Avoid using non-portable debian.h
[yaz-moved-to-github.git] / src / zoom-sru.c
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) Index Data
3  * See the file LICENSE for details.
4  */
5 /**
6  * \file zoom-sru.c
7  * \brief Implements ZOOM SRU
8  */
9 #if HAVE_CONFIG_H
10 #include <config.h>
11 #endif
12
13 #include <assert.h>
14 #include <string.h>
15 #include <errno.h>
16 #include "zoom-p.h"
17
18 #include <yaz/log.h>
19 #include <yaz/pquery.h>
20
21 #if YAZ_HAVE_XML2
22 static void set_SRU_error(ZOOM_connection c, Z_SRW_diagnostic *d)
23 {
24     const char *uri = d->uri;
25     if (uri)
26         ZOOM_set_dset_error(c, ZOOM_uri_to_code(uri), uri, d->details, 0);
27 }
28 #endif
29
30
31 #if YAZ_HAVE_XML2
32 static zoom_ret send_srw(ZOOM_connection c, Z_SRW_PDU *sr)
33 {
34     Z_GDU *gdu;
35     const char *database =  ZOOM_options_get(c->options, "databaseName");
36
37     gdu = z_get_HTTP_Request_uri(c->odr_out, c->host_port,
38                                  database, c->proxy_mode);
39
40     if (c->sru_mode == zoom_sru_get)
41     {
42         yaz_sru_get_encode(gdu->u.HTTP_Request, sr, c->odr_out, c->charset);
43     }
44     else if (c->sru_mode == zoom_sru_post)
45     {
46         yaz_sru_post_encode(gdu->u.HTTP_Request, sr, c->odr_out, c->charset);
47     }
48     else if (c->sru_mode == zoom_sru_soap)
49     {
50         yaz_sru_soap_encode(gdu->u.HTTP_Request, sr, c->odr_out, c->charset);
51     }
52     else if (c->sru_mode == zoom_sru_solr)
53     {
54         yaz_solr_encode_request(gdu->u.HTTP_Request, sr, c->odr_out, c->charset);
55     }
56     return ZOOM_send_GDU(c, gdu);
57 }
58 #endif
59
60 #if YAZ_HAVE_XML2
61 static Z_SRW_PDU *ZOOM_srw_get_pdu(ZOOM_connection c, int type)
62 {
63     Z_SRW_PDU *sr = yaz_srw_get_pdu(c->odr_out, type, c->sru_version);
64     if (c->url_authentication && c->user)
65     {
66         Z_SRW_extra_arg **ea = &sr->extra_args;
67         while (*ea)
68             ea = &(*ea)->next;
69         *ea = (Z_SRW_extra_arg *) odr_malloc(c->odr_out, sizeof(**ea));
70         (*ea)->name = "x-username";
71         (*ea)->value = c->user;
72         ea = &(*ea)->next;
73         if (c->password)
74         {
75             *ea = (Z_SRW_extra_arg *) odr_malloc(c->odr_out, sizeof(**ea));
76             (*ea)->name = "x-password";
77             (*ea)->value = c->password;
78             ea = &(*ea)->next;
79         }
80         *ea = 0;
81     }
82     else
83     {
84         sr->username = c->user;
85         sr->password = c->password;
86     }
87     return sr;
88 }
89 #endif
90
91 #if YAZ_HAVE_XML2
92 zoom_ret ZOOM_connection_srw_send_scan(ZOOM_connection c)
93 {
94     ZOOM_scanset scan;
95     Z_SRW_PDU *sr = 0;
96     const char *option_val = 0;
97     Z_Query *z_query;
98
99     if (!c->tasks)
100         return zoom_complete;
101     assert (c->tasks->which == ZOOM_TASK_SCAN);
102     scan = c->tasks->u.scan.scan;
103
104     sr = ZOOM_srw_get_pdu(c, Z_SRW_scan_request);
105
106     z_query = ZOOM_query_get_Z_Query(scan->query);
107     /* SRU scan can only carry CQL and PQF */
108     if (z_query->which == Z_Query_type_104)
109     {
110         sr->u.scan_request->queryType = "cql";
111         sr->u.scan_request->scanClause =
112             odr_strdup(c->odr_out, ZOOM_query_get_query_string(scan->query));
113     }
114     else if (z_query->which == Z_Query_type_1
115              || z_query->which == Z_Query_type_101)
116     {
117         sr->u.scan_request->queryType = "pqf";
118         sr->u.scan_request->scanClause =
119             odr_strdup(c->odr_out, ZOOM_query_get_query_string(scan->query));
120     }
121     else
122     {
123         ZOOM_set_error(c, ZOOM_ERROR_UNSUPPORTED_QUERY, 0);
124         return zoom_complete;
