X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=test%2Ftsticonv.c;h=20424d622b0a51ddbafb2d37e077d838bb1b1e88;hb=bc4b50939a936dbb214c783c18a44039ea048fdd;hp=33b99449a724f48e99eb2283396269ca56b54180;hpb=d38ee71c31b49ad13164039140ed47d18e9432cb;p=yaz-moved-to-github.git diff --git a/test/tsticonv.c b/test/tsticonv.c index 33b9944..20424d6 100644 --- a/test/tsticonv.c +++ b/test/tsticonv.c @@ -1,8 +1,6 @@ -/* - * Copyright (C) 1995-2007, Index Data ApS +/* This file is part of the YAZ toolkit. + * Copyright (C) 1995-2008 Index Data * See the file LICENSE for details. - * - * $Id: tsticonv.c,v 1.29 2007-03-20 21:37:32 adam Exp $ */ #if HAVE_CONFIG_H @@ -17,6 +15,8 @@ #include #include +#define ESC "\x1b" + static int compare_buffers(char *msg, int no, int expect_len, const char *expect_buf, int got_len, const char *got_buf) @@ -87,11 +87,12 @@ static int tst_convert_l(yaz_iconv_t cd, size_t in_len, const char *in_buf, outbuf - outbuf0, outbuf0); } -static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf) +static int tst_convert_x(yaz_iconv_t cd, const char *buf, const char *cmpbuf, + int expect_error) { - int ret = 0; + int ret = 1; WRBUF b = wrbuf_alloc(); - char outbuf[12]; + char outbuf[16]; size_t inbytesleft = strlen(buf); const char *inp = buf; int rounds = 0; @@ -106,7 +107,12 @@ static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf) { int e = yaz_iconv_error(cd); if (e != YAZ_ICONV_E2BIG) + { + if (expect_error != -1) + if (e != expect_error) + ret = 0; break; + } } else { @@ -114,18 +120,43 @@ static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf) char *outp = outbuf; r = yaz_iconv(cd, 0, 0, &outp, &outbytesleft); wrbuf_write(b, outbuf, outp - outbuf); + if (expect_error != -1) + if (expect_error) + ret = 0; break; } } if (wrbuf_len(b) == strlen(cmpbuf) && !memcmp(cmpbuf, wrbuf_buf(b), wrbuf_len(b))) - ret = 1; + ; else - yaz_log(YLOG_LOG, "GOT (%.*s)", wrbuf_len(b), wrbuf_buf(b)); + { + WRBUF w = wrbuf_alloc(); + + ret = 0; + wrbuf_rewind(w); + wrbuf_puts_escaped(w, buf); + yaz_log(YLOG_LOG, "input %s", wrbuf_cstr(w)); + + wrbuf_rewind(w); + wrbuf_write_escaped(w, wrbuf_buf(b), wrbuf_len(b)); + yaz_log(YLOG_LOG, "got %s", wrbuf_cstr(w)); + + wrbuf_rewind(w); + wrbuf_puts_escaped(w, cmpbuf); + yaz_log(YLOG_LOG, "exp %s", wrbuf_cstr(w)); + + wrbuf_destroy(w); + } + wrbuf_destroy(b); return ret; } +static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf) +{ + return tst_convert_x(cd, buf, cmpbuf, 0); +} /* some test strings in ISO-8859-1 format */ static const char *iso_8859_1_a[] = { @@ -375,10 +406,24 @@ static void tst_marc8_to_utf8(void) YAZ_CHECK(tst_convert(cd, "Cours de mathâe", "Cours de