src/solr.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 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 #include <yaz/proto.h>
  22
  23 #include "sru-p.h"
  24
  25 #if YAZ_HAVE_XML2
  26 #include <libxml/parser.h>
  27 #include <libxml/tree.h>
  28
  29 static void extract_text_node(xmlNodePtr node, WRBUF wrbuf)
  30 {
  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 =
  91                 odr_strdupn(o, (const char *) buf->content, buf->use);
  92             record->recordPosition = odr_intdup(o, start + offset + 1);
  93
  94             xmlBufferFree(buf);
  95
  96             offset++;
  97             i++;
  98         }
  99     }
 100 }
 101
 102 static int yaz_solr_decode_result(ODR o, xmlNodePtr ptr,
 103                                   Z_SRW_searchRetrieveResponse *sr)
 104 {
 105     Odr_int start = 0;
 106     struct _xmlAttr *attr;
 107     for (attr = ptr->properties; attr; attr = attr->next)
 108         if (attr->children && attr->children->type == XML_TEXT_NODE)
 109         {
 110             if (!strcmp((const char *) attr->name, "numFound"))
 111             {
 112                 sr->numberOfRecords = odr_intdup(o, odr_atoi(
 113                         (const char *) attr->children->content));
 114             }
 115             else if (!strcmp((const char *) attr->name, "start"))
 116             {
 117                 start = odr_atoi((const char *) attr->children->content);
 118             }
 119         }
 120     if (sr->numberOfRecords && *sr->numberOfRecords > 0)
 121         yaz_solr_decode_result_docs(o, ptr, start, sr);
 122     if (sr->numberOfRecords)
 123         return 0;
 124     return -1;
 125 }
 126
 127 static const char *get_facet_term_count(xmlNodePtr node, Odr_int *freq)
 128 {
 129     const char *term = yaz_element_attribute_value_get(node, "int", "name");
 130     xmlNodePtr child;
 131     WRBUF wrbuf = wrbuf_alloc();
 132     if (!term)
 133         return term;
 134
 135     for (child = node->children; child ; child = child->next)
 136     {
 137         if (child->type == XML_TEXT_NODE)
 138             wrbuf_puts(wrbuf, (const char *) child->content);
 139     }
 140     *freq = odr_atoi(wrbuf_cstr(wrbuf));
 141     wrbuf_destroy(wrbuf);
 142     return term;
 143 }
 144
 145 Z_FacetField *yaz_solr_decode_facet_field(ODR o, xmlNodePtr ptr,
 146                                           Z_SRW_searchRetrieveResponse *sr)
 147
 148 {
 149     Z_AttributeList *list;
 150     Z_FacetField *facet_field;
 151     int num_terms = 0;
 152     int index = 0;
 153     xmlNodePtr node;
 154     // USE attribute
 155     const char* name = yaz_element_attribute_value_get(ptr, "lst", "name");
 156     list = zget_AttributeList_use_string(o, name);
 157     for (node = ptr->children; node; node = node->next)
 158         num_terms++;
 159     facet_field = facet_field_create(o, list, num_terms);
 160     index = 0;
 161     for (node = ptr->children; node; node = node->next)
 162     {
 163         Odr_int count = 0;
 164         const char *term = get_facet_term_count(node, &count);
 165         facet_field_term_set(o, facet_field,
 166                              facet_term_create_cstr(o, term, count), index);
 167         index++;
 168     }
 169     return facet_field;
 170 }
 171
 172 static int yaz_solr_decode_facet_counts(ODR o, xmlNodePtr root,
 173                                         Z_SRW_searchRetrieveResponse *sr)
 174 {
 175     xmlNodePtr ptr;
 176     for (ptr = root->children; ptr; ptr = ptr->next)
 177     {
 178         if (match_xml_node_attribute(ptr, "lst", "name", "facet_fields"))
 179         {
 180             xmlNodePtr node;
 181             Z_FacetList *facet_list;
 182             int num_facets = 0;
 183             for (node = ptr->children; node; node= node->next)
 184             {
 185                 num_facets++;
 186             }
 187             facet_list = facet_list_create(o, num_facets);
 188             num_facets = 0;
 189             for (node = ptr->children; node; node= node->next)
 190             {
