X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fmarcdisp.c;h=50447bed9460ceb1dbab97752b536adafdd03f81;hp=a5084ce35a02b81b66b4101200cab5d713f60a6e;hb=887a9c6336d4731d898e185b5bd08cc6424114a8;hpb=a5ac9b8b5662012c69c5e7e5287c5ffc0c7a1f25 diff --git a/src/marcdisp.c b/src/marcdisp.c index a5084ce..50447be 100644 --- a/src/marcdisp.c +++ b/src/marcdisp.c @@ -1,5 +1,5 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2008 Index Data + * Copyright (C) 1995-2012 Index Data * See the file LICENSE for details. */ @@ -20,7 +20,6 @@ #include #include #include -#include #include #include #include @@ -87,7 +86,7 @@ struct yaz_marc_subfield { struct yaz_marc_t_ { WRBUF m_wr; NMEM nmem; - int xml; + int output_format; int debug; int write_using_libxml2; enum yaz_collection_state enable_collection; @@ -103,7 +102,7 @@ struct yaz_marc_t_ { yaz_marc_t yaz_marc_create(void) { yaz_marc_t mt = (yaz_marc_t) xmalloc(sizeof(*mt)); - mt->xml = YAZ_MARC_LINE; + mt->output_format = YAZ_MARC_LINE; mt->debug = 0; mt->write_using_libxml2 = 0; mt->enable_collection = no_collection; @@ -140,7 +139,12 @@ static void marc_iconv_reset(yaz_marc_t mt, WRBUF wr) static int marc_exec_leader(const char *leader_spec, char *leader, size_t size); - +#if YAZ_HAVE_XML2 +static int yaz_marc_write_xml_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr, + const char *ns, + const char *format, + const char *type); +#endif static struct yaz_marc_node *yaz_marc_add_node(yaz_marc_t mt) { @@ -161,6 +165,16 @@ void yaz_marc_add_controlfield_xml(yaz_marc_t mt, const xmlNode *ptr_tag, n->u.controlfield.tag = nmem_text_node_cdata(ptr_tag, mt->nmem); n->u.controlfield.data = nmem_text_node_cdata(ptr_data, mt->nmem); } + +void yaz_marc_add_controlfield_xml2(yaz_marc_t mt, char *tag, + const xmlNode *ptr_data) +{ + struct yaz_marc_node *n = yaz_marc_add_node(mt); + n->which = YAZ_MARC_CONTROLFIELD; + n->u.controlfield.tag = tag; + n->u.controlfield.data = nmem_text_node_cdata(ptr_data, mt->nmem); +} + #endif @@ -230,6 +244,42 @@ void yaz_marc_add_datafield(yaz_marc_t mt, const char *tag, mt->subfield_pp = &n->u.datafield.subfields; } +/** \brief adds a attribute value to the element name if it is plain chars + + If not, and if the attribute name is not null, it will append a + attribute element with the value if attribute name is null it will + return a non-zero value meaning it couldnt handle the value. +*/ +static int element_name_append_attribute_value( + yaz_marc_t mt, WRBUF buffer, + const char *attribute_name, char *code_data, size_t code_len) +{ + /* TODO Map special codes to something possible for XML ELEMENT names */ + + int encode = 0; + int index = 0; + int success = 0; + for (index = 0; index < code_len; index++) + { + if (!((code_data[index] >= '0' && code_data[index] <= '9') || + (code_data[index] >= 'a' && code_data[index] <= 'z') || + (code_data[index] >= 'A' && code_data[index] <= 'Z'))) + encode = 1; + } + /* Add as attribute */ + if (encode && attribute_name) + wrbuf_printf(buffer, " %s=\"", attribute_name); + + if (!encode || attribute_name) + wrbuf_iconv_write_cdata(buffer, mt->iconv_cd, code_data, code_len); + else + success = -1; + + if (encode && attribute_name) + wrbuf_printf(buffer, "\""); /* return error if we couldn't handle it.