diff options
author | Jon Bratseth <bratseth@gmail.com> | 2021-05-05 21:18:50 +0200 |
---|---|---|
committer | Jon Bratseth <bratseth@gmail.com> | 2021-05-05 21:18:50 +0200 |
commit | 346c5a523fec4ec7cecb373e480e30846f5e33e5 (patch) | |
tree | d8788eab6bb91d0b06a15d50f93f19d9d42ba96c | |
parent | a2c9cd4bc04f1a3eaa31524b3970b96be5c2eda9 (diff) |
Revert "Merge pull request #17754 from vespa-engine/revert-17747-bratseth/special-tokens-take-2"
This reverts commit a2c9cd4bc04f1a3eaa31524b3970b96be5c2eda9, reversing
changes made to 8c61a373af0066fbdf1cca354c24b197c7347321.
24 files changed, 434 insertions, 438 deletions
diff --git a/config-model/src/main/java/com/yahoo/vespa/model/HostResource.java b/config-model/src/main/java/com/yahoo/vespa/model/HostResource.java index 78d9dd473b3..3bc07db9507 100644 --- a/config-model/src/main/java/com/yahoo/vespa/model/HostResource.java +++ b/config-model/src/main/java/com/yahoo/vespa/model/HostResource.java @@ -3,6 +3,7 @@ package com.yahoo.vespa.model; import com.yahoo.config.application.api.DeployLogger; import com.yahoo.config.model.api.HostInfo; +import com.yahoo.config.provision.ClusterSpec; import com.yahoo.config.provision.HostSpec; import com.yahoo.config.provision.NodeResources; diff --git a/config-model/src/main/java/com/yahoo/vespa/model/container/docproc/DocprocChain.java b/config-model/src/main/java/com/yahoo/vespa/model/container/docproc/DocprocChain.java index ee246b5e485..2b2b17c76c3 100644 --- a/config-model/src/main/java/com/yahoo/vespa/model/container/docproc/DocprocChain.java +++ b/config-model/src/main/java/com/yahoo/vespa/model/container/docproc/DocprocChain.java @@ -23,8 +23,9 @@ public class DocprocChain extends Chain<DocumentProcessor> { } /** - * The field name schema map that applies to this whole chain - * @return doctype,from → to + * The field name schema map that applies to this whole chain. + * + * @return doctype, from → to */ public Map<Pair<String,String>,String> fieldNameSchemaMap() { return fieldNameSchemaMap; diff --git a/config-model/src/main/java/com/yahoo/vespa/model/search/IndexingDocprocChain.java b/config-model/src/main/java/com/yahoo/vespa/model/search/IndexingDocprocChain.java index 2c1d979e2c4..8fe6b51f2b4 100644 --- a/config-model/src/main/java/com/yahoo/vespa/model/search/IndexingDocprocChain.java +++ b/config-model/src/main/java/com/yahoo/vespa/model/search/IndexingDocprocChain.java @@ -2,14 +2,12 @@ package com.yahoo.vespa.model.search; import com.yahoo.component.ComponentId; -import com.yahoo.component.ComponentSpecification; import com.yahoo.component.chain.Phase; import com.yahoo.component.chain.model.ChainSpecification; import com.yahoo.vespa.configdefinition.SpecialtokensConfig; import com.yahoo.vespa.model.container.docproc.DocprocChain; import java.util.ArrayList; -import java.util.Collections; import java.util.HashMap; import java.util.List; import java.util.Set; diff --git a/container-search/abi-spec.json b/container-search/abi-spec.json index b5933936adf..74ed9d33f04 100644 --- a/container-search/abi-spec.json +++ b/container-search/abi-spec.json @@ -5574,8 +5574,8 @@ "public com.yahoo.search.query.parser.ParserEnvironment setIndexFacts(com.yahoo.prelude.IndexFacts)", "public com.yahoo.language.Linguistics getLinguistics()", "public com.yahoo.search.query.parser.ParserEnvironment setLinguistics(com.yahoo.language.Linguistics)", - "public com.yahoo.prelude.query.parser.SpecialTokens getSpecialTokens()", - "public com.yahoo.search.query.parser.ParserEnvironment setSpecialTokens(com.yahoo.prelude.query.parser.SpecialTokens)", + "public com.yahoo.language.process.SpecialTokens getSpecialTokens()", + "public com.yahoo.search.query.parser.ParserEnvironment setSpecialTokens(com.yahoo.language.process.SpecialTokens)", "public static com.yahoo.search.query.parser.ParserEnvironment fromExecutionContext(com.yahoo.search.searchchain.Execution$Context)", "public static com.yahoo.search.query.parser.ParserEnvironment fromParserEnvironment(com.yahoo.search.query.parser.ParserEnvironment)" ], @@ -7765,7 +7765,7 @@ "final" ], "methods": [ - "public void <init>(com.yahoo.search.searchchain.SearchChainRegistry, com.yahoo.prelude.IndexFacts, com.yahoo.prelude.query.parser.SpecialTokenRegistry, com.yahoo.search.rendering.RendererRegistry, com.yahoo.language.Linguistics)", + "public void <init>(com.yahoo.search.searchchain.SearchChainRegistry, com.yahoo.prelude.IndexFacts, com.yahoo.language.process.SpecialTokenRegistry, com.yahoo.search.rendering.RendererRegistry, com.yahoo.language.Linguistics)", "public static com.yahoo.search.searchchain.Execution$Context createContextStub()", "public static com.yahoo.search.searchchain.Execution$Context createContextStub(com.yahoo.prelude.IndexFacts)", "public static com.yahoo.search.searchchain.Execution$Context createContextStub(com.yahoo.search.searchchain.SearchChainRegistry, com.yahoo.prelude.IndexFacts)", @@ -7779,8 +7779,8 @@ "public void setIndexFacts(com.yahoo.prelude.IndexFacts)", "public com.yahoo.search.searchchain.SearchChainRegistry searchChainRegistry()", "public com.yahoo.search.rendering.RendererRegistry rendererRegistry()", - "public com.yahoo.prelude.query.parser.SpecialTokenRegistry getTokenRegistry()", - "public void setTokenRegistry(com.yahoo.prelude.query.parser.SpecialTokenRegistry)", + "public com.yahoo.language.process.SpecialTokenRegistry getTokenRegistry()", + "public void setTokenRegistry(com.yahoo.language.process.SpecialTokenRegistry)", "public void setDetailedDiagnostics(boolean)", "public boolean getDetailedDiagnostics()", "public boolean getBreakdown()", diff --git a/container-search/src/main/java/com/yahoo/prelude/query/parser/AbstractParser.java b/container-search/src/main/java/com/yahoo/prelude/query/parser/AbstractParser.java index 902be7e15dd..732466748eb 100644 --- a/container-search/src/main/java/com/yahoo/prelude/query/parser/AbstractParser.java +++ b/container-search/src/main/java/com/yahoo/prelude/query/parser/AbstractParser.java @@ -19,7 +19,6 @@ import java.util.*; * @author bratseth * @author Steinar Knutsen */ -@SuppressWarnings("deprecation") public abstract class AbstractParser implements CustomParser { /** The current submodes of this parser */ @@ -48,7 +47,7 @@ public abstract class AbstractParser implements CustomParser { * of these may be active at the same time. SubModes are activated or * deactivated by specifying special indexes in the query. */ - final class Submodes { + static final class Submodes { /** * Url mode allows "_" and "-" as word characters. Default is false diff --git a/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokenRegistry.java b/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokenRegistry.java deleted file mode 100644 index be2d9f9f68b..00000000000 --- a/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokenRegistry.java +++ /dev/null @@ -1,137 +0,0 @@ -// Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. -package com.yahoo.prelude.query.parser; - -import com.yahoo.config.subscription.ConfigGetter; -import com.yahoo.config.subscription.ConfigSubscriber; -import com.yahoo.vespa.configdefinition.SpecialtokensConfig; -import com.yahoo.vespa.configdefinition.SpecialtokensConfig.Tokenlist; -import com.yahoo.vespa.configdefinition.SpecialtokensConfig.Tokenlist.Tokens; - -import java.util.