feedjira 3.2.6 → 4.0.0

data/lib/feedjira/parser/json_feed_item.rb

@@ -30,13 +30,13 @@ module Feedjira
       def parse_published(date_published)
         return nil unless date_published
 
-        Time.parse_safely(date_published)
+        Feedjira::Util::ParseTime.call(date_published)
       end
 
       def parse_updated(date_modified)
         return nil unless date_modified
 
-        Time.parse_safely(date_modified)
+        Feedjira::Util::ParseTime.call(date_modified)
       end
 
       # Convenience method to return the included content type.

data/lib/feedjira/preprocessor.rb

@@ -15,7 +15,7 @@ module Feedjira
 
     def process_content
       content_nodes.each do |node|
-        node.content = raw_html(node) unless node.cdata?
+        node.content = raw_html(node)
       end
     end
 

data/lib/feedjira/util/parse_time.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+require "time"
+require "date"
+
+module Feedjira
+  module Util
+    # Module for safely parsing time strings
+    module ParseTime
+      # Parse a time string and convert it to UTC without raising errors.
+      # Parses a flattened 14-digit time (YYYYmmddHHMMMSS) as UTC.
+      #
+      # === Parameters
+      # [dt<String or Time>] Time definition to be parsed.
+      #
+      # === Returns
+      # A Time instance in UTC or nil if there were errors while parsing.
+      def self.call(datetime)
+        if datetime.is_a?(Time)
+          datetime.utc
+        elsif datetime.respond_to?(:to_datetime)
+          datetime.to_time.utc
+        else
+          parse_string_safely datetime.to_s
+        end
+      rescue StandardError => e
+        Feedjira.logger.debug("Failed to parse time #{datetime}")
+        Feedjira.logger.debug(e)
+        nil
+      end
+
+      # Parse a string safely, handling special 14-digit format
+      #
+      # === Parameters
+      # [string<String>] String to be parsed as time.
+      #
+      # === Returns
+      # A Time instance in UTC or nil if there were errors while parsing.
+      def self.parse_string_safely(string)
+        return nil if string.empty?
+
+        if /\A\d{14}\z/.match?(string)
+          Time.parse("#{string}Z", true)
+        else
+          Time.parse(string).utc
+        end
+      end
+
+      private_class_method :parse_string_safely
+    end
+  end
+end

data/lib/feedjira/util.rb

@@ -0,0 +1,7 @@
+# frozen_string_literal: true
+
+module Feedjira
+  # Utility modules and helper functions
+  module Util
+  end
+end

data/lib/feedjira/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Feedjira
-  VERSION = "3.2.6"
+  VERSION = "4.0.0"
 end

data/lib/feedjira.rb

@@ -6,7 +6,8 @@ require "loofah"
 require "logger"
 require "json"
 
-require_relative "feedjira/core_ext"
+require_relative "feedjira/util"
+require_relative "feedjira/util/parse_time"
 require_relative "feedjira/configuration"
 require_relative "feedjira/feed_entry_utilities"
 require_relative "feedjira/feed_utilities"

data/spec/feedjira/atom_entry_utilities_spec.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+RSpec.describe Feedjira::AtomEntryUtilities do
+  def klass
+    Class.new do
+      include SAXMachine
+      include Feedjira::AtomEntryUtilities
+    end
+  end
+
+  describe "#title" do
+    it "returns the title when set" do
+      entry = klass.new
+      entry.title = "My Title"
+
+      expect(entry.title).to eq "My Title"
+    end
+
+    it "returns a sanitized version of the raw title when present" do
+      entry = klass.new
+      entry.raw_title = "My <b>Raw</b> \tTitle"
+
+      expect(entry.title).to eq "My Raw Title"
+    end
+
+    it "returns nil when no raw title is present" do
+      entry = klass.new
+
+      expect(entry.title).to be_nil
+    end
+  end
+
+  describe "#url" do
+    it "returns the url when set" do
+      entry = klass.new
+      entry.url = "http://exampoo.com/feed"
+
+      expect(entry.url).to eq "http://exampoo.com/feed"
+    end
+
+    it "returns the first link when not set" do
+      entry = klass.new
+      entry.links = ["http://exampoo.com/feed"]
+
+      expect(entry.url).to eq "http://exampoo.com/feed"
+    end
+  end
+end

