Examples with IndexFacts com.yahoo.prelude.IndexFacts used on opensource projects

Example 26 with IndexFacts

use of com.yahoo.prelude.IndexFacts in project vespa by vespa-engine.

the class TokenizerTestCase method testExactMatchTokenizationEndsByColon.

@Test
public void testExactMatchTokenizationEndsByColon() {
    Index index1 = new Index("testexact1");
    index1.setExact(true, null);
    Index index2 = new Index("testexact2");
    index2.setExact(true, "()/aa*::*&");
    IndexFacts facts = new IndexFacts();
    facts.addIndex("testsd", index1);
    facts.addIndex("testsd", index2);
    Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics());
    IndexFacts.Session session = facts.newSession(Collections.emptySet(), Collections.emptySet());
    List<?> tokens = tokenizer.tokenize("normal a:b (normal testexact1:!/%#%&+-+ ) testexact2:ho_/&%&/()/aa*::*&b:", session);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "a"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "b"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    assertEquals(new Token(LBRACE, "("), tokens.get(6));
    assertEquals(new Token(WORD, "normal"), tokens.get(7));
    assertEquals(new Token(SPACE, " "), tokens.get(8));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(9));
    assertEquals(new Token(COLON, ":"), tokens.get(10));
    assertEquals(new Token(WORD, "!/%#%&+-+"), tokens.get(11));
    assertEquals(new Token(SPACE, " "), tokens.get(12));
    assertEquals(new Token(RBRACE, ")"), tokens.get(13));
    assertEquals(new Token(SPACE, " "), tokens.get(14));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(15));
    assertEquals(new Token(COLON, ":"), tokens.get(16));
    assertEquals(new Token(WORD, "ho_/&%&/"), tokens.get(17));
    assertEquals(new Token(WORD, "b"), tokens.get(18));
    assertEquals(new Token(COLON, ":"), tokens.get(19));
}

Example 27 with IndexFacts

use of com.yahoo.prelude.IndexFacts in project vespa by vespa-engine.

the class TokenizerTestCase method testExactMatchTokenizationWithTerminatorTerminatedByEndOfString.

@Test
public void testExactMatchTokenizationWithTerminatorTerminatedByEndOfString() {
    Index index1 = new Index("testexact1");
    index1.setExact(true, null);
    Index index2 = new Index("testexact2");
    index2.setExact(true, "()/aa*::*&");
    IndexFacts facts = new IndexFacts();
    facts.addIndex("testsd", index1);
    facts.addIndex("testsd", index2);
    Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics());
    IndexFacts.Session session = facts.newSession(Collections.emptySet(), Collections.emptySet());
    List<?> tokens = tokenizer.tokenize("normal a:b (normal testexact1:/,%#%&+-+ ) testexact2:ho_/&%&/()/aa*::*", session);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "a"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "b"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    assertEquals(new Token(LBRACE, "("), tokens.get(6));
    assertEquals(new Token(WORD, "normal"), tokens.get(7));
    assertEquals(new Token(SPACE, " "), tokens.get(8));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(9));
    assertEquals(new Token(COLON, ":"), tokens.get(10));
    assertEquals(new Token(WORD, "/,%#%&+-+"), tokens.get(11));
    assertEquals(new Token(SPACE, " "), tokens.get(12));
    assertEquals(new Token(RBRACE, ")"), tokens.get(13));
    assertEquals(new Token(SPACE, " "), tokens.get(14));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(15));
    assertEquals(new Token(COLON, ":"), tokens.get(16));
    assertEquals(new Token(WORD, "ho_/&%&/()/aa*::*"), tokens.get(17));
    assertTrue(((Token) tokens.get(17)).isSpecial());
}

Example 28 with IndexFacts

use of com.yahoo.prelude.IndexFacts in project vespa by vespa-engine.

the class TokenizerTestCase method testExactMatchHeuristics.

