|
|
@ -102,6 +102,7 @@ public final class QueryParams {
|
|
|
|
final Bitfield constraint,
|
|
|
|
final Bitfield constraint,
|
|
|
|
final Segment indexSegment,
|
|
|
|
final Segment indexSegment,
|
|
|
|
final RankingProfile ranking) {
|
|
|
|
final RankingProfile ranking) {
|
|
|
|
|
|
|
|
|
|
|
|
if ((queryString.length() == 12) && (Base64Order.enhancedCoder.wellformed(queryString.getBytes()))) {
|
|
|
|
if ((queryString.length() == 12) && (Base64Order.enhancedCoder.wellformed(queryString.getBytes()))) {
|
|
|
|
this.queryString = null;
|
|
|
|
this.queryString = null;
|
|
|
|
this.queryHashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
|
this.queryHashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
@ -164,6 +165,7 @@ public final class QueryParams {
|
|
|
|
final boolean specialRights,
|
|
|
|
final boolean specialRights,
|
|
|
|
final Segment indexSegment,
|
|
|
|
final Segment indexSegment,
|
|
|
|
final RankingProfile ranking) {
|
|
|
|
final RankingProfile ranking) {
|
|
|
|
|
|
|
|
|
|
|
|
this.queryString = queryString;
|
|
|
|
this.queryString = queryString;
|
|
|
|
this.queryHashes = queryHashes;
|
|
|
|
this.queryHashes = queryHashes;
|
|
|
|
this.excludeHashes = excludeHashes;
|
|
|
|
this.excludeHashes = excludeHashes;
|
|
|
@ -177,7 +179,7 @@ public final class QueryParams {
|
|
|
|
this.urlMask = Pattern.compile(urlMask);
|
|
|
|
this.urlMask = Pattern.compile(urlMask);
|
|
|
|
this.urlMask_isCatchall = this.urlMask.toString().equals(catchall_pattern.toString());
|
|
|
|
this.urlMask_isCatchall = this.urlMask.toString().equals(catchall_pattern.toString());
|
|
|
|
this.prefer = Pattern.compile(prefer);
|
|
|
|
this.prefer = Pattern.compile(prefer);
|
|
|
|
this.prefer_isMatchnothing = this.prefer.toString().equals(matchnothing_pattern.toString());;
|
|
|
|
this.prefer_isMatchnothing = this.prefer.toString().equals(matchnothing_pattern.toString());
|
|
|
|
assert language != null;
|
|
|
|
assert language != null;
|
|
|
|
this.targetlang = language;
|
|
|
|
this.targetlang = language;
|
|
|
|
this.navigators = navigators;
|
|
|
|
this.navigators = navigators;
|
|
|
@ -228,30 +230,32 @@ public final class QueryParams {
|
|
|
|
|
|
|
|
|
|
|
|
public static HandleSet hashes2Set(final String query) {
|
|
|
|
public static HandleSet hashes2Set(final String query) {
|
|
|
|
final HandleSet keyhashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
|
final HandleSet keyhashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
|
if (query == null) return keyhashes;
|
|
|
|
if (query != null) {
|
|
|
|
for (int i = 0; i < (query.length() / Word.commonHashLength); i++) try {
|
|
|
|
for (int i = 0; i < (query.length() / Word.commonHashLength); i++) try {
|
|
|
|
keyhashes.put(query.substring(i * Word.commonHashLength, (i + 1) * Word.commonHashLength).getBytes());
|
|
|
|
keyhashes.put(query.substring(i * Word.commonHashLength, (i + 1) * Word.commonHashLength).getBytes());
|
|
|
|
} catch (RowSpaceExceededException e) {
|
|
|
|
} catch (RowSpaceExceededException e) {
|
|
|
|
Log.logException(e);
|
|
|
|
Log.logException(e);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
return keyhashes;
|
|
|
|
return keyhashes;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public static HandleSet hashes2Handles(final String query) {
|
|
|
|
public static HandleSet hashes2Handles(final String query) {
|
|
|
|
final HandleSet keyhashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
|
final HandleSet keyhashes = new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0);
|
|
|
|
if (query == null) return keyhashes;
|
|
|
|
if (query != null) {
|
|
|
|
for (int i = 0; i < (query.length() / Word.commonHashLength); i++) try {
|
|
|
|
for (int i = 0; i < (query.length() / Word.commonHashLength); i++) try {
|
|
|
|
keyhashes.put(query.substring(i * Word.commonHashLength, (i + 1) * Word.commonHashLength).getBytes());
|
|
|
|
keyhashes.put(query.substring(i * Word.commonHashLength, (i + 1) * Word.commonHashLength).getBytes());
