e121aa0c5e369fb0b8c6783aff4bba9a919a0282
[yaz-moved-to-github.git] / src / marc_read_xml.c
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) 1995-2012 Index Data
3  * See the file LICENSE for details.
4  */
5
6 /**
7  * \file marc_read_xml.c
8  * \brief Implements reading of MARC as XML
9  */
10
11 #if HAVE_CONFIG_H
12 #include <config.h>
13 #endif
14
15 #ifdef WIN32
16 #include <windows.h>
17 #endif
18
19 #include <stdio.h>
20 #include <string.h>
21 #include <yaz/marcdisp.h>
22 #include <yaz/wrbuf.h>
23 #include <yaz/yaz-util.h>
24 #include <yaz/nmem_xml.h>
25
26 #if YAZ_HAVE_XML2
27 #include <libxml/tree.h>
28 #endif
29
30 #if YAZ_HAVE_XML2
31 int yaz_marc_read_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
32 {
33     NMEM nmem = yaz_marc_get_nmem(mt);
34     for (; ptr; ptr = ptr->next)
35     {
36         if (ptr->type == XML_ELEMENT_NODE)
37         {
38             if (!strcmp((const char *) ptr->name, "subfield"))
39             {
40                 size_t ctrl_data_len = 0;
41                 char *ctrl_data_buf = 0;
42                 const xmlNode *p = 0, *ptr_code = 0;
43                 struct _xmlAttr *attr;
44                 for (attr = ptr->properties; attr; attr = attr->next)
45                     if (!strcmp((const char *)attr->name, "code"))
46                         ptr_code = attr->children;
47                     else
48                     {
49                         yaz_marc_cprintf(
50                             mt, "Bad attribute '%.80s' for 'subfield'",
51                             attr->name);
52                         return -1;
53                     }
54                 if (!ptr_code)
55                 {
56                     yaz_marc_cprintf(
57                         mt, "Missing attribute 'code' for 'subfield'" );
58                     return -1;
59                 }
60                 if (ptr_code->type == XML_TEXT_NODE)
61                 {
62                     ctrl_data_len =
63                         strlen((const char *)ptr_code->content);
64                 }
65                 else
66                 {
67                     yaz_marc_cprintf(
68                         mt, "Missing value for 'code' in 'subfield'" );
69                     return -1;
70                 }
71                 for (p = ptr->children; p ; p = p->next)
72                     if (p->type == XML_TEXT_NODE)
73                         ctrl_data_len += strlen((const char *)p->content);
74                 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
75                 strcpy(ctrl_data_buf, (const char *)ptr_code->content);
76                 for (p = ptr->children; p ; p = p->next)
77                     if (p->type == XML_TEXT_NODE)
78                         strcat(ctrl_data_buf, (const char *)p->content);
79                 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
80             }
81             else
82             {
83                 yaz_marc_cprintf(
84                     mt, "Expected element 'subfield', got '%.80s'", ptr->name);
85                 return -1;
86             }
87         }
88     }
89     return 0;
90 }
91
92 const char *tag_value_extract(const char *name, char tag_buffer[5])
93 {
94     size_t length = strlen(name);
95     if (length == 3)
96     {
97         strcpy(tag_buffer, name);
98         return tag_buffer;
99     }
100     return 0;
101 }
102
103 // Given a xmlNode ptr,  extract a value from either a element name or from a given attribute
104 char *element_attribute_value_extract(const xmlNode *ptr,
105                                       const char *attribute_name,
106                                       NMEM nmem)
107 {
108     const char *name = (const char *) ptr->name;
109     size_t length = strlen(name);
110     xmlAttr *attr;
111     if (length > 1 )
112         return nmem_strdup(nmem, name+1);
113     // TODO Extract from attribute where matches attribute_name
114     for (attr = ptr->properties; attr; attr = attr->next)
115         if (!strcmp((const char *)attr->name, attribute_name))
116             return nmem_text_node_cdata(attr->children, nmem);
117     return 0;
118 }
119
120
121 int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
122 {
123     for (; ptr; ptr = ptr->next)
124     {
125         if (ptr->type == XML_ELEMENT_NODE)
126         {
127             if (!strncmp((const char *) ptr->name, "s", 1))
128             {
129                 NMEM nmem = yaz_marc_get_nmem(mt);
130                 xmlNode *p;
131                 size_t ctrl_data_len = 0;
132                 char *ctrl_data_buf = 0;
133                 const char *tag_value = element_attribute_value_extract(ptr, "code", nmem);
134                 if (!tag_value)
135                 {
136                     yaz_marc_cprintf(
137                         mt, "Missing 'code' value for 'subfield'" );
138                     return -1;
139                 }
140
141                 ctrl_data_len = strlen((const char *) tag_value);
142                 // Extract (length) from CDATA
143                 for (p = ptr->children; p ; p = p->next)
144                     if (p->type == XML_TEXT_NODE)
145                         ctrl_data_len += strlen((const char *)p->content);