125     }
126
127     sr->u.scan_request->maximumTerms = odr_intdup(
128         c->odr_out, ZOOM_options_get_int(scan->options, "number", 10));
129
130     sr->u.scan_request->responsePosition = odr_intdup(
131         c->odr_out, ZOOM_options_get_int(scan->options, "position", 1));
132
133     option_val = ZOOM_options_get(scan->options, "extraArgs");
134     yaz_encode_sru_extra(sr, c->odr_out, option_val);
135     return send_srw(c, sr);
136 }
137 #else
138 zoom_ret ZOOM_connection_srw_send_scan(ZOOM_connection c)
139 {
140     return zoom_complete;
141 }
142 #endif
143
144 #if YAZ_HAVE_XML2
145 zoom_ret ZOOM_connection_srw_send_search(ZOOM_connection c)
146 {
147     int i;
148     int *start, *count;
149     ZOOM_resultset resultset = 0;
150     Z_SRW_PDU *sr = 0;
151     const char *option_val = 0;
152     const char *schema = 0;
153     Z_Query *z_query;
154     Z_FacetList *facet_list = 0;
155
156     if (c->error)                  /* don't continue on error */
157         return zoom_complete;
158     assert(c->tasks);
159     if (c->tasks->which != ZOOM_TASK_SEARCH)
160         return zoom_complete;
161
162     resultset = c->tasks->u.search.resultset;
163
164     ZOOM_memcached_search(c, resultset);
165
166     if (!resultset->setname)
167         resultset->setname = odr_strdup(resultset->odr, "default");
168     ZOOM_options_set(resultset->options, "setname", resultset->setname);
169     start = &c->tasks->u.search.start;
170     count = &c->tasks->u.search.count;
171     if (resultset->req_facets)
172         facet_list = yaz_pqf_parse_facet_list(c->odr_out,
173                                               resultset->req_facets);
174     schema = c->tasks->u.search.schema;
175
176     if (resultset->live_set)
177     {
178         if (*start >= resultset->size)
179             return zoom_complete;
180         if (*start + *count > resultset->size)
181             *count = resultset->size - *start;
182     }
183     for (i = 0; i < *count; i++)
184     {
185         ZOOM_record rec =
186             ZOOM_record_cache_lookup(resultset, i + *start,
187                                      c->tasks->u.search.syntax,
188                                      c->tasks->u.search.elementSetName,
189                                      schema);
190         if (!rec)
191             break;
192     }
193     *start += i;
194     *count -= i;
195
196     if (*count == 0 && resultset->live_set)
197         return zoom_complete;
198
199     assert(resultset->query);
200
201     sr = ZOOM_srw_get_pdu(c, Z_SRW_searchRetrieve_request);
202     z_query = ZOOM_query_get_Z_Query(resultset->query);
203
204     if (z_query->which == Z_Query_type_104
205         && z_query->u.type_104->which == Z_External_CQL)
206     {
207         sr->u.request->queryType = "cql";
208         sr->u.request->query = z_query->u.type_104->u.cql;
209     }
210     else if (z_query->which == Z_Query_type_1 && z_query->u.type_1)
211     {
212         sr->u.request->queryType = "pqf";
213         sr->u.request->query =
214             odr_strdup(c->odr_out,
215                        ZOOM_query_get_query_string(resultset->query));
216     }
217     else
218     {
219         ZOOM_set_error(c, ZOOM_ERROR_UNSUPPORTED_QUERY, 0);
220         return zoom_complete;
221     }
222
223     option_val = ZOOM_query_get_sru11(resultset->query);
224     if (option_val)
225     {
226         sr->u.request->sort_type = Z_SRW_sort_type_sort;
227         sr->u.request->sort.sortKeys = odr_strdup(c->odr_out, option_val);
228     }
229     sr->u.request->startRecord = odr_intdup(c->odr_out, *start + 1);
230     sr->u.request->maximumRecords = odr_intdup(
231         c->odr_out, (resultset->step > 0 && resultset->step < *count) ?