 191                 facet_list_field_set(o, facet_list,
 192                                      yaz_solr_decode_facet_field(o, node, sr),
 193                                      num_facets);
 194                 num_facets++;
 195             }
 196             sr->facetList = facet_list;
 197             break;
 198         }
 199     }
 200     return 0;
 201 }
 202
 203 static void yaz_solr_decode_suggestion_values(xmlNodePtr listPptr, WRBUF wrbuf)
 204 {
 205     xmlNodePtr node;
 206     for (node = listPptr; node; node= node->next)
 207         if (!strcmp((char*) node->name, "lst"))
 208         {
 209             xmlNodePtr child;
 210             for (child = node->children; child; child= child->next)
 211             {
 212                 if (match_xml_node_attribute(child, "str", "name", "word"))
 213                 {
 214                     wrbuf_puts(wrbuf, "<suggestion>");
 215                     extract_text_node(child, wrbuf);
 216                     wrbuf_puts(wrbuf, "</suggestion>\n");
 217                 }
 218             }
 219         }
 220 }
 221
 222 static void yaz_solr_decode_suggestion_lst(xmlNodePtr lstPtr, WRBUF wrbuf)
 223 {
 224     xmlNodePtr node;
 225     for (node = lstPtr; node; node= node->next)
 226         if (match_xml_node_attribute(node, "arr", "name", "suggestion"))
 227             yaz_solr_decode_suggestion_values(node->children, wrbuf);
 228 }
 229
 230 static void yaz_solr_decode_misspelled(xmlNodePtr lstPtr, WRBUF wrbuf)
 231 {
 232     xmlNodePtr node;
 233     for (node = lstPtr; node; node= node->next)
 234     {
 235         if (!strcmp((const char*) node->name, "lst"))
 236         {
 237             const char *misspelled =
 238                 yaz_element_attribute_value_get(node, "lst", "name");
 239             if (misspelled)
 240             {
 241                 wrbuf_printf(wrbuf, "<misspelled term=\"%s\">\n", misspelled);
 242                 yaz_solr_decode_suggestion_lst(node->children, wrbuf);
 243                 wrbuf_puts(wrbuf, "</misspelled>\n");
 244             }
 245         }
 246     }
 247 }
 248
 249 static int yaz_solr_decode_spellcheck(ODR o, xmlNodePtr spellcheckPtr, Z_SRW_searchRetrieveResponse *sr)
 250 {
 251     xmlNodePtr ptr;
 252     WRBUF wrbuf = wrbuf_alloc();
 253     wrbuf_puts(wrbuf, "");
 254     for (ptr = spellcheckPtr->children; ptr; ptr = ptr->next)
 255     {
 256         if (match_xml_node_attribute(ptr, "lst", "name", "suggestions"))
 257         {
 258             yaz_solr_decode_misspelled(ptr->children, wrbuf);
 259         }
 260     }
 261     sr->suggestions = odr_strdup(o, wrbuf_cstr(wrbuf));
 262     return 0;
 263 }
 264
 265 static int yaz_solr_decode_scan_result(ODR o, xmlNodePtr ptr,
 266                                        Z_SRW_scanResponse *scr)
 267 {
 268     xmlNodePtr node;
 269     char *pos;
 270     int i = 0;
 271
 272     /* find the actual list */
 273     for (node = ptr->children; node; node = node->next)
 274         if (node->type == XML_ELEMENT_NODE)
 275         {
 276             ptr = node;
 277             break;
 278         }
 279
 280     scr->num_terms = 0;
 281     for (node = ptr->children; node; node = node->next)
 282         if (node->type == XML_ELEMENT_NODE &&
 283             !strcmp((const char *) node->name, "int"))
 284             scr->num_terms++;
 285
 286     if (scr->num_terms)
 287         scr->terms = odr_malloc(o, sizeof(*scr->terms) * scr->num_terms);
 288
 289     for (node = ptr->children; node; node = node->next)
 290     {
 291         if (node->type == XML_ELEMENT_NODE &&
 292             !strcmp((const char *) node->name, "int"))
 293         {
 294             Z_SRW_scanTerm *term = scr->terms + i;
 295
 296             Odr_int count = 0;
 297             const char *val = get_facet_term_count(node, &count);
 298
 299             term->numberOfRecords = odr_intdup(o, count);
 300
 301             /* if val contains a ^ then it is probably term<^>display term so separate them. This is due to
 302              * SOLR not being able to encode them into 2 separate attributes.