146                 // Allocate memory for code value (1 character (can be multi-byte) and data
147                 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
148                 // Build a string with "<Code><data>"
149                 strcpy(ctrl_data_buf, (const char *) tag_value);
150                 for (p = ptr->children; p ; p = p->next)
151                     if (p->type == XML_TEXT_NODE)
152                         strcat(ctrl_data_buf, (const char *)p->content);
153                 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
154             }
155             else
156             {
157                 yaz_marc_cprintf(
158                     mt, "Expected element 'subfield', got '%.80s'", ptr->name);
159                 return -1;
160             }
161         }
162     }
163     return 0;
164 }
165
166
167 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
168 {
169     int indicator_length;
170     int identifier_length;
171     int base_address;
172     int length_data_entry;
173     int length_starting;
174     int length_implementation;
175     const char *leader = 0;
176     const xmlNode *ptr = *ptr_p;
177
178     for(; ptr; ptr = ptr->next)
179         if (ptr->type == XML_ELEMENT_NODE)
180         {
181             if ( !strcmp( (const char *) ptr->name, "leader") ||
182                  (!strncmp((const char *) ptr->name, "l", 1) ))
183             {
184                 xmlNode *p = ptr->children;
185                 for(; p; p = p->next)
186                     if (p->type == XML_TEXT_NODE)
187                         leader = (const char *) p->content;
188                 break;
189             }
190             else
191             {
192                 yaz_marc_cprintf(
193                     mt, "Expected element 'leader', got '%.80s'", ptr->name);
194             }
195         }
196     if (!leader)
197     {
198         yaz_marc_cprintf(mt, "Missing element 'leader'");
199         return -1;
200     }
201     if (strlen(leader) != 24)
202     {
203         yaz_marc_cprintf(mt, "Bad length %d of leader data."
204                          " Must have length of 24 characters", strlen(leader));
205         return -1;
206     }
207     yaz_marc_set_leader(mt, leader,
208                         &indicator_length,
209                         &identifier_length,
210                         &base_address,
211                         &length_data_entry,
212                         &length_starting,
213                         &length_implementation);
214     *ptr_p = ptr;
215     return 0;
216 }
217
218 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
219 {
220     for(; ptr; ptr = ptr->next)
221         if (ptr->type == XML_ELEMENT_NODE)
222         {
223             if (!strcmp( (const char *) ptr->name, "controlfield"))
224             {
225                 const xmlNode *ptr_tag = 0;
226                 struct _xmlAttr *attr;
227                 for (attr = ptr->properties; attr; attr = attr->next)
228                     if (!strcmp((const char *)attr->name, "tag"))
229                         ptr_tag = attr->children;
230                     else
231                     {
232                         yaz_marc_cprintf(
233                             mt, "Bad attribute '%.80s' for 'controlfield'",
234                             attr->name);
235                         return -1;
236                     }
237                 if (!ptr_tag)
238                 {
239                     yaz_marc_cprintf(
240                         mt, "Missing attribute 'tag' for 'controlfield'" );
241                     return -1;
242                 }
243                 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
244             }
245             else if (!strcmp((const char *) ptr->name, "datafield"))
246             {
247                 char indstr[11]; /* 0(unused), 1,....9, + zero term */
248                 const xmlNode *ptr_tag = 0;
249                 struct _xmlAttr *attr;
250                 int i;
251                 for (i = 0; i<11; i++)
252                     indstr[i] = '\0';
253                 for (attr = ptr->properties; attr; attr = attr->next)
254                     if (!strcmp((const char *)attr->name, "tag"))
255                         ptr_tag = attr->children;
256                     else if (strlen((const char *)attr->name) == 4 &&
257                              !memcmp(attr->name, "ind", 3))
258                     {
259                         int no = atoi((const char *)attr->name+3);
260                         if (attr->children
261                             && attr->children->type == XML_TEXT_NODE)
262                             indstr[no] = attr->children->content[0];
263                     }
264                     else
265                     {
266                         yaz_marc_cprintf(
267                             mt, "Bad attribute '%.80s' for 'datafield'",
268                             attr->name);
269                     }
270                 if (!ptr_tag)
271                 {
272                     yaz_marc_cprintf(
273                         mt, "Missing attribute 'tag' for 'datafield'" );
274                     return -1;
