fix: split ssml to correct chunks (#225)

* fix: split ssml to correct chunks

* fix: split ssml to correct chunks

* fixed: eslint

* fixed: eslint

* fixed: add comment to testcase

* fixed: review comments

* fixed: review comments

* fixed: review comments

* fixed: review comments

* fixed: review comments

Co-authored-by: Quan HL <quanluuhoang8@gmail.com>

test/data/bad/bad-say-ssml.json

@@ -0,0 +1,9 @@
+{
+  "say": {
+    "text": "<speak>I already told you <emphasis level=\"strong\">I already told you I already told you I already told you I already told you! I already told you I already told you I already told you I already told you? I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you told I already told you I already told you told I already told you I already told you. I already told you <break time=\"3s\"/> I really like that person!</emphasis> this is another long text.</speak>",
+    "synthesizer": {
+      "vendor": "google",
+      "language": "en-US"
+    }
+  }
+}

test/data/good/say-ssml.json

@@ -0,0 +1,9 @@
+{
+  "say": {
+    "text": "<speak>I already told you I already told you I already told you I already told you I already told you! I already told you I already told you I already told you I already told you? I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you I already told you told I already told you I already told you told I already told you I already told you. I already told you <break time=\"3s\"/> I <emphasis level=\"strong\">really like that person!</emphasis> this is another long text.</speak>",
+    "synthesizer": {
+      "vendor": "google",
+      "language": "en-US"
+    }
+  }
+}

test/unit-tests.js

@@ -44,7 +44,22 @@ test('unit tests', (t) => {
 
   task = makeTask(logger, require('./data/good/say-text-array'));
   t.ok(task.name === 'say', 'parsed say with multiple segments');
+  
+  task = makeTask(logger, require('./data/good/say-ssml'));
+  // the ssml is more than 1000 chars, 
+  // expecting first chunk is length > 100, stop at ? instead of first .
+  // 2nd chunk is long text < 1000 char, stop at .
+  // 3rd chunk is the rest.
+  t.ok(task.text.length === 3 &&
+    task.text[0].length === 187 &&
+    task.text[1].length === 882 &&
+    task.text[2].length === 123, 'parsed say');
 
+  task = makeTask(logger, require('./data/bad/bad-say-ssml'));
+  t.ok(task.text.length === 1 &&
+    task.text[0].length === 1162, 'parsed bad say');
+
+  
   const alt = require('./data/good/alternate-syntax');
   const normalize = require('../lib/utils/normalize-jambones');
   normalize(logger, alt).forEach((t) => {