public void doGet(HttpServletRequest request, HttpServletResponse response)
throws ServletException, IOException {
response.setContentType(XML_RESPONSE_HEADER); // Talkback happens in XML form.
response.setCharacterEncoding("UTF-8"); // Unicode++
request.setCharacterEncoding("UTF-8");
PrintWriter out = null; // The talkback buffer.
// handle startrecord
Integer startRecord = 0;
if (!(request.getParameter("startRecord") == null)) {
try {
startRecord = Integer.parseInt(request.getParameter("startRecord")) - 1;
} catch (NumberFormatException e) {
startRecord = 0;
}
}
// maximumrecords
Integer maximumRecords = Integer.parseInt(this.config.getProperty("default_maximumRecords"));
if (!(request.getParameter("maximumRecords") == null)) {
maximumRecords = Integer.parseInt(request.getParameter("maximumRecords"));
}
// operation
String operation = request.getParameter("operation");
// x_collection
String x_collection = request.getParameter("x-collection");
if (x_collection == null) x_collection = this.config.getProperty("default_x_collection");
if (x_collection == null) operation = null;
// sortkeys
String sortKeys = request.getParameter("sortKeys");
// sortorder
String sortOrder = request.getParameter("sortOrder");
// recordschema
String recordSchema = request.getParameter("recordSchema");
if (recordSchema == null) recordSchema = "dc";
if (recordSchema.equalsIgnoreCase("dcx")) {
recordSchema = "dcx";
}
if (recordSchema.equalsIgnoreCase("solr")) {
recordSchema = "solr";
}
// query request
String query = request.getParameter("query");
String q = request.getParameter("q");
// who is requestor ?
String remote_ip = request.getHeader("X-FORWARDED-FOR");
if (remote_ip == null) {
remote_ip = request.getRemoteAddr().trim();
} else {
remote_ip = request.getHeader("X-FORWARDED-FOR");
}
// handle debug
Boolean debug = Boolean.parseBoolean(request.getParameter("debug"));
if (!debug) {
out = new PrintWriter(new OutputStreamWriter(response.getOutputStream(), "UTF8"), true);
}
// handle query
if ((query == null) && (q != null)) {
query = q;
} else {
if ((query != null) && (q == null)) {
q = query;
} else {
operation = null;
}
}
// handle operation
if (operation == null) {
if (query != null) {
operation = "searchRetrieve";
} else {
operation = "explain";
}
}
// searchRetrieve
if (operation.equalsIgnoreCase("searchRetrieve")) {
if (query == null) {
operation = "explain";
log.debug(operation + ":" + query);
}
}
// start talking back.
String[] sq = {""};
String solrquery = "";
// facet
String facet = null;
List<FacetField> fct = null;
if (request.getParameter("facet") != null) {
facet = request.getParameter("facet");
log.debug("facet : " + facet);
}
if (operation == null) {
operation = "searchretrieve";
} else { // explain response
if (operation.equalsIgnoreCase("explain")) {
log.debug("operation = explain");
out.write("<srw:explainResponse xmlns:srw=\"http://www.loc.gov/zing/srw/\">");
out.write("</srw:explainResponse>");
} else { // DEBUG routine
operation = "searchretrieve";
String triplequery = null;
if (query.matches(".*?\\[.+?\\].*?")) { // New symantic syntax
triplequery = symantic_query(query);
query = query.split("\\[")[0] + " " + triplequery;
log.fatal(triplequery);
solrquery = CQLtoLucene.translate(query, log, config);
} else {
solrquery = CQLtoLucene.translate(query, log, config);
}
log.debug(solrquery);
if (debug == true) {
response.setContentType(HTML_RESPONSE_HEADER);
