; @@ -1001,6 +1357,8 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode) 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; #if HAVE_WCHAR_H else if (!yaz_matchstr(fromcode, "WCHAR_T")) cd->read_handle = yaz_read_wchar_t; @@ -1009,7 +1367,10 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode) if (!yaz_matchstr(tocode, "UTF8")) cd->write_handle = yaz_write_UTF8; else if (!yaz_matchstr(tocode, "ISO88591")) + { cd->write_handle = yaz_write_ISO8859_1; + cd->flush_handle = yaz_flush_ISO8859_1; + } else if (!yaz_matchstr (tocode, "UCS4")) cd->write_handle = yaz_write_UCS4; else if (!yaz_matchstr(tocode, "UCS4LE")) @@ -1024,6 +1385,10 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode) cd->write_handle = yaz_write_marc8; cd->flush_handle = yaz_flush_marc8; } + else if (!yaz_matchstr(tocode, "advancegreek")) + { + cd->write_handle = yaz_write_advancegreek; + } #if HAVE_WCHAR_H else if (!yaz_matchstr(tocode, "WCHAR_T")) cd->write_handle = yaz_write_wchar_t; @@ -1123,6 +1488,20 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, } cd->init_flag = 0; + if (!inbuf || !*inbuf) + { + if (outbuf && *outbuf) + { + if (cd->unget_x) + r = (*cd->write_handle)(cd, cd->unget_x, outbuf, outbytesleft); + if (cd->flush_handle) + r = (*cd->flush_handle)(cd, outbuf, outbytesleft); + } + if (r == 0) + cd->init_flag = 1; + cd->unget_x = 0; + return r; + } while (1) { unsigned long x; @@ -1133,34 +1512,24 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft, x = cd->unget_x; no_read = cd->no_read_x; } - else if (inbuf && *inbuf) + else { if (*inbytesleft == 0) { r = *inbuf - inbuf0; break; } - x = (cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft, - &no_read); + x = (*cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft, + &no_read); if (no_read == 0) { r = (size_t)(-1); break; } } - else - { - r = 0; - if (cd->flush_handle && outbuf && *outbuf) - r = (*cd->flush_handle)(cd, outbuf, outbytesleft); - if (r == 0) - cd->init_flag = 1; - break; - } if (x) { - r = (cd->write_handle)(cd, x, outbuf, outbytesleft, - (*inbytesleft - no_read) == 0 ? 1 : 0); + r = (*cd->write_handle)(cd, x, outbuf, outbytesleft); if (r) { /* unable to write it. save it because read_handle cannot