| 1 | /*- | |
| 2 | * #%L | |
| 3 | * io.earcam.utilitarian.site.search.offline | |
| 4 | * %% | |
| 5 | * Copyright (C) 2017 earcam | |
| 6 | * %% | |
| 7 | * SPDX-License-Identifier: (BSD-3-Clause OR EPL-1.0 OR Apache-2.0 OR MIT) | |
| 8 | * | |
| 9 | * You <b>must</b> choose to accept, in full - any individual or combination of | |
| 10 | * the following licenses: | |
| 11 | * <ul> | |
| 12 | * <li><a href="https://opensource.org/licenses/BSD-3-Clause">BSD-3-Clause</a></li> | |
| 13 | * <li><a href="https://www.eclipse.org/legal/epl-v10.html">EPL-1.0</a></li> | |
| 14 | * <li><a href="https://www.apache.org/licenses/LICENSE-2.0">Apache-2.0</a></li> | |
| 15 | * <li><a href="https://opensource.org/licenses/MIT">MIT</a></li> | |
| 16 | * </ul> | |
| 17 | * #L% | |
| 18 | */ | |
| 19 | package io.earcam.utilitarian.site.search.offline; | |
| 20 | ||
| 21 | import static io.earcam.unexceptional.Closing.closeAfterAccepting; | |
| 22 | import static io.earcam.unexceptional.Exceptional.apply; | |
| 23 | import static io.earcam.utilitarian.site.search.offline.Component.getOrDefault; | |
| 24 | import static io.earcam.utilitarian.site.search.offline.Component.mandatory; | |
| 25 | import static io.earcam.utilitarian.site.search.offline.Javascript.createJavascriptEngine; | |
| 26 | import static io.earcam.utilitarian.site.search.offline.Javascript.invokeFunction; | |
| 27 | import static io.earcam.utilitarian.site.search.offline.Resources.SCRIPT_INDEX; | |
| 28 | import static io.earcam.utilitarian.site.search.offline.Resources.SCRIPT_SEARCH; | |
| 29 | import static io.earcam.utilitarian.site.search.offline.Resources.getResource; | |
| 30 | import static java.nio.charset.StandardCharsets.UTF_8; | |
| 31 | import static java.util.Arrays.stream; | |
| 32 | import static java.util.Collections.emptyMap; | |
| 33 | import static java.util.function.Function.identity; | |
| 34 | import static java.util.stream.Collectors.joining; | |
| 35 | import static java.util.stream.Collectors.toMap; | |
| 36 | ||
| 37 | import java.io.FileOutputStream; | |
| 38 | import java.io.IOException; | |
| 39 | import java.io.InputStream; | |
| 40 | import java.io.OutputStream; | |
| 41 | import java.nio.charset.Charset; | |
| 42 | import java.nio.file.Path; | |
| 43 | import java.nio.file.Paths; | |
| 44 | import java.util.HashMap; | |
| 45 | import java.util.Map; | |
| 46 | import java.util.Objects; | |
| 47 | import java.util.SortedSet; | |
| 48 | import java.util.TreeSet; | |
| 49 | import java.util.function.BiConsumer; | |
| 50 | import java.util.function.Consumer; | |
| 51 | import java.util.stream.Stream; | |
| 52 | import java.util.zip.GZIPOutputStream; | |
| 53 | ||
| 54 | import javax.annotation.WillNotClose; | |
| 55 | import javax.script.Invocable; | |
| 56 | import javax.script.ScriptException; | |
| 57 | ||
| 58 | import org.slf4j.Logger; | |
| 59 | import org.slf4j.LoggerFactory; | |
| 60 | ||
| 61 | //TODO test maven SOURCE filtering to replace VERSION_* constants ... but will IDE/Eclipse do the replacement before test exec? | |
| 62 | public class DefaultIndexer implements Indexer { | |
| 63 | ||
| 64 | private static final Logger LOG = LoggerFactory.getLogger(DefaultIndexer.class); | |
| 65 | ||
| 66 | public static final String BASEDIR_WEBJARS_RESOURCES = "META-INF/resources/webjars/"; | |
| 67 | public static final String VERSION_LUNR_JS = "2.1.0"; | |
| 68 | public static final String LUNR = BASEDIR_WEBJARS_RESOURCES + "lunr.js/" + VERSION_LUNR_JS + "/lunr.js"; | |
| 69 | ||
| 70 | public static final String OUTPUT_FILE = "outputFile"; | |
| 71 | public static final String FIELDS = "fields"; | |
| 72 | public static final String MAP_TITLES = "mapTitles"; | |
| 73 | public static final String GENERATE_AUTOCOMPLETE = "generateAutocomplete"; | |
| 74 | public static final String OUTPUT_CHARSET = "outputCharset"; | |
| 75 | ||
| 76 | private Path outputFile; | |
| 77 | ||
| 78 | @SuppressWarnings("squid:S1845") | |
| 79 | private String[] fields; | |
| 80 | private String refUrl; | |
| 81 | private Charset outputCharset; | |
| 82 | ||
| 83 | private Map<String, String> titlesMap = new HashMap<>(); | |