232         resultset->step : *count);
233     sr->u.request->recordSchema = odr_strdup_null(c->odr_out, schema);
234     sr->u.request->facetList = facet_list;
235
236     option_val = ZOOM_resultset_option_get(resultset, "recordPacking");
237     if (option_val)
238         sr->u.request->recordPacking = odr_strdup(c->odr_out, option_val);
239
240     option_val = ZOOM_resultset_option_get(resultset, "extraArgs");
241     yaz_encode_sru_extra(sr, c->odr_out, option_val);
242     return send_srw(c, sr);
243 }
244 #else
245 zoom_ret ZOOM_connection_srw_send_search(ZOOM_connection c)
246 {
247     return zoom_complete;
248 }
249 #endif
250
251 #if YAZ_HAVE_XML2
252 static zoom_ret handle_srw_response(ZOOM_connection c,
253                                     Z_SRW_searchRetrieveResponse *res)
254 {
255     ZOOM_resultset resultset = 0;
256     int *start, *count;
257     int i;
258     NMEM nmem;
259     ZOOM_Event event;
260     const char *syntax, *elementSetName, *schema;
261
262     if (!c->tasks)
263         return zoom_complete;
264
265     if (c->tasks->which != ZOOM_TASK_SEARCH)
266         return zoom_complete;
267
268     resultset = c->tasks->u.search.resultset;
269     start = &c->tasks->u.search.start;
270     count = &c->tasks->u.search.count;
271     syntax = c->tasks->u.search.syntax;
272     elementSetName = c->tasks->u.search.elementSetName;
273     schema = c->tasks->u.search.schema;
274
275     if (resultset->live_set == 0)
276     {
277         event = ZOOM_Event_create(ZOOM_EVENT_RECV_SEARCH);
278         ZOOM_connection_put_event(c, event);
279     }
280     if (res->facetList)
281         ZOOM_handle_facet_list(resultset, res->facetList);
282
283     resultset->size = 0;
284
285     if (res->resultSetId)
286         ZOOM_resultset_option_set(resultset, "resultSetId", res->resultSetId);
287
288     yaz_log(c->log_details, "%p handle_srw_response got SRW response OK", c);
289
290     if (res->num_diagnostics > 0)
291     {
292         resultset->live_set = 2;
293         set_SRU_error(c, &res->diagnostics[0]);
294     }
295     else
296     {
297         if (res->numberOfRecords)
298         {
299             Z_OtherInformation *oi = 0;
300             if (res->facetList)
301             {
302                 ODR o = c->odr_in;
303                 Z_External *ext = (Z_External *)
304                     odr_malloc(o, sizeof(*ext));
305
306                 ext->which = Z_External_userFacets;
307                 ext->u.facetList = res->facetList;
308                 ext->direct_reference =
309                     odr_oiddup(o, yaz_oid_userinfo_facet_1);
310                 ext->indirect_reference = 0;
311                 ext->descriptor = 0;
312                 oi = (Z_OtherInformation *) odr_malloc(o, sizeof(*oi));
313                 oi->num_elements = 1;
314                 oi->list = (Z_OtherInformationUnit **)
315                     odr_malloc(o, sizeof(*oi->list));
316                 oi->list[0] = (Z_OtherInformationUnit *)
317                     odr_malloc(o, sizeof(**oi->list));
318                 oi->list[0]->category = 0;
319                 oi->list[0]->which = Z_OtherInfo_externallyDefinedInfo;
320                 oi->list[0]->information.externallyDefinedInfo = ext;
321             }
322             resultset->size = *res->numberOfRecords;
323             ZOOM_memcached_hitcount(c, resultset, oi,
324                                     res->resultCountPrecision ?