out = new PrintWriter(new OutputStreamWriter(response.getOutputStream(), "UTF8"), true);
out.write("<html><body>\n\n");
out.write("'" + remote_ip + "'<br>\n");
out.write("<form action='http://www.kbresearch.nl/kbSRU'>");
out.write("<input type=text name=q value='" + query + "' size=120>");
out.write("<input type=hidden name=debug value=True>");
out.write("<input type=submit>");
out.write("<table border=1><tr><td>");
out.write("q</td><td>" + query + "</td></tr><tr>");
out.write("<td>query out</td><td>" + URLDecoder.decode(solrquery) + "</td></tr>");
out.write(
"<tr><td>SOLR_URL</td><td> <a href='"
+ this.config.getProperty(
"collection." + x_collection.toLowerCase() + ".solr_baseurl")
+ "/?q="
+ solrquery
+ "'>"
+ this.config.getProperty(
"collection." + x_collection.toLowerCase() + ".solr_baseurl")
+ "/select/?q="
+ solrquery
+ "</a><br>"
+ this.config.getProperty("solr_url")
+ solrquery
+ "</td></tr>");
out.write(
"<b>SOLR_QUERY</b> : <BR> <iframe width=900 height=400 src='"
+ this.config.getProperty(
"collection." + x_collection.toLowerCase() + ".solr_baseurl")
+ "/../?q="
+ solrquery
+ "'></iframe><BR>");
out.write(
"<b>SRU_QUERY</b> : <BR> <a href="
+ this.config.getProperty("baseurl")
+ "?q="
+ query
+ "'>"
+ this.config.getProperty("baseurl")
+ "?q="
+ query
+ "</a><br><iframe width=901 height=400 src='http://www.kbresearch.nl/kbSRU/?q="
+ query
+ "'></iframe><BR>");
out.write(
"<br><b>JSRU_QUERY</b> : <BR><a href='http://jsru.kb.nl/sru/?query="
+ query
+ "&x-collection="
+ x_collection
+ "'>http://jsru.kb.nl/sru/?query="
+ query
+ "&x-collection=GGC</a><br><iframe width=900 height=400 src='http://jsru.kb.nl/sru/?query="
+ query
+ "&x-collection=GGC'></iframe>");
} else { // XML SearchRetrieve response
String url =
this.config.getProperty("collection." + x_collection.toLowerCase() + ".solr_baseurl");
String buffer = "";
CommonsHttpSolrServer server = null;
server = new CommonsHttpSolrServer(url);
log.fatal("URSING " + url);
server.setParser(new XMLResponseParser());
int numfound = 0;
try {
SolrQuery do_query = new SolrQuery();
do_query.setQuery(solrquery);
do_query.setRows(maximumRecords);
do_query.setStart(startRecord);
if ((sortKeys != null) && (sortKeys.length() > 1)) {
if (sortOrder != null) {
if (sortOrder.equals("asc")) {
do_query.setSortField(sortKeys, SolrQuery.ORDER.asc);
}
if (sortOrder.equals("desc")) {
do_query.setSortField(sortKeys, SolrQuery.ORDER.desc);
}
} else {
for (String str : sortKeys.trim().split(",")) {
str = str.trim();
if (str.length() > 1) {
if (str.equals("date")) {
do_query.setSortField("date_date", SolrQuery.ORDER.desc);
log.debug("SORTORDERDEBUG | DATE! " + str + " | ");
break;
} else {
do_query.setSortField(str + "_str", SolrQuery.ORDER.asc);
log.debug("SORTORDERDEBUG | " + str + " | ");
break;
}
}
}
}
}
if (facet != null) {
if (facet.indexOf(",") > 1) {
for (String str : facet.split(",")) {
if (str.indexOf("date") > 1) {
do_query.addFacetField(str);
} else {
do_query.addFacetField(str);
}
// do_query.setParam("facet.method", "enum");
}
// q.setFacetSort(false);
} else {
do_query.addFacetField(facet);
}
do_query.setFacet(true);
do_query.setFacetMinCount(1);
do_query.setFacetLimit(-1);
}
log.fatal(solrquery);
QueryResponse rsp = null;
boolean do_err = false;
boolean do_sugg = false;
SolrDocumentList sdl = null;
String diag = "";
StringBuffer suggest = new StringBuffer("");
String content = "1";
SolrQuery spellq = do_query;
try {
rsp = server.query(do_query);
