65c1098e51883e74192cb8f9de269f61757be35a
[yaz-moved-to-github.git] / src / retrieval.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  * \file retrieval.c
7  * \brief Retrieval utility
8  */
9
10 #if HAVE_CONFIG_H
11 #include <config.h>
12 #endif
13
14 #include <string.h>
15 #include <yaz/retrieval.h>
16 #include <yaz/wrbuf.h>
17 #include <yaz/xmalloc.h>
18 #include <yaz/nmem.h>
19 #include <yaz/tpath.h>
20 #include <yaz/match_glob.h>
21 #include <yaz/proto.h>
22 #include <yaz/oid_db.h>
23
24 #if YAZ_HAVE_XML2
25 #include <libxml/parser.h>
26 #include <libxml/tree.h>
27 #include <libxml/xinclude.h>
28
29 /** \brief The internal structure for yaz_retrieval_t */
30 struct yaz_retrieval_struct {
31     /** \brief ODR memory for configuration */
32     ODR odr;
33
34     /** \brief odr's NMEM memory (odr->mem) */
35     NMEM nmem;
36
37     /** \brief string buffer for error messages */
38     WRBUF wr_error;
39
40     /** \brief path for opening files  */
41     char *path;
42
43     /** \brief retrieval list */
44     struct yaz_retrieval_elem *list;
45
46     /** \brief last pointer in retrieval list */
47     struct yaz_retrieval_elem **list_p;
48 };
49
50 /** \brief information per 'retrieval' construct */
51 struct yaz_retrieval_elem {
52     /** \brief schema identifier */
53     const char *identifier;
54     /** \brief schema name , short-hand such as "dc" */
55     const char *name;
56     /** \brief record syntax */
57     Odr_oid *syntax;
58
59     /** \brief backend name */
60     const char *backend_name;
61     /** \brief backend syntax */
62     Odr_oid *backend_syntax;
63
64     /** \brief record conversion */
65     yaz_record_conv_t record_conv;
66
67     /** \brief next element in list */
68     struct yaz_retrieval_elem *next;
69 };
70
71 static void yaz_retrieval_reset(yaz_retrieval_t p);
72
73 yaz_retrieval_t yaz_retrieval_create()
74 {
75     yaz_retrieval_t p = (yaz_retrieval_t) xmalloc(sizeof(*p));
76     p->odr = odr_createmem(ODR_ENCODE);
77     p->nmem = odr_getmem(p->odr);
78     p->wr_error = wrbuf_alloc();
79     p->list = 0;
80     p->path = 0;
81     yaz_retrieval_reset(p);
82     return p;
83 }
84
85 void yaz_retrieval_destroy(yaz_retrieval_t p)
86 {
87     if (p)
88     {
89         yaz_retrieval_reset(p);
90         odr_destroy(p->odr);
91         wrbuf_destroy(p->wr_error);
92         xfree(p->path);
93         xfree(p);
94     }
95 }
96
97 void yaz_retrieval_reset(yaz_retrieval_t p)
98 {
99     struct yaz_retrieval_elem *el = p->list;
100     for(; el; el = el->next)
101         yaz_record_conv_destroy(el->record_conv);
102
103     wrbuf_rewind(p->wr_error);
104     odr_reset(p->odr);
105
106     p->list = 0;
107     p->list_p = &p->list;
108 }
109
110 /** \brief parse retrieval XML config */
111 static int conf_retrieval(yaz_retrieval_t p, const xmlNode *ptr)
112 {
113
114     struct _xmlAttr *attr;
115     struct yaz_retrieval_elem *el = (struct yaz_retrieval_elem *)
116         nmem_malloc(p->nmem, sizeof(*el));
117
118     el->syntax = 0;
119     el->identifier = 0;
120     el->name = 0;
121     el->backend_name = 0;
122     el->backend_syntax = 0;
123
124     el->next = 0;
125
126     for (attr = ptr->properties; attr; attr = attr->next)
127     {
128         if (!xmlStrcmp(attr->name, BAD_CAST "syntax") &&
129             attr->children && attr->children->type == XML_TEXT_NODE)
130         {
131             el->syntax = yaz_string_to_oid_odr(
132                 yaz_oid_std(),
133                 CLASS_RECSYN,
134                 (const char *) attr->children->content,
135                 p->odr);
136             if (!el->syntax)
137             {
138                 wrbuf_printf(p->wr_error, "Element <retrieval>: "
139                              " unknown attribute value syntax='%s'",
140                              (const char *) attr->children->content);
141                 return -1;
142             }
143         }
144         else if (!xmlStrcmp(attr->name, BAD_CAST "identifier") &&
145                  attr->children && attr->children->type == XML_TEXT_NODE)
146             el->identifier =
147                 nmem_strdup(p->nmem, (const char *) attr->children->content);