*; -import java.util.logging.Logger; - - -/** - * A <i>registry</i> which is responsible for knowing the current - * set of special tokens. The default registry returns empty token lists - * for all names. Usage of this registry is multithread safe. - * - * @author bratseth - */ -public class SpecialTokenRegistry { - - /** The log of this */ - private static final Logger log = Logger.getLogger(SpecialTokenRegistry.class.getName()); - - private static final SpecialTokens nullSpecialTokens = new SpecialTokens(); - - /** - * The current authorative special token lists, indexed on name. - * These lists are unmodifiable and used directly by clients of this - */ - private Map<String,SpecialTokens> specialTokenMap = new HashMap<>(); - - private boolean frozen = false; - - /** - * Creates an empty special token registry which - * does not subscribe to any configuration - */ - public SpecialTokenRegistry() {} - - /** - * Create a special token registry which subscribes to the specialtokens - * configuration. Only used for testing. - */ - public SpecialTokenRegistry(String configId) { - try { - build(new ConfigGetter<>(SpecialtokensConfig.class).getConfig(configId)); - } catch (Exception e) { - log.config( - "No special tokens are configured (" + e.getMessage() + ")"); - } - } - - /** - * Create a special token registry from a configuration object. This is the production code path. - */ - public SpecialTokenRegistry(SpecialtokensConfig config) { - if (config != null) { - build(config); - } - freeze(); - } - - private void freeze() { - frozen = true; - } - - private void build(SpecialtokensConfig config) { - List<SpecialTokens> list = new ArrayList<>(); - for (Iterator<Tokenlist> i = config.tokenlist().iterator(); i.hasNext();) { - Tokenlist tokenList = i.next(); - SpecialTokens tokens = new SpecialTokens(tokenList.name()); - - for (Iterator<Tokens> j = tokenList.tokens().iterator(); j.hasNext();) { - Tokens token = j.next(); - tokens.addSpecialToken(token.token(), token.replace()); - } - tokens.freeze(); - list.add(tokens); - } - addSpecialTokens(list); - } - - /** - * Adds a SpecialTokens instance to the registry. That is, add the - * tokens contained for the name of the SpecialTokens instance - * given. - * - * @param specialTokens the SpecialTokens object to add - */ - public void addSpecialTokens(SpecialTokens specialTokens) { - ensureNotFrozen(); - List<SpecialTokens> list = new ArrayList<>(); - list.add(specialTokens); - addSpecialTokens(list); - - } - - private void ensureNotFrozen() { - if (frozen) { - throw new IllegalStateException("Tried to modify a frozen SpecialTokenRegistry instance."); - } - } - - private void addSpecialTokens(List<SpecialTokens> list) { - HashMap<String,SpecialTokens> tokens = new HashMap<>(specialTokenMap); - for(SpecialTokens t: list) { - tokens.put(t.getName(),t); - } - specialTokenMap = tokens; - } - - - /** - * Returns the currently authorative list of special tokens for - * a given name. - * - * @param name the name of the special tokens to return - * null, the empth string or the string "default" returns - * the default ones - * @return a read-only list of SpecialToken instances, an empty list if this name - * has no special tokens - */ - public SpecialTokens getSpecialTokens(String name) { - if (name == null || name.trim().equals("")) { - name = "default"; - } - SpecialTokens specialTokens = specialTokenMap.get(name); - - if (specialTokens == null) { - return nullSpecialTokens; - } - return specialTokens; - } - -} diff --git a/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokens.java b/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokens.java deleted file mode 100644 index f45ecefefa6..00000000000 --- a/container-search/src/main/java/com/yahoo/prelude/query/parser/SpecialTokens.java +++ /dev/null @@ -1,167 +0,0 @@ -// Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. -package com.yahoo.prelude.query.parser; - -import java.util.logging.Level; -import com.yahoo.prelude.query.Substring; - -import java.util.*; -import java.util.logging.Logger; - -import static com.yahoo.language.LinguisticsCase.toLowerCase; - -/** - * A list of special tokens - string that should be treated as word - * no matter what they contain. Special tokens are case insensitive. - * - * @author bratseth - */ -public class SpecialTokens { - - private static final Logger log = Logger.getLogger(SpecialTokens.class.getName()); - - private final String name; - - private final List<SpecialToken> specialTokens = new ArrayList<>(); - - private boolean frozen = false; - - private int currentMaximumLength = 0; - - /** Creates a null list of special tokens */ - public SpecialTokens() { - this.name = "(null)"; - } - - public SpecialTokens(String name) { - this.name = name; - } - - /** Returns the name of this special tokens list */ - public String getName() { - return name; - } - - /** - * Adds a special token to this - * - * @param token the special token string to add - * @param replace the token to replace instances of the special token with, or null to keep the token - */ - public void addSpecialToken(String token, String replace) { - ensureNotFrozen(); - if (!caseIndependentLength(token)) { - return; - } - // TODO are special tokens correctly unicode normalized in reagards to query parsing? - final SpecialToken specialTokenToAdd = new SpecialToken(token, replace); - currentMaximumLength = Math.max(currentMaximumLength, specialTokenToAdd.token.length()); - specialTokens.add(specialTokenToAdd); - Collections.sort(specialTokens); - } - - private boolean caseIndependentLength(String token) { - // XXX not fool proof length test, should test codepoint by codepoint for mixed case user input? not even that will necessarily be 100% robust... - String asLow = toLowerCase(token); - // TODO put along with the global toLowerCase - String asHigh = token.toUpperCase(Locale.ENGLISH); - if (asLow.length() != token.length() || asHigh.length() != token.length()) { - log.log(Level.SEVERE, "Special token '" + token + "' has case sensitive length. Ignoring the token." - + " Please report this message in a bug to the Vespa team."); - return false; - } else { - return true; - } - } - - /** - * Returns the special token starting at the start of the given string, or null if no - * special token starts at this string - * - * @param string the string to search for a special token at the start position - * @param substring true to allow