Example usage for edu.stanford.nlp.ling.tokensregex TokenSequencePattern compile

Introduction

In this page you can find the example usage for edu.stanford.nlp.ling.tokensregex TokenSequencePattern compile.

Prototype

public static TokenSequencePattern compile(SequencePattern.PatternExpr nodeSequencePattern) 

Source Link

Document

Compiles a PatternExpr into a TokenSequencePattern.

Usage

From source file:org.wso2.extension.siddhi.gpl.execution.nlp.TokensRegexPatternStreamProcessor.java

License:Open Source License

@Override
protected List<Attribute> init(AbstractDefinition inputDefinition,
        ExpressionExecutor[] attributeExpressionExecutors, ConfigReader configReader,
        SiddhiAppContext siddhiAppContext) {
    if (logger.isDebugEnabled()) {
        logger.debug("Initializing Query ...");
    }//  w  ww  . j a  v  a2s .co m

    if (attributeExpressionLength < 2) {
        throw new SiddhiAppCreationException(
                "Query expects at least two parameters. Received only " + attributeExpressionLength
                        + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    String regex;
    try {
        if (attributeExpressionExecutors[0] instanceof ConstantExpressionExecutor) {
            regex = (String) attributeExpressionExecutors[0].execute(null);
        } else {
            throw new SiddhiAppCreationException("First parameter should be a constant."
                    + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
        }
    } catch (ClassCastException e) {
        throw new SiddhiAppCreationException("First parameter should be of type string. Found "
                + attributeExpressionExecutors[0].getReturnType()
                + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    try {
        regexPattern = TokenSequencePattern.compile(regex);
    } catch (Exception e) {
        throw new SiddhiAppCreationException("Cannot parse given regex " + regex, e);
    }

    if (!(attributeExpressionExecutors[1] instanceof VariableExpressionExecutor)) {
        throw new SiddhiAppCreationException("Second parameter should be a variable."
                + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    if (logger.isDebugEnabled()) {
        logger.debug(String.format("Query parameters initialized. Regex: %s Stream Parameters: %s", regex,
                inputDefinition.getAttributeList()));
    }

    initPipeline();

    ArrayList<Attribute> attributes = new ArrayList<Attribute>(1);

    attributes.add(new Attribute("match", Attribute.Type.STRING));
    attributeCount = regexPattern.getTotalGroups();
    for (int i = 1; i < attributeCount; i++) {
        attributes.add(new Attribute(groupPrefix + i, Attribute.Type.STRING));
    }
    return attributes;
}

---

From source file:org.wso2.gpl.siddhi.extension.nlp.TokensRegexPatternStreamProcessor.java

License:Open Source License

@Override
protected List<Attribute> init(AbstractDefinition abstractDefinition,
        ExpressionExecutor[] attributeExpressionExecutors, ExecutionPlanContext executionPlanContext) {
    if (logger.isDebugEnabled()) {
        logger.debug("Initializing Query ...");
    }/*from   www .  ja  va2  s .c o m*/

    if (attributeExpressionLength < 2) {
        throw new ExecutionPlanCreationException(
                "Query expects at least two parameters. Received only " + attributeExpressionLength
                        + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    String regex;
    try {
        if (attributeExpressionExecutors[0] instanceof ConstantExpressionExecutor) {
            regex = (String) attributeExpressionExecutors[0].execute(null);
        } else {
            throw new ExecutionPlanCreationException("First parameter should be a constant."
                    + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
        }
    } catch (ClassCastException e) {
        throw new ExecutionPlanCreationException("First parameter should be of type string. Found "
                + attributeExpressionExecutors[0].getReturnType()
                + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    try {
        regexPattern = TokenSequencePattern.compile(regex);
    } catch (Exception e) {
        throw new ExecutionPlanCreationException("Cannot parse given regex " + regex, e);
    }

    if (!(attributeExpressionExecutors[1] instanceof VariableExpressionExecutor)) {
        throw new ExecutionPlanCreationException("Second parameter should be a variable."
                + ".\nUsage: #nlp.findTokensRegexPattern(regex:string, text:string-variable)");
    }

    if (logger.isDebugEnabled()) {
        logger.debug(String.format("Query parameters initialized. Regex: %s Stream Parameters: %s", regex,
                abstractDefinition.getAttributeList()));
    }

    initPipeline();

    ArrayList<Attribute> attributes = new ArrayList<Attribute>(1);

    attributes.add(new Attribute("match", Attribute.Type.STRING));
    attributeCount = regexPattern.getTotalGroups();
    for (int i = 1; i < attributeCount; i++) {
        attributes.add(new Attribute(groupPrefix + i, Attribute.Type.STRING));
    }
    return attributes;
}

---

From source file:org.wso2.siddhi.extension.nlp.TokensRegexPatternTransformProcessor.java

License:Open Source License

@Override
protected void init(Expression[] expressions, List<ExpressionExecutor> expressionExecutors,
        StreamDefinition inStreamDefinition, StreamDefinition outStreamDefinition, String elementId,
        SiddhiContext siddhiContext) {/*from   ww w .  jav a 2  s . c  o  m*/
    if (logger.isDebugEnabled()) {
        logger.debug("Initializing Query ...");
    }

    if (expressions.length < 2) {
        throw new QueryCreationException("Query expects at least two parameters. Received only "
                + expressions.length + ".\n" + "Usage: findTokensRegexPattern(regex:string, text:string)");
    }

    String regex;
    try {
        regex = ((StringConstant) expressions[0]).getValue();
    } catch (ClassCastException e) {
        logger.error("Error in reading parameter regex", e);
        throw new QueryCreationException(
                "First parameter should be of type string. Found " + Constants.getType(expressions[0]) + ".\n"
                        + "Usage: findTokensRegexPattern(regex:string, text:string)");
    }

    try {
        regexPattern = TokenSequencePattern.compile(regex);
    } catch (Exception e) {
        logger.error("Error in parsing token regex pattern", e);
        throw new QueryCreationException(
                "Cannot parse given regex " + regex + " Error: [" + e.getMessage() + "]");
    }

    if (expressions[1] instanceof Variable) {
        inStreamParamPosition = inStreamDefinition
                .getAttributePosition(((Variable) expressions[1]).getAttributeName());
    } else {
        throw new QueryCreationException(
                "Second parameter should be a variable. Found " + Constants.getType(expressions[1]) + ".\n"
                        + "Usage: findTokensRegexPattern(regex:string, text:string)");
    }

    if (logger.isDebugEnabled()) {
        logger.debug(String.format("Query parameters initialized. Regex: %s Stream Parameters: %s", regex,
                inStreamDefinition.getAttributeList()));
    }

    initPipeline();

    if (outStreamDefinition == null) {
        this.outStreamDefinition = new StreamDefinition().name("tokensRegexPatternMatchStream");

        this.outStreamDefinition.attribute("match", Attribute.Type.STRING);

        // Get all groups in the regular expression and add them to the output stream definition attributes
        attributeCount = regexPattern.getTotalGroups();
        for (int i = 1; i < regexPattern.getTotalGroups(); i++) {
            this.outStreamDefinition.attribute(groupPrefix + i, Attribute.Type.STRING);
        }

        for (Attribute strDef : inStreamDefinition.getAttributeList()) {
            this.outStreamDefinition.attribute(strDef.getName(), strDef.getType());
        }
    }
}

---