|
|
|
|
} catch (RowSpaceExceededException e) {
|
|
|
|
} catch (RowSpaceExceededException e) {
|
|
|
|
Log.logException(e);
|
|
|
|
Log.logException(e);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
return keyhashes;
|
|
|
|
return keyhashes;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public static String hashSet2hashString(final HandleSet hashes) {
|
|
|
|
public static String hashSet2hashString(final HandleSet hashes) {
|
|
|
|
final byte[] bb = new byte[hashes.size() * Word.commonHashLength];
|
|
|
|
final byte[] bb = new byte[hashes.size() * Word.commonHashLength];
|
|
|
|
int p = 0;
|
|
|
|
int p = 0;
|
|
|
|
for (byte[] b : hashes) {
|
|
|
|
for (final byte[] b : hashes) {
|
|
|
|
assert b.length == Word.commonHashLength : "hash = " + new String(b);
|
|
|
|
assert b.length == Word.commonHashLength : "hash = " + new String(b);
|
|
|
|
System.arraycopy(b, 0, bb, p, Word.commonHashLength);
|
|
|
|
System.arraycopy(b, 0, bb, p, Word.commonHashLength);
|
|
|
|
p += Word.commonHashLength;
|
|
|
|
p += Word.commonHashLength;
|
|
|
@ -284,9 +288,12 @@ public final class QueryParams {
|
|
|
|
* @return true if the query matches with the given text
|
|
|
|
* @return true if the query matches with the given text
|
|
|
|
*/
|
|
|
|
*/
|
|
|
|
public final boolean matchesText(final String text) {
|
|
|
|
public final boolean matchesText(final String text) {
|
|
|
|
|
|
|
|
boolean ret = false;
|
|
|
|
final HandleSet wordhashes = Word.words2hashesHandles(Condenser.getWords(text).keySet());
|
|
|
|
final HandleSet wordhashes = Word.words2hashesHandles(Condenser.getWords(text).keySet());
|
|
|
|
if (SetTools.anymatch(wordhashes, this.excludeHashes)) return false;
|
|
|
|
if (!SetTools.anymatch(wordhashes, this.excludeHashes)) {
|
|
|
|
return SetTools.totalInclusion(this.queryHashes, wordhashes);
|
|
|
|
ret = SetTools.totalInclusion(this.queryHashes, wordhashes);
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
protected static final boolean anymatch(final String text, final HandleSet keyhashes) {
|
|
|
|
protected static final boolean anymatch(final String text, final HandleSet keyhashes) {
|
|
|
@ -305,13 +312,15 @@ public final class QueryParams {
|
|
|
|
final TreeSet<String> exclude = new TreeSet<String>(NaturalOrder.naturalComparator);
|
|
|
|
final TreeSet<String> exclude = new TreeSet<String>(NaturalOrder.naturalComparator);
|
|
|
|
final TreeSet<String> fullquery = new TreeSet<String>(NaturalOrder.naturalComparator);
|
|
|
|
final TreeSet<String> fullquery = new TreeSet<String>(NaturalOrder.naturalComparator);
|
|
|
|
|
|
|
|
|
|
|
|
if ((querystring == null) || (querystring.length() == 0)) return new TreeSet[]{query, exclude, fullquery};
|
|
|
|
if ((querystring != null) && (querystring.length() > 0)) {
|
|
|
|
|
|
|
|
|
|
|
|
// convert Umlaute
|
|
|
|
// convert Umlaute
|
|
|
|
querystring = AbstractScraper.stripAll(querystring).toLowerCase().trim();
|
|
|
|
querystring = AbstractScraper.stripAll(querystring).toLowerCase().trim();
|
|
|
|
int c;
|
|
|
|
int c;
|
|
|
|
for (int i = 0; i < seps.length(); i++) {
|
|
|
|
for (int i = 0; i < seps.length(); i++) {
|
|
|
|
while ((c = querystring.indexOf(seps.charAt(i))) >= 0) { querystring = querystring.substring(0, c) + (((c + 1) < querystring.length()) ? (" " + querystring.substring(c + 1)) : ""); }
|
|
|
|
while ((c = querystring.indexOf(seps.charAt(i))) >= 0) {
|
|
|
|
|
|
|
|
querystring = querystring.substring(0, c) + (((c + 1) < querystring.length()) ? (" " + querystring.substring(c + 1)) : "");
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
String s;
|
|
|
|
String s;
|
|
|
@ -325,23 +334,27 @@ public final class QueryParams {
|
|
|
|
while ((c = a[i].indexOf('-')) >= 0) {
|
|
|
|
while ((c = a[i].indexOf('-')) >= 0) {
|
|
|
|
s = a[i].substring(0, c);
|
|
|
|
s = a[i].substring(0, c);
|
|
|
|
l = s.length();
|
|
|
|
l = s.length();
|
|
|
|
if (l >= Condenser.wordminsize) query.add(s);
|
|
|
|
if (l >= Condenser.wordminsize) {query.add(s);}