*/ + return success; +} + #if YAZ_HAVE_XML2 void yaz_marc_add_datafield_xml(yaz_marc_t mt, const xmlNode *ptr_tag, const char *indicator, size_t indicator_len) @@ -244,6 +294,24 @@ void yaz_marc_add_datafield_xml(yaz_marc_t mt, const xmlNode *ptr_tag, /* make subfield_pp the current (last one) */ mt->subfield_pp = &n->u.datafield.subfields; } + +void yaz_marc_add_datafield_xml2(yaz_marc_t mt, char *tag_value, char *indicators) +{ + struct yaz_marc_node *n = yaz_marc_add_node(mt); + n->which = YAZ_MARC_DATAFIELD; + n->u.datafield.tag = tag_value; + n->u.datafield.indicator = indicators; + n->u.datafield.subfields = 0; + + /* make subfield_pp the current (last one) */ + mt->subfield_pp = &n->u.datafield.subfields; +} + +void yaz_marc_datafield_set_indicators(struct yaz_marc_node *n, char *indicator) +{ + n->u.datafield.indicator = indicator; +} + #endif void yaz_marc_add_subfield(yaz_marc_t mt, @@ -274,16 +342,6 @@ void yaz_marc_add_subfield(yaz_marc_t mt, } } -int atoi_n_check(const char *buf, int size, int *val) -{ - int i; - for (i = 0; i < size; i++) - if (!isdigit(i[(const unsigned char *) buf])) - return 0; - *val = atoi_n(buf, size); - return 1; -} - void yaz_marc_set_leader(yaz_marc_t mt, const char *leader_c, int *indicator_length, int *identifier_length, @@ -425,7 +483,7 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr) case YAZ_MARC_COMMENT: wrbuf_iconv_write(wr, mt->iconv_cd, n->u.comment, strlen(n->u.comment)); - wrbuf_puts(wr, ")\n"); + wrbuf_puts(wr, "\n"); break; default: break; @@ -434,6 +492,17 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr) return 0; } +static size_t get_subfield_len(yaz_marc_t mt, const char *data, + int identifier_length) +{ + /* if identifier length is 2 (most MARCs) or less (probably an error), + the code is a single character .. However we've + seen multibyte codes, so see how big it really is */ + if (identifier_length > 2) + return identifier_length - 1; + else + return cdata_one_character(mt, data); +} int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr) { @@ -463,13 +532,8 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr) n->u.datafield.indicator); for (s = n->u.datafield.subfields; s; s = s->next) { - /* if identifier length is 2 (most MARCs), - the code is a single character .. However we've - seen multibyte codes, so see how big it really is */ - size_t using_code_len = - (identifier_length > 2) ? identifier_length - 1 - : - cdata_one_character(mt, s->code_data); + size_t using_code_len = get_subfield_len(mt, s->code_data, + identifier_length); wrbuf_puts (wr, mt->subfield_str); wrbuf_iconv_write(wr, mt->iconv_cd, s->code_data, @@ -507,9 +571,10 @@ int yaz_marc_write_trailer(yaz_marc_t mt, WRBUF wr) { if (mt->enable_collection == collection_second) { - switch(mt->xml) + switch(mt->output_format) { case YAZ_MARC_MARCXML: + case YAZ_MARC_TURBOMARC: wrbuf_printf(wr, "\n"); break; case YAZ_MARC_XCHANGE: @@ -527,12 +592,14 @@ void yaz_marc_enable_collection(yaz_marc_t mt) int