} catch (SolrServerException e) {
String header = this.SRW_HEADER.replaceAll("\\$numberOfRecords", "0");
out.write(header);
diag = this.SRW_DIAG.replaceAll("\\$error", e.getMessage());
do_err = true;
rsp = null;
}
log.fatal("query done..");
if (!(do_err)) { // XML dc response
SolrDocumentList docs = rsp.getResults();
numfound = (int) docs.getNumFound();
int count = startRecord;
String header =
this.SRW_HEADER.replaceAll("\\$numberOfRecords", Integer.toString(numfound));
out.write(header);
out.write("<srw:records>");
Iterator<SolrDocument> iter = rsp.getResults().iterator();
while (iter.hasNext()) {
count += 1;
if (recordSchema.equalsIgnoreCase("dc")) {
SolrDocument resultDoc = iter.next();
content = (String) resultDoc.getFieldValue("id");
out.write("<srw:record>");
out.write("<srw:recordPacking>xml</srw:recordPacking>");
out.write("<srw:recordSchema>info:srw/schema/1/dc-v1.1</srw:recordSchema>");
out.write(
"<srw:recordData xmlns:srw_dc=\"info:srw/schema/1/dc-v1.1\" xmlns:mods=\"http://www.loc.gov/mods\" xmlns:dcterms=\"http://purl.org/dc/terms/\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xmlns:dcx=\"http://krait.kb.nl/coop/tel/handbook/telterms.html\" xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:marcrel=\"http://www.loc.gov/loc.terms/relators/OTH\" xmlns:facets=\"info:srw/extension/4/facets\" >");
StringBuffer result = new StringBuffer("");
construct_lucene_dc(result, resultDoc);
out.write(result.toString());
out.write("</srw:recordData>");
out.write(
"<srw:recordPosition>" + Integer.toString(count) + "</srw:recordPosition>");
out.write("</srw:record>");
}
if (recordSchema.equalsIgnoreCase("solr")) {
SolrDocument resultDoc = iter.next();
content = (String) resultDoc.getFieldValue("id");
out.write("<srw:record>");
out.write("<srw:recordPacking>xml</srw:recordPacking>");
out.write("<srw:recordSchema>info:srw/schema/1/solr</srw:recordSchema>");
out.write("<srw:recordData xmlns:expand=\"http://www.kbresearch.nl/expand\">");
StringBuffer result = new StringBuffer("");
construct_lucene_solr(result, resultDoc);
out.write(result.toString());
out.write("</srw:recordData>");
out.write(
"<srw:recordPosition>" + Integer.toString(count) + "</srw:recordPosition>");
out.write("</srw:record>");
}
if (recordSchema.equalsIgnoreCase("dcx")) { // XML dcx response
out.write("<srw:record>");
out.write("<srw:recordPacking>xml</srw:recordPacking>");
out.write("<srw:recordSchema>info:srw/schema/1/dc-v1.1</srw:recordSchema>");
out.write(
"<srw:recordData><srw_dc:dc xmlns:srw_dc=\"info:srw/schema/1/dc-v1.1\" xmlns:mods=\"http://www.loc.gov/mods\" xmlns:dcterms=\"http://purl.org/dc/terms/\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xmlns:dcx=\"http://krait.kb.nl/coop/tel/handbook/telterms.html\" xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:marcrel=\"http://www.loc.gov/marc.relators/\" xmlns:expand=\"http://www.kbresearch.nl/expand\" xmlns:skos=\"http://www.w3.org/2004/02/skos/core#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\" >");
SolrDocument resultDoc = iter.next();
content = (String) resultDoc.getFieldValue("id");
String dcx_data =
helpers.getOAIdcx(
"http://services.kb.nl/mdo/oai?verb=GetRecord&identifier=" + content,
log);
if (x_collection.equalsIgnoreCase("ggc-thes")) {
dcx_data =
helpers.getOAIdcx(
"http://serviceso.kb.nl/mdo/oai?verb=GetRecord&identifier=" + content,
log);
}
if (!(dcx_data.length() == 0)) {
out.write(dcx_data);
} else {
// Should not do this!!