 303              */
 304             pos = strchr(val, '^');
 305             if (pos != NULL)
 306             {
 307                 term->displayTerm = odr_strdup(o, pos + 1);
 308                 *pos = '\0';
 309                 term->value = odr_strdup(o, val);
 310                 *pos = '^';
 311             }
 312             else
 313             {
 314                 term->value = odr_strdup(o, val);
 315                 term->displayTerm = NULL;
 316             }
 317             term->whereInList = NULL;
 318             i++;
 319         }
 320     }
 321
 322     if (scr->num_terms)
 323         return 0;
 324     return -1;
 325 }
 326 #endif
 327
 328 int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup)
 329 {
 330     int ret = -1;
 331 #if YAZ_HAVE_XML2
 332     const char *content_buf = hres->content_buf;
 333     int content_len = hres->content_len;
 334     xmlDocPtr doc = xmlParseMemory(content_buf, content_len);
 335     Z_SRW_PDU *pdu = 0;
 336
 337     if (doc)
 338     {
 339         Z_SRW_searchRetrieveResponse *sr = NULL;
 340         Z_SRW_scanResponse *scr = NULL;
 341         xmlNodePtr ptr;
 342         xmlNodePtr root = xmlDocGetRootElement(doc);
 343         if (root && !strcmp((const char *) root->name, "response"))
 344         {
 345             ret = 0;
 346             for (ptr = root->children; ptr && !ret; ptr = ptr->next)
 347             {
 348                 if (ptr->type == XML_ELEMENT_NODE &&
 349                     !strcmp((const char *) ptr->name, "result"))
 350                 {
 351                     pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
 352                     sr = pdu->u.response;
 353                     ret = yaz_solr_decode_result(o, ptr, sr);
 354
 355                 }
 356                 if (ptr->type == XML_ELEMENT_NODE &&
 357                     match_xml_node_attribute(ptr, "lst", "name", "terms"))
 358                 {
 359                     pdu = yaz_srw_get(o, Z_SRW_scan_response);
 360                     scr = pdu->u.scan_response;
 361                     ret = yaz_solr_decode_scan_result(o, ptr, scr);
 362                 }
 363                 /* The check on hits is a work-around to avoid garbled
 364                    facets on zero results from the SOLR server.
 365                    The work-around works because the results is before
 366                    the facets in the xml.
 367                 */
 368                 if (sr && *sr->numberOfRecords > 0 &&
 369                     match_xml_node_attribute(ptr, "lst", "name",
 370                                              "facet_counts"))
 371                     ret = yaz_solr_decode_facet_counts(o, ptr, sr);
 372                 if (sr && *sr->numberOfRecords == 0 &&
 373                     match_xml_node_attribute(ptr, "lst", "name",
 374                                              "spellcheck"))
 375                     ret = yaz_solr_decode_spellcheck(o, ptr, sr);
 376             }
 377         }
 378         xmlFreeDoc(doc);
 379     }
 380     if (ret == 0)
 381         *pdup = pdu;
 382 #endif
 383     return ret;
 384 }
 385
 386 static int yaz_solr_encode_facet_field(
 387     ODR encode, char **name, char **value, int *i,
 388     Z_FacetField *facet_field)
 389 {
 390     Z_AttributeList *attribute_list = facet_field->attributes;
 391     struct yaz_facet_attr attr_values;
 392     yaz_facet_attr_init(&attr_values);
 393     yaz_facet_attr_get_z_attributes(attribute_list, &attr_values);
 394
 395     if (attr_values.errcode)
 396         return -1;
 397     if (attr_values.useattr)
 398     {
 399         WRBUF wrbuf = wrbuf_alloc();
 400         yaz_add_name_value_str(encode, name, value, i,
 401                                "facet.field",
 402                                odr_strdup(encode, attr_values.useattr));
 403
 404         if (attr_values.limit > 0)
 405         {
 406             Odr_int v = attr_values.limit;
 407             wrbuf_rewind(wrbuf);
