X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fmarcdisp.c;h=32718898f7817e4f5c2ee2906b5b7947efa68ab2;hp=6eeccb3ae10005cf700cd526441f3d559b8166b7;hb=f36abe0a776fa52b972a3fec1b46b94e26814f22;hpb=021f5586a328c6600460aa9f9be664ba19ba20d4 diff --git a/src/marcdisp.c b/src/marcdisp.c index 6eeccb3..3271889 100644 --- a/src/marcdisp.c +++ b/src/marcdisp.c @@ -1,8 +1,6 @@ -/* - * Copyright (C) 1995-2007, Index Data ApS +/* This file is part of the YAZ toolkit. + * Copyright (C) Index Data * See the file LICENSE for details. - * - * $Id: marcdisp.c,v 1.47 2007-02-23 10:15:01 adam Exp $ */ /** @@ -20,8 +18,8 @@ #include #include +#include #include -#include #include #include #include @@ -33,9 +31,15 @@ #include #endif +enum yaz_collection_state { + no_collection, + collection_first, + collection_second +}; + /** \brief node types for yaz_marc_node */ enum YAZ_MARC_NODE_TYPE -{ +{ YAZ_MARC_DATAFIELD, YAZ_MARC_CONTROLFIELD, YAZ_MARC_COMMENT, @@ -82,9 +86,10 @@ 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; yaz_iconv_t iconv_cd; char subfield_str[8]; char endline_str[8]; @@ -97,9 +102,10 @@ 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; mt->m_wr = wrbuf_alloc(); mt->iconv_cd = 0; mt->leader_spec = 0; @@ -116,7 +122,7 @@ void yaz_marc_destroy(yaz_marc_t mt) if (!mt) return ; nmem_destroy(mt->nmem); - wrbuf_free(mt->m_wr, 1); + wrbuf_destroy(mt->m_wr); xfree(mt->leader_spec); xfree(mt); } @@ -128,24 +134,22 @@ NMEM yaz_marc_get_nmem(yaz_marc_t mt) static void marc_iconv_reset(yaz_marc_t mt, WRBUF wr) { - if (mt->iconv_cd) - { - char outbuf[12]; - size_t outbytesleft = sizeof(outbuf); - char *outp = outbuf; - size_t r = yaz_iconv(mt->iconv_cd, 0, 0, &outp, &outbytesleft); - if (r != (size_t) (-1)) - wrbuf_write(wr, outbuf, outp - outbuf); - } + wrbuf_iconv_reset(wr, mt->iconv_cd); } 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) { - struct yaz_marc_node *n = nmem_malloc(mt->nmem, sizeof(*n)); + struct yaz_marc_node *n = (struct yaz_marc_node *) + nmem_malloc(mt->nmem, sizeof(*n)); n->next = 0; *mt->nodes_pp = n; mt->nodes_pp = &n->next; @@ -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; + size_t 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, @@ -264,7 +332,8 @@ void yaz_marc_add_subfield(yaz_marc_t mt, if (mt->subfield_pp) { - struct yaz_marc_subfield *n = nmem_malloc(mt->nmem, sizeof(*n)); + struct yaz_marc_subfield *n = (struct yaz_marc_subfield *) + nmem_malloc(mt->nmem, sizeof(*n)); n->code_data = nmem_strdupn(mt->nmem, code_data, code_data_len); n->next = 0; /* mark subfield_pp to point to this one, so we append here next */ @@ -273,14 +342,16 @@ void yaz_marc_add_subfield(yaz_marc_t mt, } } -int atoi_n_check(const char *buf, int size, int *val) +static void check_ascii(yaz_marc_t mt, char *leader, int offset, + int ch_default) { - int i; - for (i = 0; i < size; i++) - if (!isdigit(i[(const unsigned char *) buf])) - return 0; - *val = atoi_n(buf, size); - return 1; + if (leader[offset] < ' ' || leader[offset] > 127) + { + yaz_marc_cprintf(mt, + "Leader character at offset %d is non-ASCII. " + "Setting value to '%c'", offset, ch_default); + leader[offset] = ch_default; + } } void yaz_marc_set_leader(yaz_marc_t mt, const char *leader_c, @@ -295,53 +366,57 @@ void yaz_marc_set_leader(yaz_marc_t mt, const char *leader_c, memcpy(leader, leader_c, 24); - if (!atoi_n_check(leader+10, 1, indicator_length)) + check_ascii(mt, leader, 5, 'a'); + check_ascii(mt, leader, 6, 'a'); + check_ascii(mt, leader, 7, 'a'); + check_ascii(mt, leader, 