extend regular expression to also match images with "." characters in the filename

Reto Ryter committed Sep 18, 2015

commit 8f9214c4851703e4530ab48d300fac0f5c4a2d20

Showing 2 changed files with 30 additions and 2 deletions

locomotive/wagon/commands/pull_sub_commands/concerns/assets_concern.rb b/lib/locomotive/wagon/commands/pull_sub_commands/concerns/assets_concern.rb +2 -2

@@	@@ -4,7 +4,7 @@ module Locomotive::Wagon

	module AssetsConcern

-	ASSET_REGEX = /\/sites\/[0-9a-f]{24}\/(assets\|pages\|theme\|content_entry[0-9a-f]{24})\/(([^;.]+)\/)*([a-zA-Z_\-0-9]+)\.([a-z]{2,3})/
+	REGEX = /\/sites\/[0-9a-f]{24}\/(assets\|pages\|theme\|content_entry[0-9a-f]{24})\/(([^;.]+)\/)*([a-zA-Z_\-0-9.]+)\.([A-Za-z]{2,3})/

	# The content assets on the remote engine follows the format: /sites/<id>/assets/<type>/<file>
	# This method replaces these urls by their local representation. <type>/<file>
@@	@@ -14,7 +14,7 @@ module Locomotive::Wagon
	def replace_asset_urls(content)
	return '' if content.blank?

-	content.force_encoding('utf-8').gsub(ASSET_REGEX) do \|url\|
+	content.force_encoding('utf-8').gsub(REGEX) do \|url\|
	filename = "#{$4}.#{$5}"
	folder = case $1
	when 'assets', 'pages' then File.join('samples', $1)

spec/unit/commands/pull_sub_commands/concerns/assets_concern_spec.rb +28 -0

@@	@@ -0,0 +1,28 @@
+	# encoding: utf-8
+
+	require 'spec_helper'
+
+	require 'locomotive/wagon/commands/pull_sub_commands/concerns/assets_concern'
+
+	describe Locomotive::Wagon::AssetsConcern do
+	describe "calling replace_asset_urls" do
+
+	let(:assets_concern) { Class.new { include Locomotive::Wagon::AssetsConcern } }
+	it "identifies the assets which should be downloaded when pulling" do
+
+	input = <<-INPUT
+	<td><img src="https://somedomain.com/sites/51277e8fc0864503db000007/assets/534d2a28e6439a92180009a6/UPPERCASE.CH.FILE.jpg"/></td>
+	<td><img src="https://somedomain.com/sites/51277e8fc0864503db000007/assets/532a8eeacf7727c1bb00012e/lowercase.ch.file.jpg" /></td>
+	<td><img src="https://somedomain.com/sites/51277e8fc0864503db000007/assets/532a8eeac086452e35000169/some_name.jpg" /></td>
+	<td><img src="https://somedomain.com/sites/51277e8fc0864503db000007/assets/532a8eeac086452e35000169/some_name.JPG" /></td>
+	INPUT
+
+	parsed_urls = []
+	input.force_encoding('utf-8').gsub(assets_concern::REGEX) do \|url\|
+	parsed_urls.push url
+	end
+
+	expect(parsed_urls.count).to eq(4)
+	end
+	end
+	end