APP-3330: Fixed regex in SymMessageParser

lib/SymMessageParser/index.js

@@ -2,31 +2,23 @@ var SymBotAuth = require('../SymBotAuth')
 
 var SymMessageParser = {}
 
-const regex = /(<div.+?>)(.*)(<\/div>)/g
 const regexTagStart = /<\w+>/g
 const regexTagEnd = /<\/\w+>/g
 
 const HASHTAG_TYPE = 'org.symphonyoss.taxonomy.hashtag'
 const CASHTAG_TYPE = 'org.symphonyoss.fin.security.id.ticker'
 const MENTION_TYPE = 'com.symphony.user.userId'
-const EntityRegex = 'id=\"\\d+\">(.*?)<'
 
 SymMessageParser.parse = events => {
   const arrParsedMessages = []
 
   try {
     events.forEach(event => {
       if (event.type === 'MESSAGESENT' && event.initiator.user.userId !== SymBotAuth.botUser.id) {
-        let match
         const message = event.payload.messageSent.message
-        while ((match = regex.exec(message.message)) !== null) {
-          if (match.index === regex.lastIndex) {
-            regex.lastIndex++
-          }
-          let textMessage = match[2].replace(regexTagStart, '')
-          textMessage = textMessage.replace(regexTagEnd, '')
-          message.messageText = textMessage
-        }
+
+        message.messageText = extractTextFromPresentationML(message.message)
+
         arrParsedMessages.push(message)
       }
     })
@@ -43,6 +35,25 @@ SymMessageParser.parse = events => {
   return arrParsedMessages
 }
 
+function extractTextFromPresentationML(presentationMl) {
+  const trimmedPresentationMl = presentationMl.trim()
+
+  let indexOfFirstDivTag = trimmedPresentationMl.indexOf('<div')
+  if (indexOfFirstDivTag == -1) {
+    return undefined;
+  }
+
+  let firstIndex = trimmedPresentationMl.indexOf('>', indexOfFirstDivTag) // index of first <div> tag
+
+  let lastIndex = trimmedPresentationMl.lastIndexOf('</div>') // index of last </div> tag
+  if(lastIndex == -1) {
+    return undefined
+  }
+
+  let insideDiv = trimmedPresentationMl.substring(firstIndex + 1, lastIndex) // extract what is inside <div></div>
+  return insideDiv.replace(regexTagStart, '').replace(regexTagEnd, '') //remove inside tags and extract text
+}
+
 function getTags (message, type) {
   const arrTags = []
 

tests/SymMessageParser/index.test.js

@@ -0,0 +1,96 @@
+const SymMessageParser = require('../../lib/SymMessageParser');
+
+describe('Message parsing', () => {
+  beforeAll(() => {
+    const SymBotAuth = require('../../lib/SymBotAuth')
+    SymBotAuth.botUser = { id: -1};
+  });
+
+  test('invalid opening tag', () => {
+    assertMessageParsed('<divsa data-format="PresentationML" data-version="2.0">Hello, World</div>',
+      'Hello, World');
+  });
+
+  test('no opening tag', () => {
+    assertMessageParsed(' data-format="PresentationML" data-version="2.0">Hello, World</>',
+      undefined);
+  });
+
+  test('no closing tag', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0">Hello, World',
+      undefined);
+  });
+
+  test('simple message', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0">Hello, World</div>',
+      'Hello, World');
+  });
+
+  test('simple message with spaces around div', () => {
+    assertMessageParsed('   <div data-format="PresentationML" data-version="2.0">Hello, World</div>   ',
+      'Hello, World');
+  });
+
+  test('message enclosed in markups', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0"><b>Hello, World</b></div>',
+      'Hello, World');
+  });
+
+  test('message enclosed in div markups', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0"><div>Hello, World</div></div>',
+      'Hello, World');
+  });
+
+  test('message enclosed in markups with spaces around div', () => {
+    assertMessageParsed('  <div data-format="PresentationML" data-version="2.0"><b>Hello, World</b></div>  ',
+      'Hello, World');
+  });
+
+  test('message partially enclosed in markups', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0">Hello <b>wonderful</b> world</div>',
+      'Hello wonderful world');
+  });
+
+  test('message partially enclosed in div markups', () => {
+    assertMessageParsed('<div data-format="PresentationML" data-version="2.0">Hello <div>wonderful</div> world</div>',
+      'Hello wonderful world');
+  });
+
+  test('message partially enclosed in markups with spaces around div', () => {
+    assertMessageParsed('  <div data-format="PresentationML" data-version="2.0">Hello <b>wonderful</b> world</div>  ',
+      'Hello wonderful world');
+  });
+
+  test('message with several levels of markups', () => {
+    assertMessageParsed(
+      '<div data-format="PresentationML" data-version="2.0">Hello <b>wonderful <c>and</c> beautiful</b> world</div>',
+      'Hello wonderful and beautiful world');
+  });
+
+  test('message with several levels of markups, innermost div', () => {
+    assertMessageParsed(
+      '<div data-format="PresentationML" data-version="2.0">Hello <b>wonderful <div>and</div> beautiful</b> world</div>',
+      'Hello wonderful and beautiful world');
+  });
+
+  test('message with several levels of markups with spaces around div', () => {
+    assertMessageParsed(
+      '   <div data-format="PresentationML" data-version="2.0">Hello <b>wonderful <c>and</c> beautiful</b> world</div>  ',
+      'Hello wonderful and beautiful world');
+  });
+});
+
+function assertMessageParsed(incomingMessage, expectedParsedMessage) {
+  const event = {
+    type: 'MESSAGESENT',
+    initiator: {user: {id: 1234}},
+    payload: {messageSent: {message: {message: incomingMessage}}}
+  };
+
+  let parsedMessages = SymMessageParser.parse([event]);
+
+  expect(Array.isArray(parsedMessages)).toBeTruthy();
+  expect(parsedMessages.length).toBe(1);
+
+  expect(parsedMessages[0].messageText).toBe(expectedParsedMessage);
+}