X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Frecord_render.c;h=1291858a0f292e4861fc5b1acca1b1ea59b7dce7;hp=79050162535ff2e7f3defe8d81fbde620d8e5f60;hb=7fc72f3ae149e416a297ef1f55c09271056e98f1;hpb=77c5a4fca8b516fd39b8ba213daed17a465a6b2a diff --git a/src/record_render.c b/src/record_render.c index 7905016..1291858 100644 --- a/src/record_render.c +++ b/src/record_render.c @@ -1,5 +1,5 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2012 Index Data + * Copyright (C) Index Data * See the file LICENSE for details. */ /** @@ -19,6 +19,13 @@ #include #include #include +#include +#include + +#if YAZ_HAVE_XML2 +#include +#include +#endif static yaz_iconv_t iconv_create_charset(const char *record_charset, yaz_iconv_t *cd2) @@ -27,14 +34,14 @@ static yaz_iconv_t iconv_create_charset(const char *record_charset, yaz_iconv_t cd = 0; char *from_set1 = 0; char *from_set2 = 0; - char *to_set = 0; + char *to_set = "utf-8"; if (record_charset && *record_charset) { char *cp = charset_buf; - + strncpy(charset_buf, record_charset, sizeof(charset_buf)-1); charset_buf[sizeof(charset_buf)-1] = '\0'; - + from_set1 = cp; while (*cp && *cp != ',' && *cp != '/') cp++; @@ -53,13 +60,13 @@ static yaz_iconv_t iconv_create_charset(const char *record_charset, cp++; } } - + if (from_set1) - cd = yaz_iconv_open(to_set ? to_set : "UTF-8", from_set1); + cd = yaz_iconv_open(to_set, from_set1); if (cd2) { if (from_set2) - *cd2 = yaz_iconv_open(to_set ? to_set : "UTF-8", from_set2); + *cd2 = yaz_iconv_open(to_set, from_set2); else *cd2 = 0; } @@ -81,8 +88,7 @@ static const char *return_marc_record(WRBUF wrbuf, yaz_marc_xml(mt, marc_type); if (yaz_marc_decode_wrbuf(mt, buf, sz, wrbuf) > 0) { - if (len) - *len = wrbuf_len(wrbuf); + *len = wrbuf_len(wrbuf); ret_string = wrbuf_cstr(wrbuf); } yaz_marc_destroy(mt); @@ -109,15 +115,14 @@ static const char *return_opac_record(WRBUF wrbuf, yaz_opac_decode_wrbuf2(mt, opac_rec, wrbuf, cd2); else yaz_opac_decode_wrbuf(mt, opac_rec, wrbuf); - + yaz_marc_destroy(mt); if (cd) yaz_iconv_close(cd); if (cd2) yaz_iconv_close(cd2); - if (len) - *len = wrbuf_len(wrbuf); + *len = wrbuf_len(wrbuf); return wrbuf_cstr(wrbuf); } @@ -137,8 +142,7 @@ static const char *return_string_record(WRBUF wrbuf, sz = wrbuf_len(wrbuf); yaz_iconv_close(cd); } - if (len) - *len = sz; + *len = sz; return buf; } @@ -163,7 +167,10 @@ static const char *return_record_wrbuf(WRBUF wrbuf, int *len, charset); else if (r->which == Z_External_octet) { - if (yaz_oid_is_iso2709(oid)) + if (oid_oidcmp(oid, yaz_oid_recsyn_xml) + && oid_oidcmp(oid, yaz_oid_recsyn_application_xml) + && oid_oidcmp(oid, yaz_oid_recsyn_mab) + && oid_oidcmp(oid, yaz_oid_recsyn_html)) { const char *ret_buf = return_marc_record( wrbuf, marctype, len, @@ -172,8 +179,8 @@ static const char *return_record_wrbuf(WRBUF wrbuf, int *len, charset); if (ret_buf) return ret_buf; - /* bad ISO2709. Return fail unless raw (ISO2709) is wanted */ - if (marctype != YAZ_MARC_ISO2709) + /* not ISO2709. Return fail unless raw (ISO2709) is wanted */ + if (yaz_oid_is_iso2709(oid) && marctype != YAZ_MARC_ISO2709) return 0; } return return_string_record(wrbuf, len, @@ -191,7 +198,7 @@ static const char *return_record_wrbuf(WRBUF wrbuf, int *len, } return 0; } - + static const char *get_record_format(WRBUF wrbuf, int *len, Z_NamePlusRecord *npr, int marctype, const char *charset, @@ -199,7 +206,7 @@ static const char *get_record_format(WRBUF wrbuf, int *len, { const char *res = return_record_wrbuf(wrbuf, len, npr, marctype, charset); #if YAZ_HAVE_XML2 - if (*format == '1' && len) + if (*format == '1') { /* try to XML format res */ xmlDocPtr doc; @@ -216,21 +223,133 @@ static const char *get_record_format(WRBUF wrbuf, int *len, xmlFreeDoc(doc); res = wrbuf_cstr(wrbuf); *len = wrbuf_len(wrbuf); - } + } } #endif return res; } +#if YAZ_HAVE_XML2 +static int replace_node(NMEM nmem, xmlNode *ptr, + const char *type_spec, char *record_buf) +{ + int ret = -1; + const char *res; + int len; + int m_len; + WRBUF wrbuf = wrbuf_alloc(); + ODR odr = odr_createmem(ODR_ENCODE); + Z_NamePlusRecord *npr = odr_malloc(odr, sizeof(*npr)); + npr->which = Z_NamePlusRecord_databaseRecord; + + if (atoi_n_check(record_buf, 5, &m_len)) + npr->u.databaseRecord = + z_ext_record_usmarc(odr, record_buf, strlen(record_buf)); + else + npr->u.databaseRecord = + z_ext_record_xml(odr, record_buf, strlen(record_buf)); + res = yaz_record_render(npr, 0, wrbuf, type_spec, &len); + if (res) + { + xmlDoc *doc = xmlParseMemory(res, strlen(res)); + if (doc) + { + xmlNode *nptr = xmlCopyNode(xmlDocGetRootElement(doc), 1); + xmlReplaceNode(ptr, nptr); + xmlFreeDoc(doc); + } + else + { + xmlNode *nptr = xmlNewText(BAD_CAST res); + xmlReplaceNode(ptr, nptr); + } + ret = 0; + } + wrbuf_destroy(wrbuf); + odr_destroy(odr); + return ret; +} +#endif + +static const char *base64_render(NMEM nmem, WRBUF wrbuf, + const char *buf, int *len, + const char *expr, const char *type_spec) +{ +#if YAZ_HAVE_XML2 + xmlDocPtr doc = xmlParseMemory(buf, *len); + if (doc) + { + xmlChar *buf_out; + int len_out; + xmlXPathContextPtr xpathCtx = xmlXPathNewContext(doc); + if (xpathCtx) + { + xmlXPathObjectPtr xpathObj = + xmlXPathEvalExpression((const xmlChar *) expr, xpathCtx); + if (xpathObj) + { + xmlNodeSetPtr nodes = xpathObj->nodesetval; + if (nodes) + { + int i; + for (i = 0; i < nodes->nodeNr; i++) + { + xmlNode *ptr = nodes->nodeTab[i]; + if (ptr->type == XML_TEXT_NODE) + { + const char *input = + nmem_text_node_cdata(ptr, nmem); + char *output = nmem_malloc( + nmem, strlen(input) + 1); + if (yaz_base64decode(input, output) == 0) + { + if (!replace_node(nmem, ptr, type_spec, output)) + { + /* replacement OK */ + xmlFreeNode(ptr); + /* unset below to avoid a bad reference in + xmlXPathFreeObject below */ + nodes->nodeTab[i] = 0; + } + } + } + } + } + xmlXPathFreeObject(xpathObj); + } + xmlXPathFreeContext(xpathCtx); + } + xmlDocDumpMemory(doc, &buf_out, &len_out); + if (buf_out) + { + wrbuf_rewind(wrbuf); + wrbuf_write(wrbuf, (const char *) buf_out, len_out); + buf = wrbuf_cstr(wrbuf); + *len = len_out; + } + xmlFreeDoc(doc); + xmlFree(buf_out); + } +#endif + return buf; +} + const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema, WRBUF wrbuf, const char *type_spec, int *len) { + const char *ret = 0; + NMEM nmem = 0; + char *base64_xpath = 0; size_t i; char type[40]; char charset[40]; char format[3]; const char *cp = type_spec; + int len0; + + if (!len) + len = &len0; for (i = 0; cp[i] && cp[i] != ';' && cp[i] != ' ' && i < sizeof(type)-1; i++) @@ -251,6 +370,8 @@ const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema, { size_t j = 0; i = i + 