8, '#'); + check_ascii(mt, leader, 9, '#'); + if (!atoi_n_check(leader+10, 1, indicator_length) || *indicator_length == 0) { - yaz_marc_cprintf(mt, - "Indicator length at offset 10 should hold a digit." - " Assuming 2"); + yaz_marc_cprintf(mt, "Indicator length at offset 10 should" + " hold a number 1-9. Assuming 2"); leader[10] = '2'; *indicator_length = 2; } - if (!atoi_n_check(leader+11, 1, identifier_length)) + if (!atoi_n_check(leader+11, 1, identifier_length) || *identifier_length == 0) { - yaz_marc_cprintf(mt, - "Identifier length at offset 11 should hold a digit." - " Assuming 2"); + yaz_marc_cprintf(mt, "Identifier length at offset 11 should " + " hold a number 1-9. Assuming 2"); leader[11] = '2'; *identifier_length = 2; } if (!atoi_n_check(leader+12, 5, base_address)) { - yaz_marc_cprintf(mt, - "Base address at offsets 12..16 should hold a number." - " Assuming 0"); + yaz_marc_cprintf(mt, "Base address at offsets 12..16 should" + " hold a number. Assuming 0"); *base_address = 0; } - if (!atoi_n_check(leader+20, 1, length_data_entry)) + check_ascii(mt, leader, 17, '#'); + check_ascii(mt, leader, 18, '#'); + check_ascii(mt, leader, 19, '#'); + if (!atoi_n_check(leader+20, 1, length_data_entry) || + *length_data_entry < 3) { - yaz_marc_cprintf(mt, - "Length data entry at offset 20 should hold a digit." - " Assuming 4"); + yaz_marc_cprintf(mt, "Length data entry at offset 20 should" + " hold a number 3-9. Assuming 4"); *length_data_entry = 4; leader[20] = '4'; } - if (!atoi_n_check(leader+21, 1, length_starting)) + if (!atoi_n_check(leader+21, 1, length_starting) || *length_starting < 4) { - yaz_marc_cprintf(mt, - "Length starting at offset 21 should hold a digit." - " Assuming 5"); + yaz_marc_cprintf(mt, "Length starting at offset 21 should" + " hold a number 4-9. Assuming 5"); *length_starting = 5; leader[21] = '5'; } if (!atoi_n_check(leader+22, 1, length_implementation)) { - yaz_marc_cprintf(mt, - "Length implementation at offset 22 should hold a digit." - " Assuming 0"); + yaz_marc_cprintf(mt, "Length implementation at offset 22 should" + " hold a number. Assuming 0"); *length_implementation = 0; leader[22] = '0'; } + check_ascii(mt, leader, 23, '0'); if (mt->debug) { @@ -383,14 +458,24 @@ static size_t cdata_one_character(yaz_marc_t mt, const char *buf) size_t inbytesleft = i; size_t r = yaz_iconv(mt->iconv_cd, (char**) &inp, &inbytesleft, &outp, &outbytesleft); + yaz_iconv(mt->iconv_cd, 0, 0, &outp, &outbytesleft); if (r != (size_t) (-1)) return i; /* got a complete sequence */ } return 1; /* giving up */ } + else + { + int error = 0; + size_t no_read = 0; + (void) yaz_read_UTF8_char((const unsigned char *) buf, strlen(buf), + &no_read, &error); + if (error == 0 && no_read > 0) + return no_read; + } return 1; /* we don't know */ } - + void yaz_marc_reset(yaz_marc_t mt) { nmem_reset(mt->nmem); @@ -411,7 +496,7 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr) leader = n->u.leader; break; } - + if (!leader) return -1; if (!atoi_n_check(leader+11, 1, &identifier_length)) @@ -422,9 +507,9 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr) switch(n->which) { case YAZ_MARC_COMMENT: - wrbuf_iconv_write(wr, mt->iconv_cd, + 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; @@ -433,6 +518,17 @@ 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) { @@ -446,7 +542,7 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr) leader = n->u.leader; break; } - + if (!leader) return -1; if (!atoi_n_check(leader+11, 1, &identifier_length)) @@ -462,19 +558,14 @@ 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); - - wrbuf_puts (wr, mt->subfield_str); - wrbuf_iconv_write(wr, mt->iconv_cd, 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, using_code_len); wrbuf_iconv_puts(wr, mt->iconv_cd, " "); - wrbuf_iconv_puts(wr, mt->iconv_cd, + wrbuf_iconv_puts(wr, mt->iconv_cd, s->code_data + using_code_len); marc_iconv_reset(mt, wr); } @@ -489,8 +580,9 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr) break; case YAZ_MARC_COMMENT: wrbuf_puts(wr, "("); - wrbuf_iconv_write(wr, mt->iconv_cd, + wrbuf_iconv_write(wr, mt->iconv_cd, n->u.comment, strlen(n->u.comment)); + marc_iconv_reset(mt, wr); wrbuf_puts(wr, ")\n"); break; case YAZ_MARC_LEADER: @@ -501,35 +593,73 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr) return 0; } +int yaz_marc_write_trailer(yaz_marc_t mt, WRBUF wr) +{ + if (mt->enable_collection == collection_second) + { + switch(mt->output_format) + { + case YAZ_MARC_MARCXML: + case YAZ_MARC_TURBOMARC: + wrbuf_printf(wr, "\n"); + break; + case YAZ_MARC_XCHANGE: + wrbuf_printf(wr, "\n"); + break; + } + } + return 0; +} + +void yaz_marc_enable_collection(yaz_marc_t mt) +{ + mt->enable_collection = collection_first; +} + 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: return yaz_marc_write_iso2709(mt, wr); case YAZ_MARC_CHECK: return yaz_marc_write_check(mt, wr); + case YAZ_MARC_JSON: + return yaz_marc_write_json(mt, 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; @@ -541,13 +671,25 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr, leader = n->u.leader; break; } - + if (!leader) return -1; if (!atoi_n_check(leader+11, 1, &identifier_length)) return -1; - wrbuf_printf(wr, "enable_collection != no_collection) + { + if (mt->enable_collection == collection_first) + { + wrbuf_printf(wr, "\n", ns); + mt->enable_collection = collection_second; + } + wrbuf_printf(wr, "<%s", record_name[turbo]); + } + else + { + wrbuf_printf(wr, "<%s xmlns=\"%s\"", record_name[turbo], ns); + } if (format) wrbuf_printf(wr, " format=\"%.80s\"", format); if (type) @@ -560,55 +702,85 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr, switch(n->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_iconv_puts(wr, mt->iconv_cd, n->u.controlfield.data); - + 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 *ns, const char *format, - const char *type) + const char *type, + int turbo) { if (mt->write_using_libxml2) { @@ -638,7 +811,10 @@ 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; @@ -659,15 +835,28 @@ 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) { + /* 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.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, @@ -675,17 +864,148 @@ 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 *ns, const char *format, const char *type) { -#if YAZ_HAVE_XML2 struct yaz_marc_node *n; int identifier_length; const char *leader = 0; @@ -699,7 +1019,7 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, leader = n->u.leader; break; } - + if (!leader) return -1; if (!atoi_n_check(leader+11, 1, &identifier_length)) @@ -744,20 +1064,14 @@ 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); @@ -771,11 +1085,11 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, wrbuf_rewind(wr_cdata); wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, n->u.controlfield.data); marc_iconv_reset(mt, wr_cdata); - + ptr = xmlNewTextChild(record_ptr, ns_record, BAD_CAST "controlfield", BAD_CAST