/**
* return a set of rules based on this CharGer file. All Rules are different from each other. If
* the LHS of the file contains a lemma with multiple choices, like "lemma1/lemma2/lemma3...",
* then one rule is created per lemma option otherwise, naturally, one rule is created
*/
public List<RuleWithConfidenceAndDescription<Info, BasicNode>> makeRules(File file)
throws EntailmentCompilationException {
// first add all the rules into a set, to eliminate duplicates. Then put 'em in a list
Set<RuleWithConfidenceAndDescription<Info, BasicNode>> setOfRulesWithCD =
new LinkedHashSet<RuleWithConfidenceAndDescription<Info, BasicNode>>();
String description = RuleCompilerUtils.getDescription(file);
String origText;
try {
origText = FileUtils.loadFileToString(file);
} catch (IOException e) {
throw new EntailmentCompilationException("error reading " + file, e);
}
try {
List<String> expandedTexts =
CgxMultipleChoiceExpander.expandMultipleChoiceParameters(origText, null);
for (String text : expandedTexts) {
setOfRulesWithCD.add(
new RuleWithConfidenceAndDescription<Info, BasicNode>(
cgxRuleCompiler.makeRule(text), Constants.RULE_CONFIDENCE, description));
}
EntailmentRuleBuildingUtils.addReversedRules(setOfRulesWithCD, expandedTexts.get(0));
} catch (Exception e) {
throw new EntailmentCompilationException("Error in file " + file, e);
}
List<RuleWithConfidenceAndDescription<Info, BasicNode>> rulesWithCD =
new Vector<RuleWithConfidenceAndDescription<Info, BasicNode>>(setOfRulesWithCD);
return rulesWithCD;
}
/**
* @param args
* @throws UnsupportedPosTagStringException
* @throws EntailmentCompilationException
* @throws IOException
* @throws FileNotFoundException
* @throws ConfigurationException
*/
public static void main(String[] args)
throws UnsupportedPosTagStringException, EntailmentCompilationException,
FileNotFoundException, IOException, ConfigurationException {
if (args.length < 1)
throw new EntailmentCompilationException(
"usage: EntailmentRuleCompiler configurationFile.xml");
ConfigurationFile confFile = new ConfigurationFile(new File(args[0]));
confFile.setExpandingEnvironmentVariables(true);
ConfigurationParams compilationParams =
confFile.getModuleConfiguration(RuleCompilerParameterNames.RULE_COMPILER_PARAMS_MODULE);
// ConfigurationParams applictionParams =
// confFile.getModuleConfiguration(KnowledgeResource.SYNTACTIC.getModuleName());
ConfigurationParams applictionParams =
confFile.getModuleConfiguration(RuleCompilerParameterNames.SYNTACTIC_PARAMS_MODULE);
File dir =
compilationParams.getDirectory(
RuleCompilerParameterNames
.ENTAILMENT_RULES_DIRECTORY); // new
// File(props.getProperty("directoryName").trim());
final String ruleFileSuffix =
compilationParams.get(
RuleCompilerParameterNames
.RULE_FILE_SUFFIX); // props.getProperty("graphFileSuffix").trim();
// create an english node rule compliler
EntailmentRuleCompiler compiler = new EntailmentRuleCompiler();
List<RuleWithConfidenceAndDescription<Info, BasicNode>> rulesWithCD;
rulesWithCD = compiler.compileFolder(dir, ruleFileSuffix);
// EnglishRulesViewer rv = new EnglishRulesViewer(rulesWithCD);
// ExtendedRulesViewer rv = new ExtendedRulesViewer(rulesWithCD.subList(0,1));
// rv.view();
// serialize rules to file
Set<RuleWithConfidenceAndDescription<Info, BasicNode>> rules =
new LinkedHashSet<RuleWithConfidenceAndDescription<Info, BasicNode>>(rulesWithCD);
String outFile =
applictionParams.get(TransformationsConfigurationParametersNames.SYNTACTIC_RULES_FILE);
try {
RuleCompilerUtils.serializeToFile(rules, outFile);
} catch (CompilationException e) {
throw new EntailmentCompilationException("see nested", e);
}
System.out.println("\n\nMade " + rules.size() + " rules.");
System.out.println("Serialized them into " + outFile);
}