yaz_marc_write_mode(yaz_marc_t mt, WRBUF wr) { - switch(mt->xml) + switch(mt->output_format) { case YAZ_MARC_LINE: return yaz_marc_write_line(mt, wr); case YAZ_MARC_MARCXML: return yaz_marc_write_marcxml(mt, wr); + case YAZ_MARC_TURBOMARC: + return yaz_marc_write_turbomarc(mt, wr); case YAZ_MARC_XCHANGE: return yaz_marc_write_marcxchange(mt, wr, 0, 0); /* no format, type */ case YAZ_MARC_ISO2709: @@ -543,17 +610,28 @@ int yaz_marc_write_mode(yaz_marc_t mt, WRBUF wr) return -1; } -/** \brief common MARC XML/Xchange writer +static const char *record_name[2] = { "record", "r"}; +static const char *leader_name[2] = { "leader", "l"}; +static const char *controlfield_name[2] = { "controlfield", "c"}; +static const char *datafield_name[2] = { "datafield", "d"}; +static const char *indicator_name[2] = { "ind", "i"}; +static const char *subfield_name[2] = { "subfield", "s"}; + +/** \brief common MARC XML/Xchange/turbomarc writer \param mt handle \param wr WRBUF output \param ns XMLNS for the elements \param format record format (e.g. "MARC21") \param type record type (e.g. "Bibliographic") + \param turbo =1 for turbomarc + \retval 0 OK + \retval -1 failure */ -static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr, - const char *ns, - const char *format, - const char *type) +static int yaz_marc_write_marcxml_wrbuf(yaz_marc_t mt, WRBUF wr, + const char *ns, + const char *format, + const char *type, + int turbo) { struct yaz_marc_node *n; int identifier_length; @@ -574,13 +652,15 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr, if (mt->enable_collection != no_collection) { if (mt->enable_collection == collection_first) + { wrbuf_printf(wr, "\n", ns); - mt->enable_collection = collection_second; - wrbuf_printf(wr, "enable_collection = collection_second; + } + wrbuf_printf(wr, "<%s", record_name[turbo]); } else { - wrbuf_printf(wr, "which) { case YAZ_MARC_DATAFIELD: - wrbuf_printf(wr, " iconv_cd, n->u.datafield.tag, strlen(n->u.datafield.tag)); - wrbuf_printf(wr, "\""); - if (n->u.datafield.indicator) - { - int i; - for (i = 0; n->u.datafield.indicator[i]; i++) - { - wrbuf_printf(wr, " ind%d=\"", i+1); + if (!turbo) + wrbuf_printf(wr, "\""); + if (n->u.datafield.indicator) + { + int i; + for (i = 0; n->u.datafield.indicator[i]; i++) + { + wrbuf_printf(wr, " %s%d=\"", indicator_name[turbo], i+1); wrbuf_iconv_write_cdata(wr, mt->iconv_cd, - n->u.datafield.indicator+i, 1); + n->u.datafield.indicator+i, 1); wrbuf_iconv_puts(wr, mt->iconv_cd, "\""); } } wrbuf_printf(wr, ">\n"); for (s = n->u.datafield.subfields; s; s = s->next) { - /* if identifier length is 2 (most MARCs), - the code is a single character .. However we've - seen multibyte codes, so see how big it really is */ - size_t using_code_len = - (identifier_length > 2) ? identifier_length - 1 - : - cdata_one_character(mt, s->code_data); - - wrbuf_iconv_puts(wr, mt->iconv_cd, " iconv_cd, - s->code_data, using_code_len); - wrbuf_iconv_puts(wr, mt->iconv_cd, "\">"); + size_t using_code_len = get_subfield_len(mt, s->code_data, + identifier_length); + wrbuf_printf(wr, " <%s", subfield_name[turbo]); + if (!turbo) + { + wrbuf_printf(wr, " code=\""); + wrbuf_iconv_write_cdata(wr, mt->iconv_cd, + s->code_data, using_code_len); + wrbuf_iconv_puts(wr, mt->iconv_cd, "\">"); + } + else + { + element_name_append_attribute_value(mt, wr, "code", s->code_data, using_code_len); + wrbuf_puts(wr, ">"); + } wrbuf_iconv_write_cdata(wr, mt->iconv_cd, s->code_data + using_code_len, strlen(s->code_data + using_code_len)); marc_iconv_reset(mt, wr); - wrbuf_iconv_puts(wr, mt->iconv_cd, ""); - wrbuf_puts(wr, "\n"); + wrbuf_printf(wr, "code_data, using_code_len); + wrbuf_puts(wr, ">\n"); } - wrbuf_printf(wr, " \n"); + wrbuf_printf(wr, " iconv_cd, n->u.datafield.tag, + strlen(n->u.datafield.tag)); + wrbuf_printf(wr, ">\n"); break; case YAZ_MARC_CONTROLFIELD: - wrbuf_printf(wr, " iconv_cd, n->u.controlfield.tag, - strlen(n->u.controlfield.tag)); - wrbuf_iconv_puts(wr, mt->iconv_cd, "\">"); + wrbuf_printf(wr, " <%s", controlfield_name[turbo]); + if (!turbo) + { + wrbuf_printf(wr, " tag=\""); + wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.tag, + strlen(n->u.controlfield.tag)); + wrbuf_iconv_puts(wr, mt->iconv_cd, "\">"); + } + else + { + /* TODO convert special */ + wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.tag, + strlen(n->u.controlfield.tag)); + wrbuf_iconv_puts(wr, mt->iconv_cd, ">"); + } wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.data, strlen(n->u.controlfield.data)); - marc_iconv_reset(mt, wr); - wrbuf_iconv_puts(wr, mt->iconv_cd, ""); - wrbuf_puts(wr, "\n"); + wrbuf_printf(wr, "iconv_cd, n->u.controlfield.tag, + strlen(n->u.controlfield.tag)); + wrbuf_puts(wr, ">\n"); break; case YAZ_MARC_COMMENT: wrbuf_printf(wr, "\n"); break; case YAZ_MARC_LEADER: - wrbuf_printf(wr, " "); - wrbuf_iconv_write_cdata(wr, - 0 /* no charset conversion for leader */, + wrbuf_printf(wr, " <%s>", leader_name[turbo]); + wrbuf_iconv_write_cdata(wr, + 0 , /* no charset conversion for leader */ n->u.leader, strlen(n->u.leader)); - wrbuf_printf(wr, "\n"); + wrbuf_printf(wr, "\n", leader_name[turbo]); } } - wrbuf_puts(wr, "\n"); + wrbuf_printf(wr, "\n", record_name[turbo]); return 0; } static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr, const char *ns, const char *format, - const char *type) + const char *type, + int turbo) { if (mt->write_using_libxml2) { @@ -674,7 +783,10 @@ static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr, int ret; xmlNode *root_ptr; - ret = yaz_marc_write_xml(mt, &root_ptr, ns, format, type); + if (!turbo) + ret = yaz_marc_write_xml(mt, &root_ptr, ns, format, type); + else + ret = yaz_marc_write_xml_turbo_xml(mt, &root_ptr, ns, format, type); if (ret == 0) { xmlChar *buf_out; @@ -695,7 +807,7 @@ static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr, #endif } else - return yaz_marc_write_marcxml_ns1(mt, wr, ns, format, type); + return yaz_marc_write_marcxml_wrbuf(mt, wr, ns, format, type, turbo); } int yaz_marc_write_marcxml(yaz_marc_t mt, WRBUF wr) @@ -704,8 +816,19 @@ int yaz_marc_write_marcxml(yaz_marc_t mt, WRBUF wr) /* http://www.loc.gov/marc/bibliographic/ecbdldrd.html#mrcblea */ if (!mt->leader_spec) yaz_marc_modify_leader(mt, 9, "a"); - return