/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2011 Index Data
+ * Copyright (C) 1995-2013 Index Data
* See the file LICENSE for details.
*/
/**
* \file solr.c
- * \brief Implements SOAP Webservice decoding/encoding
+ * \brief Implements Solr decoding/encoding
*/
#if HAVE_CONFIG_H
#include <config.h>
#include <libxml/parser.h>
#include <libxml/tree.h>
+static void extract_text_node(xmlNodePtr node, WRBUF wrbuf) {
+ xmlNodePtr child;
+ for (child = node->children; child ; child = child->next)
+ {
+ if (child->type == XML_TEXT_NODE)
+ wrbuf_puts(wrbuf, (const char *) child->content);
+ }
+}
+
static int match_xml_node_attribute(
xmlNodePtr ptr,
const char *node_name, const char *attribute_name, const char *value)
if (node->type == XML_ELEMENT_NODE)
sr->num_records++;
- sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records);
+ if (sr->num_records)
+ sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records);
for (node = ptr->children; node; node = node->next)
{
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);
{
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);
return -1;
}
-static const char *get_facet_term_count(xmlNodePtr node, int *freq)
+static const char *get_facet_term_count(xmlNodePtr node, Odr_int *freq)
{
const char *term = yaz_element_attribute_value_get(node, "int", "name");
xmlNodePtr child;
if (child->type == XML_TEXT_NODE)
wrbuf_puts(wrbuf, (const char *) child->content);
}
- *freq = atoi(wrbuf_cstr(wrbuf));
+ *freq = odr_atoi(wrbuf_cstr(wrbuf));
wrbuf_destroy(wrbuf);
return term;
}
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++;
index = 0;
for (node = ptr->children; node; node = node->next)
{
- int count = 0;
+ Odr_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);
+ facet_term_create_cstr(o, term, count), index);
index++;
}
return facet_field;
return 0;
}
+static void yaz_solr_decode_suggestion_values(xmlNodePtr listPptr, WRBUF wrbuf)
+{
+ xmlNodePtr node;
+ for (node = listPptr; node; node= node->next) {
+ if (!strcmp((char*) node->name, "lst")) {
+ xmlNodePtr child;
+ for (child = node->children; child; child= child->next) {
+ if (match_xml_node_attribute(child, "str", "name", "word")) {
+ wrbuf_puts(wrbuf, "<suggestion>");
+ extract_text_node(child, wrbuf);
+ wrbuf_puts(wrbuf, "</suggestion>\n");
+ }
+ }
+ }
+ }
+}
+
+static void yaz_solr_decode_suggestion_lst(xmlNodePtr lstPtr, WRBUF wrbuf)
+{
+ xmlNodePtr node;
+ for (node = lstPtr; node; node= node->next) {
+ if (match_xml_node_attribute(node, "arr", "name", "suggestion")) {
+ yaz_solr_decode_suggestion_values(node->children, wrbuf);
+ }
+ }
+}
+
+static void yaz_solr_decode_misspelled(xmlNodePtr lstPtr, WRBUF wrbuf)
+{
+ xmlNodePtr node;
+ for (node = lstPtr; node; node= node->next)
+ {
+ if (!strcmp((const char*) node->name, "lst")) {
+ const char *misspelled = yaz_element_attribute_value_get(node, "lst", "name");
+ if (misspelled) {
+ wrbuf_printf(wrbuf, "<misspelled term=\"%s\">\n", misspelled);
+ yaz_solr_decode_suggestion_lst(node->children, wrbuf);
+ wrbuf_puts(wrbuf, "</misspelled>\n");
+ }
+ }
+ }
+}
+
+static int yaz_solr_decode_spellcheck(ODR o, xmlNodePtr spellcheckPtr, Z_SRW_searchRetrieveResponse *sr)
+{
+ xmlNodePtr ptr;
+ WRBUF wrbuf = wrbuf_alloc();
+ wrbuf_puts(wrbuf, "");
+ for (ptr = spellcheckPtr->children; ptr; ptr = ptr->next)
+ {
+ if (match_xml_node_attribute(ptr, "lst", "name", "suggestions"))
+ {
+ yaz_solr_decode_misspelled(ptr->children, wrbuf);
+ }
+ }
+ sr->suggestions = odr_strdup(o, wrbuf_cstr(wrbuf));
+ return 0;
+}
+
+static int yaz_solr_decode_scan_result(ODR o, xmlNodePtr ptr,
+ Z_SRW_scanResponse *scr)
+{
+ xmlNodePtr node;
+ char *pos;
+ int i = 0;
+
+ /* find the actual list */
+ for (node = ptr->children; node; node = node->next)
+ if (node->type == XML_ELEMENT_NODE) {
+ ptr = node;
+ break;
+ }
+
+ scr->num_terms = 0;
+ for (node = ptr->children; node; node = node->next)
+ if (node->type == XML_ELEMENT_NODE && !strcmp((const char *) node->name, "int"))
+ scr->num_terms++;
+
+ if (scr->num_terms)
+ scr->terms = odr_malloc(o, sizeof(*scr->terms) * scr->num_terms);
+
+ for (node = ptr->children; node; node = node->next)
+ {
+ if (node->type == XML_ELEMENT_NODE && !strcmp((const char *) node->name, "int"))
+ {
+ Z_SRW_scanTerm *term = scr->terms + i;
+
+ Odr_int count = 0;
+ const char *val = get_facet_term_count(node, &count);
+
+ term->numberOfRecords = odr_intdup(o, count);
+
+ /* if val contains a ^ then it is probably term<^>display term so separate them. This is due to
+ * SOLR not being able to encode them into 2 separate attributes.
