Packages
Package	Description
org.apache.lucene.analysis.icu	Analysis components based on ICU
org.apache.lucene.analysis.icu.segmentation	Tokenizer that breaks text into words with the Unicode Text Segmentation algorithm.
org.apache.lucene.analysis.icu.tokenattributes	Additional ICU-specific Attributes for text analysis.
org.apache.lucene.collation	Unicode Collation support.
org.apache.lucene.collation.tokenattributes	Custom `AttributeImpl` for indexing collation keys as index terms.

This module exposes functionality from ICU to Apache Lucene. ICU4J is a Java library that enhances Java's internationalization support by improving performance, keeping current with the Unicode Standard, and providing richer APIs.

For an introduction to Lucene's analysis API, see the org.apache.lucene.analysis package documentation.

This module exposes the following functionality:

Text Segmentation: Tokenizes text based on properties and rules defined in Unicode.
Collation: Compare strings according to the conventions and standards of a particular language, region or country.
Normalization: Converts text to a unique, equivalent form.
Case Folding: Removes case distinctions with Unicode's Default Caseless Matching algorithm.
Search Term Folding: Removes distinctions (such as accent marks) between similar characters for a loose or fuzzy search.
Text Transformation: Transforms Unicode text in a context-sensitive fashion: e.g. mapping Traditional to Simplified Chinese

Text Segmentation

Text Segmentation (Tokenization) divides document and query text into index terms (typically words). Unicode provides special properties and rules so that this can be done in a manner that works well with most languages.

Text Segmentation implements the word segmentation specified in Unicode Text Segmentation. Additionally the algorithm can be tailored based on writing system, for example text in the Thai script is automatically delegated to a dictionary-based segmentation algorithm.

Use Cases

As a more thorough replacement for StandardTokenizer that works well for most languages.

Example Usages

Tokenizing multilanguage text

  /**
   * This tokenizer will work well in general for most languages.
   */
  Tokenizer tokenizer = new ICUTokenizer(reader);

Collation

ICUCollationKeyAnalyzer converts each token into its binary CollationKey using the provided Collator, allowing it to be stored as an index term.

ICUCollationKeyAnalyzer depends on ICU4J to produce the CollationKeys.

Use Cases

Efficient sorting of terms in languages that use non-Unicode character orderings. (Lucene Sort using a Locale can be very slow.)
Efficient range queries over fields that contain terms in languages that use non-Unicode character orderings. (Range queries using a Locale can be very slow.)
Effective Locale-specific normalization (case differences, diacritics, etc.). (LowerCaseFilter and ASCIIFoldingFilter provide these services in a generic way that doesn't take into account locale-specific needs.)

Example Usages

Farsi Range Queries

  Collator collator = Collator.getInstance(new ULocale("ar"));
  ICUCollationKeyAnalyzer analyzer = new ICUCollationKeyAnalyzer(collator);
  Path indexPath = Files.createTempDirectory("tempIndex");
  Directory dir = FSDirectory.open(indexPath);
  IndexWriter writer = new IndexWriter(dir, new IndexWriterConfig(analyzer));
  Document doc = new Document();
  doc.add(new Field("content", "\u0633\u0627\u0628", 
                    Field.Store.YES, Field.Index.ANALYZED));
  writer.addDocument(doc);
  writer.close();
  IndexSearcher is = new IndexSearcher(dir, true);

  QueryParser aqp = new QueryParser("content", analyzer);
  aqp.setAnalyzeRangeTerms(true);
    
  // Unicode order would include U+0633 in [ U+062F - U+0698 ], but Farsi
  // orders the U+0698 character before the U+0633 character, so the single
  // indexed Term above should NOT be returned by a ConstantScoreRangeQuery
  // with a Farsi Collator (or an Arabic one for the case when Farsi is not
  // supported).
  ScoreDoc[] result
    = is.search(aqp.parse("[ \u062F TO \u0698 ]"), null, 1000).scoreDocs;
  assertEquals("The index Term should not be included.", 0, result.length);