yaz_marc_write_marcxml_ns(mt, wr, "http://www.loc.gov/MARC21/slim", - 0, 0); + return yaz_marc_write_marcxml_ns(mt, wr, + "http://www.loc.gov/MARC21/slim", + 0, 0, 0); +} + +int yaz_marc_write_turbomarc(yaz_marc_t mt, WRBUF wr) +{ + /* set leader 09 to 'a' for UNICODE */ + /* http://www.loc.gov/marc/bibliographic/ecbdldrd.html#mrcblea */ + if (!mt->leader_spec) + yaz_marc_modify_leader(mt, 9, "a"); + return yaz_marc_write_marcxml_ns(mt, wr, + "http://www.indexdata.com/turbomarc", 0, 0, 1); } int yaz_marc_write_marcxchange(yaz_marc_t mt, WRBUF wr, @@ -713,12 +836,143 @@ int yaz_marc_write_marcxchange(yaz_marc_t mt, WRBUF wr, const char *type) { return yaz_marc_write_marcxml_ns(mt, wr, - "http://www.bs.dk/standards/MarcXchange", - 0, 0); + "info:lc/xmlns/marcxchange-v1", + 0, 0, 0); } - #if YAZ_HAVE_XML2 + +void add_marc_datafield_turbo_xml(yaz_marc_t mt, struct yaz_marc_node *n, + xmlNode *record_ptr, + xmlNsPtr ns_record, WRBUF wr_cdata, + int identifier_length) +{ + xmlNode *ptr; + struct yaz_marc_subfield *s; + WRBUF subfield_name = wrbuf_alloc(); + + /* TODO consider if safe */ + char field[10]; + field[0] = 'd'; + strncpy(field + 1, n->u.datafield.tag, 3); + field[4] = '\0'; + ptr = xmlNewChild(record_ptr, ns_record, BAD_CAST field, 0); + + if (n->u.datafield.indicator) + { + int i; + for (i = 0; n->u.datafield.indicator[i]; i++) + { + char ind_str[6]; + char ind_val[2]; + + ind_val[0] = n->u.datafield.indicator[i]; + ind_val[1] = '\0'; + sprintf(ind_str, "%s%d", indicator_name[1], i+1); + xmlNewProp(ptr, BAD_CAST ind_str, BAD_CAST ind_val); + } + } + for (s = n->u.datafield.subfields; s; s = s->next) + { + int not_written; + xmlNode *ptr_subfield; + size_t using_code_len = get_subfield_len(mt, s->code_data, + identifier_length); + wrbuf_rewind(wr_cdata); + wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, s->code_data + using_code_len); + marc_iconv_reset(mt, wr_cdata); + + wrbuf_rewind(subfield_name); + wrbuf_puts(subfield_name, "s"); + not_written = element_name_append_attribute_value(mt, subfield_name, 0, s->code_data, using_code_len) != 0; + ptr_subfield = xmlNewTextChild(ptr, ns_record, + BAD_CAST wrbuf_cstr(subfield_name), + BAD_CAST wrbuf_cstr(wr_cdata)); + if (not_written) + { + /* Generate code attribute value and add */ + wrbuf_rewind(wr_cdata); + wrbuf_iconv_write(wr_cdata, mt->iconv_cd,s->code_data, using_code_len); + xmlNewProp(ptr_subfield, BAD_CAST "code", BAD_CAST wrbuf_cstr(wr_cdata)); + } + } + wrbuf_destroy(subfield_name); +} + +static int yaz_marc_write_xml_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr, + const char *ns, + const char *format, + const char *type) +{ + struct yaz_marc_node *n; + int identifier_length; + const char *leader = 0; + xmlNode *record_ptr; + xmlNsPtr ns_record; + WRBUF wr_cdata = 0; + + for (n = mt->nodes; n; n = n->next) + if (n->which == YAZ_MARC_LEADER) + { + leader = n->u.leader; + break; + } + + if (!leader) + return -1; + if (!atoi_n_check(leader+11, 1, &identifier_length)) + return -1; + + wr_cdata = wrbuf_alloc(); + + record_ptr = xmlNewNode(0, BAD_CAST "r"); + *root_ptr = record_ptr; + + ns_record = xmlNewNs(record_ptr, BAD_CAST ns, 