Examples with ComplexDocument com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument used on opensource projects

Example 1 with ComplexDocument

use of com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument in project fdb-record-layer by FoundationDB.

the class TextIndexTest method queryComplexDocumentsWithAdditionalFilters.

@Test
public void queryComplexDocumentsWithAdditionalFilters() throws Exception {
    final List<String> textSamples = Arrays.asList(TextSamples.ANGSTROM, TextSamples.ROMEO_AND_JULIET_PROLOGUE, TextSamples.AETHELRED, TextSamples.FRENCH, TextSamples.GERMAN, TextSamples.ROMEO_AND_JULIET_PROLOGUE, TextSamples.YIDDISH, "Napoleon and the Duke of Wellington met in Waterloo in 1815.");
    final List<ComplexDocument> documents = IntStream.range(0, textSamples.size()).mapToObj(i -> ComplexDocument.newBuilder().setDocId(i).setGroup(i % 2).setText(textSamples.get(i)).addTag("3:" + (i % 3)).setScore(i).build()).collect(Collectors.toList());
    final Index rankIndex = new Index("Complex$rank(score)", field("score").groupBy(field("group")), IndexTypes.RANK);
    try (FDBRecordContext context = openContext()) {
        openRecordStore(context, metaDataBuilder -> {
            final RecordTypeBuilder complexDocRecordType = metaDataBuilder.getRecordType(COMPLEX_DOC);
            metaDataBuilder.addIndex(complexDocRecordType, COMPLEX_TEXT_BY_GROUP);
            metaDataBuilder.addIndex(complexDocRecordType, rankIndex);
        });
        documents.forEach(recordStore::saveRecord);
        assertEquals(Collections.singletonList(Tuple.from(1L, 5L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.field("tag").oneOfThem().equalsValue("3:2"), 1, 758136568));
        assertEquals(Arrays.asList(Tuple.from(1L, 1L), Tuple.from(1L, 5L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.field("text").text().containsPrefix("continu"), 1, -1043653062));
        assertEquals(Collections.singletonList(Tuple.from(1L, 7L)), queryComplexDocumentsWithIndex(Query.field("text").text().contains("napoleon"), Query.and(Query.field("text").text().containsPrefix("th"), Query.field("text").text().contains("waterloo")), 1, -754900112));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.not(Query.field("tag").oneOfThem().equalsValue("3:2")), 1, 758136569));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.not(Query.field("tag").oneOfThem().equalsValue("3:2")), 1, 758136569));
        assertEquals(Arrays.asList(Tuple.from(0L, 0L), Tuple.from(0L, 6L)), queryComplexDocumentsWithOr((OrComponent) Query.or(Query.field("text").text().containsAll("unit named after"), Query.field("text").text().containsPhrase("אן ארמיי און פלאט")), 0, -1558384887));
        assertEquals(Collections.singletonList(Tuple.from(1L, 5L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.or(Query.field("tag").oneOfThem().equalsValue("3:2"), Query.field("tag").oneOfThem().equalsValue("3:0")), true, 1, -27568755));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("fearful passage love", 7), Query.rank(field("score").groupBy(field("group"))).lessThan(2L), true, 1, -2132208833));
        assertEquals(Collections.singletonList(Tuple.from(1L, 5L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAllPrefixes("fear pass love", true), Query.field("tag").oneOfThem().equalsValue("3:2"), true, 1, -419325379));
        assertEquals(Collections.singletonList(Tuple.from(1L, 5L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAllPrefixes("fear pass love", false), Query.field("tag").oneOfThem().equalsValue("3:2"), false, 1, -1902024530));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAllPrefixes("fear pass love"), Query.rank(field("score").groupBy(field("group"))).lessThan(2L), true, 1, 669157421));
        commit(context);
    }
}

Example 2 with ComplexDocument

use of com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument in project fdb-record-layer by FoundationDB.

the class TextIndexTest method saveComplexMultiDocuments.

