AnalyzerUtilpublic class AnalyzerUtil extends Object | Various fulltext analysis utilities avoiding redundant code in several
classes. |
| Fields Summary |
|---|
| private static final Pattern | PARAGRAPHS(Line terminator followed by zero or more whitespace) two or more times | | private static final Pattern | SENTENCESDivides text into sentences; Includes inverted spanish exclamation and question mark |
| Constructors Summary |
|---|
private AnalyzerUtil()
|
| Methods Summary |
|---|
| public static org.apache.lucene.analysis.Analyzer | getLoggingAnalyzer(org.apache.lucene.analysis.Analyzer child, java.io.PrintStream log, java.lang.String logName)Returns a simple analyzer wrapper that logs all tokens produced by the
underlying child analyzer to the given log stream (typically System.err);
Otherwise behaves exactly like the child analyzer, delivering the very
same tokens; useful for debugging purposes on custom indexing and/or
querying.
if (child == null)
throw new IllegalArgumentException("child analyzer must not be null");
if (log == null)
throw new IllegalArgumentException("logStream must not be null");
return new Analyzer() {
public TokenStream tokenStream(final String fieldName, Reader reader) {
return new TokenFilter(child.tokenStream(fieldName, reader)) {
private int position = -1;
public Token next() throws IOException {
Token token = input.next(); // from filter super class
log.println(toString(token));
return token;
}
private String toString(Token token) {
if (token == null) return "[" + logName + ":EOS:" + fieldName + "]\n";
position += token.getPositionIncrement();
return "[" + logName + ":" + position + ":" + fieldName + ":"
+ token.termText() + ":" + token.startOffset()
+ "-" + token.endOffset() + ":" + token.type()
+ "]";
}
};
}
};
| | public static org.apache.lucene.analysis.Analyzer | getMaxTokenAnalyzer(org.apache.lucene.analysis.Analyzer child, int maxTokens)Returns an analyzer wrapper that returns at most the first
maxTokens tokens from the underlying child analyzer,
ignoring all remaining tokens.
if (child == null)
throw new IllegalArgumentException("child analyzer must not be null");
if (maxTokens < 0)
throw new IllegalArgumentException("maxTokens must not be negative");
if (maxTokens == Integer.MAX_VALUE)
return child; // no need to wrap
return new Analyzer() {
public TokenStream tokenStream(String fieldName, Reader reader) {
return new TokenFilter(child.tokenStream(fieldName, reader)) {
private int todo = maxTokens;
public Token next() throws IOException {
return --todo >= 0 ? input.next() : null;
}
};
}
};
| | public static java.lang.String[] | getMostFrequentTerms(org.apache.lucene.analysis.Analyzer analyzer, java.lang.String text, int limit)Returns (frequency:term) pairs for the top N distinct terms (aka words),
sorted descending by frequency (and ascending by term, if tied).
Example XQuery:
declare namespace util = "java:org.apache.lucene.index.memory.AnalyzerUtil";
declare namespace analyzer = "java:org.apache.lucene.index.memory.PatternAnalyzer";
for $pair in util:get-most-frequent-terms(
analyzer:EXTENDED_ANALYZER(), doc("samples/shakespeare/othello.xml"), 10)
return <word word="{substring-after($pair, ':')}" frequency="{substring-before($pair, ':')}"/>
if (analyzer == null)
throw new IllegalArgumentException("analyzer must not be null");
if (text == null)
throw new IllegalArgumentException("text must not be null");
if (limit <= 0) limit = Integer.MAX_VALUE;
// compute frequencies of distinct terms
HashMap map = new HashMap();
TokenStream stream = analyzer.tokenStream("", new StringReader(text));
try {
Token token;
while ((token = stream.next()) != null) {
MutableInteger freq = (MutableInteger) map.get(token.termText());
if (freq == null) {
freq = new MutableInteger(1);
map.put(token.termText(), freq);
} else {
freq.setValue(freq.intValue() + 1);
}
}
} catch (IOException e) {
throw new RuntimeException(e);
} finally {
try {
stream.close();
} catch (IOException e2) {
throw new RuntimeException(e2);
}
}
// sort by frequency, text
Map.Entry[] entries = new Map.Entry[map.size()];
map.entrySet().toArray(entries);
Arrays.sort(entries, new Comparator() {
public int compare(Object o1, Object o2) {
Map.Entry e1 = (Map.Entry) o1;
Map.Entry e2 = (Map.Entry) o2;
int f1 = ((MutableInteger) e1.getValue()).intValue();
int f2 = ((MutableInteger) e2.getValue()).intValue();
if (f2 - f1 != 0) return f2 - f1;
String s1 = (String) e1.getKey();
String s2 = (String) e2.getKey();
return s1.compareTo(s2);
}
});
// return top N entries
int size = Math.min(limit, entries.length);
String[] pairs = new String[size];
for (int i=0; i < size; i++) {
pairs[i] = entries[i].getValue() + ":" + entries[i].getKey();
}
return pairs;
| | public static java.lang.String[] | getParagraphs(java.lang.String text, int limit)Returns at most the first N paragraphs of the given text. Delimiting
characters are excluded from the results. Each returned paragraph is
whitespace-trimmed via String.trim(), potentially an empty string.