out.write("<srw:record>");
out.write("<srw:recordPacking>xml</srw:recordPacking>");
out.write("<srw:recordSchema>info:srw/schema/1/dc-v1.1</srw:recordSchema>");
out.write(
"<srw:recordData xmlns:srw_dc=\"info:srw/schema/1/dc-v1.1\" xmlns:mods=\"http://www.loc.gov/mods\" xmlns:dcterms=\"http://purl.org/dc/terms/\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xmlns:dcx=\"http://krait.kb.nl/coop/tel/handbook/telterms.html\" xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:marcrel=\"http://www.loc.gov/loc.terms/relators/OTH\" >");
StringBuffer result = new StringBuffer("");
construct_lucene_dc(result, resultDoc);
out.write(result.toString());
out.write("</srw:recordData>");
out.write(
"<srw:recordPosition>" + Integer.toString(count) + "</srw:recordPosition>");
out.write("</srw:record>");
}
out.write("</srw_dc:dc>");
StringBuffer expand_data;
boolean expand = false;
if (content.startsWith("GGC-THES:AC:")) {
String tmp_content = "";
tmp_content = content.replaceFirst("GGC-THES:AC:", "");
log.fatal("calling get");
expand_data =
new StringBuffer(
helpers.getExpand(
"http://www.kbresearch.nl/general/lod_new/get/"
+ tmp_content
+ "?format=rdf",
log));
log.fatal("get finini");
if (expand_data.toString().length() > 4) {
out.write(
"<srw_dc:expand xmlns:srw_dc=\"info:srw/schema/1/dc-v1.1\" xmlns:expand=\"http://www.kbresearch.nl/expand\" xmlns:skos=\"http://www.w3.org/2004/02/skos/core#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\" >");
out.write(expand_data.toString());
expand = true;
}
} else {
expand_data =
new StringBuffer(
helpers.getExpand(
"http://www.kbresearch.nl/ANP.cgi?q=" + content, log));
if (expand_data.toString().length() > 0) {
if (!expand) {
out.write(
"<srw_dc:expand xmlns:srw_dc=\"info:srw/schema/1/dc-v1.1\" xmlns:expand=\"http://www.kbresearch.nl/expand\" xmlns:skos=\"http://www.w3.org/2004/02/skos/core#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\" >");
expand = true;
}
out.write(expand_data.toString());
}
}
if (expand) {
out.write("</srw_dc:expand>");
}
out.write("</srw:recordData>");
out.write(
"<srw:recordPosition>" + Integer.toString(count) + "</srw:recordPosition>");
out.write("</srw:record>");
}
}
}
if ((do_err) || (numfound == 0)) {
log.fatal("I haz suggestions");
try {
spellq.setParam("spellcheck", true);
spellq.setQueryType("/spell");
server = new CommonsHttpSolrServer(url);
rsp = server.query(spellq);
sdl = rsp.getResults();
SpellCheckResponse spell;
spell = rsp.getSpellCheckResponse();
List<SpellCheckResponse.Suggestion> suggestions = spell.getSuggestions();
if (suggestions.isEmpty() == false) {
suggest.append("<srw:extraResponseData>");
suggest.append("<suggestions>");
for (SpellCheckResponse.Suggestion sugg : suggestions) {
suggest.append("<suggestionfor>" + sugg.getToken() + "</suggestionfor>");
for (String item : sugg.getSuggestions()) {
suggest.append("<suggestion>" + item + "</suggestion>");
}
suggest.append("</suggestions>");
suggest.append("</srw:extraResponseData>");
}
do_sugg = true;
}
} catch (Exception e) {
rsp = null;
// log.fatal(e.toString());
}
;
}
;
if (!do_err) {
if (facet != null) {
try {
fct = rsp.getFacetFields();
out.write("<srw:facets>");
for (String str : facet.split(",")) {
out.write("<srw:facet>");
out.write("<srw:facetType>");
out.write(str);
out.write("</srw:facetType>");
for (FacetField f : fct) {
log.debug(f.getName());
// if (f.getName().equals(str+"_str") || (f.getName().equals(str+"_date")) ) {
List<FacetField.Count> facetEnties = f.getValues();