@Test
public void saveComplexMultiDocuments() throws Exception {
    final ComplexDocument shakespeareDocument = ComplexDocument.newBuilder().setGroup(0).setDocId(1623L).setText(TextSamples.ROMEO_AND_JULIET_PROLOGUE).addTag("a").addTag("b").build();
    // First, repeated key comes *before* text.
    try (FDBRecordContext context = openContext()) {
        openRecordStore(context, metaDataBuilder -> metaDataBuilder.addIndex(COMPLEX_DOC, COMPLEX_MULTI_TAG_INDEX));
        recordStore.saveRecord(shakespeareDocument);
        int indexKeys = getSaveIndexKeyCount(recordStore);
        assertEquals(164, indexKeys);
        List<Map.Entry<Tuple, List<Integer>>> entryList = scanMapEntries(recordStore, COMPLEX_MULTI_TAG_INDEX, Tuple.from("a", "civil"));
        assertEquals(Collections.singletonList(entryOf(Tuple.from(0L, 1623L), Arrays.asList(22, 25))), entryList);
        commit(context);
    }
    try (FDBRecordContext context = openContext()) {
        FDBRecordStore.deleteStore(context, recordStore.getSubspace());
        commit(context);
    }
    // Then, repeated key comes *after* text.
    try (FDBRecordContext context = openContext()) {
        openRecordStore(context, metaDataBuilder -> metaDataBuilder.addIndex(COMPLEX_DOC, COMPLEX_THEN_TAG_INDEX));
        recordStore.saveRecord(shakespeareDocument);
        int indexKeys = getSaveIndexKeyCount(recordStore);
        assertEquals(164, indexKeys);
        List<Map.Entry<Tuple, List<Integer>>> entryList = scanMapEntries(recordStore, COMPLEX_THEN_TAG_INDEX, Tuple.from("civil"));
        assertEquals(Arrays.asList(entryOf(Tuple.from("a", 0L, 1623L), Arrays.asList(22, 25)), entryOf(Tuple.from("b", 0L, 1623L), Arrays.asList(22, 25))), entryList);
        commit(context);
    }
}

Example 3 with ComplexDocument

use of com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument in project fdb-record-layer by FoundationDB.

the class TextIndexTest method saveComplexWithAggressiveConflictRanges.

@Test
public void saveComplexWithAggressiveConflictRanges() throws Exception {
    // These two documents are in different groups, so even with aggressive conflict
    // ranges, they should be able to be committed concurrently.
    final ComplexDocument zeroGroupDocument = ComplexDocument.newBuilder().setGroup(0).setDocId(1623L).setText(TextSamples.ROMEO_AND_JULIET_PROLOGUE).build();
    final ComplexDocument oneGroupDocument = ComplexDocument.newBuilder().setGroup(1).setDocId(1623L).setText(TextSamples.ROMEO_AND_JULIET_PROLOGUE).build();
    final RecordMetaDataHook hook = metaDataBuilder -> {
        final Index newIndex = new Index(COMPLEX_TEXT_BY_GROUP.getName(), COMBINED_TEXT_BY_GROUP.getRootExpression(), IndexTypes.TEXT, ImmutableMap.of(IndexOptions.TEXT_ADD_AGGRESSIVE_CONFLICT_RANGES_OPTION, "true"));
        metaDataBuilder.addIndex(COMPLEX_DOC, newIndex);
    };
    saveTwoRecordsConcurrently(hook, zeroGroupDocument, oneGroupDocument, true);
}

Example 4 with ComplexDocument

use of com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument in project fdb-record-layer by FoundationDB.

the class TextIndexTest method saveComplexDocuments.

@Test
public void saveComplexDocuments() throws Exception {
    ComplexDocument complexDocument = ComplexDocument.newBuilder().setGroup(0).setDocId(1066L).setText("Very complex. Not to be trifled with.").build();
    ComplexDocument shakespeareDocument = ComplexDocument.newBuilder().setGroup(0).setDocId(1623L).setText(TextSamples.ROMEO_AND_JULIET_PROLOGUE).addTag("a").addTag("b").build();
    ComplexDocument yiddishDocument = ComplexDocument.newBuilder().setGroup(1).setDocId(1944L).setText(TextSamples.YIDDISH).addTag("c").build();
    try (FDBRecordContext context = openContext()) {
        openRecordStore(context, metaDataBuilder -> metaDataBuilder.addIndex(COMPLEX_DOC, COMPLEX_TEXT_BY_GROUP));
        recordStore.saveRecord(complexDocument);
        int firstKeys = getSaveIndexKeyCount(recordStore);
        assertEquals(complexDocument.getText().split(" ").length, firstKeys);
        recordStore.saveRecord(shakespeareDocument);
        int secondKeys = getSaveIndexKeyCount(recordStore) - firstKeys;
        assertEquals(82, secondKeys);
        recordStore.saveRecord(yiddishDocument);
        int thirdKeys = getSaveIndexKeyCount(recordStore) - secondKeys - firstKeys;
        assertEquals(9, thirdKeys);
        List<Map.Entry<Tuple, List<Integer>>> entryList = scanMapEntries(recordStore, COMPLEX_TEXT_BY_GROUP, Tuple.from(0L, "to"));
        assertEquals(Arrays.asList(entryOf(Tuple.from(1066L), Collections.singletonList(3)), entryOf(Tuple.from(1623L), Arrays.asList(18, 108))), entryList);
        List<Message> recordList = recordStore.scanIndexRecords(COMPLEX_TEXT_BY_GROUP.getName(), BY_TEXT_TOKEN, TupleRange.allOf(Tuple.from(0L, "to")), null, ScanProperties.FORWARD_SCAN).map(FDBIndexedRecord::getRecord).asList().get();
        assertEquals(Arrays.asList(complexDocument, shakespeareDocument), recordList);
        entryList = toMapEntries(scanIndex(recordStore, COMPLEX_TEXT_BY_GROUP, TupleRange.prefixedBy("א").prepend(Tuple.from(1L))), null);
        assertEquals(Arrays.asList(entryOf(Tuple.from(1L, "א", 1944L), Arrays.asList(0, 3)), entryOf(Tuple.from(1L, "און", 1944L), Collections.singletonList(8)), entryOf(Tuple.from(1L, "איז", 1944L), Collections.singletonList(2)), entryOf(Tuple.from(1L, "אן", 1944L), Collections.singletonList(6)), entryOf(Tuple.from(1L, "ארמיי", 1944L), Collections.singletonList(7))), entryList);
        // Read the whole store and make sure the values come back in a somewhat sensible way
        entryList = toMapEntries(scanIndex(recordStore, COMPLEX_TEXT_BY_GROUP, TupleRange.ALL), null);
        assertEquals(firstKeys + secondKeys + thirdKeys, entryList.size());
        int i = 0;
        String last = null;
        for (Map.Entry<Tuple, List<Integer>> entry : entryList) {
            assertEquals(3, entry.getKey().size());
            if (i < firstKeys + secondKeys) {
                assertEquals(0L, entry.getKey().getLong(0));
                assertThat(entry.getKey().getLong(2), anyOf(is(1066L), is(1623L)));
            } else {
                assertEquals(1L, entry.getKey().getLong(0));
                assertEquals(1944L, entry.getKey().getLong(2));
                if (i == firstKeys + secondKeys) {
                    last = null;
                }
            }
            if (last != null) {
                assertThat(entry.getKey().getString(1), greaterThanOrEqualTo(last));
            }
            last = entry.getKey().getString(1);
            i++;
        }
        commit(context);
    }
}