the special token to be followed by a character which does not - * mark the end of a token - */ - public SpecialToken tokenize(String string, boolean substring) { - // XXX detonator pattern token.length may be != the length of the - // matching data in string, ref caseIndependentLength(String) - final String input = toLowerCase(string.substring(0, Math.min(string.length(), currentMaximumLength))); - for (Iterator<SpecialToken> i = specialTokens.iterator(); i.hasNext();) { - SpecialTokens.SpecialToken special = i.next(); - - if (input.startsWith(special.token())) { - if (string.length() == special.token().length() || substring || tokenEndsAt(special.token().length(), string)) - return special; - } - } - return null; - } - - private boolean tokenEndsAt(int position,String string) { - return !Character.isLetterOrDigit(string.charAt(position)); - } - - /** Returns the number of special tokens in this */ - public int size() { - return specialTokens.size(); - } - - private void ensureNotFrozen() { - if (frozen) { - throw new IllegalStateException("Tried to modify a frozen SpecialTokens instance."); - } - } - - public void freeze() { - frozen = true; - } - - /** An immutable special token */ - public final static class SpecialToken implements Comparable<SpecialToken> { - - private String token; - - private String replace; - - public SpecialToken(String token, String replace) { - this.token = toLowerCase(token); - if (replace == null || replace.trim().equals("")) { - this.replace = this.token; - } else { - this.replace = toLowerCase(replace); - } - } - - /** Returns the special token */ - public String token() { - return token; - } - - /** Returns the right replace value, never null or an empty string */ - public String replace() { - return replace; - } - - @Override - public int compareTo(SpecialToken other) { - if (this.token().length() < other.token().length()) return 1; - if (this.token().length() == other.token().length()) return 0; - return -1; - } - - @Override - public boolean equals(Object other) { - if (other == this) return true; - if ( ! (other instanceof SpecialToken)) return false; - return Objects.equals(this.token, ((SpecialToken)other).token); - } - - @Override - public int hashCode() { return token.hashCode(); } - - public Token toToken(int start, String rawSource) { - return new Token(Token.Kind.WORD, replace(), true, new Substring(start, start + token.length(), rawSource)); // XXX: Unsafe? - } - - } - -} diff --git a/container-search/src/main/java/com/yahoo/prelude/query/parser/Tokenizer.java b/container-search/src/main/java/com/yahoo/prelude/query/parser/Tokenizer.java index 2dc2254df68..b71bd57539f 100644 --- a/container-search/src/main/java/com/yahoo/prelude/query/parser/Tokenizer.java +++ b/container-search/src/main/java/com/yahoo/prelude/query/parser/Tokenizer.java @@ -3,6 +3,7 @@ package com.yahoo.prelude.query.parser; import com.yahoo.language.Linguistics; import com.yahoo.language.process.CharacterClasses; +import com.yahoo.language.process.SpecialTokens; import com.yahoo.prelude.Index; import com.yahoo.prelude.IndexFacts; import com.yahoo.prelude.query.Substring; @@ -200,7 +201,7 @@ public final class Tokenizer { } StringBuilder tmp = new StringBuilder(); for (int i = 0; i < tokencnt; i++) { - Token useToken = tokens.get(backtrack+i); + Token useToken = tokens.get(backtrack + i); tmp.append(useToken.image); } String indexName = tmp.toString(); @@ -216,20 +217,20 @@ public final class Tokenizer { } private int consumeSpecialToken(int start) { - SpecialTokens.SpecialToken specialToken=getSpecialToken(start); - if (specialToken==null) return start; - tokens.add(specialToken.toToken(start,source)); - return start + specialToken.token().length(); + SpecialTokens.Token token = getSpecialToken(start); + if (token == null) return start; + tokens.add(toToken(token, start, source)); + return start + token.token().length(); } - private SpecialTokens.SpecialToken getSpecialToken(int start) { + private SpecialTokens.Token getSpecialToken(int start) { if (specialTokens == null) return null; return specialTokens.tokenize(source.substring(start), substringSpecialTokens); } private int consumeExact(int start,Index index) { if (index.getExactTerminator() == null) return consumeHeuristicExact(start); - return consumeToTerminator(start,index.getExactTerminator()); + return consumeToTerminator(start, index.getExactTerminator()); } private boolean looksLikeExactEnd(int end) { @@ -467,7 +468,7 @@ public final class Tokenizer { /** Consumes a word or number <i>and/or possibly</i> a special token starting within this word or number */ private int consumeWordOrNumber(int start, Index currentIndex) { int tokenEnd = start; - SpecialTokens.SpecialToken substringSpecialToken = null; + SpecialTokens.Token substringToken = null; boolean digitsOnly = true; // int underscores = 0; // boolean underscoresOnly = true; @@ -475,8 +476,8 @@ public final class Tokenizer { while (tokenEnd < source.length()) { if (substringSpecialTokens) { - substringSpecialToken = getSpecialToken(tokenEnd); - if (substringSpecialToken != null) break; + substringToken = getSpecialToken(tokenEnd); + if (substringToken != null) break; } int c = source.codePointAt(tokenEnd); @@ -524,11 +525,11 @@ public final class Tokenizer { } } - if (substringSpecialToken == null) + if (substringToken == null) return --tokenEnd; // TODO: test the logic around tokenEnd with friends - addToken(substringSpecialToken.toToken(tokenEnd, source)); - return --tokenEnd + substringSpecialToken.token().length(); + addToken(toToken(substringToken, tokenEnd, source)); + return --tokenEnd + substringToken.token().length(); } private void addToken(Token.Kind kind, String word, int start, int end) { @@ -539,4 +540,11 @@ public final class Tokenizer { tokens.add(token); } + public Token toToken(SpecialTokens.Token specialToken, int start, String rawSource) { + return new Token(Token.Kind.WORD, + specialToken.replacement(), + true, + new Substring(start, start + specialToken.token().length(), rawSource)); // XXX: Unsafe? + } + } diff --git a/container-search/src/main/java/com/yahoo/search/query/parser/ParserEnvironment.java b/container-search/src/main/java/com/yahoo/search/query/parser/ParserEnvironment.java index 94b9bf6ce65..df96d314455 100644 --- a/container-search/src/main/java/com/yahoo/search/query/parser/ParserEnvironment.java +++ b/container-search/src/main/java/com/yahoo/search/query/parser/ParserEnvironment.java @@ -4,7 +4,7 @@ package com.yahoo.search.query.parser; import com.yahoo.language.Linguistics; import com.yahoo.language.simple.SimpleLinguistics; import com.yahoo.prelude.IndexFacts; -import com.yahoo.prelude.query.parser.SpecialTokens; +import com.yahoo.language.process.SpecialTokens; import com.yahoo.search.Searcher; import com.yahoo.search.searchchain.Execution; @@ -18,7 +18,7 @@ public final class