325                                     res->resultCountPrecision : "exact");
326         }
327         resultset->live_set = 2;
328         if (res->suggestions)
329             ZOOM_resultset_option_set(resultset, "suggestions",
330                                       res->suggestions);
331         for (i = 0; i < res->num_records; i++)
332         {
333             int pos = c->tasks->u.search.start + i;
334             Z_SRW_record *sru_rec;
335             Z_SRW_diagnostic *diag = 0;
336             int num_diag;
337
338             /* only trust recordPosition if >= calculated position */
339             if (res->records[i].recordPosition &&
340                 *res->records[i].recordPosition >= pos + 1)
341                 pos = *res->records[i].recordPosition - 1;
342
343             if (!ZOOM_record_cache_lookup(resultset,
344                                           pos,
345                                           syntax, elementSetName, schema))
346             {
347                 Z_NamePlusRecord *npr = (Z_NamePlusRecord *)
348                     odr_malloc(c->odr_in, sizeof(Z_NamePlusRecord));
349                 sru_rec = &res->records[i];
350
351                 npr->databaseName = 0;
352                 npr->which = Z_NamePlusRecord_databaseRecord;
353                 npr->u.databaseRecord = (Z_External *)
354                     odr_malloc(c->odr_in, sizeof(Z_External));
355                 npr->u.databaseRecord->descriptor = 0;
356                 npr->u.databaseRecord->direct_reference =
357                     odr_oiddup(c->odr_in, yaz_oid_recsyn_xml);
358                 npr->u.databaseRecord->indirect_reference = 0;
359                 npr->u.databaseRecord->which = Z_External_octet;
360
361                 npr->u.databaseRecord->u.octet_aligned =
362                     odr_create_Odr_oct(c->odr_in,
363                                        sru_rec->recordData_buf,
364                                    sru_rec->recordData_len);
365                 if (sru_rec->recordSchema
366                     && !strcmp(sru_rec->recordSchema,
367                                "info:srw/schema/1/diagnostics-v1.1"))
368                 {
369                     sru_decode_surrogate_diagnostics(sru_rec->recordData_buf,
370                                                      sru_rec->recordData_len,
371                                                      &diag, &num_diag,
372                                                      resultset->odr);
373                 }
374                 ZOOM_record_cache_add(resultset, npr,
375                                       pos, syntax, elementSetName,
376                                       schema, diag);
377             }
378         }
379         *count -= i;
380         if (*count < 0)
381             *count = 0;
382         *start += i;
383         nmem = odr_extract_mem(c->odr_in);
384         nmem_transfer(odr_getmem(resultset->odr), nmem);
385         nmem_destroy(nmem);
386
387         return ZOOM_connection_srw_send_search(c);
388     }
389     return zoom_complete;
390 }
391 #endif
392
393 #if YAZ_HAVE_XML2
394 static zoom_ret handle_srw_scan_response(ZOOM_connection c,
395                                          Z_SRW_scanResponse *res)
396 {
397     NMEM nmem = odr_extract_mem(c->odr_in);
398     ZOOM_scanset scan;
399
400     if (!c->tasks || c->tasks->which != ZOOM_TASK_SCAN)
401         return zoom_complete;
402     scan = c->tasks->u.scan.scan;
403
404     if (res->num_diagnostics > 0)
405         set_SRU_error(c, &res->diagnostics[0]);