|
|
|
|
if (l > 0) fullquery.add(s);
|
|
|
|
if (l > 0) {fullquery.add(s);}
|
|
|
|
a[i] = a[i].substring(c + 1);
|
|
|
|
a[i] = a[i].substring(c + 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
l = a[i].length();
|
|
|
|
l = a[i].length();
|
|
|
|
if (l >= Condenser.wordminsize) query.add(a[i]);
|
|
|
|
if (l >= Condenser.wordminsize) {query.add(a[i]);}
|
|
|
|
if (l > 0) fullquery.add(a[i]);
|
|
|
|
if (l > 0) {fullquery.add(a[i]);}
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return new TreeSet[]{query, exclude, fullquery};
|
|
|
|
return new TreeSet[]{query, exclude, fullquery};
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public String queryString(final boolean encodeHTML) {
|
|
|
|
public String queryString(final boolean encodeHTML) {
|
|
|
|
|
|
|
|
final String ret;
|
|
|
|
if (encodeHTML){
|
|
|
|
if (encodeHTML){
|
|
|
|
return CharacterCoding.unicode2html(this.queryString, true);
|
|
|
|
ret =CharacterCoding.unicode2html(this.queryString, true);
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
|
|
|
ret = this.queryString;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return this.queryString;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public TreeSet<String>[] queryWords() {
|
|
|
|
public TreeSet<String>[] queryWords() {
|
|
|
@ -356,9 +369,9 @@ public final class QueryParams {
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
public final Map<MultiProtocolURI, String> separateMatches(Map<MultiProtocolURI, String> links) {
|
|
|
|
public final Map<MultiProtocolURI, String> separateMatches(final Map<MultiProtocolURI, String> links) {
|
|
|
|
Map<MultiProtocolURI, String> matcher = new HashMap<MultiProtocolURI, String>();
|
|
|
|
final Map<MultiProtocolURI, String> matcher = new HashMap<MultiProtocolURI, String>();
|
|
|
|
Iterator <Map.Entry<MultiProtocolURI, String>> i = links.entrySet().iterator();
|
|
|
|
final Iterator <Map.Entry<MultiProtocolURI, String>> i = links.entrySet().iterator();
|
|
|
|
Map.Entry<MultiProtocolURI, String> entry;
|
|
|
|
Map.Entry<MultiProtocolURI, String> entry;
|
|
|
|
MultiProtocolURI url;
|
|
|
|
MultiProtocolURI url;
|
|
|
|
String anchorText;
|
|
|
|
String anchorText;
|
|
|
@ -375,23 +388,40 @@ public final class QueryParams {
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public String id(final boolean anonymized) {
|
|
|
|
public String id(final boolean anonymized) {
|
|
|
|
|
|
|
|
final String asterisk = "*";
|
|
|
|
|
|
|
|
|
|
|
|
// generate a string that identifies a search so results can be re-used in a cache
|
|
|
|
// generate a string that identifies a search so results can be re-used in a cache
|
|
|
|
String context =
|
|
|
|
final StringBuilder context = new StringBuilder();
|
|
|
|
"*" + this.domType +
|
|
|
|
context.append(asterisk);
|
|
|
|
"*" + this.contentdom +
|
|
|
|
context.append(this.domType);
|
|
|
|
"*" + this.zonecode +
|
|
|
|
context.append(asterisk);
|
|
|
|
"*" + new String(Word.word2hash(this.ranking.toExternalString())) +
|
|
|
|
context.append(this.contentdom);
|
|
|
|
"*" + this.prefer +
|
|
|
|
context.append(asterisk);
|
|
|
|
"*" + this.urlMask +
|
|
|
|
context.append(this.zonecode);
|
|
|
|
"*" + this.sitehash +
|
|
|
|
context.append(asterisk);
|
|
|
|
"*" + this.authorhash +
|
|
|
|
context.append(Word.word2hash(this.ranking.toExternalString()));
|
|
|
|
"*" + this.targetlang +
|
|
|
|
context.append(asterisk);
|
|
|
|
"*" + this.constraint +
|
|
|
|
context.append(this.prefer);
|
|
|
|
"*" + this.maxDistance;
|
|
|
|
context.append(asterisk);
|
|
|
|
if (anonymized)
|
|
|
|
context.append(this.urlMask);
|
|
|
|
return anonymizedQueryHashes(this.queryHashes) + "-" + anonymizedQueryHashes(this.excludeHashes) + context;
|
|
|
|
context.append(asterisk);
|
|
|
|
else
|
|
|
|
context.append(this.sitehash);
|
|
|
|
return hashSet2hashString(this.queryHashes) + "-" + hashSet2hashString(this.excludeHashes) + context;
|
|
|
|
context.append(asterisk);
|
|
|
|
|
|
|
|
context.append(this.authorhash);
