Examples with JLanguageTool org.languagetool.JLanguageTool used on opensource projects

Example 1 with JLanguageTool

use of org.languagetool.JLanguageTool in project languagetool by languagetool-org.

the class FalseFriendRuleTest method testHintsForGermanSpeakers.

@Test
public void testHintsForGermanSpeakers() throws IOException, ParserConfigurationException, SAXException {
    JLanguageTool langTool = new JLanguageTool(new English(), new German());
    List<RuleMatch> matches = assertErrors(1, "We will berate you.", langTool);
    assertEquals(matches.get(0).getSuggestedReplacements().toString(), "[provide advice, give advice]");
    assertErrors(0, "We will give you advice.", langTool);
    assertErrors(1, "I go to high school in Foocity.", langTool);
    List<RuleMatch> matches2 = assertErrors(1, "The chef", langTool);
    assertEquals("[boss, chief]", matches2.get(0).getSuggestedReplacements().toString());
}

Example 2 with JLanguageTool

use of org.languagetool.JLanguageTool in project languagetool by languagetool-org.

the class RuleNumberScalabilityTest method main.

public static void main(String[] args) throws IOException {
    if (args.length != 2) {
        System.out.println("Usage: " + RuleNumberScalabilityTest.class.getSimpleName() + " <languageCode> <text_file>");
        System.exit(1);
    }
    JLanguageTool langTool = new JLanguageTool(Languages.getLanguageForShortCode(args[0]));
    String text = StringTools.readStream(new FileInputStream(args[1]), "utf-8");
    System.out.println("Warmup...");
    langTool.check(text);
    langTool.check(text);
    long baselineTime = getBaselineTime(langTool, text);
    System.out.println("Baseline: " + baselineTime + "ms (time with no pattern rules active)");
    int ruleNumber = langTool.getAllActiveRules().size();
    System.out.println("Total rules: " + ruleNumber);
    int steps = 5;
    int prevActiveRules = -1;
    long prevCleanRunTime = -1;
    for (int i = steps; i > 0; i--) {
        int targetActiveRules = ruleNumber / i;
        deactivateAllRules(langTool);
        for (Rule rule : langTool.getAllRules()) {
            langTool.enableRule(rule.getId());
            if (langTool.getAllActiveRules().size() > targetActiveRules) {
                break;
            }
        }
        int activeRules = langTool.getAllActiveRules().size();
        long startTime = System.currentTimeMillis();
        langTool.check(text);
        long runTime = System.currentTimeMillis() - startTime;
        long cleanRunTime = runTime - baselineTime;
        if (prevActiveRules != -1 && prevCleanRunTime != -1) {
            float ruleFactor = (float) activeRules / prevActiveRules;
            float cleanRuntimeFactor = (float) cleanRunTime / prevCleanRunTime;
            System.out.println("Active rules: " + activeRules + ", runtime: " + runTime + "ms, cleanRunTime: " + cleanRunTime + ", ruleFactor: " + ruleFactor + ", cleanRuntimeFactor: " + cleanRuntimeFactor);
        } else {
            System.out.println("Active rules: " + activeRules + ", runtime: " + runTime + "ms, cleanRunTime: " + cleanRunTime);
        }
        prevActiveRules = activeRules;
        prevCleanRunTime = cleanRunTime;
    }
    System.out.println("ruleFactor = the number of rules compared to the previous run");
    System.out.println("cleanRuntimeFactor = the runtime (without baseline) compared to the previous run");
    System.out.println(" => cleanRuntimeFactor should not grow much more than ruleFactor, otherwise we scale");
    System.out.println(" => badly with respect to the number of rules");
}

Example 3 with JLanguageTool

use of org.languagetool.JLanguageTool in project languagetool by languagetool-org.

the class MultiWordChunkerTest method testDisambiguate.

@Test
public void testDisambiguate() throws Exception {
    Disambiguator chunker = new MultiWordChunker("/pl/multiwords.txt");
    JLanguageTool lt = new JLanguageTool(new English());
    AnalyzedSentence analyzedSentence = lt.getAnalyzedSentence("A test... More.");
    AnalyzedSentence disambiguated = chunker.disambiguate(analyzedSentence);
    AnalyzedTokenReadings[] tokens = disambiguated.getTokens();
    assertTrue(tokens[4].getReadings().toString().contains("<ELLIPSIS>"));
    assertTrue(tokens[6].getReadings().toString().contains("</ELLIPSIS>"));
}

Example 4 with JLanguageTool

use of org.languagetool.JLanguageTool in project languagetool by languagetool-org.

the class MultiWordChunkerTest method testDisambiguateMultiSpace.

@Test
public void testDisambiguateMultiSpace() throws Exception {
    Disambiguator chunker = new MultiWordChunker("/uk/multiwords.txt");
    JLanguageTool lt = new JLanguageTool(new Ukrainian());
    AnalyzedSentence analyzedSentence = lt.getAnalyzedSentence("для  годиться.");
    AnalyzedSentence disambiguated = chunker.disambiguate(analyzedSentence);
    AnalyzedTokenReadings[] tokens = disambiguated.getTokens();
    assertTrue(tokens[1].getReadings().toString().contains("<adv>"));
    assertTrue(tokens[4].getReadings().toString().contains("</adv>"));
}

Example 5 with JLanguageTool

use of org.languagetool.JLanguageTool in project languagetool by languagetool-org.

the class RuleTest method testJavaRules.

@Test
public void testJavaRules() throws IOException {
    Set<String> ids = new HashSet<>();
    Set<Class> ruleClasses = new HashSet<>();
    if (Languages.getWithDemoLanguage().size() <= 1) {
        System.err.println("***************************************************************************");
        System.err.println("WARNING: found only these languages - the tests might not be complete:");
        System.err.println(Languages.getWithDemoLanguage());
        System.err.println("***************************************************************************");
    }
    for (Language language : Languages.getWithDemoLanguage()) {
        JLanguageTool lt = new JLanguageTool(language);
        List<Rule> allRules = lt.getAllRules();
        for (Rule rule : allRules) {
            if (!(rule instanceof AbstractPatternRule)) {
                assertIdUniqueness(ids, ruleClasses, language, rule);
                assertIdValidity(language, rule);
                assertTrue(rule.supportsLanguage(language));
                testExamples(rule, lt);
            }
        }
    }
}