for (FacetField.Count fcount : facetEnties) {
out.write("<srw:facetValue>");
out.write("<srw:valueString>");
out.write(helpers.xmlEncode(fcount.getName()));
out.write("</srw:valueString>");
out.write("<srw:count>");
out.write(Double.toString(fcount.getCount()));
out.write("</srw:count>");
out.write("</srw:facetValue>");
// }
}
}
out.write("</srw:facet>");
}
out.write("</srw:facets>");
startRecord += 1;
} catch (Exception e) {
}
// log.fatal(e.toString()); }
}
} else {
out.write(diag);
}
out.write("</srw:records>"); // SearchRetrieve response footer
String footer = this.SRW_FOOTER.replaceAll("\\$query", helpers.xmlEncode(query));
footer = footer.replaceAll("\\$startRecord", (startRecord).toString());
footer = footer.replaceAll("\\$maximumRecords", maximumRecords.toString());
footer = footer.replaceAll("\\$recordSchema", recordSchema);
if (do_sugg) {
out.write(suggest.toString());
}
out.write(footer);
} catch (MalformedURLException e) {
out.write(e.getMessage());
} catch (IOException e) {
out.write("TO ERR is Human");
}
}
}
}
out.close();
}
/** Integrate spelling suggestions from the various shards in a distributed environment. */
public SpellingResult mergeSuggestions(
SpellCheckMergeData mergeData, int numSug, int count, boolean extendedResults) {
float min = 0.5f;
try {
min = getAccuracy();
} catch (UnsupportedOperationException uoe) {
// just use .5 as a default
}
StringDistance sd = null;
try {
sd = getStringDistance() == null ? new LevensteinDistance() : getStringDistance();
} catch (UnsupportedOperationException uoe) {
sd = new LevensteinDistance();
}
SpellingResult result = new SpellingResult();
for (Map.Entry<String, HashSet<String>> entry : mergeData.origVsSuggested.entrySet()) {
String original = entry.getKey();
// Only use this suggestion if all shards reported it as misspelled,
// unless it was not a term original to the user's query
// (WordBreakSolrSpellChecker can add new terms to the response, and we want to keep these)
Integer numShards = mergeData.origVsShards.get(original);
if (numShards < mergeData.totalNumberShardResponses
&& mergeData.isOriginalToQuery(original)) {
continue;
}
HashSet<String> suggested = entry.getValue();
SuggestWordQueue sugQueue = new SuggestWordQueue(numSug);
for (String suggestion : suggested) {
SuggestWord sug = mergeData.suggestedVsWord.get(suggestion);
sug.score = sd.getDistance(original, sug.string);
if (sug.score < min) continue;
sugQueue.insertWithOverflow(sug);
if (sugQueue.size() == numSug) {
// if queue full, maintain the minScore score
min = sugQueue.top().score;
}
}
// create token
SpellCheckResponse.Suggestion suggestion = mergeData.origVsSuggestion.get(original);
Token token = new Token(original, suggestion.getStartOffset(), suggestion.getEndOffset());
// get top 'count' suggestions out of 'sugQueue.size()' candidates
SuggestWord[] suggestions = new SuggestWord[Math.min(count, sugQueue.size())];
// skip the first sugQueue.size() - count elements
for (int k = 0; k < sugQueue.size() - count; k++) sugQueue.pop();
// now collect the top 'count' responses
for (int k = Math.min(count, sugQueue.size()) - 1; k >= 0; k--) {
suggestions[k] = sugQueue.pop();
}
if (extendedResults) {
Integer o = mergeData.origVsFreq.get(original);
if (o != null) result.addFrequency(token, o);
for (SuggestWord word : suggestions) result.add(token, word.string, word.freq);
} else {
List<String> words = new ArrayList<>(sugQueue.size());
for (SuggestWord word : suggestions) words.add(word.string);
result.add(token, words);
}
}
return result;
}