mathe\xcc\x81")); + YAZ_CHECK(tst_convert(cd, "\xea" "a", "a\xcc\x8a")); + YAZ_CHECK(tst_convert(cd, "a" "\xea" "\x1e", "a" "\x1e\xcc\x8a")); + YAZ_CHECK(tst_convert(cd, "a" "\xea" "p", "a" "p\xcc\x8a")); + + YAZ_CHECK(tst_convert_x(cd, "a\xea", "a", YAZ_ICONV_EINVAL)); + YAZ_CHECK(tst_convert(cd, "p", "\xcc\x8a")); /* note: missing p */ + yaz_iconv(cd, 0, 0, 0, 0); /* incomplete. so we have to reset */ + + /* bug #2115 */ + YAZ_CHECK(tst_convert(cd, ESC "(N" ESC ")Qp" ESC "(B", "\xd0\x9f")); - YAZ_CHECK(tst_convert(cd, "a\xea\x1e", "a\x1e\xcc\x8a")); + YAZ_CHECK(tst_convert_x(cd, ESC , "", YAZ_ICONV_EINVAL)); + YAZ_CHECK(tst_convert_x(cd, ESC "(", "", YAZ_ICONV_EINVAL)); + YAZ_CHECK(tst_convert_x(cd, ESC "(B", "", 0)); + + YAZ_CHECK(tst_convert(cd, ESC "(B" "\x31", "1")); /* ASCII in G0 */ + YAZ_CHECK(tst_convert(cd, ESC ")B" "\xB1", "1")); /* ASCII in G1 */ - YAZ_CHECK(tst_convert(cd, "a\xea", "a")); yaz_iconv_close(cd); } @@ -442,9 +487,9 @@ static void tst_marc8_to_latin1(void) yaz_iconv_close(cd); } -static void tst_utf8_to_marc8(void) +static void tst_utf8_to_marc8(const char *marc8_type) { - yaz_iconv_t cd = yaz_iconv_open("MARC8", "UTF-8"); + yaz_iconv_t cd = yaz_iconv_open(marc8_type, "UTF-8"); YAZ_CHECK(cd); if (!cd) @@ -458,7 +503,7 @@ static void tst_utf8_to_marc8(void) /** Pure ASCII. 12 characters (sizeof(outbuf)) */ YAZ_CHECK(tst_convert(cd, "Cours de math", "Cours de math")); - /** Pure ASCII. 13 characters (sizeof(outbuf)) */ + /** Pure ASCII. 13 characters (sizeof(outbuf)+1) */ YAZ_CHECK(tst_convert(cd, "Cours de math.", "Cours de math.")); /** UPPERCASE SCANDINAVIAN O */ @@ -487,11 +532,38 @@ static void tst_utf8_to_marc8(void) "\033$1" "\x21\x2B\x3B" "\033(B" "o" )); + /** Ideographic space per ANSI Z39.64 */ + YAZ_CHECK(tst_convert(cd, + "\xe3\x80\x80" "o", /* UTF-8 */ + "\033$1" "\x21\x23\x21" "\033(B" "o" )); + /** Superscript 0 . bug #642 */ YAZ_CHECK(tst_convert(cd, "(\xe2\x81\xb0)", /* UTF-8 */ "(\033p0\x1bs)")); + + /** bug #1778 */ + YAZ_CHECK(tst_convert(cd, + /* offset 0x530 in UTF-8 rec marccol4.u8.marc */ + "\xE3\x83\xB3" "\xE3\x82\xBF" + "\xCC\x84" "\xCC\x84" "\xE3\x83\xBC" /* UTF-8 */, + "\x1B\x24\x31" "\x69\x25\x73" + "\x1B\x28\x42" "\xE5\xE5" "\x1B\x24\x31" + "\x69\x25\x3F" + "\x69\x21\x3C" "\x1B\x28\x42")); + + + /** bug #2120 */ + YAZ_CHECK(tst_convert(cd, + "\xCE\x94\xCE\xB5\xCF\x84" + "\xCE\xBF\xCF\x81\xCE\xB1" + "\xCE\xBA\xCE\xB7\xCF\x82\x2C", + + "\x1B\x28\x53\x45\x66\x78\x72\x75" + "\x61\x6D\x6A\x77" + "\x1B\x28\x42\x2C" + )); { char *inbuf0 = "\xe2\x81\xb0"; @@ -528,6 +600,7 @@ static void tst_utf8_to_marc8(void) #endif YAZ_CHECK(strcmp("\033p0\x1bs", outbuf0) == 0); } + yaz_iconv(cd, 0, 0, 0, 0); yaz_iconv_close(cd); } @@ -623,7 +696,9 @@ int main (int argc, char **argv) tst_advance_to_utf8(); tst_utf8_to_advance(); - tst_utf8_to_marc8(); + tst_utf8_to_marc8("marc8"); + tst_utf8_to_marc8("marc8lossy"); + tst_utf8_to_marc8("marc8lossless"); tst_latin1_to_marc8();