data/spec/feedjira/feed_utilities_entry_spec.rb

@@ -13,7 +13,7 @@ describe Feedjira::FeedUtilities do
     it "parses an ISO 8601 formatted datetime into Time" do
       time = @klass.new.parse_datetime("2008-02-20T8:05:00-010:00")
       expect(time.class).to eq Time
-      expect(time).to eq Time.parse_safely("Wed Feb 20 18:05:00 UTC 2008")
+      expect(time).to eq Feedjira::Util::ParseTime.call("Wed Feb 20 18:05:00 UTC 2008")
     end
 
     it "parses a ISO 8601 with milliseconds into Time" do
@@ -23,6 +23,78 @@ describe Feedjira::FeedUtilities do
     end
   end
 
+  describe "updated= method" do
+    it "sets updated when no existing updated value and parsed date is valid" do
+      instance = @klass.new
+      instance.updated = "2023-01-01T10:00:00Z"
+      expect(instance["updated"]).to eq Time.parse("2023-01-01T10:00:00Z").utc
+    end
+
+    it "updates to newer date when existing updated value is older" do
+      instance = @klass.new
+      instance.updated = "2023-01-01T10:00:00Z"
+      instance.updated = "2023-01-02T10:00:00Z"
+      expect(instance["updated"]).to eq Time.parse("2023-01-02T10:00:00Z").utc
+    end
+
+    it "keeps existing updated value when new date is older" do
+      instance = @klass.new
+      instance.updated = "2023-01-02T10:00:00Z"
+      instance.updated = "2023-01-01T10:00:00Z"
+      expect(instance["updated"]).to eq Time.parse("2023-01-02T10:00:00Z").utc
+    end
+
+    it "does not set updated when date parsing fails" do
+      instance = @klass.new
+      instance.updated = "invalid-date"
+      expect(instance["updated"]).to be_nil
+    end
+
+    it "does not change existing updated when new date is invalid" do
+      instance = @klass.new
+      instance.updated = "2023-01-01T10:00:00Z"
+      original_updated = instance["updated"]
+      instance.updated = "invalid-date"
+      expect(instance["updated"]).to eq original_updated
+    end
+  end
+
+  describe "published= method" do
+    it "sets published when no existing published value and parsed date is valid" do
+      instance = @klass.new
+      instance.published = "2023-01-01T10:00:00Z"
+      expect(instance["published"]).to eq Time.parse("2023-01-01T10:00:00Z").utc
+    end
+
+    it "updates to older date when existing published value is newer" do
+      instance = @klass.new
+      instance.published = "2023-01-02T10:00:00Z"
+      instance.published = "2023-01-01T10:00:00Z"
+      expect(instance["published"]).to eq Time.parse("2023-01-01T10:00:00Z").utc
+    end
+
+    it "keeps existing published value when new date is newer" do
+      instance = @klass.new
+      instance.published = "2023-01-01T10:00:00Z"
+      instance.published = "2023-01-02T10:00:00Z"
+      expect(instance["published"]).to eq Time.parse("2023-01-01T10:00:00Z").utc
+    end
+
+    it "does not set published when date parsing fails" do
+      instance = @klass.new
+      instance.published = "invalid-date"
+      expect(instance["published"]).to be_nil
+    end
+
+    it "does not change existing published when new date is invalid" do
+      instance = @klass.new
+      instance.published = "2023-01-01T10:00:00Z"
+      original_published = instance["published"]
+      instance.published = "invalid-date"
+      expect(instance["published"]).to eq original_published
+    end
+  end
+
   describe "sanitizing" do
     before do
       @feed = Feedjira.parse(sample_atom_feed)
@@ -37,7 +109,7 @@ describe Feedjira::FeedUtilities do
       new_title = "<script>this is not safe</script>#{@entry.title}"
       @entry.title = new_title
       scrubbed_title = Loofah.scrub_fragment(new_title, :prune).to_s
-      expect(@entry.title.sanitize).to eq scrubbed_title
+      expect(Loofah.scrub_fragment(@entry.title, :prune).to_s).to eq scrubbed_title
     end
 
     it "sanitizes content in place" do
@@ -46,7 +118,7 @@ describe Feedjira::FeedUtilities do
 
       scrubbed_content = Loofah.scrub_fragment(new_content, :prune).to_s
 
-      expect(@entry.content.sanitize!).to eq scrubbed_content
+      @entry.sanitize!
       expect(@entry.content).to eq scrubbed_content
     end
 

data/spec/feedjira/feed_utilities_spec.rb

@@ -7,6 +7,8 @@ describe Feedjira::FeedUtilities do
     @klass = Class.new do
       include SAXMachine
       include Feedjira::FeedUtilities
+
+      elements :item, as: :entries, class: Feedjira::Parser::RSSEntry
     end
   end
 
@@ -28,6 +30,23 @@ describe Feedjira::FeedUtilities do
     end
   end
 
+  describe "#preprocess" do
+    it "returns the xml without changes when not overridden" do
+      expect(@klass.preprocess(sample_rss_feed)).to eq sample_rss_feed
+    end
+  end
+
+  describe "#sanitize_entries!" do
+    it "sanitizes all entries" do
+      result = @klass.parse(sample_rss_feed)
+
+      entry = result.entries.first
+
+      expect { result.sanitize_entries! }.to change(entry, :content)
+        .from(/onclick="javascript/)
+    end
+  end
+
   describe "when configured to strip whitespace" do
     context "when strip_whitespace config is true" do
       it "strips all XML whitespace" do

data/spec/feedjira/parser/atom_entry_spec.rb

@@ -39,7 +39,7 @@ describe Feedjira::Parser::AtomEntry do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "Fri Jan 16 18:21:00 UTC 2009"
+    published = Feedjira::Util::ParseTime.call "Fri Jan 16 18:21:00 UTC 2009"
     expect(@entry.published).to eq published
   end
 
@@ -48,7 +48,7 @@ describe Feedjira::Parser::AtomEntry do
   end
 
   it "parses the updated date" do
-    updated = Time.parse_safely "Fri Jan 16 18:21:00 UTC 2009"
+    updated = Feedjira::Util::ParseTime.call "Fri Jan 16 18:21:00 UTC 2009"
     expect(@entry.updated).to eq updated
   end
 

data/spec/feedjira/parser/atom_feed_burner_entry_spec.rb

@@ -46,7 +46,7 @@ describe Feedjira::Parser::AtomFeedBurnerEntry do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "Thu Jan 22 15:50:22 UTC 2009"
+    published = Feedjira::Util::ParseTime.call "Thu Jan 22 15:50:22 UTC 2009"
     expect(@entry.published).to eq published
   end
 

data/spec/feedjira/parser/atom_google_alerts_entry_spec.rb

@@ -14,8 +14,23 @@ describe Feedjira::Parser::AtomGoogleAlertsEntry do
     expect(@entry.title_type).to eq "html"
   end
 
-  it "parses the url" do
-    expect(@entry.url).to eq "https://www.aglobalmarketresearch.com/report-offers-prediction-of-automotive-slack-market-by-top-key-players-like-haldex-meritor-bendix-mei-wabco-accuride-stemco-tbk-febi-aydinsan/"
+  it "parses the url out of the params when the host is google" do
+    url = "https://www.exampoo.com"
+    entry = described_class.new(url: "https://www.google.com/url?url=#{url}")
+
+    expect(entry.url).to eq url
+  end
+
+  it "returns nil when the url is not present" do
+    entry = described_class.new
+
+    expect(entry.url).to be_nil
+  end
+
+  it "returns nil when the host is not google" do
+    entry = described_class.new(url: "https://www.exampoo.com")
+
+    expect(entry.url).to be_nil
   end
 
   it "parses the content" do
@@ -23,12 +38,12 @@ describe Feedjira::Parser::AtomGoogleAlertsEntry do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "2019-07-10T11:53:37Z"
+    published = Feedjira::Util::ParseTime.call "2019-07-10T11:53:37Z"
     expect(@entry.published).to eq published
   end
 
   it "parses the updated date" do
-    updated = Time.parse_safely "2019-07-10T11:53:37Z"
+    updated = Feedjira::Util::ParseTime.call "2019-07-10T11:53:37Z"
     expect(@entry.updated).to eq updated
   end
 end