| 84 | private SortedSet<String> autocomplete = new TreeSet<>(); | |
| 85 | ||
| 86 | private BiConsumer<String, String> titleMapper = titlesMap::put; | |
| 87 | private Consumer<String> autocompleter = autocomplete::add; | |
| 88 | ||
| 89 | private Invocable engine; | |
| 90 | private Object javascriptIndexBuilder; | |
| 91 | ||
| 92 | ||
| 93 | @Override | |
| 94 | public void configure(Map<String, String> configuration) | |
| 95 | { | |
| 96 | outputCharset = getOrDefault(configuration, OUTPUT_CHARSET, UTF_8); | |
| 97 | refUrl = mandatory(configuration, Document.REF_URL); | |
| 98 | outputFile = Paths.get(mandatory(configuration, OUTPUT_FILE)); | |
| 99 | fields = mandatory(configuration, FIELDS).split(","); | |
| 100 | ||
| 101 |
1
1. configure : negated conditional → KILLED |
if(!getOrDefault(configuration, MAP_TITLES, true)) { |
| 102 | titleMapper = (u, t) -> { /* noop */ }; | |
| 103 | } | |
| 104 | ||
| 105 |
1
1. configure : negated conditional → KILLED |
if(!getOrDefault(configuration, GENERATE_AUTOCOMPLETE, true)) { |
| 106 | autocompleter = d -> { /* noop */ }; | |
| 107 | } | |
| 108 | ||
| 109 |
1
1. configure : removed call to io/earcam/utilitarian/site/search/offline/DefaultIndexer::initialize → KILLED |
initialize(); |
| 110 | } | |
| 111 | ||
| 112 | ||
| 113 | private void initialize() | |
| 114 | { | |
| 115 | engine = createSearchEngine(SCRIPT_INDEX); | |
| 116 | ||
| 117 |
1
1. lambda$initialize$2 : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::lambda$initialize$2 to ( if (x != null) null else throw new RuntimeException ) → KILLED |
Map<String, Map<Object, Object>> fieldConfigurations = stream(fields).collect(toMap(identity(), v -> emptyMap())); |
| 118 | javascriptIndexBuilder = invokeFunction(engine, "createIndexBuilder", refUrl, fieldConfigurations); | |
| 119 | } | |
| 120 | ||
| 121 | ||
| 122 | static Invocable createSearchEngine(String script) | |
| 123 | { | |
| 124 | InputStream lunr = getResource(LUNR); | |
| 125 | InputStream indexScript = getResource(script); | |
| 126 | Objects.requireNonNull(lunr, "Could not load lunrjs lib"); | |
| 127 | Objects.requireNonNull(indexScript, "Could not load indexScript"); | |
| 128 |
1
1. createSearchEngine : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::createSearchEngine to ( if (x != null) null else throw new RuntimeException ) → KILLED |
return createJavascriptEngine(lunr, indexScript); |
| 129 | } | |
| 130 | ||
| 131 | ||
| 132 | @Override | |
| 133 | public synchronized Indexer add(Stream<Document> documents) | |
| 134 | { | |
| 135 | invokeFunction(engine, "addDocuments", javascriptIndexBuilder, documents | |
| 136 | .filter(Document::hasTokens) | |
| 137 |
1
1. lambda$add$3 : removed call to java/util/function/BiConsumer::accept → KILLED |
.peek(d -> titleMapper.accept(d.refUrl(), d.title())) |
| 138 |
1
1. lambda$add$4 : removed call to java/util/List::forEach → KILLED |
.peek(d -> d.tokens().forEach(autocompleter::accept)) |
| 139 | .map(Document::asMap) | |
| 140 | .iterator()); | |
| 141 |
1
1. add : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::add to ( if (x != null) null else throw new RuntimeException ) → SURVIVED |
return this; |
| 142 | } | |
| 143 | ||
| 144 | ||
| 145 | @Override | |
| 146 | public void writeJson() | |
| 147 | { | |
| 148 | outputFile.getParent().toFile().mkdirs(); | |
| 149 | ||
| 150 |
1
1. writeJson : negated conditional → KILLED |
if(isGzip()) { |
| 151 | FileOutputStream fos = apply(FileOutputStream::new, outputFile.toFile()); | |
| 152 |
1
1. writeJson : removed call to io/earcam/unexceptional/Closing::closeAfterAccepting → KILLED |
closeAfterAccepting(GZIPOutputStream::new, fos, this::writeJson); |
| 153 | } else { | |
| 154 |
1
1. writeJson : removed call to io/earcam/unexceptional/Closing::closeAfterAccepting → KILLED |
closeAfterAccepting(FileOutputStream::new, outputFile.toFile(), this::writeJson); |
| 155 | } | |
| 156 | } | |
| 157 | ||
| 158 | ||
| 159 | private boolean isGzip() | |
| 160 | { | |
| 161 |
1
1. isGzip : replaced return of integer sized value with (x == 0 ? 1 : 0) → KILLED |
return outputFile.getFileName().toString().endsWith(".gz"); |
