1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2010 Index Data
3 * See the file LICENSE for details.
7 * \file marc_read_xml.c
8 * \brief Implements reading of MARC as XML
22 #include <yaz/marcdisp.h>
23 #include <yaz/wrbuf.h>
24 #include <yaz/yaz-util.h>
25 #include <yaz/nmem_xml.h>
28 #include <libxml/tree.h>
32 int yaz_marc_read_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
34 NMEM nmem = yaz_marc_get_nmem(mt);
35 for (; ptr; ptr = ptr->next)
37 if (ptr->type == XML_ELEMENT_NODE)
39 if (!strcmp((const char *) ptr->name, "subfield"))
41 size_t ctrl_data_len = 0;
42 char *ctrl_data_buf = 0;
43 const xmlNode *p = 0, *ptr_code = 0;
44 struct _xmlAttr *attr;
45 for (attr = ptr->properties; attr; attr = attr->next)
46 if (!strcmp((const char *)attr->name, "code"))
47 ptr_code = attr->children;
51 mt, "Bad attribute '%.80s' for 'subfield'",
58 mt, "Missing attribute 'code' for 'subfield'" );
61 if (ptr_code->type == XML_TEXT_NODE)
64 strlen((const char *)ptr_code->content);
69 mt, "Missing value for 'code' in 'subfield'" );
72 for (p = ptr->children; p ; p = p->next)
73 if (p->type == XML_TEXT_NODE)
74 ctrl_data_len += strlen((const char *)p->content);
75 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
76 strcpy(ctrl_data_buf, (const char *)ptr_code->content);
77 for (p = ptr->children; p ; p = p->next)
78 if (p->type == XML_TEXT_NODE)
79 strcat(ctrl_data_buf, (const char *)p->content);
80 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
85 mt, "Expected element 'subfield', got '%.80s'", ptr->name);
93 const char *tag_value_extract(const char *name, char tag_buffer[5]) {
94 size_t length = strlen(name);
96 strcpy(tag_buffer, name);
102 // pattern <one character or -AB[CD]
103 const char *code_value_extract(const char *name, char tag_buffer[5]) {
104 size_t length = strlen(name);
108 if (length > 2 && length < 6) {
109 if (name[0] != '-') {
113 const char *ptr = name+1;
115 for (index = 0; index < length/2; index++) {
118 strncpy(temp, ptr + 2*index, 2);
119 sscanf(temp, "%02X", &value);
120 tag_buffer[index] = (unsigned char) value;
122 tag_buffer[index] = '\0';
130 int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
132 NMEM nmem = yaz_marc_get_nmem(mt);
133 for (; ptr; ptr = ptr->next)
135 if (ptr->type == XML_ELEMENT_NODE)
138 if (!strncmp((const char *) ptr->name, "s", 1))
140 NMEM nmem = yaz_marc_get_nmem(mt);
141 char *buffer = (char *) nmem_malloc(nmem, 5);
142 const char *tag_value = code_value_extract((ptr->name+1), buffer);
146 mt, "Missing 'code' value for 'subfield'" );
150 size_t ctrl_data_len = 0;
151 char *ctrl_data_buf = 0;
152 ctrl_data_len = strlen((const char *) tag_value);
153 // Extract (length) from CDATA
155 for (p = ptr->children; p ; p = p->next)
156 if (p->type == XML_TEXT_NODE)
157 ctrl_data_len += strlen((const char *)p->content);
158 // Allocate memory for code value (1 character (can be multi-byte) and data
159 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
160 // Build a string with "<Code><data>"
161 strcpy(ctrl_data_buf, (const char *) tag_value);
162 for (p = ptr->children; p ; p = p->next)
163 if (p->type == XML_TEXT_NODE)
164 strcat(ctrl_data_buf, (const char *)p->content);
165 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
170 mt, "Expected element 'subfield', got '%.80s'", ptr->name);
179 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
181 int indicator_length;
182 int identifier_length;
184 int length_data_entry;
186 int length_implementation;
187 const char *leader = 0;
188 const xmlNode *ptr = *ptr_p;
190 for(; ptr; ptr = ptr->next)
191 if (ptr->type == XML_ELEMENT_NODE)
193 if ( !strcmp( (const char *) ptr->name, "leader") ||
194 (!strncmp((const char *) ptr->name, "l", 1) ))
196 xmlNode *p = ptr->children;
197 for(; p; p = p->next)
198 if (p->type == XML_TEXT_NODE)
199 leader = (const char *) p->content;
205 mt, "Expected element 'leader', got '%.80s'", ptr->name);
210 yaz_marc_cprintf(mt, "Missing element 'leader'");
213 if (strlen(leader) != 24)
215 yaz_marc_cprintf(mt, "Bad length %d of leader data."