8; /* skip charset= */ + while (cp[i] == ' ') + i++; for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++) { if (j < sizeof(charset)-1) @@ -260,31 +381,44 @@ const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema, } else if (!strncmp(cp + i, "format=", 7)) { - size_t j = 0; + size_t j = 0; i = i + 7; + while (cp[i] == ' ') + i++; for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++) { if (j < sizeof(format)-1) format[j++] = cp[i]; } format[j] = '\0'; - } + } + else if (!strncmp(cp + i, "base64=", 7)) + { + size_t i0; + i = i + 7; + while (cp[i] == ' ') + i++; + i0 = i; + while (cp[i] && cp[i] != ';') + i++; + + nmem = nmem_create(); + base64_xpath = nmem_strdupn(nmem, cp + i0, i - i0); + } } if (!strcmp(type, "database")) { - if (len) - *len = (npr->databaseName ? strlen(npr->databaseName) : 0); - return npr->databaseName; + *len = (npr->databaseName ? strlen(npr->databaseName) : 0); + ret = npr->databaseName; } else if (!strcmp(type, "schema")) { - if (len) - *len = schema ? strlen(schema) : 0; - return schema; + *len = schema ? strlen(schema) : 0; + ret = schema; } else if (!strcmp(type, "syntax")) { - const char *desc = 0; + const char *desc = 0; if (npr->which == Z_NamePlusRecord_databaseRecord) { Z_External *r = (Z_External *) npr->u.databaseRecord; @@ -292,45 +426,61 @@ const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema, } if (!desc) desc = "none"; - if (len) - *len = strlen(desc); - return desc; + *len = strlen(desc); + ret = desc; } if (npr->which != Z_NamePlusRecord_databaseRecord) - return 0; - - /* from now on - we have a database record .. */ - if (!strcmp(type, "render")) + ; + else if (!strcmp(type, "render")) { - return get_record_format(wrbuf, len, npr, YAZ_MARC_LINE, charset, format); + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_LINE, charset, format); } else if (!strcmp(type, "xml")) { - return get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset, - format); + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset, + format); } else if (!strcmp(type, "txml")) { - return get_record_format(wrbuf, len, npr, YAZ_MARC_TURBOMARC, charset, - format); + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_TURBOMARC, charset, + format); + } + else if (!strcmp(type, "json")) + { + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_JSON, charset, + format); } else if (!strcmp(type, "raw")) { - return get_record_format(wrbuf, len, npr, YAZ_MARC_ISO2709, charset, - format); + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_ISO2709, charset, + format); } else if (!strcmp(type, "ext")) { - if (len) *len = -1; - return (const char *) npr->u.databaseRecord; + *len = -1; + ret = (const char *) npr->u.databaseRecord; } else if (!strcmp(type, "opac")) { if (npr->u.databaseRecord->which == Z_External_OPAC) - return get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset, - format); + ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset, + format); } - return 0; + + if (base64_xpath && *len != -1) + { + char *type_spec = nmem_malloc(nmem, + strlen(type) + strlen(charset) + 11); + strcpy(type_spec, type); + if (*charset) + { + strcat(type_spec, "; charset="); + strcat(type_spec, charset); + } + ret = base64_render(nmem, wrbuf, ret, len, base64_xpath, type_spec); + } + nmem_destroy(nmem); + return ret; } /*