 408             wrbuf_printf(wrbuf, "f.%s.facet.limit", attr_values.useattr);
 409             yaz_add_name_value_int(encode, name, value, i,
 410                                    odr_strdup(encode, wrbuf_cstr(wrbuf)),
 411                                    &v);
 412         }
 413         if (attr_values.start > 1)
 414         {
 415             Odr_int v = attr_values.start - 1;
 416             wrbuf_rewind(wrbuf);
 417             wrbuf_printf(wrbuf, "f.%s.facet.offset", attr_values.useattr);
 418             yaz_add_name_value_int(encode, name, value, i,
 419                                    odr_strdup(encode, wrbuf_cstr(wrbuf)),
 420                                    &v);
 421         }
 422         if (attr_values.sortorder == 1)
 423         {
 424             wrbuf_rewind(wrbuf);
 425             wrbuf_printf(wrbuf, "f.%s.facet.sort", attr_values.useattr);
 426             yaz_add_name_value_str(encode, name, value, i,
 427                                    odr_strdup(encode, wrbuf_cstr(wrbuf)),
 428                                    "index");
 429         }
 430         wrbuf_destroy(wrbuf);
 431     }
 432     else
 433     {
 434         if (attr_values.limit > 0)
 435         {
 436             Odr_int v = attr_values.limit;
 437             yaz_add_name_value_int(encode, name, value, i, "facet.limit", &v);
 438         }
 439         if (attr_values.start > 1)
 440         {
 441             Odr_int v = attr_values.start - 1;
 442             yaz_add_name_value_int(encode, name, value, i, "facet.offset", &v);
 443         }
 444         if (attr_values.sortorder == 1)
 445         {
 446             yaz_add_name_value_str(encode, name, value, i, "facet.sort",
 447                                    "index");
 448         }
 449     }
 450     return 0;
 451 }
 452
 453 static int yaz_solr_encode_facet_list(
 454     ODR encode, char **name, char **value,
 455     int *i, Z_FacetList *facet_list)
 456 {
 457     int index;
 458     for (index = 0; index < facet_list->num; index++)
 459     {
 460         int r = yaz_solr_encode_facet_field(encode, name, value, i,
 461                                             facet_list->elements[index]);
 462         if (r)
 463             return -1;
 464
 465     }
 466     return 0;
 467 }
 468
 469 int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu,
 470                             ODR encode, const char *charset)
 471 {
 472     const char *solr_op = 0;
 473     char **name, **value;
 474     char *uri_args;
 475     char *path;
 476     char *q;
 477     char *cp;
 478     const char *path_args = 0;
 479     int i = 0;
 480     int no_parms = 20; /* safe upper limit of args without extra_args */
 481     Z_SRW_extra_arg *ea;
 482
 483     for (ea = srw_pdu->extra_args; ea; ea = ea->next)
 484         no_parms++;
 485     name = (char **) odr_malloc(encode, sizeof(*name) * no_parms);
 486     value = (char **) odr_malloc(encode, sizeof(*value) * no_parms);
 487
 488     for (ea = srw_pdu->extra_args; ea; ea = ea->next)
 489     {
 490         name[i] = ea->name;
 491         value[i] = ea->value;
 492         i++;
 493     }
 494
 495     z_HTTP_header_add_basic_auth(encode, &hreq->headers,
 496                                  srw_pdu->username, srw_pdu->password);
 497     if (srw_pdu->which == Z_SRW_searchRetrieve_request)
 498     {
 499         Z_SRW_searchRetrieveRequest *request = srw_pdu->u.request;
 500         solr_op = "select";
 501         if (!srw_pdu->u.request->query)
 502             return -1;
 503         yaz_add_name_value_str(encode, name, value, &i, "defType", "lucene");
 504         yaz_add_name_value_str(encode, name, value, &i, "q", request->query);
 505         if (srw_pdu->u.request->startRecord)
 506         {
 507             Odr_int start = *request->startRecord - 1;
 508             yaz_add_name_value_int(encode, name, value, &i,
 509                                    "start", &start);
 510         }
 511         yaz_add_name_value_int(encode, name, value, &i,
 512                                "rows", request->maximumRecords);
