klipbook 0.3.0 → 1.0.0

data/spec/lib/klipbook/sources/kindle_device/entry_parser_spec.rb

@@ -0,0 +1,275 @@
+require 'spec_helper'
+
+describe Klipbook::Sources::KindleDevice::EntryParser do
+
+  describe '#build_entry' do
+    subject { Klipbook::Sources::KindleDevice::EntryParser.new.build_entry(entry_text) }
+
+    context 'passed an empty entry' do
+      let(:entry_text) { "  " }
+
+      it { should be_nil }
+    end
+
+    context 'passed an incomplete entry' do
+      let(:entry_text) do
+        "Not long enough"
+      end
+
+      it { should be_nil }
+    end
+
+    context 'passed an entry with the title line "Book title (Author\'s Name)"' do
+      let (:entry_text) do
+        "Book title (Author's Name)\n" +
+        "- Highlight Loc. 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Highlight text\n"
+      end
+
+      its(:title) { should == 'Book title' }
+
+      its(:author) { should == "Author's Name" }
+    end
+
+    context 'passed an entry with the title line "Book title (sub title) (Author\'s Name)"' do
+      let (:entry_text) do
+        "Book title (sub title) (Author's Name)\n" +
+        "- Highlight Loc. 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Highlight text\n"
+      end
+
+      its(:title) { should == 'Book title (sub title)' }
+
+      its(:author) { should == "Author's Name" }
+    end
+
+    context 'passed an entry with the title line "Book title"' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight Loc. 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Highlight text\n"
+      end
+
+      its(:title) { should == 'Book title' }
+
+      its (:author) { should be_nil }
+    end
+
+    context 'passed an entry that is a highlight' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight Loc. 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      its(:type) { should == :highlight }
+
+      it 'extracts the highlighted text' do
+        subject.text.should == "The first line of the highlight\nThe second line of the highlight"
+      end
+    end
+
+    context 'passed an entry that is a 4th-generation highlight' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Highlight Location 466-69 | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      its(:type) { should == :highlight }
+
+      it 'extracts the highlighted text' do
+        subject.text.should == "The first line of the highlight\nThe second line of the highlight"
+      end
+    end
+
+    context 'passed an entry that is a note' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Note Loc. 623  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The note text"
+      end
+
+      its(:type) { should == :note }
+
+      it 'extracts the note text' do
+        subject.text.should == "The note text"
+      end
+    end
+
+    context 'passed an entry that is a 4th-generation note' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Note Location 623  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The note text"
+      end
+
+      its(:type) { should == :note }
+
+      it 'extracts the note text' do
+        subject.text.should == "The note text"
+      end
+    end
+
+    context 'passed an entry with a bookmark' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Bookmark on Page 1 | Loc. 406  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "\n"
+      end
+
+      its(:type) { should == :bookmark }
+
+      its(:text) { should == '' }
+    end
+
+    context 'passed an entry with a 4th-generation bookmark' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Bookmark on Page 1 | Location 406  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "\n"
+      end
+
+      its(:type) { should == :bookmark }
+
+      its(:text) { should == '' }
+    end
+
+    context 'passed an entry with a single location value' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight Loc. 465 | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      its(:location) { should == 465 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed a 4th-generation entry with a single location value' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Highlight Location 465 | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      its(:location) { should == 465 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed an entry with a location range' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight Loc. 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      it 'extracts the first element of the location range' do
+        subject.location.should == 466
+      end
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed a 4th-generation entry with a location range' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Highlight Location 466-69  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "The first line of the highlight\n" +
+        "The second line of the highlight"
+      end
+
+      it 'extracts the first element of the location range' do
+        subject.location.should == 466
+      end
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed an entry with a page number and location range' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight on Page 171 | Loc. 1858-59  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Some highlighted text\n" +
+        "\n"
+      end
+
+      it 'extracts the first element of the location range' do
+        subject.location.should == 1858
+      end
+
+      its(:page) { should == 171 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed a 4th-generation entry with a page number and location range' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Highlight on Page 171 | Location 1858-59  | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Some highlighted text\n" +
+        "\n"
+      end
+
+      it 'extracts the first element of the location range' do
+        subject.location.should == 1858
+      end
+
+      its(:page) { should == 171 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed an entry with a page number and no location' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Highlight on Page 9 | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Clipping"
+      end
+
+      its(:location) { should == 0 }
+
+      its(:page) { should == 9 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+
+    context 'passed a 4th-generation entry with a page number and no location' do
+      let (:entry_text) do
+        "Book title\n" +
+        "- Your Highlight on Page 9 | Added on Thursday, April 21, 2011, 07:31 AM\n" +
+        "\n" +
+        "Clipping"
+      end
+
+      its(:location) { should == 0 }
+
+      its(:page) { should == 9 }
+
+      its(:added_on) { should == DateTime.parse('Thursday, April 21, 2011, 07:31 AM') }
+    end
+  end
+end

