X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fsolr.c;h=3458a0b9037c7b8e86e8f6e3b3dc0ca4fa0180c8;hp=1783440362f188a0f9accea0f35a6a6a8e4ce9db;hb=1fbd038d97d6e1bfe97270ecc17cca123bcc435c;hpb=96e65728eefb6d7efef6e2f74a3b0d64f92b3c23 diff --git a/src/solr.c b/src/solr.c index 1783440..3458a0b 100644 --- a/src/solr.c +++ b/src/solr.c @@ -1,23 +1,184 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2010 Index Data + * Copyright (C) 1995-2011 Index Data * See the file LICENSE for details. */ /** - * \file srwutil.c - * \brief Implements SRW/SRU utilities. + * \file solr.c + * \brief Implements SOAP Webservice decoding/encoding */ +#if HAVE_CONFIG_H +#include +#endif #include #include #include #include #include +#include +#include +#include #include "sru-p.h" +#define SOLR_MAX_PARAMETERS 100 + #if YAZ_HAVE_XML2 #include #include + +static int match_xml_node_attribute(xmlNodePtr ptr, const char *node_name, const char *attribute_name, const char *value) +{ + const char *attribute_value; + // check if the node name matches + if (strcmp((const char*) ptr->name, node_name)) + return 0; + if (attribute_name) { + attribute_value = yaz_element_attribute_value_get(ptr, node_name, attribute_name); + if (attribute_value && !strcmp(attribute_value, value)) + return 1; + } + else /* No attribute to check */ + return 1; + return 0; +} + +static void yaz_solr_decode_result_docs(ODR o, xmlNodePtr ptr, Odr_int start, Z_SRW_searchRetrieveResponse *sr) { + xmlNodePtr node; + int offset = 0; + int i = 0; + + sr->num_records = 0; + for (node = ptr->children; node; node = node->next) + if (node->type == XML_ELEMENT_NODE) + sr->num_records++; + + sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records); + + for (node = ptr->children; node; node = node->next) + { + if (node->type == XML_ELEMENT_NODE) + { + Z_SRW_record *record = sr->records + i; + xmlBufferPtr buf = xmlBufferCreate(); + xmlNode *tmp = xmlCopyNode(node, 1); + + xmlNodeDump(buf, tmp->doc, tmp, 0, 0); + + xmlFreeNode(tmp); + + record->recordSchema = 0; + record->recordPacking = Z_SRW_recordPacking_XML; + record->recordData_len = buf->use; + record->recordData_buf = odr_malloc(o, buf->use + 1); + memcpy(record->recordData_buf, buf->content, buf->use); + record->recordData_buf[buf->use] = '\0'; + // TODO Solve the real problem in zoom-sru, that doesnt work with 0-based indexes. + // Work-around: Making the recordPosition 1-based. + record->recordPosition = odr_intdup(o, start + offset + 1); + + xmlBufferFree(buf); + + offset++; + i++; + } + } +} + +static int yaz_solr_decode_result(ODR o, xmlNodePtr ptr, Z_SRW_searchRetrieveResponse *sr) { + Odr_int start = 0; + struct _xmlAttr *attr; + for (attr = ptr->properties; attr; attr = attr->next) + if (attr->children && attr->children->type == XML_TEXT_NODE) { + if (!strcmp((const char *) attr->name, "numFound")) { + sr->numberOfRecords = odr_intdup(o, odr_atoi( + (const char *) attr->children->content)); + } + else if (!strcmp((const char *) attr->name, "start")) { + start = odr_atoi((const char *) attr->children->content); + } + } + if (sr->numberOfRecords && *sr->numberOfRecords > 0) + yaz_solr_decode_result_docs(o, ptr, start, sr); + if (sr->numberOfRecords) + return 0; + return -1; +} + +static const char *get_facet_term_count(xmlNodePtr node, int *freq) { + + const char *term = yaz_element_attribute_value_get(node, "int", "name"); + xmlNodePtr child; + WRBUF wrbuf = wrbuf_alloc(); + if (!term) + return term; + + for (child = node->children; child ; child = child->next) { + if (child->type == XML_TEXT_NODE) + wrbuf_puts(wrbuf, (const char *) child->content); + } + *freq = atoi(wrbuf_cstr(wrbuf)); + wrbuf_destroy(wrbuf); + return term; +} + +Z_FacetField *yaz_solr_decode_facet_field(ODR o, xmlNodePtr