ParserEnvironment { private IndexFacts indexFacts = new IndexFacts(); private Linguistics linguistics = new SimpleLinguistics(); - private SpecialTokens specialTokens = new SpecialTokens(); + private SpecialTokens specialTokens = SpecialTokens.empty(); public IndexFacts getIndexFacts() { return indexFacts; diff --git a/container-search/src/main/java/com/yahoo/search/searchchain/Execution.java b/container-search/src/main/java/com/yahoo/search/searchchain/Execution.java index 84fe88d0292..0574fc660c3 100644 --- a/container-search/src/main/java/com/yahoo/search/searchchain/Execution.java +++ b/container-search/src/main/java/com/yahoo/search/searchchain/Execution.java @@ -6,7 +6,7 @@ import com.yahoo.language.Linguistics; import com.yahoo.prelude.IndexFacts; import com.yahoo.prelude.Ping; import com.yahoo.prelude.Pong; -import com.yahoo.prelude.query.parser.SpecialTokenRegistry; +import com.yahoo.language.process.SpecialTokenRegistry; import com.yahoo.processing.Processor; import com.yahoo.processing.Request; import com.yahoo.processing.Response; @@ -17,8 +17,6 @@ import com.yahoo.search.cluster.PingableSearcher; import com.yahoo.search.rendering.RendererRegistry; import com.yahoo.search.statistics.TimeTracker; -import java.util.logging.Logger; - /** * <p>An execution of a search chain. This keeps track of the call state for an execution (in the calling thread) * of the searchers of a search chain.</p> @@ -111,7 +109,7 @@ public class Execution extends com.yahoo.processing.execution.Execution { public Context(SearchChainRegistry searchChainRegistry, IndexFacts indexFacts, SpecialTokenRegistry tokenRegistry, RendererRegistry rendererRegistry, Linguistics linguistics) { - owner=null; + owner = null; // The next time something is added here, compose into wrapper objects. Many arguments... // Four methods need to be updated when adding something: diff --git a/container-search/src/main/java/com/yahoo/search/searchchain/ExecutionFactory.java b/container-search/src/main/java/com/yahoo/search/searchchain/ExecutionFactory.java index 31b6d06f78e..a813229c984 100644 --- a/container-search/src/main/java/com/yahoo/search/searchchain/ExecutionFactory.java +++ b/container-search/src/main/java/com/yahoo/search/searchchain/ExecutionFactory.java @@ -13,7 +13,7 @@ import com.yahoo.language.Linguistics; import com.yahoo.language.simple.SimpleLinguistics; import com.yahoo.prelude.IndexFacts; import com.yahoo.prelude.IndexModel; -import com.yahoo.prelude.query.parser.SpecialTokenRegistry; +import com.yahoo.language.process.SpecialTokenRegistry; import com.yahoo.processing.rendering.Renderer; import com.yahoo.search.Searcher; import com.yahoo.search.config.IndexInfoConfig; diff --git a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParseTestCase.java b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParseTestCase.java index 6afea895f3a..cef8ae1751c 100644 --- a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParseTestCase.java +++ b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParseTestCase.java @@ -18,16 +18,14 @@ import com.yahoo.prelude.query.PhraseSegmentItem; import com.yahoo.prelude.query.PrefixItem; import com.yahoo.prelude.query.RankItem; import com.yahoo.prelude.query.SubstringItem; -import com.yahoo.prelude.query.SubstringItem; import com.yahoo.prelude.query.SuffixItem; import com.yahoo.prelude.query.TaggableItem; import com.yahoo.prelude.query.WordItem; -import com.yahoo.prelude.query.parser.SpecialTokens; +import com.yahoo.language.process.SpecialTokens; import com.yahoo.prelude.query.parser.TestLinguistics; import com.yahoo.search.Query; import org.junit.Test; -import java.util.Collections; import java.util.Iterator; import static org.junit.Assert.assertEquals; @@ -1639,7 +1637,7 @@ public class ParseTestCase { @Test public void testNonSpecialTokenParsing() { - ParsingTester customTester = new ParsingTester(new SpecialTokens("default")); + ParsingTester customTester = new ParsingTester(SpecialTokens.empty()); customTester.assertParsed("OR c or c with (AND tcp ip)", "c# or c++ with tcp/ip", Query.Type.ANY); } diff --git a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParsingTester.java b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParsingTester.java index 17155fff5de..fd7e4cbe0e6 100644 --- a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParsingTester.java +++ b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/ParsingTester.java @@ -11,8 +11,8 @@ import com.yahoo.prelude.IndexFacts; import com.yahoo.prelude.IndexModel; import com.yahoo.prelude.query.Item; import com.yahoo.prelude.query.NullItem; -import com.yahoo.prelude.query.parser.SpecialTokenRegistry; -import com.yahoo.prelude.query.parser.SpecialTokens; +import com.yahoo.language.process.SpecialTokenRegistry; +import com.yahoo.language.process.SpecialTokens; import com.yahoo.search.Query; import com.yahoo.search.config.IndexInfoConfig; import com.yahoo.search.query.parser.Parsable; @@ -20,6 +20,9 @@ import com.yahoo.search.query.parser.Parser; import com.yahoo.search.query.parser.ParserEnvironment; import com.yahoo.search.query.parser.ParserFactory; +import java.util.ArrayList; +import java.util.List; + import static org.junit.Assert.assertEquals; import static org.junit.Assert.assertNotNull; import static org.junit.Assert.assertTrue; @@ -32,7 +35,7 @@ import static org.junit.Assert.assertTrue; public class ParsingTester { private static final Linguistics linguistics = new SimpleLinguistics(); - private IndexFacts indexFacts; + private final IndexFacts indexFacts; private SpecialTokenRegistry tokenRegistry; public ParsingTester() { @@ -49,11 +52,10 @@ public class ParsingTester { public ParsingTester(IndexFacts indexFacts, SpecialTokens specialTokens) { indexFacts.freeze(); - specialTokens.freeze(); this.indexFacts = indexFacts; tokenRegistry = new SpecialTokenRegistry(); - tokenRegistry.addSpecialTokens(specialTokens); + tokenRegistry = new SpecialTokenRegistry(List.of(specialTokens)); } /** @@ -72,13 +74,13 @@ public class ParsingTester { * This can be used to add new tokens and passing the resulting special tokens to the constructor of this. */ public static SpecialTokens createSpecialTokens() { - SpecialTokens tokens = new SpecialTokens("default"); - tokens.addSpecialToken("c++", null); - tokens.addSpecialToken(".net", "dotnet"); - tokens.addSpecialToken("tcp/ip", null); - tokens.addSpecialToken("c#", null); - tokens.addSpecialToken("special-token-fs","firstsecond"); - return tokens; + List<SpecialTokens.Token> tokens = new ArrayList<>(); + tokens.add(new SpecialTokens.Token("c++")); + tokens.add(new SpecialTokens.Token(".net", "dotnet")); + tokens.add(new SpecialTokens.Token("tcp/ip")); + tokens.add(new SpecialTokens.Token("c#")); + tokens.add(new SpecialTokens.Token("special-token-fs","firstsecond")); + return new SpecialTokens("default", tokens); } /** diff --git a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/TokenizerTestCase.java b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/TokenizerTestCase.java index aa2e9dbcf75..e10fbd71c72 100644 --- a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/TokenizerTestCase.java +++ b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/TokenizerTestCase.java @@ -6,12 +6,13 @@ import com.yahoo.prelude.Index; import com.yahoo.prelude.IndexFacts; import com.yahoo.prelude.IndexModel; import com.yahoo.prelude.SearchDefinition; -import com.yahoo.prelude.query.parser.SpecialTokenRegistry; -import com.yahoo.prelude.query.parser.SpecialTokens; +import com.yahoo.language.process.SpecialTokenRegistry; +import com.yahoo.language.process.SpecialTokens; import com.yahoo.prelude.query.parser.Token; import com.yahoo.prelude.query.parser.Tokenizer; import org.junit.Test; +import java.util.ArrayList; import java.util.Collections; import java.util.List; @@ -39,13 +40,11 @@ import static org.junit.Assert.assertTrue; */ public class TokenizerTestCase { - private SpecialTokenRegistry defaultRegistry = new SpecialTokenRegistry("file:src/test/java/com/yahoo/prelude/query/parser/test/replacingtokens.cfg"); - @Test public void testPlainTokenization() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("drive (to hwy88, 88) +or language:en ugcapi_1 & &a"); assertEquals(new Token(WORD, "drive"), tokens.get(0)); @@ -87,7 +86,7 @@ public class TokenizerTestCase { public void testOneSpecialToken() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("c++ lovers, please apply"); assertEquals(new Token(WORD, "c++"), tokens.get(0)); @@ -97,7 +96,7 @@ public class TokenizerTestCase { public void testSpecialTokenCombination() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("c#, c++ or .net know, not tcp/ip"); assertEquals(new Token(WORD, "c#"), tokens.get(0)); @@ -123,10 +122,9 @@ public class TokenizerTestCase { */ @Test public void testSpecialTokenCJK() { - assertEquals("Special tokens configured", 6, defaultRegistry.getSpecialTokens("default").size()); Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); tokenizer.setSubstringSpecialTokens(true); - tokenizer.setSpecialTokens(defaultRegistry.getSpecialTokens("default")); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("replacing")); List<?> tokens = tokenizer.tokenize("fooc#bar,c++with spacebarknowknowknow,knowknownot know"); assertEquals(new Token(WORD, "foo"), tokens.get(0)); @@ -151,7 +149,7 @@ public class TokenizerTestCase { public void testSpecialTokenCaseInsensitive() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("The AS/400 is great"); assertEquals(new Token(WORD, "The"), tokens.get(0)); @@ -167,7 +165,7 @@ public class TokenizerTestCase { public void testSpecialTokenNonMatch() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("c++ c+ aS/400 i/o .net i/ooo ap.net"); assertEquals(new Token(WORD, "c++"), tokens.get(0)); @@ -190,18 +188,9 @@ public class TokenizerTestCase { @Test public void testSpecialTokenConfigurationDefault() { - String tokenFile = "file:src/test/java/com/yahoo/prelude/query/parser/test/specialtokens.cfg"; - - SpecialTokenRegistry r = new SpecialTokenRegistry(tokenFile); - assertEquals("Special tokens configured", 6, - r.getSpecialTokens("default").size()); - assertEquals("Special tokens configured", 4, - r.getSpecialTokens("other").size()); - Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens( - r.getSpecialTokens("default")); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize( "with space, c++ or .... know, not b.s.d."); @@ -224,18 +213,9 @@ public class TokenizerTestCase { @Test public void testSpecialTokenConfigurationOther() { - String tokenFile = "file:src/test/java/com/yahoo/prelude/query/parser/test/specialtokens.cfg"; - - SpecialTokenRegistry r = new SpecialTokenRegistry(tokenFile); - assertEquals("Special tokens configured", 6, - r.getSpecialTokens("default").size()); - assertEquals("Special tokens configured", 4, - r.getSpecialTokens("other").size()); - Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens( - r.getSpecialTokens("other")); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("other")); List<?> tokens = tokenizer.tokenize( "with space,!!!*** [huh] or ------ " + "know, &&&%%% b.s.d."); @@ -267,26 +247,9 @@ public class TokenizerTestCase { } @Test - public void testSpecialTokenConfigurationMissing() { - String tokenFile = "file:source/bogus/specialtokens.cfg"; - - SpecialTokenRegistry r = new SpecialTokenRegistry(tokenFile); - - Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - - tokenizer.setSpecialTokens(r.getSpecialTokens("other")); - List<?> tokens = tokenizer.tokenize("c++"); - - assertEquals(new Token(WORD, "c"), tokens.get(0)); - assertEquals(new Token(PLUS, "+"), tokens.get(1)); - assertEquals(new Token(PLUS, "+"), tokens.get(2)); - } - - @Test public void testTokenReplacing() { - assertEquals("Special tokens configured", 6, defaultRegistry.getSpecialTokens("default").size()); Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(defaultRegistry.getSpecialTokens("default")); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("replacing")); List<?> tokens = tokenizer.tokenize("with space, c++ or .... know, not b.s.d."); assertEquals(new Token(WORD, "with-space"), tokens.get(0)); @@ -745,7 +708,7 @@ public class TokenizerTestCase { public void testSingleQuoteAsWordCharacter() { Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics()); - tokenizer.setSpecialTokens(createSpecialTokens()); + tokenizer.setSpecialTokens(createSpecialTokens().getSpecialTokens("default")); List<?> tokens = tokenizer.tokenize("drive (to hwy88, 88) +or language:en nalle:a'a ugcapi_1 'a' 'a a'"); assertEquals(new Token(WORD, "drive"), tokens.get(0)); @@ -781,17 +744,38 @@ public class TokenizerTestCase { assertEquals(new Token(WORD, "a'"), tokens.get(30)); } - private SpecialTokens createSpecialTokens() { - SpecialTokens tokens = new SpecialTokens("default"); - - tokens.addSpecialToken("c+", null); - tokens.addSpecialToken("c++", null); - tokens.addSpecialToken(".net", null); - tokens.addSpecialToken("tcp/ip", null); - tokens.addSpecialToken("i/o", null); - tokens.addSpecialToken("c#", null); - tokens.addSpecialToken("AS/400", null); - return tokens; + private SpecialTokenRegistry createSpecialTokens() { + List<SpecialTokens.Token> tokens = new ArrayList<>(); + tokens.add(new SpecialTokens.Token("c+")); + tokens.add(new SpecialTokens.Token("c++")); + tokens.add(new SpecialTokens.Token(".net")); + tokens.add(new SpecialTokens.Token("tcp/ip")); + tokens.add(new SpecialTokens.Token("i/o")); + tokens.add(new SpecialTokens.Token("c#")); + tokens.add(new SpecialTokens.Token("AS/400")); + tokens.add(new SpecialTokens.Token("....")); + tokens.add(new SpecialTokens.Token("b.s.d.")); + tokens.add(new SpecialTokens.Token("with space")); + tokens.add(new SpecialTokens.Token("dvd\\xB1r")); + SpecialTokens defaultTokens = new SpecialTokens("default", tokens); + + tokens = new ArrayList<>(); + tokens.add(new SpecialTokens.Token("[huh]")); + tokens.add(new SpecialTokens.Token("&&&%%%")); + tokens.add(new SpecialTokens.Token("------")); + tokens.add(new SpecialTokens.Token("!!!