|
|
|
|
|
|
|
|
context.append(asterisk);
|
|
|
|
|
|
|
|
context.append(this.targetlang);
|
|
|
|
|
|
|
|
context.append(asterisk);
|
|
|
|
|
|
|
|
context.append(this.constraint);
|
|
|
|
|
|
|
|
context.append(asterisk);
|
|
|
|
|
|
|
|
context.append(this.maxDistance);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
final String ret;
|
|
|
|
|
|
|
|
if (anonymized) {
|
|
|
|
|
|
|
|
ret = anonymizedQueryHashes(this.queryHashes) + "-" + anonymizedQueryHashes(this.excludeHashes) + context;
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
|
|
|
ret = hashSet2hashString(this.queryHashes) + "-" + hashSet2hashString(this.excludeHashes) + context;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
/**
|
|
|
@ -403,19 +433,64 @@ public final class QueryParams {
|
|
|
|
* @param addToQuery
|
|
|
|
* @param addToQuery
|
|
|
|
* @return
|
|
|
|
* @return
|
|
|
|
*/
|
|
|
|
*/
|
|
|
|
public static String navurl(String ext, final int page, final int display, final QueryParams theQuery, final String originalUrlMask, String addToQuery, String nav) {
|
|
|
|
public static String navurl(final String ext, final int page, final int display, final QueryParams theQuery, final String originalUrlMask, final String addToQuery, final String nav) {
|
|
|
|
return
|
|
|
|
|
|
|
|
"/yacysearch." + ext + "?display=" + display +
|
|
|
|
final String ampersand = "&";
|
|
|
|
"&query=" + theQuery.queryString(true).replace(' ', '+') + ((addToQuery == null) ? "" : "+" + addToQuery) +
|
|
|
|
|
|
|
|
"&maximumRecords="+ theQuery.displayResults() +
|
|
|
|
final StringBuilder sb = new StringBuilder();
|
|
|
|
"&startRecord=" + (page * theQuery.displayResults()) +
|
|
|
|
sb.append("/yacysearch.");
|
|
|
|
"&resource=" + ((theQuery.isLocal()) ? "local" : "global") +
|
|
|
|
sb.append(ext);
|
|
|
|
"&verify=" + (theQuery.snippetCacheStrategy.mustBeOffline() ? "false" : "true") +
|
|
|
|
sb.append("?display=");
|
|
|
|
"&nav=" + nav +
|
|
|
|
sb.append(display);
|
|
|
|
"&urlmaskfilter=" + originalUrlMask +
|
|
|
|
|
|
|
|
"&prefermaskfilter=" + theQuery.prefer +
|
|
|
|
sb.append(ampersand);
|
|
|
|
"&cat=href&constraint=" + ((theQuery.constraint == null) ? "" : theQuery.constraint.exportB64()) +
|
|
|
|
sb.append("query=");
|
|
|
|
"&contentdom=" + theQuery.contentdom() +
|
|
|
|
sb.append(theQuery.queryString(true).replace(' ', '+'));
|
|
|
|
"&former=" + theQuery.queryString(true);
|
|
|
|
sb.append((addToQuery == null) ? "" : "+" + addToQuery);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("maximumRecords=");
|
|
|
|
|
|
|
|
sb.append(theQuery.displayResults());
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("startRecord=");
|
|
|
|
|
|
|
|
sb.append(page * theQuery.displayResults());
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("resource=");
|
|
|
|
|
|
|
|
sb.append((theQuery.isLocal()) ? "local" : "global");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("verify=");
|
|
|
|
|
|
|
|
sb.append(theQuery.snippetCacheStrategy.mustBeOffline() ? "false" : "true");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("nav=");
|
|
|
|
|
|
|
|
sb.append(nav);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("urlmaskfilter=");
|
|
|
|
|
|
|
|
sb.append(originalUrlMask);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("prefermaskfilter=");
|
|
|
|
|
|
|
|
sb.append(theQuery.prefer);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("cat=href");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("constraint=");
|
|
|
|
|
|
|
|
sb.append((theQuery.constraint == null) ? "" : theQuery.constraint.exportB64());
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("contentdom=");
|
|
|
|
|
|
|
|
sb.append(theQuery.contentdom());
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sb.append(ampersand);
|
|
|
|
|
|
|
|
sb.append("former=");
|
|
|
|
|
|
|
|
sb.append(theQuery.queryString(true));
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return sb.toString();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|