1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2010 Index Data
3 * See the file LICENSE for details.
7 * \brief Implements SRW/SRU utilities.
13 #include <yaz/matchstr.h>
14 #include <yaz/yaz-iconv.h>
16 #include <yaz/facet.h>
17 #include <yaz/wrbuf.h>
22 #include <libxml/parser.h>
23 #include <libxml/tree.h>
25 #define SOLR_MAX_PARAMETERS 100
27 const char *xml_node_attribute_value_get(xmlNodePtr ptr, const char *node_name, const char *attribute_name) {
29 struct _xmlAttr *attr;
30 // check if the node name matches
31 if (strcmp((const char*) ptr->name, node_name))
33 // check if the attribute name and return the value
34 for (attr = ptr->properties; attr; attr = attr->next)
35 if (attr->children && attr->children->type == XML_TEXT_NODE) {
36 if (!strcmp((const char *) attr->name, attribute_name))
37 return (const char *) attr->children->content;
43 static int match_xml_node_attribute(xmlNodePtr ptr, const char *node_name, const char *attribute_name, const char *value)
45 const char *attribute_value;
46 // check if the node name matches
47 if (strcmp((const char*) ptr->name, node_name))
49 attribute_value = xml_node_attribute_value_get(ptr, node_name, attribute_name);
50 if (attribute_value && !strcmp(attribute_value, value))
55 static void yaz_solr_decode_result_docs(ODR o, xmlNodePtr ptr, Odr_int start, Z_SRW_searchRetrieveResponse *sr) {
61 for (node = ptr->children; node; node = node->next)
62 if (node->type == XML_ELEMENT_NODE)
65 sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records);
67 for (node = ptr->children; node; node = node->next)
69 if (node->type == XML_ELEMENT_NODE)
71 Z_SRW_record *record = sr->records + i;
72 xmlBufferPtr buf = xmlBufferCreate();
73 xmlNode *tmp = xmlCopyNode(node, 1);
75 xmlNodeDump(buf, tmp->doc, tmp, 0, 0);
79 record->recordSchema = 0;
80 record->recordPacking = Z_SRW_recordPacking_XML;
81 record->recordData_len = buf->use;
82 record->recordData_buf = odr_malloc(o, buf->use + 1);
83 memcpy(record->recordData_buf, buf->content, buf->use);
84 record->recordData_buf[buf->use] = '\0';
85 // TODO Solve the real problem in zoom-sru, that doesnt work with 0-based indexes.
86 // Work-around: Making the recordPosition 1-based.
87 record->recordPosition = odr_intdup(o, start + offset + 1);
97 static int yaz_solr_decode_result(ODR o, xmlNodePtr ptr, Z_SRW_searchRetrieveResponse *sr) {
99 struct _xmlAttr *attr;
100 for (attr = ptr->properties; attr; attr = attr->next)
101 if (attr->children && attr->children->type == XML_TEXT_NODE) {
102 if (!strcmp((const char *) attr->name, "numFound")) {
103 sr->numberOfRecords = odr_intdup(o, odr_atoi(
104 (const char *) attr->children->content));
106 else if (!strcmp((const char *) attr->name, "start")) {
107 start = odr_atoi((const char *) attr->children->content);
110 if (sr->numberOfRecords && *sr->numberOfRecords > 0)
111 yaz_solr_decode_result_docs(o, ptr, start, sr);
112 if (sr->numberOfRecords)
117 static Z_AttributeList *yaz_solr_use_atttribute_create(ODR o, const char *name) {
123 static const char *get_facet_term_count(xmlNodePtr node, int *freq) {
125 const char *term = xml_node_attribute_value_get(node, "int", "name");
127 WRBUF wrbuf = wrbuf_alloc();
131 for (child = node->children; child ; child = child->next) {
132 if (child->type == XML_TEXT_NODE)
133 wrbuf_puts(wrbuf, (const char *) child->content);
135 *freq = atoi(wrbuf_cstr(wrbuf));
136 wrbuf_destroy(wrbuf);
140 Z_FacetField *yaz_solr_decode_facet_field(ODR o, xmlNodePtr ptr, Z_SRW_searchRetrieveResponse *sr)
143 const char* name = xml_node_attribute_value_get(ptr, "lst", "name");
144 Z_AttributeList *list = yaz_solr_use_atttribute_create(o, name);
145 Z_FacetField *facet_field;
149 for (node = ptr->children; node; node = node->next) {
152 facet_field = facet_field_create(o, list, num_terms);
154 for (node = ptr->children; node; node = node->next) {
156 const char *term = get_facet_term_count(node, &count);
157 facet_field_term_set(o, facet_field, facet_term_create(o, term_create(o, term), count), index);
163 static int yaz_solr_decode_facet_counts(ODR o, xmlNodePtr root, Z_SRW_searchRetrieveResponse *sr) {
165 for (ptr = root->children; ptr; ptr = ptr->next)