0); + xmlSetNs(record_ptr, ns_record); + + if (format) + xmlNewProp(record_ptr, BAD_CAST "format", BAD_CAST format); + if (type) + xmlNewProp(record_ptr, BAD_CAST "type", BAD_CAST type); + for (n = mt->nodes; n; n = n->next) + { + xmlNode *ptr; + + char field[10]; + field[0] = 'c'; + field[4] = '\0'; + + switch(n->which) + { + case YAZ_MARC_DATAFIELD: + add_marc_datafield_turbo_xml(mt, n, record_ptr, ns_record, wr_cdata, identifier_length); + break; + case YAZ_MARC_CONTROLFIELD: + wrbuf_rewind(wr_cdata); + wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, n->u.controlfield.data); + marc_iconv_reset(mt, wr_cdata); + + strncpy(field + 1, n->u.controlfield.tag, 3); + ptr = xmlNewTextChild(record_ptr, ns_record, + BAD_CAST field, + BAD_CAST wrbuf_cstr(wr_cdata)); + break; + case YAZ_MARC_COMMENT: + ptr = xmlNewComment(BAD_CAST n->u.comment); + xmlAddChild(record_ptr, ptr); + break; + case YAZ_MARC_LEADER: + xmlNewTextChild(record_ptr, ns_record, BAD_CAST "l", + BAD_CAST n->u.leader); + break; + } + } + wrbuf_destroy(wr_cdata); + return 0; +} + + int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, const char *ns, const char *format, @@ -782,20 +1036,14 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, for (s = n->u.datafield.subfields; s; s = s->next) { xmlNode *ptr_subfield; - /* if identifier length is 2 (most MARCs), - the code is a single character .. However we've - seen multibyte codes, so see how big it really is */ - size_t using_code_len = - (identifier_length > 2) ? identifier_length - 1 - : - cdata_one_character(mt, s->code_data); - + size_t using_code_len = get_subfield_len(mt, s->code_data, + identifier_length); wrbuf_rewind(wr_cdata); wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, s->code_data + using_code_len); marc_iconv_reset(mt, wr_cdata); ptr_subfield = xmlNewTextChild( - ptr, ns_record, + ptr, ns_record, BAD_CAST "subfield", BAD_CAST wrbuf_cstr(wr_cdata)); wrbuf_rewind(wr_cdata); @@ -829,6 +1077,7 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, wrbuf_destroy(wr_cdata); return 0; } + #endif int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) @@ -939,8 +1188,7 @@ int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) switch(n->which) { case YAZ_MARC_DATAFIELD: - wrbuf_printf(wr, "%.*s", indicator_length, - n->u.datafield.indicator); + wrbuf_write(wr, n->u.datafield.indicator, indicator_length); for (s = n->u.datafield.subfields; s; s = s->next) { wrbuf_putc(wr, ISO2709_IDFS); @@ -992,8 +1240,7 @@ int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize, void yaz_marc_xml(yaz_marc_t mt, int xmlmode) { - if (mt) - mt->xml = xmlmode; + mt->output_format = xmlmode; } void yaz_marc_debug(yaz_marc_t mt, int level) @@ -1089,6 +1336,8 @@ int yaz_marc_decode_formatstr(const char *arg) mode = YAZ_MARC_ISO2709; if (!strcmp(arg, "marcxml")) mode = YAZ_MARC_MARCXML; + if (!strcmp(arg, "turbomarc")) + mode = YAZ_MARC_TURBOMARC; if (!strcmp(arg, "marcxchange")) mode = YAZ_MARC_XCHANGE; if (!strcmp(arg, "line")) @@ -1104,6 +1353,7 @@ void yaz_marc_write_using_libxml2(yaz_marc_t mt, int enable) /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab