src/solr.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  * \file solr.c
   7  * \brief Implements Solr decoding/encoding
   8  */
   9 #if HAVE_CONFIG_H
  10 #include <config.h>
  11 #endif
  12
  13 #include <stdlib.h>
  14 #include <assert.h>
  15 #include <yaz/srw.h>
  16 #include <yaz/matchstr.h>
  17 #include <yaz/yaz-iconv.h>
  18 #include <yaz/log.h>
  19 #include <yaz/facet.h>
  20 #include <yaz/wrbuf.h>
  21
  22 #include "sru-p.h"
  23
  24 #define SOLR_MAX_PARAMETERS  100
  25
  26 #if YAZ_HAVE_XML2
  27 #include <libxml/parser.h>
  28 #include <libxml/tree.h>
  29
  30 static void extract_text_node(xmlNodePtr node, WRBUF wrbuf) {
  31     xmlNodePtr child;
  32     for (child = node->children; child ; child = child->next)
  33     {
  34         if (child->type == XML_TEXT_NODE)
  35             wrbuf_puts(wrbuf, (const char *) child->content);
  36     }
  37 }
  38
  39 static int match_xml_node_attribute(
  40     xmlNodePtr ptr,
  41     const char *node_name, const char *attribute_name, const char *value)
  42 {
  43     const char *attribute_value;
  44     // check if the node name matches
  45     if (strcmp((const char*) ptr->name, node_name))
  46         return 0;
  47     if (attribute_name)
  48     {
  49         attribute_value = yaz_element_attribute_value_get(ptr, node_name,
  50                                                           attribute_name);
  51         if (attribute_value && !strcmp(attribute_value, value))
  52             return 1;
  53     }
  54     else /* No attribute to check */
  55         return 1;
  56     return 0;
  57 }
  58
  59 static void yaz_solr_decode_result_docs(ODR o, xmlNodePtr ptr,
  60                                         Odr_int start,
  61                                         Z_SRW_searchRetrieveResponse *sr)
  62 {
  63     xmlNodePtr node;
  64     int offset = 0;
  65     int i = 0;
  66
  67     sr->num_records = 0;
  68     for (node = ptr->children; node; node = node->next)
  69         if (node->type == XML_ELEMENT_NODE)
  70             sr->num_records++;
  71
  72     if (sr->num_records)
  73         sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records);
  74
  75     for (node = ptr->children; node; node = node->next)
  76     {
  77         if (node->type == XML_ELEMENT_NODE)
  78         {
  79             Z_SRW_record *record = sr->records + i;
  80             xmlBufferPtr buf = xmlBufferCreate();
  81             xmlNode *tmp = xmlCopyNode(node, 1);
  82
  83             xmlNodeDump(buf, tmp->doc, tmp, 0, 0);
  84
  85             xmlFreeNode(tmp);
  86
  87             record->recordSchema = 0;
  88             record->recordPacking = Z_SRW_recordPacking_XML;
  89             record->recordData_len = buf->use;
  90             record->recordData_buf = odr_malloc(o, buf->use + 1);
  91             memcpy(record->recordData_buf, buf->content, buf->use);
  92             record->recordData_buf[buf->use] = '\0';
  93             record->recordPosition = odr_intdup(o, start + offset + 1);
  94
  95             xmlBufferFree(buf);
  96
  97             offset++;
  98             i++;
  99         }
 100     }
 101 }
 102
 103 static int yaz_solr_decode_result(ODR o, xmlNodePtr ptr,
 104                                   Z_SRW_searchRetrieveResponse *sr)
 105 {
 106     Odr_int start = 0;
 107     struct _xmlAttr *attr;
 108     for (attr = ptr->properties; attr; attr = attr->next)
 109         if (attr->children && attr->children->type == XML_TEXT_NODE)
 110         {
 111             if (!strcmp((const char *) attr->name, "numFound"))
 112             {
 113                 sr->numberOfRecords = odr_intdup(o, odr_atoi(
 114                         (const char *) attr->children->content));
 115             }
 116             else if (!strcmp((const char *) attr->name, "start"))
 117             {
 118                 start = odr_atoi((const char *) attr->children->content);
 119             }
 120         }
 121     if (sr->numberOfRecords && *sr->numberOfRecords > 0)
 122         yaz_solr_decode_result_docs(o, ptr, start, sr);
 123     if (sr->numberOfRecords)
 124         return 0;
 125     return -1;
 126 }
 127
 128 static const char *get_facet_term_count(xmlNodePtr node, Odr_int *freq)
 129 {
 130     const char *term = yaz_element_attribute_value_get(node, "int", "name");
 131     xmlNodePtr child;
 132     WRBUF wrbuf = wrbuf_alloc();
 133     if (!term)
 134         return term;
 135
 136     for (child = node->children; child ; child = child->next)
 137     {
 138         if (child->type == XML_TEXT_NODE)
 139             wrbuf_puts(wrbuf, (const char *) child->content);
 140     }
 141     *freq = odr_atoi(wrbuf_cstr(wrbuf));
 142     wrbuf_destroy(wrbuf);
 143     return term;
 144 }
 145
 146 Z_FacetField *yaz_solr_decode_facet_field(ODR o, xmlNodePtr ptr,
 147                                           Z_SRW_searchRetrieveResponse *sr)
 148
 149 {
 150     Z_AttributeList *list;
 151     Z_FacetField *facet_field;
 152     int num_terms = 0;
 153     int index = 0;
 154     xmlNodePtr node;
 155     // USE attribute
 156     const char* name = yaz_element_attribute_value_get(ptr, "lst", "name");
 157     list = yaz_use_attribute_create(o, name);
 158     for (node = ptr->children; node; node = node->next)
 159         num_terms++;
 160     facet_field = facet_field_create(o, list, num_terms);
 161     index = 0;
 162     for (node = ptr->children; node; node = node->next)
 163     {
 164         Odr_int count = 0;
 165         const char *term = get_facet_term_count(node, &count);
 166         facet_field_term_set(o, facet_field,
 167                              facet_term_create_cstr(o, term, count), index);
 168         index++;
 169     }
 170     return facet_field;
 171 }
 172
 173 static int yaz_solr_decode_facet_counts(ODR o, xmlNodePtr root,
 174                                         Z_SRW_searchRetrieveResponse *sr)
 175 {
 176     xmlNodePtr ptr;
 177     for (ptr = root->children; ptr; ptr = ptr->next)
 178     {
 179         if (match_xml_node_attribute(ptr, "lst", "name", "facet_fields"))
 180         {
 181             xmlNodePtr node;
 182             Z_FacetList *facet_list;
 183             int num_facets = 0;
 184             for (node = ptr->children; node; node= node->next)
 185             {
 186                 num_facets++;
 187             }
 188             facet_list = facet_list_create(o, num_facets);
 189             num_facets = 0;
 190             for (node = ptr->children; node; node= node->next)
 191             {
 192                 facet_list_field_set(o, facet_list,
 193                                      yaz_solr_decode_facet_field(o, node, sr),
 194                                      num_facets);
 195                 num_facets++;
 196             }
 197             sr->facetList = facet_list;
 198             break;
 199         }
 200     }
 201     return 0;
 202 }
 203
 204 static void yaz_solr_decode_suggestion_values(xmlNodePtr listPptr, WRBUF wrbuf)
 205 {
 206     xmlNodePtr node;
 207     for (node = listPptr; node; node= node->next) {
 208         if (!strcmp((char*) node->name, "lst")) {
 209             xmlNodePtr child;
 210             for (child = node->children; child; child= child->next) {
 211                 if (match_xml_node_attribute(child, "str", "name", "word")) {
 212                     wrbuf_puts(wrbuf, "<suggestion>");
 213                     extract_text_node(child, wrbuf);
 214                     wrbuf_puts(wrbuf, "</suggestion>\n");
 215                 }
 216             }
 217         }
 218     }
 219 }
 220
 221 static void yaz_solr_decode_suggestion_lst(xmlNodePtr lstPtr, WRBUF wrbuf)
 222 {
 223     xmlNodePtr node;
 224     for (node = lstPtr; node; node= node->next) {
 225         if (match_xml_node_attribute(node, "arr", "name", "suggestion")) {
 226             yaz_solr_decode_suggestion_values(node->children, wrbuf);
 227         }
 228     }
 229 }
 230
 231 static void yaz_solr_decode_misspelled(xmlNodePtr lstPtr, WRBUF wrbuf)
 232 {
 233     xmlNodePtr node;
 234     for (node = lstPtr; node; node= node->next)
 235     {
 236         if (!strcmp((const char*) node->name, "lst")) {
 237             const char *misspelled = yaz_element_attribute_value_get(node, "lst", "name");
 238             if (misspelled) {
 239                 wrbuf_printf(wrbuf, "<misspelled term=\"%s\">\n", misspelled);
 240                 yaz_solr_decode_suggestion_lst(node->children, wrbuf);
 241                 wrbuf_puts(wrbuf, "</misspelled>\n");
 242             }
 243         }
 244     }
 245 }
 246
 247 static int yaz_solr_decode_spellcheck(ODR o, xmlNodePtr spellcheckPtr, Z_SRW_searchRetrieveResponse *sr)
 248 {
 249     xmlNodePtr ptr;
 250     WRBUF wrbuf = wrbuf_alloc();
 251     wrbuf_puts(wrbuf, "");
 252     for (ptr = spellcheckPtr->children; ptr; ptr = ptr->next)
 253     {
 254         if (match_xml_node_attribute(ptr, "lst", "name", "suggestions"))
 255         {
 256             yaz_solr_decode_misspelled(ptr->children, wrbuf);
 257         }
 258     }
 259     sr->suggestions = odr_strdup(o, wrbuf_cstr(wrbuf));
 260     return 0;
 261 }
 262
 263 static int yaz_solr_decode_scan_result(ODR o, xmlNodePtr ptr,
 264                                        Z_SRW_scanResponse *scr)
 265 {
 266     xmlNodePtr node;
 267     char *pos;
 268     int i = 0;
 269
 270     /* find the actual list */
 271     for (node = ptr->children; node; node = node->next)
 272         if (node->type == XML_ELEMENT_NODE) {
 273             ptr = node;
 274             break;
 275         }
 276
 277     scr->num_terms = 0;
 278     for (node = ptr->children; node; node = node->next)
 279         if (node->type == XML_ELEMENT_NODE && !strcmp((const char *) node->name, "int"))
 280             scr->num_terms++;
 281
 282     if (scr->num_terms)
 283         scr->terms = odr_malloc(o, sizeof(*scr->terms) * scr->num_terms);
 284
 285     for (node = ptr->children; node; node = node->next)
 286     {
 287         if (node->type == XML_ELEMENT_NODE && !strcmp((const char *) node->name, "int"))
 288         {
 289             Z_SRW_scanTerm *term = scr->terms + i;
 290
 291             Odr_int count = 0;
 292             const char *val = get_facet_term_count(node, &count);
 293
 294             term->numberOfRecords = odr_intdup(o, count);
 295
 296             /* if val contains a ^ then it is probably term<^>display term so separate them. This is due to
 297              * SOLR not being able to encode them into 2 separate attributes.
 298              */
 299             pos = strchr(val, '^');
 300             if (pos != NULL) {
 301                 term->displayTerm = odr_strdup(o, pos + 1);
 302                 *pos = '\0';
 303                 term->value = odr_strdup(o, val);
 304                 *pos = '^';
 305             } else {
 306                 term->value = odr_strdup(o, val);
 307                 term->displayTerm = NULL;
 308             }
 309             term->whereInList = NULL;
 310
 311             i++;
 312         }
 313     }
 314
 315     if (scr->num_terms)
 316         return 0;
 317     return -1;
 318 }
 319 #endif
 320
 321 int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup)
 322 {
 323 #if YAZ_HAVE_XML2
 324     const char *content_buf = hres->content_buf;
 325     int content_len = hres->content_len;
 326     xmlDocPtr doc = xmlParseMemory(content_buf, content_len);
 327     int ret = 0;
 328     xmlNodePtr ptr = 0;
 329     Z_SRW_PDU *pdu;
 330     Z_SRW_searchRetrieveResponse *sr = NULL;
 331     Z_SRW_scanResponse *scr = NULL;
 332
 333     if (!doc)
 334     {
 335         ret = -1;
 336     }
 337     if (doc)
 338     {
 339         xmlNodePtr root = xmlDocGetRootElement(doc);
 340         if (!root)
 341         {
 342             ret = -1;
 343         }
 344         else if (strcmp((const char *) root->name, "response"))
 345         {
 346             ret = -1;
 347         }
 348         else
 349         {
 350             /** look for result (required) and facets node (optional) */
 351             int rc_result = -1;
 352             int rc_facets = 0;
 353             for (ptr = root->children; ptr; ptr = ptr->next)
 354             {
 355                 if (ptr->type == XML_ELEMENT_NODE &&
 356                     !strcmp((const char *) ptr->name, "result")) {
 357                         pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
 358                         sr = pdu->u.response;
 359                         rc_result = yaz_solr_decode_result(o, ptr, sr);
 360                 }
 361                 if (ptr->type == XML_ELEMENT_NODE &&
 362                     match_xml_node_attribute(ptr, "lst", "name", "terms")) {
 363                         pdu = yaz_srw_get(o, Z_SRW_scan_response);
 364                         scr = pdu->u.scan_response;
 365                         rc_result = yaz_solr_decode_scan_result(o, ptr, scr);
 366                 }
 367                 /* TODO The check on hits is a work-around to avoid garbled facets on zero results from the SOLR server.
 368                  * The work-around works because the results is before the facets in the xml. */
 369                 if (sr) {
 370                     if (rc_result == 0 &&  *sr->numberOfRecords > 0 &&
 371                         match_xml_node_attribute(ptr, "lst", "name", "facet_counts"))
 372                             rc_facets =  yaz_solr_decode_facet_counts(o, ptr, sr);
 373                     if (rc_result == 0 &&  *sr->numberOfRecords == 0 &&
 374                         match_xml_node_attribute(ptr, "lst", "name", "spellcheck"))
 375                             rc_facets =  yaz_solr_decode_spellcheck(o, ptr, sr);
 376                 }
 377
 378             }
 379             ret = rc_result + rc_facets;
 380         }
 381     }
 382     if (doc)
 383         xmlFreeDoc(doc);
 384     if (ret == 0)
 385         *pdup = pdu;
 386     return ret;
 387 #else
 388     return -1;
 389 #endif
 390 }
 391
 392 static int yaz_solr_encode_facet_field(
 393     ODR encode, char **name, char **value, int *i,
 394     Z_FacetField *facet_field)
 395 {
 396     Z_AttributeList *attribute_list = facet_field->attributes;
 397     struct yaz_facet_attr attr_values;
 398     yaz_facet_attr_init(&attr_values);
 399     yaz_facet_attr_get_z_attributes(attribute_list, &attr_values);
 400     // TODO do we want to support server decided
 401
 402     if (attr_values.errcode)
 403         return -1;
 404     if (attr_values.useattr)
 405     {
 406         WRBUF wrbuf = wrbuf_alloc();
 407         wrbuf_puts(wrbuf, (char *) attr_values.useattr);
 408         yaz_add_name_value_str(encode, name, value, i,
 409                                "facet.field",
 410                                odr_strdup(encode, wrbuf_cstr(wrbuf)));
 411         if (attr_values.limit > 0)
 412         {
 413             WRBUF wrbuf2 = wrbuf_alloc();
 414             Odr_int olimit;
 415             wrbuf_puts(wrbuf2, "f.");
 416             wrbuf_puts(wrbuf2, wrbuf_cstr(wrbuf));
 417             wrbuf_puts(wrbuf2, ".facet.limit");
 418             olimit = attr_values.limit;
 419             yaz_add_name_value_int(encode, name, value, i,
 420                                    odr_strdup(encode, wrbuf_cstr(wrbuf2)),
 421                                    &olimit);
 422             wrbuf_destroy(wrbuf2);
 423         }
 424         wrbuf_destroy(wrbuf);
 425     }
 426     return 0;
 427 }
 428
 429 static int yaz_solr_encode_facet_list(
 430     ODR encode, char **name, char **value,
 431     int *i, Z_FacetList *facet_list)
 432 {
 433     int index;
 434     for (index = 0; index < facet_list->num; index++)
 435     {
 436         int r = yaz_solr_encode_facet_field(encode, name, value, i,
 437                                             facet_list->elements[index]);
 438         if (r)
 439             return -1;
 440
 441     }
 442     return 0;
 443 }
 444
 445 int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu,
 446                             ODR encode, const char *charset)
 447 {
 448     const char *solr_op = 0;
 449     //TODO Change. not a nice hard coded, unchecked limit.
 450     char *name[SOLR_MAX_PARAMETERS], *value[SOLR_MAX_PARAMETERS];
 451     char *uri_args;
 452     char *path;
 453     char *q;
 454     char *pos;
 455     char *cp;
 456     int i = 0;
 457
 458     z_HTTP_header_add_basic_auth(encode, &hreq->headers,
 459                                  srw_pdu->username, srw_pdu->password);
 460     if (srw_pdu->which == Z_SRW_searchRetrieve_request)
 461     {
 462         Z_SRW_searchRetrieveRequest *request = srw_pdu->u.request;
 463         solr_op = "select";
 464         switch (srw_pdu->u.request->query_type)
 465         {
 466         case Z_SRW_query_type_pqf:
 467             yaz_add_name_value_str(encode, name, value, &i,
 468                                    "q", request->query.pqf);
 469             break;
 470         case Z_SRW_query_type_cql:
 471             yaz_add_name_value_str(encode, name, value, &i,
 472                                    "q", request->query.cql);
 473             break;
 474         default:
 475             return -1;
 476         }
 477         if (srw_pdu->u.request->startRecord)
 478         {
 479             Odr_int start = *request->startRecord - 1;
 480             yaz_add_name_value_int(encode, name, value, &i,
 481                                    "start", &start);
 482         }
 483         yaz_add_name_value_int(encode, name, value, &i,
 484                                "rows", request->maximumRecords);
 485         yaz_add_name_value_str(encode, name, value, &i,
 486                                "fl", request->recordSchema);
 487
 488         switch(srw_pdu->u.request->sort_type)
 489         {
 490         case Z_SRW_sort_type_none:
 491             break;
 492         case Z_SRW_sort_type_sort:
 493             yaz_add_name_value_str(encode, name, value, &i, "sort",
 494                                    srw_pdu->u.request->sort.sortKeys);
 495             break;
 496         }
 497         if (request->facetList)
 498         {
 499             Z_FacetList *facet_list = request->facetList;
 500             yaz_add_name_value_str(encode, name, value, &i, "facet", "true");
 501             yaz_add_name_value_str(encode, name, value, &i, "facet.mincount", "1");
 502             if (yaz_solr_encode_facet_list(encode, name, value, &i, facet_list))
 503                 return -1;
 504         }
 505     }
 506     else if (srw_pdu->which == Z_SRW_scan_request) {
 507         Z_SRW_scanRequest *request = srw_pdu->u.scan_request;
 508         solr_op = "terms";
 509         switch (srw_pdu->u.scan_request->query_type)
 510         {
 511             case Z_SRW_query_type_pqf:
 512                 yaz_add_name_value_str(encode, name, value, &i,
 513                                        "terms.fl", request->scanClause.pqf);
 514                 yaz_add_name_value_str(encode, name, value, &i,
 515                                        "terms.lower", request->scanClause.pqf);
 516                 break;
 517             case Z_SRW_query_type_cql:
 518                 q = request->scanClause.cql;
 519                 pos = strchr(q, ':');
 520                 if (pos != NULL) {
 521                                         yaz_add_name_value_str(encode, name, value, &i,
 522                                                                                    "terms.lower", odr_strdup(encode, pos + 1));
 523                                         *pos = '\0';
 524                                         yaz_add_name_value_str(encode, name, value, &i,
 525                                                                                    "terms.fl", odr_strdup(encode, q));
 526                                         *pos = ':';
 527                 } else {
 528                                         yaz_add_name_value_str(encode, name, value, &i,
 529                                                                                    "terms.lower", odr_strdup(encode, q));
 530                 }
 531                 break;
 532             default:
 533                 return -1;
 534         }
 535         yaz_add_name_value_str(encode, name, value, &i,
 536                                "terms.sort", "index");
 537         yaz_add_name_value_int(encode, name, value, &i,
 538                                "terms.limit", request->maximumTerms);
 539     }
 540     else
 541         return -1;
 542
 543     if (srw_pdu->extra_args)
 544     {
 545         Z_SRW_extra_arg *ea = srw_pdu->extra_args;
 546         for (; ea && i < SOLR_MAX_PARAMETERS; ea = ea->next)
 547         {
 548             name[i] = ea->name;
 549             value[i] = ea->value;
 550             i++;
 551         }
 552     }
 553
 554     name[i++] = 0;
 555
 556     yaz_array_to_uri(&uri_args, encode, name, value);
 557
 558     hreq->method = "GET";
 559
 560     path = (char *)
 561         odr_malloc(encode, strlen(hreq->path) +
 562                    strlen(uri_args) + strlen(solr_op) + 4);
 563
 564     strcpy(path, hreq->path);
 565     cp = strrchr(path, '/');
 566     if (cp)
 567     {
 568         if (!strcmp(cp, "/select") || !strcmp(cp, "/"))
 569             *cp = '\0';
 570     }
 571     strcat(path, "/");
 572     strcat(path, solr_op);
 573     strcat(path, "?");
 574     strcat(path, uri_args);
 575     hreq->path = path;
 576
 577     z_HTTP_header_add_content_type(encode, &hreq->headers,
 578                                    "text/xml", charset);
 579     return 0;
 580 }
 581
 582
 583 /*
 584  * Local variables:
 585  * c-basic-offset: 4
 586  * c-file-style: "Stroustrup"
 587  * indent-tabs-mode: nil
 588  * End:
 589  * vim: shiftwidth=4 tabstop=8 expandtab
 590  */
 591