return tokenize(PARAGRAPHS, text, limit);
| | public static org.apache.lucene.analysis.Analyzer | getPorterStemmerAnalyzer(org.apache.lucene.analysis.Analyzer child)Returns an English stemming analyzer that stems tokens from the
underlying child analyzer according to the Porter stemming algorithm. The
child analyzer must deliver tokens in lower case for the stemmer to work
properly.
Background: Stemming reduces token terms to their linguistic root form
e.g. reduces "fishing" and "fishes" to "fish", "family" and "families" to
"famili", as well as "complete" and "completion" to "complet". Note that
the root form is not necessarily a meaningful word in itself, and that
this is not a bug but rather a feature, if you lean back and think about
fuzzy word matching for a bit.
See the Lucene contrib packages for stemmers (and stop words) for German,
Russian and many more languages.
if (child == null)
throw new IllegalArgumentException("child analyzer must not be null");
return new Analyzer() {
public TokenStream tokenStream(String fieldName, Reader reader) {
return new PorterStemFilter(
child.tokenStream(fieldName, reader));
// /* PorterStemFilter and SnowballFilter have the same behaviour,
// but PorterStemFilter is much faster. */
// return new org.apache.lucene.analysis.snowball.SnowballFilter(
// child.tokenStream(fieldName, reader), "English");
}
};
| | public static java.lang.String[] | getSentences(java.lang.String text, int limit)Returns at most the first N sentences of the given text. Delimiting
characters are excluded from the results. Each returned sentence is
whitespace-trimmed via String.trim(), potentially an empty string.
// return tokenize(SENTENCES, text, limit); // equivalent but slower
int len = text.length();
if (len == 0) return new String[] { text };
if (limit <= 0) limit = Integer.MAX_VALUE;
// average sentence length heuristic
String[] tokens = new String[Math.min(limit, 1 + len/40)];
int size = 0;
int i = 0;
while (i < len && size < limit) {
// scan to end of current sentence
int start = i;
while (i < len && !isSentenceSeparator(text.charAt(i))) i++;
if (size == tokens.length) { // grow array
String[] tmp = new String[tokens.length << 1];
System.arraycopy(tokens, 0, tmp, 0, size);
tokens = tmp;
}
// add sentence (potentially empty)
tokens[size++] = text.substring(start, i).trim();
// scan to beginning of next sentence
while (i < len && isSentenceSeparator(text.charAt(i))) i++;
}
if (size == tokens.length) return tokens;
String[] results = new String[size];
System.arraycopy(tokens, 0, results, 0, size);
return results;
| | public static org.apache.lucene.analysis.Analyzer | getSynonymAnalyzer(org.apache.lucene.analysis.Analyzer child, SynonymMap synonyms, int maxSynonyms)Returns an analyzer wrapper that wraps the underlying child analyzer's
token stream into a {@link SynonymTokenFilter}.
if (child == null)
throw new IllegalArgumentException("child analyzer must not be null");
if (synonyms == null)
throw new IllegalArgumentException("synonyms must not be null");
if (maxSynonyms < 0)
throw new IllegalArgumentException("maxSynonyms must not be negative");
if (maxSynonyms == 0)
return child; // no need to wrap
return new Analyzer() {
public TokenStream tokenStream(String fieldName, Reader reader) {
return new SynonymTokenFilter(
child.tokenStream(fieldName, reader), synonyms, maxSynonyms);
}
};
| | private static boolean | isSentenceSeparator(char c)
// regex [!\\.\\?\\xA1\\xBF]
switch (c) {
case '!": return true;
case '.": return true;
case '?": return true;
case 0xA1: return true; // spanish inverted exclamation mark
case 0xBF: return true; // spanish inverted question mark
default: return false;
}
| | private static java.lang.String[] | tokenize(java.util.regex.Pattern pattern, java.lang.String text, int limit)
String[] tokens = pattern.split(text, limit);
for (int i=tokens.length; --i >= 0; ) tokens[i] = tokens[i].trim();
return tokens;
|
|