167 if (match_xml_node_attribute(ptr, "lst", "name", "facet_fields"))
170 Z_FacetList *facet_list;
172 for (node = ptr->children; node; node= node->next)
176 facet_list = facet_list_create(o, num_facets);
178 for (node = ptr->children; node; node= node->next)
180 facet_list_field_set(o, facet_list, yaz_solr_decode_facet_field(o, node, sr), num_facets);
183 sr->facetList = facet_list;
190 static void yaz_solr_decode_facets(ODR o, xmlNodePtr ptr, Z_SRW_searchRetrieveResponse *sr) {
191 if (match_xml_node_attribute(ptr, "lst", "name", "facet_counts"))
192 yaz_solr_decode_facet_counts(o, ptr->children, sr);
196 int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup)
199 const char *content_buf = hres->content_buf;
200 int content_len = hres->content_len;
201 xmlDocPtr doc = xmlParseMemory(content_buf, content_len);
204 Z_SRW_PDU *pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
205 Z_SRW_searchRetrieveResponse *sr = pdu->u.response;
213 xmlNodePtr root = xmlDocGetRootElement(doc);
218 else if (strcmp((const char *) root->name, "response"))
224 /** look for result (required) and facets node (optional) */
227 for (ptr = root->children; ptr; ptr = ptr->next)
229 if (ptr->type == XML_ELEMENT_NODE &&
230 !strcmp((const char *) ptr->name, "result"))
231 rc_result = yaz_solr_decode_result(o, ptr, sr);
232 if (match_xml_node_attribute(ptr, "lst", "name", "facet_counts"))
233 rc_facets = yaz_solr_decode_facet_counts(o, ptr, sr);
235 ret = rc_result + rc_facets;
248 static void yaz_solr_encode_facet_field(ODR encode, char **name, char **value, int *i, Z_FacetField *facet_field, int *limit) {
249 Z_AttributeList *attribute_list = facet_field->attributes;
250 struct yaz_facet_attr attr_values;
251 yaz_facet_attr_init(&attr_values);
252 yaz_facet_attr_get_z_attributes(attribute_list, &attr_values);
253 // TODO do we want to support server decided
254 if (!attr_values.errcode && attr_values.useattr) {
255 yaz_add_name_value_str(encode, name, value, i, "facet.field", (char *) attr_values.useattr);
256 // TODO max(attr_values, *limit);
257 if (attr_values.limit > 0 && attr_values.limit > *limit) {
258 *limit = attr_values.limit;
263 static void yaz_solr_encode_facet_list(ODR encode, char **name, char **value, int *i, Z_FacetList *facet_list, int *limit) {
266 for (index = 0; index < facet_list->num; index++) {
267 yaz_solr_encode_facet_field(encode, name, value, i, facet_list->elements[index], limit);
273 int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu,
274 ODR encode, const char *charset)
276 const char *solr_op = 0;
277 //TODO Change. not a nice hard coded, unchecked limit.
278 char *name[SOLR_MAX_PARAMETERS], *value[SOLR_MAX_PARAMETERS];
283 z_HTTP_header_add_basic_auth(encode, &hreq->headers,
284 srw_pdu->username, srw_pdu->password);
286 switch (srw_pdu->which)
288 case Z_SRW_searchRetrieve_request: {
289 Z_SRW_searchRetrieveRequest *request = srw_pdu->u.request;
291 switch(srw_pdu->u.request->query_type)
293 case Z_SRW_query_type_pqf:
294 yaz_add_name_value_str(encode, name, value, &i,
295 "q", request->query.pqf);
297 case Z_SRW_query_type_cql:
298 yaz_add_name_value_str(encode, name, value, &i,
299 "q", request->query.cql);
304 if (srw_pdu->u.request->startRecord)
306 Odr_int start = *request->startRecord - 1;
307 yaz_add_name_value_int(encode, name, value, &i,
310 yaz_add_name_value_int(encode, name, value, &i,
311 "rows", request->maximumRecords);
312 yaz_add_name_value_str(encode, name, value, &i,
313 "fl", request->recordSchema);
315 if (request->facetList) {
316 Z_FacetList *facet_list = request->facetList;
319 yaz_add_name_value_str(encode, name, value, &i, "facet", "true");
320 yaz_solr_encode_facet_list(encode, name, value, &i, facet_list, &limit);
322 yaz_add_name_value_int(encode, name, value, &i, "facet.limit", &olimit);
331 yaz_array_to_uri(&uri_args, encode, name, value);
333 hreq->method = "GET";
336 odr_malloc(encode, strlen(hreq->path) +
337 strlen(uri_args) + strlen(solr_op) + 4);
339 sprintf(path, "%s/%s?%s", hreq->path, solr_op, uri_args);
342 z_HTTP_header_add_content_type(encode, &hreq->headers,
343 "text/xml", charset);
351 * c-file-style: "Stroustrup"
352 * indent-tabs-mode: nil
354 * vim: shiftwidth=4 tabstop=8 expandtab