406
407     scan->scan_response = 0;
408     scan->srw_scan_response = res;
409     nmem_transfer(odr_getmem(scan->odr), nmem);
410
411     ZOOM_options_set_int(scan->options, "number", res->num_terms);
412     nmem_destroy(nmem);
413     return zoom_complete;
414 }
415 #endif
416
417 int ZOOM_handle_sru(ZOOM_connection c, Z_HTTP_Response *hres,
418                     zoom_ret *cret, char **addinfo)
419 {
420 #if YAZ_HAVE_XML2
421     int ret = 0;
422
423     /* not redirect (normal response) */
424     if (!yaz_srw_check_content_type(hres))
425     {
426         *addinfo = "content-type";
427         ret = -1;
428     }
429     else if (c->sru_mode == zoom_sru_solr)
430     {
431         Z_SRW_PDU *sr;
432         ret = yaz_solr_decode_response(c->odr_in, hres, &sr);
433         if (ret == 0)
434         {
435             if (sr->which == Z_SRW_searchRetrieve_response)
436                 *cret = handle_srw_response(c, sr->u.response);
437             else if (sr->which == Z_SRW_scan_response)
438                 *cret = handle_srw_scan_response(c, sr->u.scan_response);
439         }
440     }
441     else
442     {
443         Z_SOAP *soap_package = 0;
444         ODR o = c->odr_in;
445         Z_SOAP_Handler soap_handlers[4] = {
446             {YAZ_XMLNS_SRU_v1_response, 0, (Z_SOAP_fun) yaz_srw_codec},
447             {YAZ_XMLNS_SRU_v2_mask, 0, (Z_SOAP_fun) yaz_srw_codec},
448             {"searchRetrieveResponse", 0, (Z_SOAP_fun) yaz_srw_codec},
449             {0, 0, 0}
450         };
451         ret = z_soap_codec(o, &soap_package,
452                            &hres->content_buf, &hres->content_len,
453                            soap_handlers);
454         if (!ret && soap_package->which == Z_SOAP_generic)
455         {
456             Z_SRW_PDU *sr = (Z_SRW_PDU*) soap_package->u.generic->p;
457
458             ZOOM_options_set(c->options, "sru_version", sr->srw_version);
459             ZOOM_options_setl(c->options, "sru_extra_response_data",
460                               sr->extraResponseData_buf, sr->extraResponseData_len);
461             if (sr->which == Z_SRW_searchRetrieve_response)
462                 *cret = handle_srw_response(c, sr->u.response);
463             else if (sr->which == Z_SRW_scan_response)
464                 *cret = handle_srw_scan_response(c, sr->u.scan_response);
465             else
466                 ret = -1;
467         }
468         else if (!ret && (soap_package->which == Z_SOAP_fault
469                           || soap_package->which == Z_SOAP_error))
470         {
471             ZOOM_set_HTTP_error(c, hres->code,
472                                 soap_package->u.fault->fault_code,
473                                 soap_package->u.fault->fault_string);
474         }
475         else
476         {
477             size_t max_chars = 1000;
478             size_t sz = hres->content_len;
479             if (sz > max_chars - 1)
480                 sz = max_chars;
481             *addinfo = odr_malloc(c->odr_in, sz + 4);
482             memcpy(*addinfo, hres->content_buf, sz);
483             if (sz == max_chars)
484                 strcpy(*addinfo + sz, "...");
485             else
486                 strcpy(*addinfo + sz, "");
487             ret = -1;
488         }
489     }
490     return ret;
491 #else
492     return -1;
493 #endif
494 }
495
496 /*
497  * Local variables:
498  * c-basic-offset: 4
499  * c-file-style: "Stroustrup"
500  * indent-tabs-mode: nil
501  * End:
502  * vim: shiftwidth=4 tabstop=8 expandtab
503  */
504