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])
95 size_t length = strlen(name);
98 strcpy(tag_buffer, name);
104 // Given a xmlNode ptr, extract a value from either a element name or from a given attribute
105 const char *element_attribute_value_extract(const xmlNode *ptr,
106 const char *attribute_name,
110 const char *name = ptr->name;
111 size_t length = strlen(name);
113 return nmem_strdup(nmem, name+1);
114 // TODO Extract from attribute where matches attribute_name
116 for (attr = ptr->properties; attr; attr = attr->next)
117 if (!strcmp((const char *)attr->name, attribute_name))
118 return nmem_text_node_cdata(attr->children, nmem);
123 int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
125 NMEM nmem = yaz_marc_get_nmem(mt);
126 for (; ptr; ptr = ptr->next)
128 if (ptr->type == XML_ELEMENT_NODE)
131 if (!strncmp((const char *) ptr->name, "s", 1))
133 NMEM nmem = yaz_marc_get_nmem(mt);
134 char *buffer = (char *) nmem_malloc(nmem, 5);
135 const char *tag_value = element_attribute_value_extract(ptr, "code", nmem);
139 mt, "Missing 'code' value for 'subfield'" );
143 size_t ctrl_data_len = 0;
144 char *ctrl_data_buf = 0;
145 ctrl_data_len = strlen((const char *) tag_value);
146 // Extract (length) from CDATA
148 for (p = ptr->children; p ; p = p->next)
149 if (p->type == XML_TEXT_NODE)
150 ctrl_data_len += strlen((const char *)p->content);
151 // Allocate memory for code value (1 character (can be multi-byte) and data
152 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
153 // Build a string with "<Code><data>"
154 strcpy(ctrl_data_buf, (const char *) tag_value);
155 for (p = ptr->children; p ; p = p->next)
156 if (p->type == XML_TEXT_NODE)
157 strcat(ctrl_data_buf, (const char *)p->content);
158 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
163 mt, "Expected element 'subfield', got '%.80s'", ptr->name);
172 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
174 int indicator_length;
175 int identifier_length;
177 int length_data_entry;
179 int length_implementation;
180 const char *leader = 0;
181 const xmlNode *ptr = *ptr_p;
183 for(; ptr; ptr = ptr->next)
184 if (ptr->type == XML_ELEMENT_NODE)
186 if ( !strcmp( (const char *) ptr->name, "leader") ||
187 (!strncmp((const char *) ptr->name, "l", 1) ))
189 xmlNode *p = ptr->children;
190 for(; p; p = p->next)
191 if (p->type == XML_TEXT_NODE)
192 leader = (const char *) p->content;
198 mt, "Expected element 'leader', got '%.80s'", ptr->name);
203 yaz_marc_cprintf(mt, "Missing element 'leader'");
206 if (strlen(leader) != 24)
208 yaz_marc_cprintf(mt, "Bad length %d of leader data."
209 " Must have length of 24 characters", strlen(leader));
212 yaz_marc_set_leader(mt, leader,
218 &length_implementation);
223 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
225 for(; ptr; ptr = ptr->next)
226 if (ptr->type == XML_ELEMENT_NODE)
228 if (!strcmp( (const char *) ptr->name, "controlfield"))
230 const xmlNode *ptr_tag = 0;
231 struct _xmlAttr *attr;
232 for (attr = ptr->properties; attr; attr = attr->next)
233 if (!strcmp((const char *)attr->name, "tag"))
234 ptr_tag = attr->children;
238 mt, "Bad attribute '%.80s' for 'controlfield'",
245 mt, "Missing attribute 'tag' for 'controlfield'" );
248 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
250 else if (!strcmp((const char *) ptr->name, "datafield"))
252 char indstr[11]; /* 0(unused), 1,....9, + zero term */
253 const xmlNode *ptr_tag = 0;
254 struct _xmlAttr *attr;
256 for (i = 0; i<11; i++)
258 for (attr = ptr->properties; attr; attr = attr->next)
259 if (!strcmp((const char *)attr->name, "tag"))
260 ptr_tag = attr->children;
261 else if (strlen((const char *)attr->name) == 4 &&
262 !memcmp(attr->name, "ind", 3))
264 int no = atoi((const char *)attr->name+3);
266 && attr->children->type == XML_TEXT_NODE)
267 indstr[no] = attr->children->content[0];
272 mt, "Bad attribute '%.80s' for 'datafield'",
278 mt, "Missing attribute 'tag' for 'datafield'" );
281 /* note that indstr[0] is unused so we use indstr[1..] */
282 yaz_marc_add_datafield_xml(mt, ptr_tag,
283 indstr+1, strlen(indstr+1));
285 if (yaz_marc_read_xml_subfields(mt, ptr->children))
291 "Expected element controlfield or datafield,"
292 " got %.80s", ptr->name);
299 void yaz_marc_add_datafield_turbo_xml(yaz_marc_t mt, char *tag_value, char *indicators);
301 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
303 for(; ptr; ptr = ptr->next)
304 if (ptr->type == XML_ELEMENT_NODE)
306 if (!strncmp( (const char *) ptr->name, "c", 1))
308 NMEM nmem = yaz_marc_get_nmem(mt);
309 const char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
313 mt, "Missing attribute 'tag' for 'controlfield'" );
316 yaz_marc_add_controlfield_turbo_xml(mt, tag_value, ptr->children);
318 else if (!strncmp((const char *) ptr->name, "d",1))
320 struct _xmlAttr *attr;
321 NMEM nmem = yaz_marc_get_nmem(mt);
322 char *indstr = nmem_malloc(nmem, 11); /* 0(unused), 1,....9, + zero term */
324 for (index = 0; index < 11; index++)
325 indstr[index] = '\0';
326 const char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
330 mt, "Missing attribute 'tag' for 'datafield'" );
333 for (attr = ptr->properties; attr; attr = attr->next)
334 if (strlen((const char *)attr->name) == 2 &&
335 attr->name[0] == 'i')
337 //extract indicator attribute from i#="Y" pattern
338 int no = atoi((const char *)attr->name+1);
340 && attr->children->type == XML_TEXT_NODE)
341 indstr[no] = attr->children->content[0];
346 mt, "Bad attribute '%.80s' for 'datafield'",
349 /* note that indstr[0] is unused so we use indstr[1..] */
350 yaz_marc_add_datafield_turbo_xml(mt, tag_value, indstr+1);
351 int rc = yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */);
358 "Expected element controlfield or datafield,"
359 " got %.80s", ptr->name);
370 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
374 for(; ptr; ptr = ptr->next)
375 if (ptr->type == XML_ELEMENT_NODE)
377 //TODO Should actually look at the namespace but...
378 if (!strcmp((const char *) ptr->name, "record"))
380 yaz_marc_set_read_format(mt, YAZ_MARC_MARCXML);
383 else if (!strcmp((const char *) ptr->name, "r"))
385 yaz_marc_set_read_format(mt, YAZ_MARC_TMARCXML);
390 mt, "Unknown element '%.80s' in MARC XML reader",
397 yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
400 /* ptr points to record node now */
402 if (yaz_marc_read_xml_leader(mt, &ptr))
405 switch (yaz_marc_get_read_format(mt))
407 case YAZ_MARC_MARCXML:
408 return yaz_marc_read_xml_fields(mt, ptr->next);
409 case YAZ_MARC_TMARCXML:
410 return yaz_marc_read_turbo_xml_fields(mt, ptr->next);
420 * c-file-style: "Stroustrup"
421 * indent-tabs-mode: nil
423 * vim: shiftwidth=4 tabstop=8 expandtab