***")); + SpecialTokens otherTokens = new SpecialTokens("other", tokens); + + tokens = new ArrayList<>(); + tokens.add(new SpecialTokens.Token("....")); + tokens.add(new SpecialTokens.Token("c++", "cpp")); + tokens.add(new SpecialTokens.Token("b.s.d.")); + tokens.add(new SpecialTokens.Token("with space", "with-space")); + tokens.add(new SpecialTokens.Token("c#")); + tokens.add(new SpecialTokens.Token("know", "knuwww")); + SpecialTokens replacingTokens = new SpecialTokens("replacing", tokens); + + return new SpecialTokenRegistry(List.of(defaultTokens, otherTokens, replacingTokens)); } } diff --git a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/replacingtokens.cfg b/container-search/src/test/java/com/yahoo/prelude/query/parser/test/replacingtokens.cfg deleted file mode 100644 index 6a189de0164..00000000000 --- a/container-search/src/test/java/com/yahoo/prelude/query/parser/test/replacingtokens.cfg +++ /dev/null @@ -1,12 +0,0 @@ -tokenlist[1] -tokenlist[0].name default -tokenlist[0].tokens[6] -tokenlist[0].tokens[0].token .... -tokenlist[0].tokens[1].token c++ -tokenlist[0].tokens[1].replace cpp -tokenlist[0].tokens[2].token b.s.d. -tokenlist[0].tokens[3].token with space -tokenlist[0].tokens[3].replace with-space -tokenlist[0].tokens[4].token c# -tokenlist[0].tokens[5].token know -tokenlist[0].tokens[5].replace knuwww diff --git a/container-search/src/test/java/com/yahoo/search/query/rewrite/RewriterFeaturesTestCase.java b/container-search/src/test/java/com/yahoo/search/query/rewrite/RewriterFeaturesTestCase.java index 5508c2a73a7..08146bbe069 100644 --- a/container-search/src/test/java/com/yahoo/search/query/rewrite/RewriterFeaturesTestCase.java +++ b/container-search/src/test/java/com/yahoo/search/query/rewrite/RewriterFeaturesTestCase.java @@ -8,7 +8,7 @@ import org.junit.Test; import com.yahoo.prelude.query.AndItem; import com.yahoo.prelude.query.CompositeItem; import com.yahoo.prelude.query.Item; -import com.yahoo.prelude.query.parser.SpecialTokenRegistry; +import com.yahoo.language.process.SpecialTokenRegistry; import com.yahoo.search.Query; import com.yahoo.search.searchchain.Execution; import com.yahoo.search.searchchain.Execution.Context; diff --git a/linguistics/abi-spec.json b/linguistics/abi-spec.json index 58b838d7332..b77b03664d4 100644 --- a/linguistics/abi-spec.json +++ b/linguistics/abi-spec.json @@ -427,6 +427,57 @@ ], "fields": [] }, + "com.yahoo.language.process.SpecialTokenRegistry": { + "superClass": "java.lang.Object", + "interfaces": [], + "attributes": [ + "public" + ], + "methods": [ + "public void <init>()", + "public void <init>(com.yahoo.vespa.configdefinition.SpecialtokensConfig)", + "public void <init>(java.util.List)", + "public com.yahoo.language.process.SpecialTokens getSpecialTokens(java.lang.String)" + ], + "fields": [] + }, + "com.yahoo.language.process.SpecialTokens$Token": { + "superClass": "java.lang.Object", + "interfaces": [ + "java.lang.Comparable" + ], + "attributes": [ + "public", + "final" + ], + "methods": [ + "public void <init>(java.lang.String)", + "public void <init>(java.lang.String, java.lang.String)", + "public java.lang.String token()", + "public java.lang.String replacement()", + "public int compareTo(com.yahoo.language.process.SpecialTokens$Token)", + "public boolean equals(java.lang.Object)", + "public int hashCode()", + "public java.lang.String toString()", + "public bridge synthetic int compareTo(java.lang.Object)" + ], + "fields": [] + }, + "com.yahoo.language.process.SpecialTokens": { + "superClass": "java.lang.Object", + "interfaces": [], + "attributes": [ + "public" + ], + "methods": [ + "public void <init>(java.lang.String, java.util.List)", + "public java.lang.String name()", + "public java.util.Map asMap()", + "public com.yahoo.language.process.SpecialTokens$Token tokenize(java.lang.String, boolean)", + "public static com.yahoo.language.process.SpecialTokens empty()" + ], + "fields": [] + }, "com.yahoo.language.process.StemList": { "superClass": "java.util.AbstractList", "interfaces": [], diff --git a/linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java b/linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java index e1185cb2457..73518876c3f 100644 --- a/linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java +++ b/linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java @@ -4,6 +4,7 @@ package com.yahoo.language.opennlp; import com.yahoo.language.Language; import com.yahoo.language.LinguisticsCase; import com.yahoo.language.process.Normalizer; +import com.yahoo.language.process.SpecialTokenRegistry; import com.yahoo.language.process.StemMode; import com.yahoo.language.process.Token; import com.yahoo.language.process.TokenType; @@ -32,15 +33,21 @@ public class OpenNlpTokenizer implements Tokenizer { private final Normalizer normalizer; private final Transformer transformer; private final SimpleTokenizer simpleTokenizer; + private final SpecialTokenRegistry specialTokenRegistry; public OpenNlpTokenizer() { this(new SimpleNormalizer(), new SimpleTransformer()); } public OpenNlpTokenizer(Normalizer normalizer, Transformer transformer) { + this(normalizer, transformer, new SpecialTokenRegistry(List.of())); + } + + public OpenNlpTokenizer(Normalizer normalizer, Transformer transformer, SpecialTokenRegistry specialTokenRegistry) { this.normalizer = normalizer; this.transformer = transformer; - simpleTokenizer = new SimpleTokenizer(normalizer, transformer); + this.specialTokenRegistry = specialTokenRegistry; + this.simpleTokenizer = new SimpleTokenizer(normalizer, transformer, specialTokenRegistry); } @Override diff --git a/linguistics/src/main/java/com/yahoo/language/process/SpecialTokenRegistry.java b/linguistics/src/main/java/com/yahoo/language/process/SpecialTokenRegistry.java new file mode 100644 index 00000000000..b6335d67967 --- /dev/null +++ b/linguistics/src/main/java/com/yahoo/language/process/SpecialTokenRegistry.java @@ -0,0 +1,72 @@ +// Copyright Verizon Media. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package com.yahoo.language.process; + +import com.yahoo.vespa.configdefinition.SpecialtokensConfig; +import com.yahoo.vespa.configdefinition.SpecialtokensConfig.Tokenlist; +import com.yahoo.vespa.configdefinition.SpecialtokensConfig.Tokenlist.Tokens; + +import java.util.ArrayList; +import java.util.Iterator; +import java.util.List; +import java.util.Map; +import java.util.stream.Collectors; + +/** + * Immutable named lists of "special tokens" - strings which should override the normal tokenizer semantics + * and be tokenized into a single token. + * + * @author bratseth + */ +public class SpecialTokenRegistry { + + /** + * The current special token lists, indexed on name. + * These lists are unmodifiable and used directly by clients of this + */ + private final Map<String, SpecialTokens> specialTokenMap; + + /** Creates an empty special token registry */ + public SpecialTokenRegistry() { + this(List.of()); + } + + /** Create a special token registry from a configuration object. */ + public SpecialTokenRegistry(SpecialtokensConfig config) { + this(specialTokensFrom(config)); + } + + public SpecialTokenRegistry(List<SpecialTokens> specialTokensList) { + specialTokenMap = specialTokensList.stream().collect(Collectors.toUnmodifiableMap(t -> t.name(), t -> t)); + } + + private static List<SpecialTokens> specialTokensFrom(SpecialtokensConfig config) { + List<SpecialTokens> specialTokensList = new ArrayList<>(); + for (Iterator<Tokenlist> i = config.tokenlist().iterator(); i.hasNext();) { + Tokenlist tokenListConfig = i.next(); + + List<SpecialTokens.Token> tokenList = new ArrayList<>(); + for (Iterator<Tokens> j = tokenListConfig.tokens().iterator(); j.hasNext();) { + Tokens tokenConfig = j.next(); + tokenList.add(new SpecialTokens.Token(tokenConfig.token(), tokenConfig.replace())); + } + specialTokensList.add(new SpecialTokens(tokenListConfig.name(), tokenList)); + } + return specialTokensList; + } + + /** + * Returns the list of special tokens for a given name. + * + * @param name the name of the special tokens to return + * null, the empty string or the string "default" returns + * the default ones + * @return a read-only list of SpecialToken instances, an empty list if this name + * has no special tokens + */ + public SpecialTokens getSpecialTokens(String name) { + if (name == null || name.trim().equals("")) + name = "default"; + return specialTokenMap.getOrDefault(name, SpecialTokens.empty()); + } + +} diff --git a/linguistics/src/main/java/com/yahoo/language/process/SpecialTokens.java b/linguistics/src/main/java/com/yahoo/language/process/SpecialTokens.java new file mode 100644 index 00000000000..465d9b754b3 --- /dev/null +++ b/linguistics/src/main/java/com/yahoo/language/process/SpecialTokens.java @@ -0,0 +1,141 @@ +// Copyright Verizon Media. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package com.yahoo.language.process; + +import java.util.ArrayList; +import java.util.Collections; +import java.util.Iterator; +import java.util.List; +import java.util.Locale; +import java.util.Map; +import java.util.Objects; +import java.util.stream.Collectors; + +import static com.yahoo.language.LinguisticsCase.toLowerCase; + +/** + * An immutable list of special tokens - strings which should override the normal tokenizer semantics + * and be tokenized into a single token. Special tokens are case insensitive. + * + * @author bratseth + */ +public class SpecialTokens { + + private static final SpecialTokens empty = new SpecialTokens("(empty)", List.of()); + + private final String name; + private final int maximumLength; + private final List<Token> tokens; + private final Map<String, String> tokenMap; + + public SpecialTokens(String name, List<Token> tokens) { + tokens.stream().peek(token -> token.validate()); + List<Token> mutableTokens = new ArrayList<>(tokens); + Collections.sort(mutableTokens); + this.name = name; + this.maximumLength = tokens.stream().mapToInt(token -> token.token().length()).max().orElse(0); + this.tokens = List.copyOf(mutableTokens); + this.tokenMap = tokens.stream().collect(Collectors.toUnmodifiableMap(t -> t.token(), t -> t.replacement())); + } + + /** Returns the name of this special tokens list */ + public String name() { + return name; + } + + /** + * Returns the tokens of this as an immutable map from token to replacement. + * Tokens which do not have a replacement token maps to themselves. + */ + public Map<String, String> asMap() { return tokenMap; } + + /** + * Returns the special token starting at the start of the given string, or null if no + * special token starts at this string + * + * @param string the string to search for a special token at the start position + * @param substring true to allow the special token to be followed by a character which does not + * mark the end of a token + */ + public Token tokenize(String string, boolean substring) { + // XXX detonator pattern token.length may be != the length of the + // matching data in string, ref caseIndependentLength(String) + String input = toLowerCase(string.substring(0, Math.min(string.length(), maximumLength))); + for (Iterator<Token> i = tokens.iterator(); i.hasNext();) { + Token special = i.next(); + + if (input.startsWith(special.token())) { + if (string.length() == special.token().length() || substring || tokenEndsAt(special.token().length(), string)) + return special; + } + } + return null; + } + + private boolean tokenEndsAt(int position, String string) { + return !Character.isLetterOrDigit(string.charAt(position)); + } + + public static SpecialTokens empty() { return empty; } + + /** An immutable special token */ + public final static class Token implements Comparable<Token> { + + private final String token; + private final String replacement; + + /** Creates a special token */ + public Token(String token) { + this(token, null); + } + + /** Creates a special token which will be represented by the given replacement token */ + public Token(String token, String replacement) { + this.token = toLowerCase(token); + if (replacement == null || replacement.trim().equals("")) + this.replacement = this.token; + else + this.replacement = toLowerCase(replacement); + } + + /** Returns the special token */ + public String token() { return token; } + + /** Returns the token to replace occurrences of this by, which equals token() unless this has a replacement. */ + public String replacement() { return replacement; } + + @Override + public int compareTo(Token other) { + if (this.token().length() < other.token().length()) return 1; + if (this.token().length() == other.token().length()) return 0; + return -1; + } + + @Override + public boolean equals(Object other) { + if (other == this) return true; + if ( ! (other instanceof Token)) return false; + return Objects.equals(this.token, ((Token)other).token); + } + + @Override + public int hashCode() { return token.hashCode(); } + + @Override + public String toString() { + return "token '" + token + "'" + (replacement.equals(token) ? "" : " replacement '" + replacement + "'"); + } + + private void validate() { + // XXX not fool proof length test, should test codepoint by codepoint for mixed case user input? not even that will necessarily be 100% robust... + String asLow = toLowerCase(token); + // TODO: Put along with the global toLowerCase + String asHigh = token.toUpperCase(Locale.ENGLISH); + if (asLow.length() != token.length() || asHigh.length() != token.length()) { + throw