@Test
public void testExactMatchHeuristics() {
    Index index1 = new Index("testexact1");
    index1.setExact(true, null);
    Index index2 = new Index("testexact2");
    index2.setExact(true, "()/aa*::*&");
    IndexFacts indexFacts = new IndexFacts();
    indexFacts.addIndex("testsd", index1);
    indexFacts.addIndex("testsd", index2);
    IndexFacts.Session facts = indexFacts.newSession(Collections.emptySet(), Collections.emptySet());
    Tokenizer tokenizer = new Tokenizer(new SimpleLinguistics());
    List<?> tokens = tokenizer.tokenize("normal a:b (normal testexact1:foo) testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "a"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "b"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    assertEquals(new Token(LBRACE, "("), tokens.get(6));
    assertEquals(new Token(WORD, "normal"), tokens.get(7));
    assertEquals(new Token(SPACE, " "), tokens.get(8));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(9));
    assertEquals(new Token(COLON, ":"), tokens.get(10));
    assertEquals(new Token(WORD, "foo"), tokens.get(11));
    assertEquals(new Token(RBRACE, ")"), tokens.get(12));
    assertEquals(new Token(SPACE, " "), tokens.get(13));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(14));
    assertEquals(new Token(COLON, ":"), tokens.get(15));
    assertEquals(new Token(WORD, "bar"), tokens.get(16));
    tokens = tokenizer.tokenize("testexact1:a*teens", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "a*teens"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1:foo\"bar", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo\"bar"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1:foo!bar", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo!bar"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1:foo! ", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(3));
    assertEquals(new Token(SPACE, " "), tokens.get(4));
    tokens = tokenizer.tokenize("testexact1:foo!! ", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(3));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    tokens = tokenizer.tokenize("testexact1:foo!100 ", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(3));
    assertEquals(new Token(NUMBER, "100"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    tokens = tokenizer.tokenize("testexact1:foo*!100 ", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    assertEquals(new Token(STAR, "*"), tokens.get(3));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(4));
    assertEquals(new Token(NUMBER, "100"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    tokens = tokenizer.tokenize("testexact1: *\"foo bar\"*!100 ", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(STAR, "*"), tokens.get(2));
    assertEquals(new Token(WORD, "foo bar"), tokens.get(3));
    assertEquals(new Token(STAR, "*"), tokens.get(4));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(5));
    assertEquals(new Token(NUMBER, "100"), tokens.get(6));
    assertEquals(new Token(SPACE, " "), tokens.get(7));
    tokens = tokenizer.tokenize("testexact1: *\"foo bar\"*!100", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(STAR, "*"), tokens.get(2));
    assertEquals(new Token(WORD, "foo bar"), tokens.get(3));
    assertEquals(new Token(STAR, "*"), tokens.get(4));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(5));
    assertEquals(new Token(NUMBER, "100"), tokens.get(6));
    tokens = tokenizer.tokenize("testexact1: *foobar*!100", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(STAR, "*"), tokens.get(2));
    assertEquals(new Token(WORD, "foobar"), tokens.get(3));
    assertEquals(new Token(STAR, "*"), tokens.get(4));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(5));
    assertEquals(new Token(NUMBER, "100"), tokens.get(6));
    tokens = tokenizer.tokenize("testexact1: *foobar*!100!", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(STAR, "*"), tokens.get(2));
    assertEquals(new Token(WORD, "foobar*!100"), tokens.get(3));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(4));
    tokens = tokenizer.tokenize("testexact1:foo(bar)", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo(bar)"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1:\"foo\"", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1: foo", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1: \"foo\"", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1: \"foo\"", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "foo"), tokens.get(2));
    tokens = tokenizer.tokenize("testexact1:vespa testexact2:resolved", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "vespa"), tokens.get(2));
    assertEquals(new Token(SPACE, " "), tokens.get(3));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(4));
    assertEquals(new Token(COLON, ":"), tokens.get(5));
    assertEquals(new Token(WORD, "resolved"), tokens.get(6));
    tokens = tokenizer.tokenize("testexact1:\"news search\" testexact2:resolved", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "news search"), tokens.get(2));
    assertEquals(new Token(SPACE, " "), tokens.get(3));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(4));
    assertEquals(new Token(COLON, ":"), tokens.get(5));
    assertEquals(new Token(WORD, "resolved"), tokens.get(6));
    tokens = tokenizer.tokenize("(testexact1:\"news search\" testexact1:vespa)", facts);