data/spec/feedjira/parser/atom_youtube_entry_spec.rb

@@ -22,11 +22,11 @@ describe Feedjira::Parser::AtomYoutubeEntry do
     end
 
     it "has the published date" do
-      expect(@entry.published).to eq Time.parse_safely("2015-05-04T00:01:27+00:00")
+      expect(@entry.published).to eq Feedjira::Util::ParseTime.call("2015-05-04T00:01:27+00:00")
     end
 
     it "has the updated date" do
-      expect(@entry.updated).to eq Time.parse_safely("2015-05-13T17:38:30+00:00")
+      expect(@entry.updated).to eq Feedjira::Util::ParseTime.call("2015-05-13T17:38:30+00:00")
     end
 
     it "has the content populated from the media:description element" do

data/spec/feedjira/parser/google_docs_atom_spec.rb

@@ -31,5 +31,41 @@ module Feedjira
         expect(@feed.entries.first).to be_a GoogleDocsAtomEntry
       end
     end
+
+    describe "#url" do
+      it "returns the url when set" do
+        feed = GoogleDocsAtom.new
+
+        feed.url = "http://exampoo.com/feed"
+
+        expect(feed.url).to eq "http://exampoo.com/feed"
+      end
+
+      it "returns the first link when not set" do
+        feed = GoogleDocsAtom.new
+
+        feed.links = ["http://exampoo.com/feed"]
+
+        expect(feed.url).to eq "http://exampoo.com/feed"
+      end
+    end
+
+    describe "#feed_url" do
+      it "returns the feed_url when set" do
+        feed = GoogleDocsAtom.new
+
+        feed.feed_url = "http://exampoo.com/feed"
+
+        expect(feed.feed_url).to eq "http://exampoo.com/feed"
+      end
+
+      it "returns the first link when not set" do
+        feed = GoogleDocsAtom.new
+
+        feed.links = ["http://exampoo.com/feed"]
+
+        expect(feed.feed_url).to eq "http://exampoo.com/feed"
+      end
+    end
   end
 end

data/spec/feedjira/parser/i_tunes_rss_category_spec.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Feedjira::Parser::ITunesRSSCategory do
+  describe "#each_subcategory" do
+    it "returns an enumerator when no block is given" do
+      category = described_class.new
+      category.text = "Technology"
+
+      result = category.each_subcategory
+      expect(result).to be_an(Enumerator)
+    end
+
+    it "yields category text and subcategories when block is given" do
+      parent_category = described_class.new
+      parent_category.text = "Technology"
+
+      subcategory = described_class.new
+      subcategory.text = "Gadgets"
+
+      parent_category.itunes_categories = [subcategory]
+
+      yielded_categories = []
+      parent_category.each_subcategory { |cat| yielded_categories << cat }
+
+      expect(yielded_categories).to eq %w[Technology Gadgets]
+    end
+  end
+
+  describe "#each_path" do
+    it "returns an enumerator when no block is given" do
+      category = described_class.new
+      category.text = "Technology"
+
+      result = category.each_path
+      expect(result).to be_an(Enumerator)
+    end
+  end
+end

data/spec/feedjira/parser/i_tunes_rss_item_spec.rb

@@ -54,7 +54,7 @@ describe Feedjira::Parser::ITunesRSSItem do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "Wed Jun 15 19:00:00 UTC 2005"
+    published = Feedjira::Util::ParseTime.call "Wed Jun 15 19:00:00 UTC 2005"
     expect(@item.published).to eq published
   end
 

data/spec/feedjira/parser/json_feed_item_spec.rb

@@ -3,6 +3,10 @@
 require "spec_helper"
 
 describe Feedjira::Parser::JSONFeedItem do
+  def params(**overrides)
+    { "id" => "my_id", "url" => "my_url", **overrides }
+  end
+
   before do
     # I don't really like doing it this way because these unit test should only
     # rely on JSONFeed, but this is actually how it should work. You would
@@ -28,10 +32,42 @@ describe Feedjira::Parser::JSONFeedItem do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "2017-06-02T22:05:47-07:00"
+    published = Feedjira::Util::ParseTime.call "2017-06-02T22:05:47-07:00"
     expect(@entry.published).to eq published
   end
 
