Fix XML config GFS do not fail if port is in use YAZ-709
[yaz-moved-to-github.git] / src / marc_read_xml.c
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) 1995-2013 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                                     int *indicator_length)
169 {
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                 ptr = ptr->next;
189             }
190             break;
191         }
192     if (!leader)
193     {
194         yaz_marc_cprintf(mt, "Missing leader. Inserting fake leader");
195         leader = "00000nam a22000000a 4500";
196     }
197     if (strlen(leader) != 24)
198     {
199         yaz_marc_cprintf(mt, "Bad length %d of leader data."
200                          " Must have length of 24 characters", strlen(leader));
201         return -1;
202     }
203     yaz_marc_set_leader(mt, leader,
204                         indicator_length,
205                         &identifier_length,
206                         &base_address,
207                         &length_data_entry,
208                         &length_starting,
209                         &length_implementation);
210     *ptr_p = ptr;
211     return 0;
212 }
213
214 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr,
215                                     int indicator_length)
216 {
217     for(; ptr; ptr = ptr->next)
218         if (ptr->type == XML_ELEMENT_NODE)
219         {
220             if (!strcmp( (const char *) ptr->name, "controlfield"))
221             {
222                 const xmlNode *ptr_tag = 0;
223                 struct _xmlAttr *attr;
224                 for (attr = ptr->properties; attr; attr = attr->next)
225                     if (!strcmp((const char *)attr->name, "tag"))
226                         ptr_tag = attr->children;
227                     else
228                     {
229                         yaz_marc_cprintf(
230                             mt, "Bad attribute '%.80s' for 'controlfield'",
231                             attr->name);
232                         return -1;
233                     }
234                 if (!ptr_tag)
235                 {
236                     yaz_marc_cprintf(
237                         mt, "Missing attribute 'tag' for 'controlfield'" );
238                     return -1;
239                 }
240                 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
241             }
242             else if (!strcmp((const char *) ptr->name, "datafield"))
243             {
244                 char indstr[11]; /* 0(unused), 1,....9, + zero term */
245                 const xmlNode *ptr_tag = 0;
246                 struct _xmlAttr *attr;
247                 int i;
248                 for (i = 0; i < indicator_length; i++)
249                     indstr[i] = ' ';
250                 indstr[i] = '\0';
251                 for (attr = ptr->properties; attr; attr = attr->next)
252                     if (!strcmp((const char *)attr->name, "tag"))
253                         ptr_tag = attr->children;
254                     else if (strlen((const char *)attr->name) == 4 &&
255                              !memcmp(attr->name, "ind", 3))
256                     {
257                         int no = atoi((const char *)attr->name + 3);
258                         if (attr->children &&
259                             attr->children->type == XML_TEXT_NODE &&
260                             no <= indicator_length && no > 0 &&
261                             attr->children->content[0])
262                         {
263                             indstr[no - 1] = attr->children->content[0];
264                         }
265                         else
266                         {
267                             yaz_marc_cprintf(
268                                 mt, "Bad attribute '%.80s' for 'datafield'",
269                                 attr->name);
270                         }
271                     }
272                     else
273                     {
274                         yaz_marc_cprintf(
275                             mt, "Bad attribute '%.80s' for 'datafield'",
276                             attr->name);
277                     }
278                 if (!ptr_tag)
279                 {
280                     yaz_marc_cprintf(
281                         mt, "Missing attribute 'tag' for 'datafield'" );
282                     return -1;
283                 }
284                 yaz_marc_add_datafield_xml(mt, ptr_tag,
285                                            indstr, indicator_length);
286                 if (yaz_marc_read_xml_subfields(mt, ptr->children))
287                     return -1;
288             }
289             else
290             {
291                 yaz_marc_cprintf(mt,
292                                  "Expected element controlfield or datafield,"
293                                  " got %.80s", ptr->name);
294                 return -1;
295             }
296         }
297     return 0;
