Adds support for .formats and fixes tests

spec/book_spec.cr

@@ -17,6 +17,8 @@ describe Muse::Dl::Book do
     book.author.should eq "Albert H. Tillson, Jr."
     book.date.should eq "2010"
     book.publisher.should eq "University of Virginia Press"
+    book.formats.should contain :pdf
+    book.formats.should_not contain :html
   end
 
   it "should parse the summary" do
@@ -50,7 +52,7 @@ describe Muse::Dl::Book do
     ]
   end
 
-  it "it should parse the DOI for 68534" do
+  it "should parse book/68534" do
     html = File.new("spec/fixtures/book-68534.html").gets_to_end
     book = Muse::Dl::Book.new html
     book.info["ISBN"].should eq "9781501737695"
@@ -60,5 +62,14 @@ describe Muse::Dl::Book do
     book.info["Language"].should eq "English"
     book.info["Open Access"].should eq "Yes"
     book.info["DOI"].should eq "10.1353/book.68534"
+    book.formats.should contain :html
+    book.formats.should_not contain :pdf
+  end
+
+  it "should note both formats for book/60322" do
+    html = File.new("spec/fixtures/book-60322.html").gets_to_end
+    book = Muse::Dl::Book.new html
+    book.formats.should contain :pdf
+    book.formats.should contain :html
   end
 end

src/fetch.cr

@@ -25,7 +25,7 @@ module Muse::Dl
         return
       end
 
-      url = "https://muse.jhu.edu/chapter/#{chapter_id}"
+      url = "https://muse.jhu.edu/chapter/#{chapter_id}/pdf"
       headers = HEADERS.merge({
         "Referer" => "https://muse.jhu.edu/verify?url=%2Fchapter%2F#{chapter_id}%2Fpdf",
       })

src/infoparser.cr

@@ -1,5 +1,12 @@
 require "myhtml"
 
+# https://github.com/kostya/myhtml/issues/19
+struct Myhtml::Node
+  def inner_html
+    String.build { |buf| children.each &.to_html(buf) }
+  end
+end
+
 module Muse::Dl
   class InfoParser
     def self.infobox(myhtml : Myhtml::Parser)
@@ -16,6 +23,13 @@ module Muse::Dl
       return info
     end
 
+    def self.id(myhtml : Myhtml::Parser)
+      searchid = myhtml.css("#search_within_book_id").first
+      if searchid
+        searchid.attribute_by("value")
+      end
+    end
+
     def self.title(myhtml : Myhtml::Parser)
       myhtml.css("#book_about_info .title").map(&.inner_text).to_a[0].strip
     end
@@ -42,8 +56,24 @@ module Muse::Dl
     end
 
     def self.summary_html(myhtml : Myhtml::Parser)
-      return "TODO"
-      myhtml.css("#book_about_info .card_summary").map(&.tag_text).to_a[0].strip
+      summary_div = myhtml.css("#book_about_info .card_summary")
+      begin
+        summary_div.first.inner_html
+      rescue e : Exception
+        "NA"
+      end
+    end
+
+    def self.formats(myhtml : Myhtml::Parser)
+      formats = Set(Symbol).new
+      myhtml.css("img.icon").each do |icon|
+        url = icon.attribute_by("src")
+        if url
+          formats.add :html if /html/i.match url
+          formats.add :pdf if /pdf/i.match url
+        end
+      end
+      formats
     end
   end
 end

src/thing.cr

@@ -13,22 +13,26 @@ module Muse::Dl
     @cover_url : String
     @thumbnail_url : String
     @h : Myhtml::Parser
+    @formats : Set(Symbol)
 
-    getter :info, :title, :author, :date, :publisher, :summary, :summary_html, :cover_url, :thumbnail_url
+    getter :info, :title, :author, :date, :publisher, :summary, :summary_html, :cover_url, :thumbnail_url, :formats
 
     private getter :h
 
     def initialize(html : String)
       @h = Myhtml::Parser.new html
       @info = InfoParser.infobox(h)
+      id : String | Nil = InfoParser.id(h)
       @title = InfoParser.title(h)
       @author = InfoParser.author(h)
       @date = InfoParser.date(h)
       @publisher = InfoParser.publisher(h)
       @summary = InfoParser.summary(h)
       @summary_html = InfoParser.summary_html(h)
-      @cover_url = "TODO"
-      @thumbnail_url = "TODO"
+      @formats = InfoParser.formats(h)
+      # TODO: Make this work for journals as well
+      @cover_url = "https://muse.jhu.edu/book/#{id}/image/front_cover.jpg"
+      @thumbnail_url = "https://muse.jhu.edu/book/#{id}/image/front_cover.jpg?format=180"
     end
   end
 end