+  it "sets the published date to nil when not present" do
+    entry = described_class.new(params)
+
+    expect(entry.published).to be_nil
+  end
+
+  it "sets updated to date_modified when present" do
+    updated = "2017-06-02T22:05:47-07:00"
+    entry = described_class.new(params("date_modified" => updated))
+
+    updated = Feedjira::Util::ParseTime.call "2017-06-02T22:05:47-07:00"
+    expect(entry.updated).to eq updated
+  end
+
+  it "sets updated to nil when date_modified is not present" do
+    entry = described_class.new(params)
+
+    expect(entry.updated).to be_nil
+  end
+
+  it "sets the author when nested author object is present" do
+    entry = described_class.new(params("author" => { "name" => "John Doe" }))
+
+    expect(entry.author).to eq "John Doe"
+  end
+
+  it "sets the author to nil when nested author object is not present" do
+    entry = described_class.new(params)
+
+    expect(entry.author).to be_nil
+  end
+
   it "supports each" do
     expect(@entry).to respond_to :each
   end

data/spec/feedjira/parser/podlove_chapter_spec.rb

@@ -16,13 +16,21 @@ describe Feedjira::Parser::PodloveChapter do
     expect(@item.chapters.last.title).to eq "Abschied"
   end
 
-  it "parses the start time" do
-    expect(@chapter.start_ntp).to eq "00:00:26.407"
-    expect(@chapter.start).to eq 26.407
-    expect(@item.chapters[1].start).to eq 50
-    expect(@item.chapters[2].start).to eq 59.12
-    expect(@item.chapters[3].start).to eq 89.201
-    expect(@item.chapters.last.start).to eq 5700.034
+  describe "#start" do
+    it "returns the start time" do
+      expect(@chapter.start_ntp).to eq "00:00:26.407"
+      expect(@chapter.start).to eq 26.407
+      expect(@item.chapters[1].start).to eq 50
+      expect(@item.chapters[2].start).to eq 59.12
+      expect(@item.chapters[3].start).to eq 89.201
+      expect(@item.chapters.last.start).to eq 5700.034
+    end
+
+    it "returns nil when start_ntp is not present" do
+      chapter = described_class.new
+
+      expect(chapter.start).to be_nil
+    end
   end
 
   it "parses the title" do

data/spec/feedjira/parser/rss_entry_spec.rb

@@ -42,7 +42,7 @@ describe Feedjira::Parser::RSSEntry do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "Thu Dec 04 17:17:49 UTC 2008"
+    published = Feedjira::Util::ParseTime.call "Thu Dec 04 17:17:49 UTC 2008"
     expect(@entry.published).to eq published
   end
 
@@ -122,4 +122,40 @@ describe Feedjira::Parser::RSSEntry do
     feed = Feedjira.parse(sample_rss_feed_with_comments)
     expect(feed.entries[0].comments).to eq "https://news.ycombinator.com/item?id=30937433"
   end
+
+  it "returns nil when no URL is available from link or guid" do
+    xml = <<~XML
+      <rss version="2.0">
+        <channel>
+          <item>
+            <title>Entry without URL</title>
+            <description>This entry has no link or guid</description>
+          </item>
+        </channel>
+      </rss>
+    XML
+
+    feed = Feedjira.parse(xml)
+    entry = feed.entries.first
+
+    expect(entry.url).to be_nil
+  end
+
+  it "returns nil when guid exists but is not a permalink" do
+    xml = <<~XML
+      <rss version="2.0">
+        <channel>
+          <item>
+            <title>Entry with non-permalink GUID</title>
+            <guid isPermaLink="false">some-guid-123</guid>
+          </item>
+        </channel>
+      </rss>
+    XML
+
+    feed = Feedjira.parse(xml)
+    entry = feed.entries.first
+
+    expect(entry.url).to be_nil
+  end
 end

data/spec/feedjira/parser/rss_feed_burner_entry_spec.rb

@@ -43,7 +43,7 @@ describe Feedjira::Parser::RSSFeedBurnerEntry do
   end
 
   it "parses the published date" do
-    published = Time.parse_safely "Wed Nov 02 17:25:27 UTC 2011"
+    published = Feedjira::Util::ParseTime.call "Wed Nov 02 17:25:27 UTC 2011"
     expect(@entry.published).to eq published
   end