| 162 | } | |
| 163 | ||
| 164 | ||
| 165 | protected void writeJson(@WillNotClose OutputStream output) throws IOException | |
| 166 | { | |
| 167 |
1
1. writeJson : removed call to io/earcam/utilitarian/site/search/offline/DefaultIndexer::writeIndex → SURVIVED |
writeIndex(output); |
| 168 |
1
1. writeJson : removed call to io/earcam/utilitarian/site/search/offline/DefaultIndexer::writeAutocomplete → KILLED |
writeAutocomplete(output); |
| 169 |
1
1. writeJson : removed call to io/earcam/utilitarian/site/search/offline/DefaultIndexer::writeTitleMap → KILLED |
writeTitleMap(output); |
| 170 | } | |
| 171 | ||
| 172 | ||
| 173 | private void writeIndex(OutputStream output) throws IOException | |
| 174 | { | |
| 175 |
1
1. writeIndex : removed call to java/io/OutputStream::write → SURVIVED |
output.write(bytes("{\n\n\"index\": ")); |
| 176 | String indexJson = serializeIndex(); | |
| 177 | byte[] bytes = bytes(indexJson); | |
| 178 |
1
1. writeIndex : removed call to java/io/OutputStream::write → SURVIVED |
output.write(bytes); |
| 179 | String id = id(); | |
| 180 | LOG.debug("{} wrote {} bytes for index to {}", id, bytes.length, outputFile); | |
| 181 | } | |
| 182 | ||
| 183 | ||
| 184 | public byte[] bytes(String text) | |
| 185 | { | |
| 186 |
1
1. bytes : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::bytes to ( if (x != null) null else throw new RuntimeException ) → KILLED |
return text.getBytes(outputCharset); |
| 187 | } | |
| 188 | ||
| 189 | ||
| 190 | public String serializeIndex() | |
| 191 | { | |
| 192 |
1
1. serializeIndex : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::serializeIndex to ( if (x != null) null else throw new RuntimeException ) → KILLED |
return (String) invokeFunction(engine, "buildSerializedIndex", javascriptIndexBuilder); |
| 193 | } | |
| 194 | ||
| 195 | ||
| 196 | private void writeAutocomplete(OutputStream output) throws IOException | |
| 197 | { | |
| 198 | byte[] bytes = bytes(autocomplete.stream().collect(joining("\", \"", ",\n\n\"autocomplete\": [\"", "\"]"))); | |
| 199 |
1
1. writeAutocomplete : removed call to java/io/OutputStream::write → KILLED |
output.write(bytes); |
| 200 | String id = id(); | |
| 201 | LOG.debug("{} wrote {} bytes for {} words for autocomplete to {}", id, bytes.length, autocomplete.size(), outputFile); | |
| 202 | } | |
| 203 | ||
| 204 | ||
| 205 | private void writeTitleMap(OutputStream output) throws IOException | |
| 206 | { | |
| 207 | byte[] bytes = bytes(titlesMap.entrySet().stream().map( | |
| 208 |
1
1. lambda$writeTitleMap$5 : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::lambda$writeTitleMap$5 to ( if (x != null) null else throw new RuntimeException ) → KILLED |
e -> new StringBuilder() |
| 209 | .append('"').append(e.getKey()).append('"') | |
| 210 | .append(':') | |
| 211 | .append('"').append(e.getValue()).append('"')) | |
| 212 | .collect(joining(", ", ",\n\n\"titleMap\": {", "}\n}"))); | |
| 213 |
1
1. writeTitleMap : removed call to java/io/OutputStream::write → KILLED |
output.write(bytes); |
| 214 | String id = id(); | |
| 215 | LOG.debug("{} wrote {} bytes for {} entries for title map to {}", id, bytes.length, titlesMap.size(), outputFile); | |
| 216 | } | |
| 217 | ||
| 218 | ||
| 219 | public static String search(String indexJson, String query) throws ScriptException, NoSuchMethodException | |
| 220 | { | |
| 221 | Invocable engine = createSearchEngine(SCRIPT_SEARCH); | |
| 222 |
1
1. search : mutated return of Object value for io/earcam/utilitarian/site/search/offline/DefaultIndexer::search to ( if (x != null) null else throw new RuntimeException ) → NO_COVERAGE |
return engine.invokeFunction("jsonSearchIndex", indexJson, query).toString(); |
| 223 | } | |
| 224 | } | |
Mutations | ||
| 101 |
1.1 |
|
| 105 |
1.1 |
|
| 109 |
1.1 |
|
| 117 |
1.1 |
|
| 128 |
1.1 |
|
| 137 |
1.1 |
|
| 138 |
1.1 |
|
| 141 |
1.1 |
|
| 150 |
1.1 |
|
| 152 |
1.1 |
|
| 154 |
1.1 |
|
| 161 |
1.1 |
|
| 167 |
1.1 |
|
| 168 |
1.1 |
|
| 169 |
1.1 |
|
| 175 |
1.1 |
|
| 178 |
1.1 |
|
| 186 |
1.1 |
|
| 192 |
1.1 |
|
| 199 |
1.1 |
|
| 208 |
1.1 |
|
| 213 |
1.1 |
|
| 222 |
1.1 |