feat: allow parser to define custom date formats (#141)

* feat: allow parser to define custom date formats

* feat: updating macrumors to test/verify format working correctly

src/cleaners/date-published.js

@@ -23,29 +23,29 @@ export function cleanDateString(dateString) {
                    .trim();
 }
 
-export function createDate(dateString, timezone) {
+export function createDate(dateString, timezone, format) {
   if (TIME_WITH_OFFSET_RE.test(dateString)) {
     return moment(new Date(dateString));
   }
 
   return timezone ?
-    moment.tz(dateString, parseFormat(dateString), timezone) :
-    moment(dateString, parseFormat(dateString));
+    moment.tz(dateString, format || parseFormat(dateString), timezone) :
+    moment(dateString, format || parseFormat(dateString));
 }
 
 // Take a date published string, and hopefully return a date out of
 // it. Return none if we fail.
-export default function cleanDatePublished(dateString, { timezone } = {}) {
+export default function cleanDatePublished(dateString, { timezone, format } = {}) {
   // If string is in milliseconds or seconds, convert to int and return
   if (MS_DATE_STRING.test(dateString) || SEC_DATE_STRING.test(dateString)) {
     return new Date(parseInt(dateString, 10)).toISOString();
   }
 
-  let date = createDate(dateString, timezone);
+  let date = createDate(dateString, timezone, format);
 
   if (!date.isValid()) {
     dateString = cleanDateString(dateString);
-    date = createDate(dateString, timezone);
+    date = createDate(dateString, timezone, format);
   }
 
   return date.isValid() ? date.toISOString() : null;

src/cleaners/date-published.test.js

@@ -30,6 +30,14 @@ describe('cleanDatePublished(dateString)', () => {
       cleanDatePublished('November 29, 2016: 8:18 AM ET', { timezone: 'America/New_York' });
     assert.equal(datePublished, '2016-11-29T13:18:00.000Z');
   });
+
+  it('accepts a custom date format', () => {
+    // The JS date parser is forgiving, but
+    // it needs am/pm separated from a time
+    const datePublished =
+      cleanDatePublished('Mon Aug 03 12:45:00 EDT 2015', { timezone: 'America/New_York', format: 'ddd MMM DD HH:mm:ss zz YYYY' });
+    assert.equal(datePublished, '2015-08-03T16:45:00.000Z');
+  });
 });
 
 describe('cleanDateString(dateString)', () => {

src/extractors/custom/www.macrumors.com/index.js

@@ -19,15 +19,8 @@ export const WwwMacrumorsComExtractor = {
       '.article .byline',
     ],
 
-    transforms: {
-      '*': ($node, $) => {
-        const $parent = $node.parent();
-        const text = $parent.text().split(' by')[0];
-        $parent.empty().text(text);
-
-        $('body').after($node);
-      },
-    },
+    // Wednesday January 18, 2017 11:44 am PST
+    format: 'dddd MMMM D, YYYY h:mm A zz',
 
     timezone: 'America/Los_Angeles',
   },