wrbuf_cstr(wr_cdata)); - + xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.controlfield.tag); break; case YAZ_MARC_COMMENT: @@ -790,11 +1104,10 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr, } wrbuf_destroy(wr_cdata); return 0; -#else - return -1; -#endif } +#endif + int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) { struct yaz_marc_node *n; @@ -807,11 +1120,11 @@ int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) const char *leader = 0; WRBUF wr_dir, wr_head, wr_data_tmp; int base_address; - + for (n = mt->nodes; n; n = n->next) if (n->which == YAZ_MARC_LEADER) leader = n->u.leader; - + if (!leader) return -1; if (!atoi_n_check(leader+10, 1, &indicator_length)) @@ -847,16 +1160,18 @@ int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) } /* write dummy FS (makes MARC-8 to become ASCII) */ wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' '); + marc_iconv_reset(mt, wr_data_tmp); data_length += wrbuf_len(wr_data_tmp); break; case YAZ_MARC_CONTROLFIELD: wrbuf_printf(wr_dir, "%.3s", n->u.controlfield.tag); wrbuf_rewind(wr_data_tmp); - wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd, + wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd, n->u.controlfield.data); marc_iconv_reset(mt, wr_data_tmp); wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' ');/* field sep */ + marc_iconv_reset(mt, wr_data_tmp); data_length += wrbuf_len(wr_data_tmp); break; case YAZ_MARC_COMMENT: @@ -887,12 +1202,12 @@ int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) wrbuf_printf(wr_head, "%05d", base_address); /* from "original" leader */ wrbuf_write(wr_head, leader+17, 7); - + wrbuf_write(wr, wrbuf_buf(wr_head), 24); wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir)); - wrbuf_free(wr_head, 1); - wrbuf_free(wr_dir, 1); - wrbuf_free(wr_data_tmp, 1); + wrbuf_destroy(wr_head); + wrbuf_destroy(wr_dir); + wrbuf_destroy(wr_data_tmp); for (n = mt->nodes; n; n = n->next) { @@ -901,8 +1216,7 @@ 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); @@ -926,6 +1240,91 @@ int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr) return 0; } +int yaz_marc_write_json(yaz_marc_t mt, WRBUF w) +{ + int identifier_length; + struct yaz_marc_node *n; + const char *leader = 0; + int first = 1; + + wrbuf_puts(w, "{\n"); + for (n = mt->nodes; n; n = n->next) + if (n->which == YAZ_MARC_LEADER) + leader = n->u.leader; + + if (!leader) + return -1; + + if (!atoi_n_check(leader+11, 1, &identifier_length)) + return -1; + + wrbuf_puts(w, "\t\"leader\":\""); + wrbuf_json_puts(w, leader); + wrbuf_puts(w, "\",\n"); + wrbuf_puts(w, "\t\"fields\":\n\t[\n"); + + for (n = mt->nodes; n; n = n->next) + { + struct yaz_marc_subfield *s; + const char *sep = ""; + switch (n->which) + { + case YAZ_MARC_LEADER: + case YAZ_MARC_COMMENT: + break; + case YAZ_MARC_CONTROLFIELD: + if (first) + first = 0; + else + wrbuf_puts(w, ",\n"); + wrbuf_puts(w, "\t\t{\n\t\t\t\""); + wrbuf_iconv_json_puts(w, mt->iconv_cd, n->u.controlfield.tag); + wrbuf_puts(w, "\":\""); + wrbuf_iconv_json_puts(w, mt->iconv_cd, n->u.controlfield.data); + wrbuf_puts(w, "\"\n\t\t}"); + break; + case YAZ_MARC_DATAFIELD: + if (first) + first = 0; + else + wrbuf_puts(w, ",\n"); + + wrbuf_puts(w, "\t\t{\n\t\t\t\""); + wrbuf_json_puts(w, n->u.datafield.tag); + wrbuf_puts(w, "\":\n\t\t\t{\n\t\t\t\t\"subfields\":\n\t\t\t\t[\n"); + for (s = n->u.datafield.subfields; s; s = s->next) + { + size_t using_code_len = get_subfield_len(mt, s->code_data, + identifier_length); + wrbuf_puts(w, sep); + sep = ",\n"; + wrbuf_puts(w, "\t\t\t\t\t{\n\t\t\t\t\t\t\""); + wrbuf_iconv_json_write(w, mt->iconv_cd, + s->code_data, using_code_len); + wrbuf_puts(w, "\":\""); + wrbuf_iconv_json_puts(w, mt->iconv_cd, + s->code_data + using_code_len); + wrbuf_puts(w, "\"\n\t\t\t\t\t}"); + } + wrbuf_puts(w, "\n\t\t\t\t]"); + if (n->u.datafield.indicator[0]) + { + int i; + for (i = 0; n->u.datafield.indicator[i]; i++) + { + wrbuf_printf(w, ",\n\t\t\t\t\"ind%d\":\"%c\"", i + 1, + n->u.datafield.indicator[i]); + } + } + wrbuf_puts(w, "\n\t\t\t}\n"); + wrbuf_puts(w, "\n\t\t}"); + break; + } + } + wrbuf_puts(w, "\n\t]\n"); + wrbuf_puts(w, "}\n"); + return 0; +} int yaz_marc_decode_wrbuf(yaz_marc_t mt, const char *buf, int bsize, WRBUF wr) { @@ -939,14 +1338,14 @@ int yaz_marc_decode_wrbuf(yaz_marc_t mt, const char *buf, int bsize, WRBUF wr) } int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize, - char **result, int *rsize) + const char **result, size_t *rsize) { int r; wrbuf_rewind(mt->m_wr); r = yaz_marc_decode_wrbuf(mt, buf, bsize, mt->m_wr); if (result) - *result = wrbuf_buf(mt->m_wr); + *result = wrbuf_cstr(mt->m_wr); if (rsize) *rsize = wrbuf_len(mt->m_wr); return r; @@ -954,8 +1353,7 @@ 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) @@ -969,6 +1367,11 @@ void yaz_marc_iconv(yaz_marc_t mt, yaz_iconv_t cd) mt->iconv_cd = cd; } +yaz_iconv_t yaz_marc_get_iconv(yaz_marc_t mt) +{ + return mt->iconv_cd; +} + void yaz_marc_modify_leader(yaz_marc_t mt, size_t off, const char *str) { struct yaz_marc_node *n; @@ -982,53 +1385,6 @@ void yaz_marc_modify_leader(yaz_marc_t mt, size_t off, const char *str) } } -/* deprecated */ -int yaz_marc_decode(const char *buf, WRBUF wr, int debug, int bsize, int xml) -{ - yaz_marc_t mt = yaz_marc_create(); - int r; - - mt->debug = debug; - mt->xml = xml; - r = yaz_marc_decode_wrbuf(mt, buf, bsize, wr); - yaz_marc_destroy(mt); - return r; -} - -/* deprecated */ -int marc_display_wrbuf (const char *buf, WRBUF wr, int debug, int bsize) -{ - return yaz_marc_decode(buf, wr, debug, bsize, 0); -} - -/* deprecated */ -int marc_display_exl (const char *buf, FILE *outf, int debug, int bsize) -{ - yaz_marc_t mt = yaz_marc_create(); - int r; - - mt->debug = debug; - r = yaz_marc_decode_wrbuf (mt, buf, bsize, mt->m_wr); - if (!outf) - outf = stdout; - if (r > 0) - fwrite (wrbuf_buf(mt->m_wr), 1, wrbuf_len(mt->m_wr), outf); - yaz_marc_destroy(mt); - return r; -} - -/* deprecated */ -int marc_display_ex (const char *buf, FILE *outf, int debug) -{ - return marc_display_exl (buf, outf, debug, -1); -} - -/* deprecated */ -int marc_display (const char *buf, FILE *outf) -{ - return marc_display_ex (buf, outf, 0); -} - int yaz_marc_leader_spec(yaz_marc_t mt, const char *leader_spec) { xfree(mt->leader_spec); @@ -1055,14 +1411,14 @@ static int marc_exec_leader(const char *leader_spec, char *leader, size_t size) no = sscanf(cp, "%d=%20[^,]%n", &pos, val, &no_read); if (no < 2 || no_read < 3) return -1; - if (pos < 0 || pos >= size) + if (pos < 0 || (size_t) pos >= size) return -1; if (*val == '\'') { const char *vp = strchr(val+1, '\''); size_t len; - + if (!vp) return -1; len = vp-val-1; @@ -1088,15 +1444,19 @@ static int marc_exec_leader(const char *leader_spec, char *leader, size_t size) int yaz_marc_decode_formatstr(const char *arg) { - int mode = -1; + int mode = -1; if (!strcmp(arg, "marc")) 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")) mode = YAZ_MARC_LINE; + if (!strcmp(arg, "json")) + mode = YAZ_MARC_JSON; return mode; } @@ -1108,6 +1468,7 @@ 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