ptr, Z_SRW_searchRetrieveResponse *sr) + +{ + Z_AttributeList *list; + Z_FacetField *facet_field; + int num_terms = 0; + int index = 0; + xmlNodePtr node; + // USE attribute + const char* name = yaz_element_attribute_value_get(ptr, "lst", "name"); + char *pos = strstr(name, "_exact"); + /* HACK */ + if (pos) { + pos[0] = 0; + } + list = yaz_use_attribute_create(o, name); + for (node = ptr->children; node; node = node->next) { + num_terms++; + } + facet_field = facet_field_create(o, list, num_terms); + index = 0; + for (node = ptr->children; node; node = node->next) { + int count = 0; + const char *term = get_facet_term_count(node, &count); + facet_field_term_set(o, facet_field, facet_term_create(o, term_create(o, term), count), index); + index++; + } + return facet_field; +} + +static int yaz_solr_decode_facet_counts(ODR o, xmlNodePtr root, Z_SRW_searchRetrieveResponse *sr) { + xmlNodePtr ptr; + for (ptr = root->children; ptr; ptr = ptr->next) + { + if (match_xml_node_attribute(ptr, "lst", "name", "facet_fields")) + { + xmlNodePtr node; + Z_FacetList *facet_list; + int num_facets = 0; + for (node = ptr->children; node; node= node->next) + { + num_facets++; + } + facet_list = facet_list_create(o, num_facets); + num_facets = 0; + for (node = ptr->children; node; node= node->next) + { + facet_list_field_set(o, facet_list, yaz_solr_decode_facet_field(o, node, sr), num_facets); + num_facets++; + } + sr->facetList = facet_list; + break; + } + } + return 0; +} + #endif int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup) @@ -28,7 +189,6 @@ int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup) xmlDocPtr doc = xmlParseMemory(content_buf, content_len); int ret = 0; xmlNodePtr ptr = 0; - Odr_int start = 0; Z_SRW_PDU *pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response); Z_SRW_searchRetrieveResponse *sr = pdu->u.response; @@ -49,77 +209,20 @@ int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup) } else { - /** look for result node */ + /** look for result (required) and facets node (optional) */ + int rc_result = -1; + int rc_facets = 0; for (ptr = root->children; ptr; ptr = ptr->next) { if (ptr->type == XML_ELEMENT_NODE && !strcmp((const char *) ptr->name, "result")) - break; - } - if (!ptr) - { - ret = -1; - } - } - } - if (ptr) - { /* got result node */ - struct _xmlAttr *attr; - - for (attr = ptr->properties; attr; attr = attr->next) - if (attr->children && attr->children->type == XML_TEXT_NODE) - { - if (!strcmp((const char *) attr->name, "numFound")) - { - sr->numberOfRecords = - odr_intdup(o, - odr_atoi( - (const char *) attr->children->content)); - } - else if (!strcmp((const char *) attr->name, "start")) - { - start = odr_atoi((const char *) attr->children->content); - } - } - } - if (ptr) - { - xmlNodePtr node; - int offset = 0; - int i = 0; - - sr->num_records = 0; - for (node = ptr->children; node; node = node->next) - if (node->type == XML_ELEMENT_NODE) - sr->num_records++; - - sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records); - - for (node = ptr->children; node; node = node->next) - { - if (node->type == XML_ELEMENT_NODE) - { - Z_SRW_record *record = sr->records + i; - xmlBufferPtr buf = xmlBufferCreate(); - xmlNode *tmp = xmlCopyNode(node, 1); - - xmlNodeDump(buf, tmp->doc, tmp, 0, 0); - - xmlFreeNode(tmp); - - record->recordSchema = 0; - record->recordPacking = Z_SRW_recordPacking_XML; - record->recordData_len = buf->use; - record->recordData_buf = odr_malloc(o, buf->use + 1); - memcpy(record->recordData_buf, buf->content, buf->use); - record->recordData_buf[buf->use] = '\0'; - record->recordPosition = odr_intdup(o, start + offset); - - xmlBufferFree(buf); - - offset++; - i++; + rc_result = yaz_solr_decode_result(o, ptr, sr); + /* TODO The check on hits is a work-around to avoid