Merge pull request #244 from chughts/sttwords

STT Customisation - Words Support

Author: chughts

README.md

@@ -7,8 +7,11 @@ Node-RED Watson Nodes for IBM Bluemix
 
 <a href="https://cla-assistant.io/watson-developer-cloud/node-red-node-watson"><img src="https://cla-assistant.io/readme/badge/watson-developer-cloud/node-red-node-watson" alt="CLA assistant" /></a>
 
+### New in version 0.4.37
+- Added support for word training in Speech to Text Customizations Node.
+
 ### New in version 0.4.36
-- New Node for Speech to Text Customizations - Initial support for corpus (sentances) training.
+- New Node for Speech to Text Customizations - Initial support for corpus (sentences) training.
 - Added option to select Customization in Speech to Text Node.
 
 ### New in version 0.4.35

package.json

@@ -1,6 +1,6 @@
 {
   "name": "node-red-node-watson",
-  "version": "0.4.36",
+  "version": "0.4.37",
   "description": "A collection of Node-RED nodes for IBM Watson services",
   "dependencies": {
     "alchemy-api": "^1.3.0",

services/language_translator_identify/v2.html

@@ -35,7 +35,7 @@
 </script>
 
 <script type="text/x-red" data-help-name="watson-language-translator-identify">
-    <p>Packaged in with release 0.4.36 of node-red-node-watson</p>
+    <p>Packaged in with release 0.4.37 of node-red-node-watson</p>
     <p>The Watson Language Translator service can be used to identify languages used in a text input. <p>
     <p>Node input : </p>
     <ul>

services/speech_to_text/v1-corpus-builder.html

@@ -48,6 +48,8 @@
             <option value="addCorpus">Add Corpus</option>
             <option value="getCorpora">Get Corpora</option>
             <option value="train">Train</option>
+            <option value="listCustomWords">List Custom Words</option>
+            <option value="addWords">Add Words</option>
             <option value="deleteCorpus">Delete Corpus</option>
         </select>
     </div>
@@ -81,6 +83,11 @@
         <input type="text" id="node-input-stt-corpus-name" placeholder="">
     </div>
 
+    <div class="form-row">
+        <label for="node-input-stt-allow-overwrite"><i class="fa fa-book"></i> Allow Overwrite</label>
+        <input type="checkbox" id="node-input-stt-allow-overwrite" placeholder="">
+    </div>
+
 
 </script>
 
@@ -125,6 +132,25 @@
         by the speech to text service.
       </p>
     </li>
+    <li>List Custom Words
+      <p>In this mode, a list of all the words from a custom model are
+        retrieved.
+      </p>
+    </li>
+    <li>Add Words
+      <p>In this mode, the specified custom words are added to the
+        customization. The input array of words should be passed in
+        as <code>msg.payload</code> in form of
+        <code>[{word: String, sounds_like: [String, ...], display_as: String}, ...]</code>
+        <br/>EG.
+        <code>
+          [ {"word": "HHonors", "sounds_like": ["hilton honors", "h honors"],
+                                "display_as": "HHonors"},
+            {"word": "IEEE", "sounds_like": ["i triple e"]}
+          ]
+        </code>
+      </p>
+    </li>
     <li>Delete Corpus
       <p>In this mode, the specified corpus is deleted.
       </p>
@@ -150,17 +176,6 @@
   sttv1qbb.mode_selected = null;
   sttv1qbb.model_selected = '';
 
-
-  sttv1qbb.LANGUAGES = { 'en-US': 'US English',
-                    'pt-BR': 'Portuguese Braziilian',
-                    'en-UK': 'UK English' ,
-                    'fr-FR': 'French',
-                    'zh-CN': 'Mandarin',
-                    'es-ES': 'Spanish',
-                    'ar-AR': 'Arablic',
-                    'ja-JP': 'Japanese'
-                  };
-
   // Function to be used at the start, as don't want to expose any fields, unless the models are
   // available. The models can only be fetched if the credentials are available.
   sttv1qbb.hideEverything = function () {
@@ -173,6 +188,7 @@
     $('input#node-input-stt-custom-model-description').parent().hide();
     $('input#node-input-stt-custom-id').parent().hide();
     $('input#node-input-stt-corpus-name').parent().hide();
+    $('input#node-input-stt-allow-overwrite').parent().hide();
   }
 