 513         yaz_add_name_value_str(encode, name, value, &i,
 514                                "fl", request->recordSchema);
 515
 516         switch(srw_pdu->u.request->sort_type)
 517         {
 518         case Z_SRW_sort_type_none:
 519             break;
 520         case Z_SRW_sort_type_sort:
 521             yaz_add_name_value_str(encode, name, value, &i, "sort",
 522                                    srw_pdu->u.request->sort.sortKeys);
 523             break;
 524         }
 525         if (request->facetList)
 526         {
 527             Z_FacetList *facet_list = request->facetList;
 528             yaz_add_name_value_str(encode, name, value, &i, "facet", "true");
 529             yaz_add_name_value_str(encode, name, value, &i, "facet.mincount", "1");
 530             if (yaz_solr_encode_facet_list(encode, name, value, &i, facet_list))
 531                 return -1;
 532         }
 533     }
 534     else if (srw_pdu->which == Z_SRW_scan_request)
 535     {
 536         Z_SRW_scanRequest *request = srw_pdu->u.scan_request;
 537         solr_op = "terms";
 538         if (!srw_pdu->u.scan_request->scanClause)
 539             return -1;
 540         if (!strcmp(srw_pdu->u.scan_request->queryType, "pqf"))
 541         {
 542             yaz_add_name_value_str(encode, name, value, &i,
 543                                    "terms.fl", request->scanClause);
 544             yaz_add_name_value_str(encode, name, value, &i,
 545                                    "terms.lower", request->scanClause);
 546         }
 547         else if (!strcmp(srw_pdu->u.scan_request->queryType, "cql"))
 548         {
 549             q = request->scanClause;
 550             cp = strchr(q, ':');
 551             if (cp != NULL)
 552             {
 553                 yaz_add_name_value_str(encode, name, value, &i,
 554                                        "terms.lower", odr_strdup(encode, cp + 1));
 555                 *cp = '\0';
 556                 yaz_add_name_value_str(encode, name, value, &i,
 557                                        "terms.fl", odr_strdup(encode, q));
 558                 *cp = ':';
 559             }
 560             else
 561                 yaz_add_name_value_str(encode, name, value, &i,
 562                                        "terms.lower", odr_strdup(encode, q));
 563         }
 564         else
 565             return -1;
 566         yaz_add_name_value_str(encode, name, value, &i,
 567                                "terms.sort", "index");
 568         yaz_add_name_value_int(encode, name, value, &i,
 569                                "terms.limit", request->maximumTerms);
 570     }
 571     else
 572         return -1;
 573
 574     name[i++] = 0;
 575
 576     yaz_array_to_uri(&uri_args, encode, name, value);
 577
 578     hreq->method = "GET";
 579
 580     path = (char *)
 581         odr_malloc(encode, strlen(hreq->path) +
 582                    strlen(uri_args) + strlen(solr_op) + 5);
 583
 584     cp = strchr(hreq->path, '#');
 585     if (cp)
 586         *cp = '\0';
 587     cp = strchr(hreq->path, '?');
 588     if (cp)
 589     {
 590         *cp = '\0'; /* args in path */
 591         path_args = cp + 1;
 592     }
 593     strcpy(path, hreq->path);
 594     cp = strrchr(path, '/');
 595     if (cp)
 596     {
 597         if (!strcmp(cp, "/select") || !strcmp(cp, "/"))
 598             *cp = '\0';
 599     }
 600     strcat(path, "/");
 601     strcat(path, solr_op);
 602     strcat(path, "?");
 603     if (path_args)
 604     {
 605         strcat(path, path_args);
 606         strcat(path, "&");
 607     }
 608     strcat(path, uri_args);
 609     hreq->path = path;
 610
 611     z_HTTP_header_add_content_type(encode, &hreq->headers,
 612                                    "text/xml", charset);
 613     return 0;
 614 }
 615
 616
 617 /*
 618  * Local variables:
 619  * c-basic-offset: 4
 620  * c-file-style: "Stroustrup"
 621  * indent-tabs-mode: nil
 622  * End:
 623  * vim: shiftwidth=4 tabstop=8 expandtab
 624  */
 625