+ */
+ pos = strchr(val, '^');
+ if (pos != NULL) {
+ term->displayTerm = odr_strdup(o, pos + 1);
+ *pos = '\0';
+ term->value = odr_strdup(o, val);
+ *pos = '^';
+ } else {
+ term->value = odr_strdup(o, val);
+ term->displayTerm = NULL;
+ }
+ term->whereInList = NULL;
+
+ i++;
+ }
+ }
+
+ if (scr->num_terms)
+ return 0;
+ return -1;
+}
#endif
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;
- Z_SRW_PDU *pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
- Z_SRW_searchRetrieveResponse *sr = pdu->u.response;
+ Z_SRW_PDU *pdu;
+ Z_SRW_searchRetrieveResponse *sr = NULL;
+ Z_SRW_scanResponse *scr = NULL;
if (!doc)
{
for (ptr = root->children; ptr; ptr = ptr->next)
{
if (ptr->type == XML_ELEMENT_NODE &&
- !strcmp((const char *) ptr->name, "result"))
+ !strcmp((const char *) ptr->name, "result")) {
+ pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
+ sr = pdu->u.response;
rc_result = yaz_solr_decode_result(o, ptr, sr);
+ }
+ if (ptr->type == XML_ELEMENT_NODE &&
+ match_xml_node_attribute(ptr, "lst", "name", "terms")) {
+ pdu = yaz_srw_get(o, Z_SRW_scan_response);
+ scr = pdu->u.scan_response;
+ rc_result = yaz_solr_decode_scan_result(o, ptr, scr);
+ }
/* 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);
+ if (sr) {
+ 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);
+ if (rc_result == 0 && *sr->numberOfRecords == 0 &&
+ match_xml_node_attribute(ptr, "lst", "name", "spellcheck"))
+ rc_facets = yaz_solr_decode_spellcheck(o, ptr, sr);
+ }
+
}
ret = rc_result + rc_facets;
}
#endif
}
-static void yaz_solr_encode_facet_field(
+static int yaz_solr_encode_facet_field(
ODR encode, char **name, char **value, int *i,
Z_FacetField *facet_field)
{
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)
+
+ if (attr_values.errcode)
+ return -1;
+ if (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)));
}
wrbuf_destroy(wrbuf);
}
+ return 0;
}
-static void yaz_solr_encode_facet_list(
+static int yaz_solr_encode_facet_list(
ODR encode, char **name, char **value,
int *i, Z_FacetList *facet_list)
{
int index;
for (index = 0; index < facet_list->num; index++)
{
- yaz_solr_encode_facet_field(encode, name, value, i,
- facet_list->elements[index]);
-
+ int r = yaz_solr_encode_facet_field(encode, name, value, i,
+ facet_list->elements[index]);
+ if (r)
+ return -1;
+
}
+ return 0;
}
int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu,
char *name[SOLR_MAX_PARAMETERS], *value[SOLR_MAX_PARAMETERS];
char *uri_args;
char *path;
+ char *q;
+ char *pos;
int i = 0;
-
- z_HTTP_header_add_basic_auth(encode, &hreq->headers,
+
+ z_HTTP_header_add_basic_auth(encode, &hreq->headers,
srw_pdu->username, srw_pdu->password);
if (srw_pdu->which == Z_SRW_searchRetrieve_request)
{
yaz_add_name_value_str(encode, name, value, &i,
"fl", request->recordSchema);
+ switch(srw_pdu->u.request->sort_type)
+ {
+ case Z_SRW_sort_type_none:
+ break;
+ case Z_SRW_sort_type_sort:
+ yaz_add_name_value_str(encode, name, value, &i, "sort",
+ srw_pdu->u.request->sort.sortKeys);
+ break;
+ }
if (request->facetList)
{
Z_FacetList *facet_list = request->facetList;
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);
- /*
- olimit = limit;
- yaz_add_name_value_int(encode, name, value, &i, "facet.limit", &olimit);
- */
-
+ if (yaz_solr_encode_facet_list(encode, name, value, &i, facet_list))
+ return -1;
+ }
+ }
+ else if (srw_pdu->which == Z_SRW_scan_request) {
+ Z_SRW_scanRequest *request = srw_pdu->u.scan_request;
+ solr_op = "terms";
+ switch (srw_pdu->u.scan_request->query_type)
+ {
+ case Z_SRW_query_type_pqf:
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.fl", request->scanClause.pqf);
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.lower", request->scanClause.pqf);
+ break;
+ case Z_SRW_query_type_cql:
+ q = request->scanClause.cql;
+ pos = strchr(q, ':');
+ if (pos != NULL) {
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.lower", odr_strdup(encode, pos + 1));
+ *pos = '\0';
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.fl", odr_strdup(encode, q));
+ *pos = ':';
+ } else {
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.lower", odr_strdup(encode, q));
+ }
+ break;
+ default:
+ return -1;
}
+ yaz_add_name_value_str(encode, name, value, &i,
+ "terms.sort", "index");
+ yaz_add_name_value_int(encode, name, value, &i,
+ "terms.limit", request->maximumTerms);
}
else
return -1;
- name[i] = 0;
+
+ if (srw_pdu->extra_args)
+ {
+ Z_SRW_extra_arg *ea = srw_pdu->extra_args;
+ for (; ea && i < SOLR_MAX_PARAMETERS; ea = ea->next)
+ {
+ name[i] = ea->name;
+ value[i] = ea->value;
+ i++;
+ }
+ }
+
+ name[i++] = 0;
+
yaz_array_to_uri(&uri_args, encode, name, value);
-
+
hreq->method = "GET";
-
+
path = (char *)
odr_malloc(encode, strlen(hreq->path) +
strlen(uri_args) + strlen(solr_op) + 4);