Example 5 with ComplexDocument

use of com.apple.foundationdb.record.TestRecordsTextProto.ComplexDocument in project fdb-record-layer by FoundationDB.

the class TextIndexTest method queryComplexDocuments.

@Test
public void queryComplexDocuments() throws Exception {
    final List<String> textSamples = Arrays.asList(TextSamples.ANGSTROM, TextSamples.ROMEO_AND_JULIET_PROLOGUE, TextSamples.AETHELRED, TextSamples.FRENCH);
    final List<ComplexDocument> documents = IntStream.range(0, textSamples.size()).mapToObj(i -> ComplexDocument.newBuilder().setDocId(i).setGroup(i % 2).setText(textSamples.get(i)).build()).collect(Collectors.toList());
    try (FDBRecordContext context = openContext()) {
        openRecordStore(context, metaDataBuilder -> {
            final RecordTypeBuilder complexDocRecordType = metaDataBuilder.getRecordType(COMPLEX_DOC);
            metaDataBuilder.addIndex(complexDocRecordType, COMPLEX_TEXT_BY_GROUP);
        });
        documents.forEach(recordStore::saveRecord);
        assertEquals(Collections.singletonList(Tuple.from(0L, 0L)), queryComplexDocumentsWithIndex(Query.field("text").text().contains("angstrom"), 0, 372972877));
        assertEquals(Collections.emptyList(), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("civil blood parents' strife"), 0L, -1615886689));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("civil blood parents' strife"), 1L, -1615886658));
        assertEquals(Collections.emptyList(), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("civil blood parents' strife", 4), 1L, -1436111364));
        assertEquals(Collections.singletonList(Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAll("civil blood parents' strife", 35), 1L, -1436081573));
        assertEquals(Arrays.asList(Tuple.from(0L, 0L), Tuple.from(0L, 2L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAny("angstrom parents king napoleons"), 0L, -1092421072));
        assertEquals(Collections.singletonList(Tuple.from(1L, 3L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsPhrase("recu un Thiers"), 1L, 1395848801));
        assertEquals(Collections.singletonList(Tuple.from(0L, 0L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsPrefix("ang"), 0L, -1013515738));
        assertEquals(Arrays.asList(Tuple.from(1L, 3L), Tuple.from(1L, 1L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsPrefix("un"), 1L, -995158140));
        assertEquals(ImmutableSet.of(Tuple.from(0L, 0L), Tuple.from(0L, 2L)), ImmutableSet.copyOf(queryComplexDocumentsWithIndex(Query.field("text").text().containsAnyPrefix("ang par nap kin"), 0L, -1089713854)));
        assertEquals(Collections.singletonList(Tuple.from(0L, 0L)), queryComplexDocumentsWithIndex(Query.field("text").text().containsAllPrefixes("ang uni name", false), 0L, 646414402));
        commit(context);
    }
}