garbled facets on zero results from the SOLR server. + * The work-around works because the results is before the facets in the xml. */ + if (rc_result == 0 && *sr->numberOfRecords > 0 && match_xml_node_attribute(ptr, "lst", "name", "facet_counts")) + rc_facets = yaz_solr_decode_facet_counts(o, ptr, sr); } + ret = rc_result + rc_facets; } } if (doc) @@ -132,11 +235,49 @@ int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup) #endif } -int yaz_solr_encode(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu, - ODR encode, const char *charset) +static void yaz_solr_encode_facet_field(ODR encode, char **name, char **value, int *i, Z_FacetField *facet_field, int *limit) { + Z_AttributeList *attribute_list = facet_field->attributes; + struct yaz_facet_attr attr_values; + yaz_facet_attr_init(&attr_values); + yaz_facet_attr_get_z_attributes(attribute_list, &attr_values); + // TODO do we want to support server decided + if (!attr_values.errcode && attr_values.useattr) { + WRBUF wrbuf = wrbuf_alloc(); + wrbuf_puts(wrbuf, (char *) attr_values.useattr); + /* Skip date field */ + if (strcmp("date", attr_values.useattr) != 0) + wrbuf_puts(wrbuf, "_exact"); + yaz_add_name_value_str(encode, name, value, i, "facet.field", odr_strdup(encode, wrbuf_cstr(wrbuf))); + if (attr_values.limit > 0) { + WRBUF wrbuf2 = wrbuf_alloc(); + Odr_int olimit; + wrbuf_puts(wrbuf2, "f."); + wrbuf_puts(wrbuf2, wrbuf_cstr(wrbuf)); + wrbuf_puts(wrbuf2, ".facet.limit"); + olimit = attr_values.limit; + yaz_add_name_value_int(encode, name, value, i, odr_strdup(encode, wrbuf_cstr(wrbuf2)), &olimit); + wrbuf_destroy(wrbuf2); + } + wrbuf_destroy(wrbuf); + } +} + +static void yaz_solr_encode_facet_list(ODR encode, char **name, char **value, int *i, Z_FacetList *facet_list, int *limit) { + + int index; + for (index = 0; index < facet_list->num; index++) { + yaz_solr_encode_facet_field(encode, name, value, i, facet_list->elements[index], limit); + + } +} + + +int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu, + ODR encode, const char *charset) { const char *solr_op = 0; - char *name[30], *value[30]; + //TODO Change. not a nice hard coded, unchecked limit. + char *name[SOLR_MAX_PARAMETERS], *value[SOLR_MAX_PARAMETERS]; char *uri_args; char *path; int i = 0; @@ -146,31 +287,47 @@ int yaz_solr_encode(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu, switch (srw_pdu->which) { - case Z_SRW_searchRetrieve_request: + case Z_SRW_searchRetrieve_request: { + Z_SRW_searchRetrieveRequest *request = srw_pdu->u.request; solr_op = "select"; - switch(srw_pdu->u.request->query_type) { case Z_SRW_query_type_pqf: yaz_add_name_value_str(encode, name, value, &i, - "q", srw_pdu->u.request->query.pqf); + "q", request->query.pqf); break; case Z_SRW_query_type_cql: yaz_add_name_value_str(encode, name, value, &i, - "q", srw_pdu->u.request->query.cql); + "q", request->query.cql); break; default: return -1; } if (srw_pdu->u.request->startRecord) { - Odr_int start = *srw_pdu->u.request->startRecord - 1; + Odr_int start = *request->startRecord - 1; yaz_add_name_value_int(encode, name, value, &i, "start", &start); } yaz_add_name_value_int(encode, name, value, &i, - "rows", srw_pdu->u.request->maximumRecords); + "rows", request->maximumRecords); + yaz_add_name_value_str(encode, name, value, &i, + "fl", request->recordSchema); + + if (request->facetList) { + Z_FacetList *facet_list = request->facetList; + int limit = 0; + yaz_add_name_value_str(encode, name, value, &i, "facet", "true"); + yaz_add_name_value_str(encode, name, value, &i, "facet.mincount", "1"); + yaz_solr_encode_facet_list(encode, name, value, &i, facet_list, &limit); + /* + olimit = limit; + yaz_add_name_value_int(encode, name, value, &i, "facet.limit", &olimit); + */ + + } break; + } default: return -1; }