Examples of LanguageRule

net.sourceforge.segment.srx.LanguageRule
Represents rule for segmenting text in some language. Contains {@link Rule}list. @author loomchild

Examples of net.sourceforge.segment.srx.LanguageRule


    Body body = srx.getBody();


    Map<String, LanguageRule> languageRuleMap = new HashMap<String, LanguageRule>();
    for (Languagerule lr : body.getLanguagerules().getLanguagerule()) {
      LanguageRule languageRule = new LanguageRule(lr
          .getLanguagerulename());
      for (net.sourceforge.segment.srx.io.bind.Rule r : lr.getRule()) {
        boolean breakRule = !"no".equals(r.getBreak());


        String before;
        if (r.getBeforebreak() != null) {
          before = r.getBeforebreak().getContent();
        } else {
          before = "";
        }


        String after;
        if (r.getAfterbreak() != null) {
          after = r.getAfterbreak().getContent();
        } else {
          after = "";
        }


        Rule rule = new Rule(breakRule, before, after);
        languageRule.addRule(rule);
      }
      languageRuleMap.put(languageRule.getName(), languageRule);
    }


    for (Languagemap lm : body.getMaprules().getLanguagemap()) {
      LanguageRule languageRule = languageRuleMap.get(lm
          .getLanguagerulename());
      if (languageRule == null) {
        log.warn("Language map \"" + lm.getLanguagepattern()
            + "\": language rule \"" + lm.getLanguagerulename()
            + "\" not found.");

View Full Code Here

Examples of net.sourceforge.segment.srx.LanguageRule

    if (ruleLength < 1) {
      throw new RuntimeException("Rule length must be greater or equal to one: " + ruleCount + ".");
    }


    SrxDocument srxDocument = new SrxDocument();
    LanguageRule languageRule = generateLanguageRule(ruleCount, ruleLength);
    srxDocument.addLanguageMap(".*", languageRule);
    return srxDocument;
  }

View Full Code Here

Examples of net.sourceforge.segment.srx.LanguageRule

    srxDocument.addLanguageMap(".*", languageRule);
    return srxDocument;
  }


  private LanguageRule generateLanguageRule(int ruleCount, int ruleLenght) {
    LanguageRule languageRule = new LanguageRule("");
    // Add rules
    for (int i = 0; i < ruleCount; ++i) {
      Rule rule = generateRule(ruleLenght);
      languageRule.addRule(rule);
    }
    // Add end of sentence rule
    languageRule.addRule(new Rule(true, "\\.", " "));
    return languageRule;
  }

View Full Code Here

Examples of net.sourceforge.segment.srx.LanguageRule

      assertTrue(document.getCascade());


      List<LanguageRule> languageRuleList = document
      .getLanguageRuleList("en");
      
      LanguageRule languageRule = languageRuleList.get(0);
      assertEquals("Default", languageRule.getName());


      List<Rule> ruleList = languageRule.getRuleList();
      assertEquals(1, ruleList.size());    


      Rule rule = ruleList.get(0);
      assertEquals("[\\.!?…]['»\"”\\)\\]\\}]?\\u0002?\\s", rule.getBeforePattern());
      assertEquals("", rule.getAfterPattern());

View Full Code Here

Examples of net.sourceforge.segment.srx.LanguageRule


    List<LanguageRule> languageRuleList = document
        .getLanguageRuleList("en");
    assertEquals(1, languageRuleList.size());


    LanguageRule languageRule = languageRuleList.get(0);
    assertEquals("Default", languageRule.getName());


    List<Rule> ruleList = languageRule.getRuleList();
    assertEquals(5, ruleList.size());


    Rule rule = ruleList.get(1);
    assertEquals("[Ee][Tt][Cc]\\.", rule.getBeforePattern());
    assertEquals("\\s[a-z]", rule.getAfterPattern());

View Full Code Here

Examples of net.sourceforge.segment.srx.LanguageRule


    List<LanguageRule> languageRuleList = document
        .getLanguageRuleList("fr_FR");
    assertEquals(2, languageRuleList.size());


    LanguageRule languageRule = languageRuleList.get(0);
    assertEquals("French", languageRule.getName());


    List<Rule> ruleList = languageRule.getRuleList();
    assertEquals(4, ruleList.size());


    Rule rule = ruleList.get(1);
    assertEquals("\\s[Mm]lles\\.", rule.getBeforePattern());
    assertEquals("\\s", rule.getAfterPattern());

View Full Code Here

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.