216 " Must have length of 24 characters", strlen(leader));
219 yaz_marc_set_leader(mt, leader,
225 &length_implementation);
230 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
232 for(; ptr; ptr = ptr->next)
233 if (ptr->type == XML_ELEMENT_NODE)
235 if (!strcmp( (const char *) ptr->name, "controlfield"))
237 const xmlNode *ptr_tag = 0;
238 struct _xmlAttr *attr;
239 for (attr = ptr->properties; attr; attr = attr->next)
240 if (!strcmp((const char *)attr->name, "tag"))
241 ptr_tag = attr->children;
245 mt, "Bad attribute '%.80s' for 'controlfield'",
252 mt, "Missing attribute 'tag' for 'controlfield'" );
255 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
257 else if (!strcmp((const char *) ptr->name, "datafield"))
259 char indstr[11]; /* 0(unused), 1,....9, + zero term */
260 const xmlNode *ptr_tag = 0;
261 struct _xmlAttr *attr;
263 for (i = 0; i<11; i++)
265 for (attr = ptr->properties; attr; attr = attr->next)
266 if (!strcmp((const char *)attr->name, "tag"))
267 ptr_tag = attr->children;
268 else if (strlen((const char *)attr->name) == 4 &&
269 !memcmp(attr->name, "ind", 3))
271 int no = atoi((const char *)attr->name+3);
273 && attr->children->type == XML_TEXT_NODE)
274 indstr[no] = attr->children->content[0];
279 mt, "Bad attribute '%.80s' for 'datafield'",
285 mt, "Missing attribute 'tag' for 'datafield'" );
288 /* note that indstr[0] is unused so we use indstr[1..] */
289 yaz_marc_add_datafield_xml(mt, ptr_tag,
290 indstr+1, strlen(indstr+1));
292 if (yaz_marc_read_xml_subfields(mt, ptr->children))
298 "Expected element controlfield or datafield,"
299 " got %.80s", ptr->name);
306 void yaz_marc_add_datafield_turbo_xml(yaz_marc_t mt, char *tag_value, char *indicators);
308 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
310 for(; ptr; ptr = ptr->next)
311 if (ptr->type == XML_ELEMENT_NODE)
313 if (!strncmp( (const char *) ptr->name, "c", 1))
315 NMEM nmem = yaz_marc_get_nmem(mt);
316 char *buffer = (char *) nmem_malloc(nmem, 5);
317 const char *tag_value = tag_value_extract((const char *)(ptr->name+1), buffer);
321 mt, "Missing attribute 'tag' for 'controlfield'" );
324 yaz_marc_add_controlfield_turbo_xml(mt, tag_value, ptr->children);
326 else if (!strncmp((const char *) ptr->name, "d",1))
328 struct _xmlAttr *attr;
329 NMEM nmem = yaz_marc_get_nmem(mt);
330 char *indstr = nmem_malloc(nmem, 11); /* 0(unused), 1,....9, + zero term */
332 for (index = 0; index < 11; index++)
333 indstr[index] = '\0';
334 char *buffer = (char *) nmem_malloc(nmem, 5);
335 char *tag_value = tag_value_extract(ptr->name+1, buffer);
339 mt, "Missing attribute 'tag' for 'datafield'" );
342 for (attr = ptr->properties; attr; attr = attr->next)
343 if (strlen((const char *)attr->name) == 2 &&
344 attr->name[0] == 'i')
346 //extract indicator attribute from i#="Y" pattern
347 int no = atoi((const char *)attr->name+1);
349 && attr->children->type == XML_TEXT_NODE)
350 indstr[no] = attr->children->content[0];
355 mt, "Bad attribute '%.80s' for 'datafield'",
358 /* note that indstr[0] is unused so we use indstr[1..] */
359 yaz_marc_add_datafield_turbo_xml(mt, tag_value, indstr+1);
360 int rc = yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */);
367 "Expected element controlfield or datafield,"
368 " got %.80s", ptr->name);
379 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
383 for(; ptr; ptr = ptr->next)
384 if (ptr->type == XML_ELEMENT_NODE)
386 //TODO Should actually look at the namespace but...
387 if (!strcmp((const char *) ptr->name, "record")) {
388 yaz_marc_set_read_format(mt, YAZ_MARC_MARCXML);
391 else if (!strcmp((const char *) ptr->name, "r")) {
392 yaz_marc_set_read_format(mt, YAZ_MARC_TMARCXML);
397 mt, "Unknown element '%.80s' in MARC XML reader",
404 yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
407 /* ptr points to record node now */
409 if (yaz_marc_read_xml_leader(mt, &ptr))
412 switch (yaz_marc_get_read_format(mt)) {
413 case YAZ_MARC_MARCXML:
414 return yaz_marc_read_xml_fields(mt, ptr->next);
415 case YAZ_MARC_TMARCXML:
416 return yaz_marc_read_turbo_xml_fields(mt, ptr->next);
426 * c-file-style: "Stroustrup"
427 * indent-tabs-mode: nil
429 * vim: shiftwidth=4 tabstop=8 expandtab