275                 }
276                 /* note that indstr[0] is unused so we use indstr[1..] */
277                 yaz_marc_add_datafield_xml(mt, ptr_tag,
278                                            indstr+1, strlen(indstr+1));
279
280                 if (yaz_marc_read_xml_subfields(mt, ptr->children))
281                     return -1;
282             }
283             else
284             {
285                 yaz_marc_cprintf(mt,
286                                  "Expected element controlfield or datafield,"
287                                  " got %.80s", ptr->name);
288                 return -1;
289             }
290         }
291     return 0;
292 }
293
294
295 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
296 {
297     for(; ptr; ptr = ptr->next)
298         if (ptr->type == XML_ELEMENT_NODE)
299         {
300             if (!strncmp( (const char *) ptr->name, "c", 1))
301             {
302                 NMEM nmem = yaz_marc_get_nmem(mt);
303                 char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
304                 if (!tag_value)
305                 {
306                     yaz_marc_cprintf(
307                         mt, "Missing attribute 'tag' for 'controlfield'" );
308                     return -1;
309                 }
310                 yaz_marc_add_controlfield_xml2(mt, tag_value, ptr->children);
311             }
312             else if (!strncmp((const char *) ptr->name, "d",1))
313             {
314                 struct _xmlAttr *attr;
315                 NMEM nmem = yaz_marc_get_nmem(mt);
316                 char *tag_value;
317                 char *indstr = nmem_malloc(nmem, 11);  /* 0(unused), 1,....9, + zero term */
318                 int index = 0;
319                 for (index = 0; index < 11; index++)
320                     indstr[index] = '\0';
321                 tag_value = element_attribute_value_extract(ptr, "tag", nmem);
322                 if (!tag_value)
323                 {
324                     yaz_marc_cprintf(
325                         mt, "Missing attribute 'tag' for 'datafield'" );
326                     return -1;
327                 }
328                 for (attr = ptr->properties; attr; attr = attr->next)
329                     if (strlen((const char *)attr->name) == 2 &&
330                         attr->name[0] == 'i')
331                     {
332                         //extract indicator attribute from i#="Y" pattern
333                         int no = atoi((const char *)attr->name+1);
334                         if (attr->children
335                             && attr->children->type == XML_TEXT_NODE)
336                             indstr[no] = attr->children->content[0];
337                     }
338                     else
339                     {
340                         yaz_marc_cprintf(
341                             mt, "Bad attribute '%.80s' for 'datafield'",
342                             attr->name);
343                     }
344                 /* note that indstr[0] is unused so we use indstr[1..] */
345                 yaz_marc_add_datafield_xml2(mt, tag_value, indstr+1);
346                 if (yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */))
347                     return -1;
348             }
349             else
350             {
351                 yaz_marc_cprintf(mt,
352                                  "Expected element controlfield or datafield,"
353                                  " got %.80s", ptr->name);
354                 return -1;
355             }
356         }
357     return 0;
358 }
359
360
361 #endif
362
363 #if YAZ_HAVE_XML2
364 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
365 {
366     int format = 0;
367     yaz_marc_reset(mt);
368
369     for(; ptr; ptr = ptr->next)
370         if (ptr->type == XML_ELEMENT_NODE)
371         {
372             if (!strcmp((const char *) ptr->name, "record"))
373             {
374                 format = YAZ_MARC_MARCXML;
375                 break;
376             }
377             else if (!strcmp((const char *) ptr->name, "r"))
378             {
379                 format = YAZ_MARC_TURBOMARC;
380                 break;
381             }
382             else
383             {
384                 yaz_marc_cprintf(
385                     mt, "Unknown element '%.80s' in MARC XML reader",
386                     ptr->name);
387                 return -1;
388             }
389         }
390     if (!ptr)
391     {
392         yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
393         return -1;
394     }
395     /* ptr points to record node now */
396     ptr = ptr->children;
397     if (yaz_marc_read_xml_leader(mt, &ptr))
398         return -1;
399
400     switch (format)
401     {
402     case YAZ_MARC_MARCXML:
403         return yaz_marc_read_xml_fields(mt, ptr->next);
404     case YAZ_MARC_TURBOMARC:
405         return yaz_marc_read_turbo_xml_fields(mt, ptr->next);
406     }
407     return -1;
408 }
409 #endif
410
411
412 /*
413  * Local variables:
414  * c-basic-offset: 4
415  * c-file-style: "Stroustrup"
416  * indent-tabs-mode: nil
417  * End:
418  * vim: shiftwidth=4 tabstop=8 expandtab
419  */
420