    assertEquals(new Token(LBRACE, "("), tokens.get(0));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(1));
    assertEquals(new Token(COLON, ":"), tokens.get(2));
    assertEquals(new Token(WORD, "news search"), tokens.get(3));
    assertEquals(new Token(SPACE, " "), tokens.get(4));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(5));
    assertEquals(new Token(COLON, ":"), tokens.get(6));
    assertEquals(new Token(WORD, "vespa"), tokens.get(7));
    assertEquals(new Token(RBRACE, ")"), tokens.get(8));
    tokens = tokenizer.tokenize("testexact1:news*", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "news"), tokens.get(2));
    assertEquals(new Token(STAR, "*"), tokens.get(3));
    tokens = tokenizer.tokenize("testexact1:\"news\"*", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "news"), tokens.get(2));
    assertEquals(new Token(STAR, "*"), tokens.get(3));
    tokens = tokenizer.tokenize("testexact1:\"news search\"!200", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "news search"), tokens.get(2));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(3));
    assertEquals(new Token(NUMBER, "200"), tokens.get(4));
    tokens = tokenizer.tokenize("testexact1:vespa!200", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(WORD, "vespa"), tokens.get(2));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(3));
    assertEquals(new Token(NUMBER, "200"), tokens.get(4));
    tokens = tokenizer.tokenize("testexact1:*\"news\"*", facts);
    assertEquals(new Token(WORD, "testexact1"), tokens.get(0));
    assertEquals(new Token(COLON, ":"), tokens.get(1));
    assertEquals(new Token(STAR, "*"), tokens.get(2));
    assertEquals(new Token(WORD, "news"), tokens.get(3));
    assertEquals(new Token(STAR, "*"), tokens.get(4));
    tokens = tokenizer.tokenize("normal(testexact1:foo) testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(LBRACE, "("), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "foo"), tokens.get(4));
    assertEquals(new Token(RBRACE, ")"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(7));
    assertEquals(new Token(COLON, ":"), tokens.get(8));
    assertEquals(new Token(WORD, "bar"), tokens.get(9));
    tokens = tokenizer.tokenize("normal testexact1:(foo testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "(foo"), tokens.get(4));
    assertEquals(new Token(SPACE, " "), tokens.get(5));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(6));
    assertEquals(new Token(COLON, ":"), tokens.get(7));
    assertEquals(new Token(WORD, "bar"), tokens.get(8));
    tokens = tokenizer.tokenize("normal testexact1:foo! testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "foo"), tokens.get(4));
    assertEquals(new Token(EXCLAMATION, "!"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(7));
    assertEquals(new Token(COLON, ":"), tokens.get(8));
    assertEquals(new Token(WORD, "bar"), tokens.get(9));
    tokens = tokenizer.tokenize("normal testexact1:foo* testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "foo"), tokens.get(4));
    assertEquals(new Token(STAR, "*"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(7));
    assertEquals(new Token(COLON, ":"), tokens.get(8));
    assertEquals(new Token(WORD, "bar"), tokens.get(9));
    tokens = tokenizer.tokenize("normal testexact1: foo* testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, "foo"), tokens.get(4));
    assertEquals(new Token(STAR, "*"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(7));
    assertEquals(new Token(COLON, ":"), tokens.get(8));
    assertEquals(new Token(WORD, "bar"), tokens.get(9));
    tokens = tokenizer.tokenize("normal testexact1:\" foo\"* testexact2:bar", facts);
    assertEquals(new Token(WORD, "normal"), tokens.get(0));
    assertEquals(new Token(SPACE, " "), tokens.get(1));
    assertEquals(new Token(WORD, "testexact1"), tokens.get(2));
    assertEquals(new Token(COLON, ":"), tokens.get(3));
    assertEquals(new Token(WORD, " foo"), tokens.get(4));
    assertEquals(new Token(STAR, "*"), tokens.get(5));
    assertEquals(new Token(SPACE, " "), tokens.get(6));
    assertEquals(new Token(WORD, "testexact2"), tokens.get(7));
    assertEquals(new Token(COLON, ":"), tokens.get(8));
    assertEquals(new Token(WORD, "bar"), tokens.get(9));
}

Example 29 with IndexFacts

use of com.yahoo.prelude.IndexFacts in project vespa by vespa-engine.

the class IndexCombinatorTestCase method setUp.

@Before
public void setUp() throws Exception {
    transformer = new IndexCombinatorSearcher();
    f = new IndexFacts();
    f.addIndex("one", "z");
    Index i = new Index("default");
    i.addCommand("match-group a i");
    f.addIndex("one", i);
}

Example 30 with IndexFacts

use of com.yahoo.prelude.IndexFacts in project vespa by vespa-engine.

the class ExecutionTestCase method testContextCacheSingleLengthSearchChain.

public void testContextCacheSingleLengthSearchChain() {
    IndexFacts[] contextsBefore = new IndexFacts[1];
    IndexFacts[] contextsAfter = new IndexFacts[1];
    List<Searcher> l = new ArrayList<>(1);
    l.add(new ContextCacheSearcher(0, contextsBefore, contextsAfter));
    Chain<Searcher> chain = new Chain<>(l);
    Query query = new Query("?mutatecontext=0");
    new Execution(chain, Execution.Context.createContextStub()).search(query);
    assertEquals(contextsBefore[0], contextsAfter[0]);
    assertSame(contextsBefore[0], contextsAfter[0]);
}