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 // Given a xmlNode ptr, extract a value from either a element name or from a given attribute
103 const char *element_attribute_value_extract(const xmlNode *ptr, const char *attribute_name, NMEM nmem) {
105 const char *name = ptr->name;
106 size_t length = strlen(name);
108 return nmem_strdup(nmem, name+1);
110 // TODO Extract from attribute where matches attribute_name
112 for (attr = ptr->properties; attr; attr = attr->next)
113 if (!strcmp((const char *)attr->name, attribute_name)) {
114 return nmem_text_node_cdata(attr->children, nmem);
120 int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
122 NMEM nmem = yaz_marc_get_nmem(mt);
123 for (; ptr; ptr = ptr->next)
125 if (ptr->type == XML_ELEMENT_NODE)
128 if (!strncmp((const char *) ptr->name, "s", 1))
130 NMEM nmem = yaz_marc_get_nmem(mt);
131 char *buffer = (char *) nmem_malloc(nmem, 5);
132 const char *tag_value = element_attribute_value_extract(ptr, "code", nmem);
136 mt, "Missing 'code' value for 'subfield'" );
140 size_t ctrl_data_len = 0;
141 char *ctrl_data_buf = 0;
142 ctrl_data_len = strlen((const char *) tag_value);
143 // Extract (length) from CDATA
145 for (p = ptr->children; p ; p = p->next)
146 if (p->type == XML_TEXT_NODE)
147 ctrl_data_len += strlen((const char *)p->content);
148 // Allocate memory for code value (1 character (can be multi-byte) and data
149 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
150 // Build a string with "<Code><data>"
151 strcpy(ctrl_data_buf, (const char *) tag_value);
152 for (p = ptr->children; p ; p = p->next)
153 if (p->type == XML_TEXT_NODE)
154 strcat(ctrl_data_buf, (const char *)p->content);
155 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
160 mt, "Expected element 'subfield', got '%.80s'", ptr->name);
169 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
171 int indicator_length;
172 int identifier_length;
174 int length_data_entry;
176 int length_implementation;
177 const char *leader = 0;
178 const xmlNode *ptr = *ptr_p;
180 for(; ptr; ptr = ptr->next)
181 if (ptr->type == XML_ELEMENT_NODE)
183 if ( !strcmp( (const char *) ptr->name, "leader") ||
184 (!strncmp((const char *) ptr->name, "l", 1) ))
186 xmlNode *p = ptr->children;
187 for(; p; p = p->next)
188 if (p->type == XML_TEXT_NODE)
189 leader = (const char *) p->content;
195 mt, "Expected element 'leader', got '%.80s'", ptr->name);
200 yaz_marc_cprintf(mt, "Missing element 'leader'");
203 if (strlen(leader) != 24)
205 yaz_marc_cprintf(mt, "Bad length %d of leader data."
206 " Must have length of 24 characters", strlen(leader));
209 yaz_marc_set_leader(mt, leader,
215 &length_implementation);
220 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
222 for(; ptr; ptr = ptr->next)
223 if (ptr->type == XML_ELEMENT_NODE)
225 if (!strcmp( (const char *) ptr->name, "controlfield"))
227 const xmlNode *ptr_tag = 0;
228 struct _xmlAttr *attr;
229 for (attr = ptr->properties; attr; attr = attr->next)
230 if (!strcmp((const char *)attr->name, "tag"))
231 ptr_tag = attr->children;
235 mt, "Bad attribute '%.80s' for 'controlfield'",
242 mt, "Missing attribute 'tag' for 'controlfield'" );
245 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
247 else if (!strcmp((const char *) ptr->name, "datafield"))
249 char indstr[11]; /* 0(unused), 1,....9, + zero term */
250 const xmlNode *ptr_tag = 0;
251 struct _xmlAttr *attr;
253 for (i = 0; i<11; i++)
255 for (attr = ptr->properties; attr; attr = attr->next)
256 if (!strcmp((const char *)attr->name, "tag"))
257 ptr_tag = attr->children;
258 else if (strlen((const char *)attr->name) == 4 &&
259 !memcmp(attr->name, "ind", 3))
261 int no = atoi((const char *)attr->name+3);
263 && attr->children->type == XML_TEXT_NODE)
264 indstr[no] = attr->children->content[0];
269 mt, "Bad attribute '%.80s' for 'datafield'",
275 mt, "Missing attribute 'tag' for 'datafield'" );
278 /* note that indstr[0] is unused so we use indstr[1..] */
279 yaz_marc_add_datafield_xml(mt, ptr_tag,
280 indstr+1, strlen(indstr+1));
282 if (yaz_marc_read_xml_subfields(mt, ptr->children))
288 "Expected element controlfield or datafield,"
289 " got %.80s", ptr->name);
296 void yaz_marc_add_datafield_turbo_xml(yaz_marc_t mt, char *tag_value, char *indicators);
298 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
300 for(; ptr; ptr = ptr->next)
301 if (ptr->type == XML_ELEMENT_NODE)
303 if (!strncmp( (const char *) ptr->name, "c", 1))
305 NMEM nmem = yaz_marc_get_nmem(mt);
306 const char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
310 mt, "Missing attribute 'tag' for 'controlfield'" );
313 yaz_marc_add_controlfield_turbo_xml(mt, tag_value, ptr->children);
315 else if (!strncmp((const char *) ptr->name, "d",1))
317 struct _xmlAttr *attr;
318 NMEM nmem = yaz_marc_get_nmem(mt);
319 char *indstr = nmem_malloc(nmem, 11); /* 0(unused), 1,....9, + zero term */
321 for (index = 0; index < 11; index++)
322 indstr[index] = '\0';
323 const char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
327 mt, "Missing attribute 'tag' for 'datafield'" );
330 for (attr = ptr->properties; attr; attr = attr->next)
331 if (strlen((const char *)attr->name) == 2 &&
332 attr->name[0] == 'i')
334 //extract indicator attribute from i#="Y" pattern
335 int no = atoi((const char *)attr->name+1);
337 && attr->children->type == XML_TEXT_NODE)
338 indstr[no] = attr->children->content[0];
343 mt, "Bad attribute '%.80s' for 'datafield'",
346 /* note that indstr[0] is unused so we use indstr[1..] */
347 yaz_marc_add_datafield_turbo_xml(mt, tag_value, indstr+1);
348 int rc = yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */);
355 "Expected element controlfield or datafield,"
356 " got %.80s", ptr->name);
367 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
371 for(; ptr; ptr = ptr->next)
372 if (ptr->type == XML_ELEMENT_NODE)
374 //TODO Should actually look at the namespace but...
375 if (!strcmp((const char *) ptr->name, "record")) {
376 yaz_marc_set_read_format(mt, YAZ_MARC_MARCXML);
379 else if (!strcmp((const char *) ptr->name, "r")) {
380 yaz_marc_set_read_format(mt, YAZ_MARC_TMARCXML);
385 mt, "Unknown element '%.80s' in MARC XML reader",
392 yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
395 /* ptr points to record node now */
397 if (yaz_marc_read_xml_leader(mt, &ptr))
400 switch (yaz_marc_get_read_format(mt)) {
401 case YAZ_MARC_MARCXML:
402 return yaz_marc_read_xml_fields(mt, ptr->next);
403 case YAZ_MARC_TMARCXML:
404 return yaz_marc_read_turbo_xml_fields(mt, ptr->next);
414 * c-file-style: "Stroustrup"
415 * indent-tabs-mode: nil
417 * vim: shiftwidth=4 tabstop=8 expandtab