148         else if (!xmlStrcmp(attr->name, BAD_CAST "name") &&
149                  attr->children && attr->children->type == XML_TEXT_NODE)
150             el->name = 
151                 nmem_strdup(p->nmem, (const char *) attr->children->content);
152         else
153         {
154             wrbuf_printf(p->wr_error, "Element <retrieval>: "
155                          " expected attributes 'syntax', identifier' or "
156                          "'name', got '%s'", attr->name);
157             return -1;
158         }
159     }
160
161     if (!el->syntax)
162     {
163         wrbuf_printf(p->wr_error, "Missing 'syntax' attribute");
164         return -1;
165     }
166
167     /* parsing backend element */
168
169     el->record_conv = 0; /* OK to have no 'backend' sub content */
170     for (ptr = ptr->children; ptr; ptr = ptr->next)
171     {
172         if (ptr->type != XML_ELEMENT_NODE)
173             continue;
174         if (strcmp((const char *) ptr->name, "backend")){
175             wrbuf_printf(p->wr_error, "Element <retrieval>: expected"
176                          " zero or one element <backend>, got <%s>",
177                          (const char *) ptr->name);
178             return -1;
179         }
180
181         else {
182
183             if (el->record_conv)
184             {
185                 wrbuf_printf(p->wr_error, "Element <retrieval>: "
186                              "only one <backend> allowed");
187                 yaz_record_conv_destroy(el->record_conv);
188                 return -1;
189             }
190
191             /* parsing attributees */
192             struct _xmlAttr *attr;
193             for (attr = ptr->properties; attr; attr = attr->next){
194             
195                 if (!xmlStrcmp(attr->name, BAD_CAST "name") 
196                          && attr->children 
197                          && attr->children->type == XML_TEXT_NODE)
198                     el->backend_name 
199                         = nmem_strdup(p->nmem, 
200                                       (const char *) attr->children->content);
201
202                 else if (!xmlStrcmp(attr->name, BAD_CAST "syntax") 
203                          && attr->children 
204                          && attr->children->type == XML_TEXT_NODE){
205                     el->backend_syntax 
206                         = yaz_string_to_oid_odr(
207                             yaz_oid_std(),
208                             CLASS_RECSYN,
209                             (const char *) attr->children->content,
210                             p->odr);
211                     if (!el->backend_syntax){
212                         wrbuf_printf(p->wr_error, 
213                                      "Element <backend syntax='%s'>: "
214                                      "attribute 'syntax' has invalid "
215                                      "value '%s'", 
216                                      attr->children->content,
217                                      attr->children->content);
218                         return -1;
219                     } 
220                 }
221                 else {
222                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
223                                  "attributes 'syntax' or 'name, got '%s'", 
224                                  attr->name);
225                     return -1;
226                 }
227             }
228           
229  
230             /* parsing internal of record conv */
231             el->record_conv = yaz_record_conv_create();
232             
233             yaz_record_conv_set_path(el->record_conv, p->path);
234
235         
236             if (yaz_record_conv_configure(el->record_conv, ptr))
237             {
238                 wrbuf_printf(p->wr_error, "%s",
239                              yaz_record_conv_get_error(el->record_conv));
240                 yaz_record_conv_destroy(el->record_conv);
241                 return -1;
242             }
243         }
244     }
245     
246     *p->list_p = el;
247     p->list_p = &el->next;
248     return 0;
249 }
250
251 int yaz_retrieval_configure(yaz_retrieval_t p, const xmlNode *ptr)
