Skip to main content
These examples show how to use extraction scripts with the FieldExtractor object to detect named entities, run XML queries, and parse address components into separate fields.

​
C# extraction script example

The following C# example runs several extraction methods on the FieldExtractor object. It matches keywords and people, writes text spans to fields, and runs an XML query for two-word dictionary phrases. 
FieldExtractor.ExtractRegularExpression( "(grant(s))|(convey to)|(grant)", "keyWords" );
FieldExtractor.ExtractWordsFromUserDictionary("dictionary", "English");
FieldExtractor.ExtractNerObjects();

// Access identified objects by collection name
IExtractedObjects personObjects = FieldExtractor.ExtractedObjects( "NerPerson" );
IExtractedObjects keyWordsObjects = FieldExtractor.ExtractedObjects( "keyWords" );

// Look for matching person to left of keyword
for( int i = 0; i < keyWordsObjects.Count; i++ ) {
    IExtractedObject keyWord = personObjects.Item( i );
    IInterval span = keyWord.Span;
    IExtractedObject grantor = personObjects.Find( span.StartPos, false );
    string grantorName = grantor.Value;
    if ( grantorName.Length > 2 && grantorName != "Doug Darrell" ) {
        FieldExtractor.PutSpanToField( grantor.Span, "NlpField1" );
    }
}

// Access source text
string sourceText = FieldExtractor.SourceText;

// Write any text span to field
FieldExtractor.PutTextToField( 0, sourceText.Length - 1, "NlpField2" );

// An XML query that looks for two-word dictionary phrases
string query = "<Request> " +
    "<Query>" +
    "<Contain MaxDistance=\"1\">" +
    "<Required>" +
    "<Form><Attributes><Attribute>dictionary1</Attribute></Attributes></Form>" +
    "</Required>" +
    "<Required>" +
    "<Form><Attributes><Attribute>dictionary2</Attribute></Attributes></Form>" +
    "</Required>" +
    "</Contain>" +
    "</Query>" +
    "</Request>";

// Run query, get collection of results, and save them to field
FieldExtractor.RunQueryAndSaveToField( query, "query1", "NlpField3");

// Run query and get collection of results
IExtractedObjects queryResults1 = FieldExtractor.RunQuery( query, "query2" );

// An alternative way of accessing query results after running the query
IExtractedObjects queryResults2 = FieldExtractor.QueryResults( "query2" );

​
Script parameters

NameTypePermissionsValue
FieldExtractorIFieldExtractorReadIdentifies fields in the text of a document

​
Sample address extraction script for the entire field

This script is called for the entire source field. It parses the address and extracts the ZIP code and street into separate fields. 
// Parse the address
this.ParseAddress();

// Extract the components into separate fields
var zip = this.ExtractedObjects( "NerZipCode" );
var street = this.ExtractedObjects( "NerStreet" );

for( var i = 0; i < zip.Count; i++ ) {
    this.PutSpanToField( zip.Item(i).Span, "ZipCode" );
}

for( var j = 0; j < street.Count; j++ ) {
    this.PutSpanToField( street.Item(j).Span, "Street" );
}

​
Sample address extraction script for part of a field

This script is called for part of the source field. It detects the NER address object, splits it into components, and saves the ZIP code and street into separate fields. 
// Detect all NER objects in the text of the field or section
this.ExtractNerObjects();

// Extract the Address object
var address = this.ExtractedObjects( "NerAddress" );

// Split the address into components
for( var addressSpanIndex = 0; addressSpanIndex < address.Count; addressSpanIndex++ ) {
    this.PutSpanToField( address.Item(addressSpanIndex).Span, "Address" );

    // Assign a unique prefix to the names of all component collections for the given address
    var collectionName = "nerAddress" + String(addressSpanIndex);

    // Extract the address
    this.ParseAddressInPosition( collectionName, address.Item(addressSpanIndex).Span.StartPos, address.Item(addressSpanIndex).Span.EndPos );

    // Save the components into separate fields
    var zip = this.ExtractedObjects( collectionName, "NerZipCode" );
    var street = this.ExtractedObjects( collectionName, "NerStreet" );
    RunQueryAndSaveToField
    for( var i = 0; i < zip.Count; i++ ) {
        this.PutSpanToField( zip.Item(i).Span, "ZipCode" );
    }
    for( var j = 0; j < street.Count; j++ ) {
        this.PutSpanToField( street.Item(j).Span, "Street" );
    }
}