initial opml parser working, needs styling

Matt Swanson · Matt Swanson · commit 87d94360482e · 2013-04-15T20:53:01.000-04:00
diff --git a/Gemfile b/Gemfile
@@ -10,6 +10,7 @@ gem "delayed_job_active_record"
 
 gem "feedzirra", github: "pauldix/feedzirra"
 gem "loofah"
+gem "nokogiri"
 
 group :production do
   gem "unicorn"
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -111,6 +111,7 @@ DEPENDENCIES
   faker
   feedzirra!
   loofah
+  nokogiri
   pg
   pry
   rack-test
diff --git a/Rakefile b/Rakefile
@@ -11,6 +11,11 @@ task :fetch_feeds do
   FetchFeeds.new(Feed.all).fetch_all
 end
 
+desc "Fetch single feed"
+task :fetch_feed => :id do
+  # TODO
+end
+
 desc "Clear the delayed_job queue."
 task :clear_jobs do
   Delayed::Job.delete_all
diff --git a/app.rb b/app.rb
@@ -35,4 +35,5 @@ def render(*args)
   end
 end
 
-require_relative "app/controllers/stories_controller"
+require_relative "app/controllers/stories_controller"
+require_relative "app/controllers/first_run_controller"
diff --git a/app/commands/feeds/import_from_opml.rb b/app/commands/feeds/import_from_opml.rb
@@ -0,0 +1,18 @@
+require_relative "../../models/feed"
+require_relative "../../utils/opml_parser"
+
+class ImportFromOpml
+  ONE_HOUR = 60 * 60
+
+  def self.import(opml_contents, should_overwrite = false)
+    feeds = OpmlParser.new.parse_feeds(opml_contents)
+
+    Feed.delete_all if should_overwrite
+
+    feeds.each do |feed|
+      Feed.create(name: feed[:title],
+                  url: feed[:url],
+                  last_fetched: Time.now - ONE_HOUR)
+    end
+  end
+end
diff --git a/app/controllers/first_run_controller.rb b/app/controllers/first_run_controller.rb
@@ -0,0 +1,13 @@
+require_relative "../commands/feeds/import_from_opml"
+
+class Stringer < Sinatra::Base
+  get "/import" do
+    erb :import
+  end
+
+  post "/import" do
+    ImportFromOpml.import(params["opml_file"][:tempfile].read, true)
+
+    redirect to("/")
+  end
+end
diff --git a/app/tasks/fetch_feed.rb b/app/tasks/fetch_feed.rb
@@ -10,14 +10,18 @@ def initialize(feed, feed_parser = Feedzirra::Feed)
   end
 
   def fetch
-    result = @parser.fetch_and_parse(@feed.url)
+    begin
+      result = @parser.fetch_and_parse(@feed.url)
 
-    unless result.last_modified < @feed.last_fetched
-      result.entries.each do |entry|
-        StoryRepository.add(entry, @feed) if is_new?(entry)
-      end
+      unless result.last_modified < @feed.last_fetched
+        result.entries.each do |entry|
+          StoryRepository.add(entry, @feed) if is_new?(entry)
+        end
 
-      FeedRepository.update_last_fetched(@feed, result.last_modified)
+        FeedRepository.update_last_fetched(@feed, result.last_modified)
+      end
+    rescue
+      puts "Something went wrong when parsing #{@feed.url}"
     end
 
     result
diff --git a/app/utils/opml_parser.rb b/app/utils/opml_parser.rb
@@ -0,0 +1,14 @@
+require "nokogiri"
+
+class OpmlParser
+  def parse_feeds(contents)
+    doc = Nokogiri.XML(contents)
+
+    doc.xpath("//body//outline").inject([]) do |feeds, outline|
+      feeds << {
+        name: outline.attributes["title"].value,
+        url: outline.attributes["xmlUrl"].value
+      }
+    end
+  end
+end
diff --git a/app/views/import.erb b/app/views/import.erb
@@ -0,0 +1,4 @@
+<form id="import" method="POST" action="/import" enctype="multipart/form-data">
+  <input id="opml_file" name="opml_file" type="file" />
+  <input type="submit" />
+</form> 
diff --git a/app/views/index.erb b/app/views/index.erb
@@ -71,7 +71,7 @@
 </div>
 <% else %>
 <div id="zen">
-  <span id="zero">&mdash; 0 &mdash;</span>
+  <span id="zero">&ndash; 0 &ndash;</span>
   <p>You&#8217;ve reached RSS Zero&trade;</p>
   <p>
     <small>
diff --git a/spec/factories/story_factory.rb b/spec/factories/story_factory.rb
@@ -9,7 +9,7 @@ def headline
     def source
       self.feed.name
     end
-  end;
+  end
 
   def self.build
     FakeStory.new(
diff --git a/spec/sample_data/subscriptions.xml b/spec/sample_data/subscriptions.xml
diff --git a/spec/utils/opml_parser_spec.rb b/spec/utils/opml_parser_spec.rb
@@ -0,0 +1,50 @@
+require "spec_helper"
+
+app_require "utils/opml_parser"
+
+describe OpmlParser do
+  describe "#parse_feeds" do
+    it "it returns a hash of feed details from an OPML file" do
+      parser = OpmlParser.new
+
+      result = parser.parse_feeds(<<-eos)
+        <?xml version="1.0" encoding="UTF-8"?>
+        <opml version="1.0">
+        <head>
+          <title>matt swanson subscriptions in Google Reader</title>
+        </head>
+        <body>
+          <outline text="a sample feed" title="a sample feed" type="rss"
+              xmlUrl="http://feeds.feedburner.com/foobar" htmlUrl="http://www.example.org/"/>
+          <outline text="lolol" title="Matt's Blog" type="rss"
+              xmlUrl="http://mdswanson.com/atom.xml" htmlUrl="http://mdswanson.com/"/>
+        </body>
+        </opml>
+      eos
+
+      result.count.should eq 2
+      result.first[:name].should eq "a sample feed"
+      result.first[:url].should eq "http://feeds.feedburner.com/foobar"
+    
+      result.last[:name].should eq "Matt's Blog"
+      result.last[:url].should eq "http://mdswanson.com/atom.xml"
+    end
+
+    it "doesn't explode when there are no feeds" do
+      parser = OpmlParser.new
+
+      result = parser.parse_feeds(<<-eos)
+        <?xml version="1.0" encoding="UTF-8"?>
+        <opml version="1.0">
+        <head>
+          <title>matt swanson subscriptions in Google Reader</title>
+        </head>
+        <body>
+        </body>
+        </opml>
+      eos
+
+      result.should be_empty
+    end
+  end
+end