252 {
253     yaz_retrieval_reset(p);
254
255     if (ptr && ptr->type == XML_ELEMENT_NODE &&
256         !strcmp((const char *) ptr->name, "retrievalinfo"))
257     {
258         for (ptr = ptr->children; ptr; ptr = ptr->next)
259         {
260             if (ptr->type != XML_ELEMENT_NODE)
261                 continue;
262             if (!strcmp((const char *) ptr->name, "retrieval"))
263             {
264                 if (conf_retrieval(p, ptr))
265                     return -1;
266             }
267             else
268             {
269                 wrbuf_printf(p->wr_error, "Element <retrievalinfo>: "
270                              "expected element <retrieval>, got <%s>", 
271                              ptr->name);
272                 return -1;
273             }
274         }
275     }
276     else
277     {
278         wrbuf_printf(p->wr_error, "Expected element <retrievalinfo>");
279         return -1;
280     }
281     return 0;
282 }
283
284 int yaz_retrieval_request(yaz_retrieval_t p,
285                           const char *schema, Odr_oid *syntax,
286                           const char **match_schema, Odr_oid **match_syntax,
287                           yaz_record_conv_t *rc,
288                           const char **backend_schema,
289                           Odr_oid **backend_syntax)
290 {
291     struct yaz_retrieval_elem *el = p->list;
292     int syntax_matches = 0;
293     int schema_matches = 0;
294
295     wrbuf_rewind(p->wr_error);
296     if (!el)
297         return 0;
298     for(; el; el = el->next)
299     {
300         int schema_ok = 0;
301         int syntax_ok = 0;
302
303         if (!schema)
304             schema_ok = 1;
305         else
306         {
307             if (el->name && yaz_match_glob(el->name, schema))
308                 schema_ok = 1;
309             if (el->identifier && !strcmp(schema, el->identifier))
310                 schema_ok = 1;
311             if (!el->name && !el->identifier)
312                 schema_ok = 1;
313         }
314         
315         if (syntax && el->syntax && !oid_oidcmp(syntax, el->syntax))
316             syntax_ok = 1;
317         if (!syntax)
318             syntax_ok = 1;
319
320         if (syntax_ok)
321             syntax_matches++;
322         if (schema_ok)
323             schema_matches++;
324         if (syntax_ok && schema_ok)
325         {
326             *match_syntax = el->syntax;
327             if (el->identifier)
328                 *match_schema = el->identifier;
329             else
330                 *match_schema = 0;
331             if (backend_schema)
332             {
333                 if (el->backend_name)
334                     *backend_schema = el->backend_name;
335                 else if (el->name)
336                     *backend_schema = el->name;                    
337                 else
338                     *backend_schema = schema;
339             }
340             if (backend_syntax)
341             {
342                 if (el->backend_syntax)
343                     *backend_syntax = el->backend_syntax;
344                 else
345                     *backend_syntax = el->syntax;
346             }
347             if (rc)
348                 *rc = el->record_conv;
349             return 0;
350         }
351     }
352     if (!syntax_matches && syntax)
353     {
354         char buf[OID_STR_MAX];
355         wrbuf_printf(p->wr_error, "%s", oid_oid_to_dotstring(syntax, buf));
356         return 2;
357     }
358     if (schema)
359         wrbuf_printf(p->wr_error, "%s", schema);
360     if (!schema_matches)
361         return 1;
362     return 3;
363 }
364
365 const char *yaz_retrieval_get_error(yaz_retrieval_t p)
366 {
367     return wrbuf_cstr(p->wr_error);
368 }
369
370 void yaz_retrieval_set_path(yaz_retrieval_t p, const char *path)
371 {
372     xfree(p->path);
373     p->path = 0;
374     if (path)
375         p->path = xstrdup(path);
376 }
377
378 #endif
379
380 /*
381  * Local variables:
382  * c-basic-offset: 4
383  * c-file-style: "Stroustrup"
384  * indent-tabs-mode: nil
385  * End:
386  * vim: shiftwidth=4 tabstop=8 expandtab
387  */
388