Merge pull request #24 from fmacpro/development

Development
fmacpro · Oct 3, 2019 · c2a6101 · c2a6101
2 parents df5ff76 + 5d3a211
commit c2a6101
Show file tree

Hide file tree

Showing 4 changed files with 2,677 additions and 1,372 deletions.
diff --git a/index.js b/index.js
@@ -14,7 +14,6 @@ const htmlToText = require('html-to-text')
 const nlp = require('compromise')
 const absolutify = require('absolutify')
 const personalDictionary = require('./personalDictionary.js')
-const htmlTags = require('./stripTags.js')
 const jsdom = require('jsdom')
 const { JSDOM } = jsdom
 const helpers = require('./helpers')
@@ -75,7 +74,7 @@ const articleParser = async function (options, socket) {
   article.lighthouse = {}
 
   if (typeof options.striptags === 'undefined') {
-    options.striptags = htmlTags
+    options.striptags = []
   }
 
   socket.emit('parse:status', 'Starting Horseman')
@@ -116,8 +115,9 @@ const articleParser = async function (options, socket) {
   socket.emit('parse:status', 'Status ' + article.status)
 
   if (article.status === 403 || article.status === 404) {
+    socket.emit('parse:status', 'Failed to fetch URL')
     await browser.close()
-    return article.status + ' Failed to fetch URL'
+    process.exit(1)
   }
 
   // Evaluate URL

diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "horseman-article-parser",
-  "version": "0.7.1",
+  "version": "0.7.2",
   "description": "Web Page Inspection Tool. Sentiment Analysis, Keyword Extraction, Named Entity Recognition & Spell Check",
   "main": "index.js",
   "scripts": {

diff --git a/stripTags.js b/stripTags.js
diff --git a/testresults.json b/testresults.json