298 }
299
300
301 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr,
302                                           int indicator_length)
303 {
304     for(; ptr; ptr = ptr->next)
305         if (ptr->type == XML_ELEMENT_NODE)
306         {
307             if (!strncmp( (const char *) ptr->name, "c", 1))
308             {
309                 NMEM nmem = yaz_marc_get_nmem(mt);
310                 char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
311                 if (!tag_value)
312                 {
313                     yaz_marc_cprintf(
314                         mt, "Missing attribute 'tag' for 'controlfield'" );
315                     return -1;
316                 }
317                 yaz_marc_add_controlfield_xml2(mt, tag_value, ptr->children);
318             }
319             else if (!strncmp((const char *) ptr->name, "d",1))
320             {
321                 struct _xmlAttr *attr;
322                 NMEM nmem = yaz_marc_get_nmem(mt);
323                 char *tag_value;
324                 char *indstr = nmem_malloc(nmem, indicator_length + 1);
325                 int i = 0;
326                 for (i = 0; i < indicator_length; i++)
327                     indstr[i] = ' ';
328                 indstr[i] = '\0';
329                 tag_value = element_attribute_value_extract(ptr, "tag", nmem);
330                 if (!tag_value)
331                 {
332                     yaz_marc_cprintf(
333                         mt, "Missing attribute 'tag' for 'datafield'" );
334                     return -1;
335                 }
336                 for (attr = ptr->properties; attr; attr = attr->next)
337                     if (strlen((const char *)attr->name) == 2 &&
338                         attr->name[0] == 'i')
339                     {
340                         //extract indicator attribute from i#="Y" pattern
341                         int no = atoi((const char *)attr->name + 1);
342                         if (attr->children &&
343                             attr->children->type == XML_TEXT_NODE &&
344                             no <= indicator_length && no > 0 &&
345                             attr->children->content[0])
346                         {
347                             indstr[no - 1] = attr->children->content[0];
348                         }
349                         else
350                         {
351                             yaz_marc_cprintf(
352                                 mt, "Bad attribute '%.80s' for 'd'",attr->name);
353                         }
354                     }
355                     else
356                     {
357                         yaz_marc_cprintf(
358                             mt, "Bad attribute '%.80s' for 'd'", attr->name);
359                     }
360                 yaz_marc_add_datafield_xml2(mt, tag_value, indstr);
361                 if (yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */))
362                     return -1;
363             }
364             else
365             {
366                 yaz_marc_cprintf(mt,
367                                  "Expected element controlfield or datafield,"
368                                  " got %.80s", ptr->name);
369                 return -1;
370             }
371         }
372     return 0;
373 }
374
375
376 #endif
377
378 #if YAZ_HAVE_XML2
379 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
380 {
381     int indicator_length = 0;
382     int format = 0;
383     yaz_marc_reset(mt);
384
385     for(; ptr; ptr = ptr->next)
386         if (ptr->type == XML_ELEMENT_NODE)
387         {
388             if (!strcmp((const char *) ptr->name, "record"))
389             {
390                 format = YAZ_MARC_MARCXML;
391                 break;
392             }
393             else if (!strcmp((const char *) ptr->name, "r"))
394             {
395                 format = YAZ_MARC_TURBOMARC;
396                 break;
397             }
398             else
399             {
400                 yaz_marc_cprintf(
401                     mt, "Unknown element '%.80s' in MARC XML reader",
402                     ptr->name);
403                 return -1;
404             }
405         }
406     if (!ptr)
407     {
408         yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
409         return -1;
410     }
411     /* ptr points to record node now */
412     ptr = ptr->children;
413     if (yaz_marc_read_xml_leader(mt, &ptr, &indicator_length))
414         return -1;
415
416     switch (format)
417     {
418     case YAZ_MARC_MARCXML:
419         return yaz_marc_read_xml_fields(mt, ptr, indicator_length);
420     case YAZ_MARC_TURBOMARC:
421         return yaz_marc_read_turbo_xml_fields(mt, ptr, indicator_length);
422     }
423     return -1;
424 }
425 #endif
426
427
428 /*
429  * Local variables:
430  * c-basic-offset: 4
431  * c-file-style: "Stroustrup"
432  * indent-tabs-mode: nil
433  * End:
434  * vim: shiftwidth=4 tabstop=8 expandtab
435  */
436