   sttv1qbb.showSelectedFields = function(fields) {
@@ -203,12 +219,15 @@
       case 'getCustomisation':
       case 'getCorpora':
       case 'train':
+      case 'listCustomWords':
+      case 'addWords':
         fields.push('#node-input-stt-custom-id');
         break;
       case 'addCorpus':
       case 'deleteCorpus':
           fields.push('#node-input-stt-custom-id'
-                         + ', #node-input-stt-corpus-name');
+                         + ', #node-input-stt-corpus-name'
+                         + ', #node-input-stt-allow-overwrite');
           break;
     }
     sttv1qbb.showSelectedFields(fields);
@@ -366,7 +385,8 @@
                 'base-model-hidden' : {value: ''},
                 'stt-custom-model-description' : {value: ''},
                 'stt-custom-id' : {value: ''},
-                'stt-corpus-name' : {value: ''}
+                'stt-corpus-name' : {value: ''},
+                'stt-allow-overwrite' : {value: false}
             },
             credentials: {
               username: {type:'text'} //,

services/speech_to_text/v1-corpus-builder.js

@@ -125,6 +125,31 @@ module.exports = function (RED) {
     });
   }
 
+  function executeGetCustomWords(node, stt, params, msg) {
+    stt.getWords(params, function (err, response) {
+      node.status({});
+      if (err) {
+        reportError(node, msg, err);
+      } else {
+        msg['words'] = response.words ? response.words : response ;
+      }
+      node.send(msg);
+    });
+  }
+
+  function executeAddWords(node, stt, params, msg) {
+    stt.addWords(params, function (err, response) {
+      node.status({});
+      if (err) {
+        reportError(node, msg, err);
+      } else {
+        msg['addwordsresponse'] = response ;
+      }
+      node.send(msg);
+    });
+  }
+
+
   function executeDeleteCorpus(node, stt, params, msg) {
     stt.deleteCorpus(params, function (err, response) {
       node.status({});
@@ -164,6 +189,12 @@ module.exports = function (RED) {
     case 'train':
       executeTrain(node, stt, params, msg);
       break;
+    case 'listCustomWords':
+      executeGetCustomWords(node, stt, params, msg);
+      break;
+    case 'addWords':
+      executeAddWords(node, stt, params, msg);
+      break;
     case 'deleteCorpus':
       executeDeleteCorpus(node, stt, params, msg);
       break;
@@ -189,10 +220,25 @@ module.exports = function (RED) {
       params['customization_id'] = config['stt-custom-id'];
     }
 
+    if ('addCorpus' === config['stt-custom-mode']) {
+      params['allow_overwrite'] = config['stt-allow-overwrite'];
+    }
+
     return params;
   }
 
-  function loadCorpusFile(node, method, params, msg) {
+  function setFileParams(method, params, msg) {
+    switch (method) {
+    case 'addCorpus':
+      params.corpus = msg.payload;
+      break;
+    case 'addWords':
+      params.words = msg.payload;
+      break;
+    }
+  }
+
+  function loadFile(node, method, params, msg) {
     temp.open({
       suffix: '.txt'
     }, function(err, info) {
@@ -217,15 +263,28 @@ module.exports = function (RED) {
           throw err;
         }
 
-        params.corpus = fs.createReadStream(info.path);
+        switch (method) {
+        case 'addCorpus':
+          params.corpus = fs.createReadStream(info.path);
+          break;
+        case 'addWords':
+          try {
+            params.words = JSON.parse(fs.readFileSync(info.path, 'utf8'));
+          } catch (err) {
+            params.words = fs.createReadStream(info.path);
+          }
+        }
+
         executeMethod(node, method, params, msg);
+        temp.cleanup();
       });
     });
   }
 
   function checkForFile(method) {
     switch (method) {
     case 'addCorpus':
+    case 'addWords':
       return true;
     }
     return false;
@@ -286,11 +345,14 @@ module.exports = function (RED) {
       }
 
       if (checkForFile(method)) {
-        loadCorpusFile(node, method, params, msg);
-      } else {
-        executeMethod(node, method, params, msg);
+        if (msg.payload instanceof Buffer) {
+          loadFile(node, method, params, msg);
+          return;
+        }
+        setFileParams(method, params, msg);
       }
 
+      executeMethod(node, method, params, msg);
     });
   }