Danish Sorting

  Analyzer analyzer 
    = new ICUCollationKeyAnalyzer(Collator.getInstance(new ULocale("da", "dk")));
  Path indexPath = Files.createTempDirectory("tempIndex");
  Directory dir = FSDirectory.open(indexPath);
  IndexWriter writer = new IndexWriter(dir, new IndexWriterConfig(analyzer));
  String[] tracer = new String[] { "A", "B", "C", "D", "E" };
  String[] data = new String[] { "HAT", "HUT", "H\u00C5T", "H\u00D8T", "HOT" };
  String[] sortedTracerOrder = new String[] { "A", "E", "B", "D", "C" };
  for (int i = 0 ; i < data.length ; ++i) {
    Document doc = new Document();
    doc.add(new Field("tracer", tracer[i], Field.Store.YES, Field.Index.NO));
    doc.add(new Field("contents", data[i], Field.Store.NO, Field.Index.ANALYZED));
    writer.addDocument(doc);
  }
  writer.close();
  IndexSearcher searcher = new IndexSearcher(dir, true);
  Sort sort = new Sort();
  sort.setSort(new SortField("contents", SortField.STRING));
  Query query = new MatchAllDocsQuery();
  ScoreDoc[] result = searcher.search(query, null, 1000, sort).scoreDocs;
  for (int i = 0 ; i < result.length ; ++i) {
    Document doc = searcher.doc(result[i].doc);
    assertEquals(sortedTracerOrder[i], doc.getValues("tracer")[0]);
  }

Turkish Case Normalization

  Collator collator = Collator.getInstance(new ULocale("tr", "TR"));
  collator.setStrength(Collator.PRIMARY);
  Analyzer analyzer = new ICUCollationKeyAnalyzer(collator);
  Path indexPath = Files.createTempDirectory("tempIndex");
  Directory dir = FSDirectory.open(indexPath);
  IndexWriter writer = new IndexWriter(dir, new IndexWriterConfig(analyzer));
  Document doc = new Document();
  doc.add(new Field("contents", "DIGY", Field.Store.NO, Field.Index.ANALYZED));
  writer.addDocument(doc);
  writer.close();
  IndexSearcher is = new IndexSearcher(dir, true);
  QueryParser parser = new QueryParser("contents", analyzer);
  Query query = parser.parse("d\u0131gy");   // U+0131: dotless i
  ScoreDoc[] result = is.search(query, null, 1000).scoreDocs;
  assertEquals("The index Term should be included.", 1, result.length);

Caveats and Comparisons

WARNING: Make sure you use exactly the same Collator at index and query time -- CollationKeys are only comparable when produced by the same Collator. Since RuleBasedCollators are not independently versioned, it is unsafe to search against stored CollationKeys unless the following are exactly the same (best practice is to store this information with the index and check that they remain the same at query time):

JVM vendor
JVM version, including patch version
The language (and country and variant, if specified) of the Locale used when constructing the collator via Collator.getInstance(java.util.Locale).
The collation strength used - see Collator.setStrength(int)

ICUCollationKeyAnalyzer uses ICU4J's Collator, which makes its version available, thus allowing collation to be versioned independently from the JVM. ICUCollationKeyAnalyzer is also significantly faster and generates significantly shorter keys than CollationKeyAnalyzer. See http://site.icu-project.org/charts/collation-icu4j-sun for key generation timing and key length comparisons between ICU4J and java.text.Collator over several languages.

CollationKeys generated by java.text.Collators are not compatible with those those generated by ICU Collators. Specifically, if you use CollationKeyAnalyzer to generate index terms, do not use ICUCollationKeyAnalyzer on the query side, or vice versa.

Normalization

ICUNormalizer2Filter normalizes term text to a Unicode Normalization Form, so that equivalent forms are standardized to a unique form.

Use Cases

Removing differences in width for Asian-language text.
Standardizing complex text with non-spacing marks so that characters are ordered consistently.

Example Usages

Normalizing text to NFC

  /**
   * Normalizer2 objects are unmodifiable and immutable.
   */
  Normalizer2 normalizer = Normalizer2.getInstance(null, "nfc", Normalizer2.Mode.COMPOSE);
  /**
   * This filter will normalize to NFC.
   */
  TokenStream tokenstream = new ICUNormalizer2Filter(tokenizer, normalizer);

Case Folding

Default caseless matching, or case-folding is more than just conversion to lowercase. For example, it handles cases such as the Greek sigma, so that "Μάϊος" and "ΜΆΪΟΣ" will match correctly.

