feat(*): add pipeConfig() in nlp to allow inquiry of active annotators

Author: sanjayaksaxena

src/doc-v2.js

@@ -456,6 +456,9 @@ var doc = function ( docData, addons ) {
 
   methods.printTokens = () => printTokens( tokens, cache );
 
+  // Enusre that we make a deep copy of config before returning to avoid corruption!
+  methods.pipeConfig = () => JSON.parse( JSON.stringify( docData.currPipe ) );
+
   return methods;
 };
 

src/wink-nlp.js

@@ -109,14 +109,9 @@ var nlp = function ( theModel, pipe ) {
   // Used to innstantiate the compiler.
   var cerMetaModel;
 
-  // Annotation stuff.
+  // Contains a list of valid annotations built from `theModel`.
   var validAnnotations = Object.create( null );
-  validAnnotations.sbd = true;
-  validAnnotations.negation = true;
-  validAnnotations.sentiment = true;
-  validAnnotations.pos = true;
-  validAnnotations.ner = true;
-  validAnnotations.cer = true;
+
   // Current pipe.
   var currPipe = Object.create( null );
   var onlyTokenization = true;
@@ -231,6 +226,9 @@ var nlp = function ( theModel, pipe ) {
     // Markings are 4-tuples of `start`, `end` **token indexes**,  and `begin & end markers`.
     // The begin & end markers are used to markup the tokens specified.
     rdd.markings = [];
+    // Publish the current annotation pipeline so that code can inquire about
+    // active annotations!
+    rdd.currPipe = currPipe;
 
     var wrappedDocData = DocDataWrapper( rdd );  // eslint-disable-line new-cap
 
@@ -395,6 +393,15 @@ var nlp = function ( theModel, pipe ) {
     throw Error( 'wink-nlp: invalid model used.' );
   }
 
+  // Build a list of valid annotations from `theModel`. This will ensure that
+  // only **available** annotations from the model can be used in the pipe.
+  validAnnotations.sbd = typeof theModel.sbd === 'function';
+  validAnnotations.negation = typeof theModel.negation === 'function';
+  validAnnotations.sentiment = typeof theModel.sa === 'function';
+  validAnnotations.pos = typeof theModel.pos === 'function';
+  validAnnotations.ner = typeof theModel.ner === 'function';
+  validAnnotations.cer = typeof theModel.metaCER === 'function';
+
   const tempPipe = ( pipe === undefined ) ? Object.keys( validAnnotations ) : pipe;
   if ( helper.isArray( tempPipe ) ) {
     tempPipe.forEach( ( at ) => {

test/wink-nlp-specs.js

@@ -155,6 +155,10 @@ describe( 'wink-nlp test-coverage and basic behavior', function () {
   it( 'should throw error if readDoc is given non-text', function () {
     expect( nlp.readDoc.bind( 1 ) ).to.throw( /^wink-nlp: expecting a valid Javascript string/ );
   } );
+
+  it( 'pipeConfig should return sbd, negation, sa, pos, ner and cer as true', function () {
+    expect( nlp.readDoc('Test').pipeConfig() ).to.deep.equal( { sbd: true, negation: true, pos: true, sentiment: true, ner: true, cer: true } );
+  } );
 } );
 
 describe( 'OOV handling', function () {
@@ -227,6 +231,7 @@ describe( 'Annotation pipe validity checks', function () {
 
 describe( 'Empty Annotation pipe should not detect any of the existing annotations', function () {
   const nlpNoAnno = winkNLP( model, [] );
+
   it( 'should still learn from patterns', function () {
     const patterns = [
       { name: 'adjectiveNounPair', patterns: [ 'ADJ' ] }
@@ -235,6 +240,11 @@ describe( 'Empty Annotation pipe should not detect any of the existing annotatio
   } );
 
   const doc = nlpNoAnno.readDoc( 'Hello World! Not happy! :-)' );
+
+  it( 'pipeConfig should return empty config', function () {
+    expect( doc.pipeConfig() ).to.deep.equal( Object.create( null ) );
+  } );
+
   it( 'should return single sentence', function () {
     expect( doc.sentences().length() ).to.equal( 1 );
     expect( doc.sentences().out()[ 0 ] ).to.equal( doc.out() );