data/spec/lib/klipbook/sources/kindle_device/file_parser_spec.rb

@@ -0,0 +1,68 @@
+# encoding: UTF-8
+
+require 'spec_helper'
+
+describe Klipbook::Sources::KindleDevice::FileParser do
+
+  let(:parser) { Klipbook::Sources::KindleDevice::FileParser.new(entry_parser) }
+
+  let(:entry_parser) do
+    mock_parser = Object.new
+    stub(mock_parser).build_entry
+    mock_parser
+  end
+
+  describe '#extract_entries' do
+    subject { parser.extract_entries(raw_text) }
+
+    context 'called with empty text' do
+
+      let(:raw_text) { '' }
+
+      it { should be_empty }
+    end
+
+    context 'called with text containing two entries' do
+      let(:raw_text) do
+        " entry one" +
+        "==========" +
+        " entry two" +
+        "=========="
+      end
+
+      it 'builds two entries with the entries parser and returns them' do
+        entry_one = Object.new
+        entry_two = Object.new
+
+        stub(entry_parser).build_entry(' entry one') { entry_one }
+        stub(entry_parser).build_entry(' entry two') { entry_two }
+
+        subject.should == [ entry_one, entry_two ]
+      end
+    end
+
+    context 'called with text containing carriage return characters' do
+      let(:raw_text) do
+        "Example \r text\r" +
+        "=========="
+      end
+
+      it 'strips carriage returns before calling the entry parser' do
+        subject
+        entry_parser.should have_received.build_entry('Example  text')
+      end
+    end
+
+    context 'called with text containing control characters' do
+      let(:raw_text) do
+        "Example \xef\xbb\xbf text\xef\xbb\xbf" +
+        "=========="
+      end
+
+      it 'strips control characters before calling the entry parser' do
+        subject
+        entry_parser.should have_received.build_entry('Example  text')
+      end
+    end
+  end
+end

data/spec/lib/klipbook/sources/kindle_device/file_spec.rb

@@ -0,0 +1,163 @@
+require 'spec_helper'
+
+describe Klipbook::Sources::KindleDevice::File do
+
+  let(:file) { Klipbook::Sources::KindleDevice::File.new('file text', max_books, file_parser) }
+
+  let (:max_books) { 30 }
+
+  let (:entries) { [] }
+
+  let (:file_parser) do
+    Object.new.tap do |fake_parser|
+      stub(fake_parser).extract_entries { entries }
+    end
+  end
+
+  describe '#books' do
+    subject { file.books }
+
+    it 'parses the file text with the file parser' do
+      subject
+      file_parser.should have_received.extract_entries('file text')
+    end
+
+    context 'with entries for three books' do
+      let(:entries) do
+        [
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :highlight
+            e.added_on = DateTime.new(2012, 10, 10)
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :highlight
+            e.added_on = DateTime.new(2012, 10, 10)
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book two'
+            e.author = 'Author two'
+            e.type = :highlight
+            e.added_on = DateTime.new(2012, 10, 12)
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book three'
+            e.author = 'Author two'
+            e.type = :highlight
+            e.added_on = DateTime.new(2012, 10, 11)
+          end
+        ]
+      end
+
+      it 'returns three books' do
+        subject.should have(3).items
+      end
+
+      it 'returns books sorted by last_update descending' do
+        subject.map(&:title).should == [ 'Book two', 'Book three', 'Book one' ]
+      end
+
+      context 'and max_books set to 2' do
+
+        let (:max_books) { 2 }
+
+        it 'returns two books' do
+          subject.should have(2).items
+        end
+      end
+    end
+
+    context 'with entries for a single book' do
+      let(:entries) do
+        [
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :bookmark
+            e.location = 1
+            e.page = 1
+            e.added_on = DateTime.new(2012, 10, 10)
+            e.text = 'First one'
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :highlight
+            e.location = 10
+            e.page = 3
+            e.added_on = DateTime.new(2012, 10, 11)
+            e.text = 'Second one'
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :highlight
+            e.location = 3
+            e.page = 2
+            e.added_on = DateTime.new(2012, 10, 1)
+            e.text = 'Third one'
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :note
+            e.location = 2
+            e.page = 1
+            e.added_on = DateTime.new(2012, 10, 21)
+            e.text = 'Fourth one'
+          end
+        ]
+      end
+
+      it 'returns a single book with the correct title and author information' do
+        subject.first.title.should == 'Book one'
+        subject.first.author.should == 'Author one'
+      end
+
+      it "returns a single book with last update equal to the latest added on date of the book's entries" do
+        subject.first.last_update.should == DateTime.new(2012, 10, 21)
+      end
+
+      it "ignores bookmarks when building the book's clipping list" do
+        subject.first.clippings.should have(3).items
+        subject.first.clippings.map(&:type).should_not include(:bookmark)
+      end
+
+      it 'returns a single book whose clippings are sorted by location' do
+        subject.first.clippings.map(&:location).should == [2, 3, 10]
+        subject.first.clippings.map(&:text).should == ['Fourth one', 'Third one', 'Second one']
+        subject.first.clippings.map(&:page).should == [1, 2, 3]
+      end
+    end
+
+    context 'with entries for a single book that are all bookmarks' do
+      let(:entries) do
+        [
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :bookmark
+            e.location = 1
+            e.page = 1
+            e.added_on = DateTime.new(2012, 10, 10)
+            e.text = 'First one'
+          end,
+          Klipbook::Sources::KindleDevice::Entry.new do |e|
+            e.title = 'Book one'
+            e.author = 'Author one'
+            e.type = :bookmark
+            e.location = 10
+            e.page = 3
+            e.added_on = DateTime.new(2012, 10, 11)
+            e.text = 'Second one'
+          end
+        ]
+      end
+
+      it { should be_empty }
+    end
+  end
+end