Case-folding is still only an approximation of the language-specific rules governing case. If the specific language is known, consider using ICUCollationKeyFilter and indexing collation keys instead. This implementation performs the "full" case-folding specified in the Unicode standard, and this may change the length of the term. For example, the German ß is case-folded to the string 'ss'.

Case folding is related to normalization, and as such is coupled with it in this integration. To perform case-folding, you use normalization with the form "nfkc_cf" (which is the default).

Use Cases

As a more thorough replacement for LowerCaseFilter that has good behavior for most languages.

Example Usages

Lowercasing text

  /**
   * This filter will case-fold and normalize to NFKC.
   */
  TokenStream tokenstream = new ICUNormalizer2Filter(tokenizer);

Search Term Folding

Search term folding removes distinctions (such as accent marks) between similar characters. It is useful for a fuzzy or loose search.

Search term folding implements many of the foldings specified in Character Foldings as a special normalization form. This folding applies NFKC, Case Folding, and many character foldings recursively.

Use Cases

As a more thorough replacement for ASCIIFoldingFilter and LowerCaseFilter that applies the same ideas to many more languages.

Example Usages

Removing accents

  /**
   * This filter will case-fold, remove accents and other distinctions, and
   * normalize to NFKC.
   */
  TokenStream tokenstream = new ICUFoldingFilter(tokenizer);

Text Transformation

ICU provides text-transformation functionality via its Transliteration API. This allows you to transform text in a variety of ways, taking context into account.

For more information, see the User's Guide and Rule Tutorial.

Use Cases

Convert Traditional to Simplified
Transliterate between different writing systems: e.g. Romanization

Example Usages

Convert Traditional to Simplified

  /**
   * This filter will map Traditional Chinese to Simplified Chinese
   */
  TokenStream tokenstream = new ICUTransformFilter(tokenizer, Transliterator.getInstance("Traditional-Simplified"));

Transliterate Serbian Cyrillic to Serbian Latin

  /**
   * This filter will map Serbian Cyrillic to Serbian Latin according to BGN rules
   */
  TokenStream tokenstream = new ICUTransformFilter(tokenizer, Transliterator.getInstance("Serbian-Latin/BGN"));

Backwards Compatibility

This module exists to provide up-to-date Unicode functionality that supports the most recent version of Unicode (currently 11.0). However, some users who wish for stronger backwards compatibility can restrict ICUNormalizer2Filter to operate on only a specific Unicode Version by using a FilteredNormalizer2.

Example Usages

Restricting normalization to Unicode 5.0

  /**
   * This filter will do NFC normalization, but will ignore any characters that
   * did not exist as of Unicode 5.0. Because of the normalization stability policy
   * of Unicode, this is an easy way to force normalization to a specific version.
   */
    Normalizer2 normalizer = Normalizer2.getInstance(null, "nfc", Normalizer2.Mode.COMPOSE);
    UnicodeSet set = new UnicodeSet("[:age=5.0:]");
    // see FilteredNormalizer2 docs, the set should be frozen or performance will suffer
    set.freeze(); 
    FilteredNormalizer2 unicode50 = new FilteredNormalizer2(normalizer, set);
    TokenStream tokenstream = new ICUNormalizer2Filter(tokenizer, unicode50);

Lucene 8.4.0 analyzers-icu API

Text Segmentation

Use Cases

Example Usages

Tokenizing multilanguage text

Collation

Use Cases

Example Usages

Farsi Range Queries

Danish Sorting

Turkish Case Normalization

Caveats and Comparisons

Normalization

Use Cases

Example Usages

Normalizing text to NFC

Case Folding

Use Cases

Example Usages

Lowercasing text

Search Term Folding

Use Cases

Example Usages

Removing accents

Text Transformation

Use Cases

Example Usages

Convert Traditional to Simplified

Transliterate Serbian Cyrillic to Serbian Latin

Backwards Compatibility

Example Usages

Restricting normalization to Unicode 5.0