Handle source dates with days of the week

lib/german-days.js

@@ -0,0 +1,18 @@
+const assert = require('assert')
+const days = ['montag','dienstag','mittwoch','donnerstag','freitag','samstag','sonntag']
+
+assert( days.length == 7 )
+
+module.exports = {
+  days,
+  nameToNumber: (day, offset = 1) => days.findIndex( (d) => d == day.toLowerCase() ) + offset,
+  removeThem: (str, suffix = false) => {
+    const asRegex = days.map((d) => {
+      if(!suffix){
+        return d
+      }
+      return d+"("+suffix+")?"
+    }).join('|')
+    return str.replace( new RegExp( asRegex, 'gi' ), '' )
+  }
+}

lib/report.js

@@ -5,6 +5,7 @@ const url = require('url')
 const preset = require('./preset')
 const {ReportParserError} = require('./errors')
 const germanMonths = require('./german-months')
+const germanDays = require('./german-days')
 const boroughs = require('./boroughs')
 const motives = require('./motives')
 const factums = require('./factums')
@@ -152,7 +153,7 @@ const parseClassNames = (classNames) => {
  */
 
 const parseSources = (sources) => {
-  const sourcesArr = sources.replace('Quelle:','')
+  const sourcesArr = germanDays.removeThem( sources.replace('Quelle:',''), ',')
     .replace(/(\(.+),(.+\))/, '$1%$2') //replace commas inside parens by %
     .replace(/\sund\s/,',') //turn " und "s into commas
     .split(',')

tests/lib/report.test.js

@@ -28,6 +28,7 @@ const sourceWithMultipleDates = fs.readFileSync('./tests/samples/sourceWithMulti
 const anonymousSource =fs.readFileSync('./tests/samples/anonymousSource.html', 'utf8')
 const monthOnly = fs.readFileSync('./tests/samples/monthOnly.html', 'utf8')
 const yearOnly = fs.readFileSync('./tests/samples/yearOnly.html', 'utf8')
+const sourceDateWithDayOfWeek = fs.readFileSync('./tests/samples/sourceDateWithDayOfWeek.html', 'utf8')
 
 
 request.mockImplementation((...args) => {
@@ -132,5 +133,15 @@ test( 'If only the year is provided record start date as 1st january of the year
   expect( validateSchema(result) ).toBeTruthy()
   expect( result.startDate ).toEqual( '2015-01-01T00:00:00.000Z' )
   expect( result.endDate ).toEqual( '2015-12-31T00:00:00.000Z' )
+})
 
+test( 'Parse date starting with day of week', async() => {
+  request.mockImplementationOnce((...args) => {
+    const cb = args.pop()
+    cb( null, {statusCode: 200}, sourceDateWithDayOfWeek )
+  })
+  const result = await scrapeReport( 'https://domain.tld/path/to/page.html' )
+  expect( validateSchema(result) ).toBeTruthy()
+  expect( result.sources[1].publishedDate ).toEqual( '2014-12-11T00:00:00.000Z' )
 })
+