new IllegalArgumentException("Special token '" + token + "' has case sensitive length. " + + "Please report this to the Vespa team."); + } + } + + } + +} diff --git a/linguistics/src/main/java/com/yahoo/language/process/TokenType.java b/linguistics/src/main/java/com/yahoo/language/process/TokenType.java index 57a5b6edb68..ad154d1b003 100644 --- a/linguistics/src/main/java/com/yahoo/language/process/TokenType.java +++ b/linguistics/src/main/java/com/yahoo/language/process/TokenType.java @@ -4,7 +4,7 @@ package com.yahoo.language.process; /** * An enumeration of token types. * - * @author <a href="mailto:mathiasm@yahoo-inc.com">Mathias Mølster Lidal</a> + * @author Mathias Mølster Lidal */ public enum TokenType { diff --git a/linguistics/src/main/java/com/yahoo/language/simple/SimpleLinguistics.java b/linguistics/src/main/java/com/yahoo/language/simple/SimpleLinguistics.java index e1a04b2985d..4ffe2a866d8 100644 --- a/linguistics/src/main/java/com/yahoo/language/simple/SimpleLinguistics.java +++ b/linguistics/src/main/java/com/yahoo/language/simple/SimpleLinguistics.java @@ -11,10 +11,14 @@ import com.yahoo.language.process.GramSplitter; import com.yahoo.language.process.Normalizer; import com.yahoo.language.process.Segmenter; import com.yahoo.language.process.SegmenterImpl; +import com.yahoo.language.process.SpecialTokenRegistry; import com.yahoo.language.process.Stemmer; import com.yahoo.language.process.StemmerImpl; import com.yahoo.language.process.Tokenizer; import com.yahoo.language.process.Transformer; +import com.yahoo.vespa.configdefinition.SpecialtokensConfig; + +import java.util.List; /** * Factory of simple linguistic processor implementations. @@ -31,6 +35,7 @@ public class SimpleLinguistics implements Linguistics { private final Detector detector; private final CharacterClasses characterClasses; private final GramSplitter gramSplitter; + private final SpecialTokenRegistry specialTokenRegistry = new SpecialTokenRegistry(List.of()); @Inject public SimpleLinguistics() { @@ -45,7 +50,7 @@ public class SimpleLinguistics implements Linguistics { public Stemmer getStemmer() { return new StemmerImpl(getTokenizer()); } @Override - public Tokenizer getTokenizer() { return new SimpleTokenizer(normalizer, transformer); } + public Tokenizer getTokenizer() { return new SimpleTokenizer(normalizer, transformer, specialTokenRegistry); } @Override public Normalizer getNormalizer() { return normalizer; } diff --git a/linguistics/src/main/java/com/yahoo/language/simple/SimpleTokenizer.java b/linguistics/src/main/java/com/yahoo/language/simple/SimpleTokenizer.java index 7df432f496d..740307c0cca 100644 --- a/linguistics/src/main/java/com/yahoo/language/simple/SimpleTokenizer.java +++ b/linguistics/src/main/java/com/yahoo/language/simple/SimpleTokenizer.java @@ -23,11 +23,13 @@ import java.util.logging.Level; */ public class SimpleTokenizer implements Tokenizer { + private static final Logger log = Logger.getLogger(SimpleTokenizer.class.getName()); private final static int SPACE_CODE = 32; + private final Normalizer normalizer; private final Transformer transformer; private final KStemmer stemmer = new KStemmer(); - private static final Logger log = Logger.getLogger(SimpleTokenizer.class.getName()); + private final SpecialTokenRegistry specialTokenRegistry; public SimpleTokenizer() { this(new SimpleNormalizer(), new SimpleTransformer()); @@ -38,8 +40,13 @@ public class SimpleTokenizer implements Tokenizer { } public SimpleTokenizer(Normalizer normalizer, Transformer transformer) { + this(normalizer, transformer, new SpecialTokenRegistry(List.of())); + } + + public SimpleTokenizer(Normalizer normalizer, Transformer transformer, SpecialTokenRegistry specialTokenRegistry) { this.normalizer = normalizer; this.transformer = transformer; + this.specialTokenRegistry = specialTokenRegistry; } @Override @@ -56,8 +63,8 @@ public class SimpleTokenizer implements Tokenizer { String original = input.substring(prev, next); String token = processToken(original, language, stemMode, removeAccents); tokens.add(new SimpleToken(original).setOffset(prev) - .setType(prevType) - .setTokenString(token)); + .setType(prevType) + .setTokenString(token)); prev = next; prevType = nextType; } @@ -67,20 +74,20 @@ public class SimpleTokenizer implements Tokenizer { } private String processToken(String token, Language language, StemMode stemMode, boolean removeAccents) { - final String original = token; - log.log(Level.FINEST, () -> "processToken '"+original+"'"); + String original = token; + log.log(Level.FINEST, () -> "processToken '" + original + "'"); token = normalizer.normalize(token); token = LinguisticsCase.toLowerCase(token); if (removeAccents) token = transformer.accentDrop(token, language); if (stemMode != StemMode.NONE) { - final String oldToken = token; + String oldToken = token; token = stemmer.stem(token); - final String newToken = token; - log.log(Level.FINEST, () -> "stem '"+oldToken+"' to '"+newToken+"'"); + String newToken = token; + log.log(Level.FINEST, () -> "stem '" + oldToken+"' to '" + newToken+"'"); } - final String result = token; - log.log(Level.FINEST, () -> "processed token is: "+result); + String result = token; + log.log(Level.FINEST, () -> "processed token is: " + result); return result; } diff --git a/linguistics/src/test/java/com/yahoo/language/process/SpecialTokensTestCase.java b/linguistics/src/test/java/com/yahoo/language/process/SpecialTokensTestCase.java new file mode 100644 index 00000000000..47c3ba7933c --- /dev/null +++ b/linguistics/src/test/java/com/yahoo/language/process/SpecialTokensTestCase.java @@ -0,0 +1,40 @@ +// Copyright Verizon Media. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package com.yahoo.language.process; + +import com.yahoo.vespa.configdefinition.SpecialtokensConfig; +import org.junit.Test; + +import static org.junit.Assert.assertEquals; + +/** + * @author bratseth + */ +public class SpecialTokensTestCase { + + @Test + public void testSpecialTokensConfig() { + var builder = new SpecialtokensConfig.Builder(); + var tokenBuilder = new SpecialtokensConfig.Tokenlist.Builder(); + tokenBuilder.name("default"); + + var tokenListBuilder1 = new SpecialtokensConfig.Tokenlist.Tokens.Builder(); + tokenListBuilder1.token("c++"); + tokenListBuilder1.replace("cpp"); + tokenBuilder.tokens(tokenListBuilder1); + + var tokenListBuilder2 = new SpecialtokensConfig.Tokenlist.Tokens.Builder(); + tokenListBuilder2.token("..."); + tokenBuilder.tokens(tokenListBuilder2); + + builder.tokenlist(tokenBuilder); + + var registry = new SpecialTokenRegistry(builder.build()); + + var defaultTokens = registry.getSpecialTokens("default"); + assertEquals("default", defaultTokens.name()); + assertEquals(2, defaultTokens.asMap().size()); + assertEquals("cpp", defaultTokens.asMap().get("c++")); + assertEquals("...", defaultTokens.asMap().get("...")); + } + +} |