From db6d1c2c822d1e864e113df270d8e247b3ea53ea Mon Sep 17 00:00:00 2001
From: Nemo <me@captnemo.in>
Date: Sun, 29 Mar 2020 02:32:18 +0530
Subject: [PATCH] Chapter parsing in place for books

---
 spec/book_spec.cr | 21 +++++++++++++++++++++
 src/book.cr       | 20 ++++++++++++++++++++
 src/thing.cr      |  6 ++++--
 3 files changed, 45 insertions(+), 2 deletions(-)

diff --git a/spec/book_spec.cr b/spec/book_spec.cr
index 9feb6e8..0a94bef 100644
--- a/spec/book_spec.cr
+++ b/spec/book_spec.cr
@@ -29,6 +29,27 @@ describe Muse::Dl::Book do
     book.thumbnail_url.should eq "https://muse.jhu.edu/book/875/image/front_cover.jpg?format=180"
   end
 
+  it "should parse the chapters" do
+    book.chapters.should eq [
+      ["16872", "Cover"],
+      ["16873", "Title Page"],
+      ["16874", "Copyright Page"],
+      ["16875", "Table of Contents"],
+      ["16876", "Acknowledgments"],
+      ["16877", "Introduction"],
+      ["16878", "Chapter 1: A Troubled Gentry"],
+      ["16879", "Chapter 2: Beyond the Plantations"],
+      ["16880", "Chapter 3: The World(s) Northern Neck Slavery Made"],
+      ["16881", "Chapter 4: The Scottish Merchants\n"],
+      ["16882", "Chapter 5: Controlling the Revolution\n"],
+      ["16883", "Chapter 6: The Evangelical Challenge"],
+      ["16884", "Chapter 7: The Preservation of Hegemony"],
+      ["16885", "Notes"],
+      ["16886", "Bibliography"],
+      ["16887", "Index"],
+    ]
+  end
+
   it "it should parse the DOI for 68534" do
     html = File.new("spec/fixtures/book-68534.html").gets_to_end
     book = Muse::Dl::Book.new html
diff --git a/src/book.cr b/src/book.cr
index 90cce6b..6da03e9 100644
--- a/src/book.cr
+++ b/src/book.cr
@@ -2,5 +2,25 @@ require "./thing.cr"
 
 module Muse::Dl
   class Book < Muse::Dl::Thing
+    @chapters : Array(Array(String))
+
+    getter :chapters
+
+    def initialize(html : String)
+      super(html)
+      @chapters = parts(@h)
+    end
+
+    def parts(myhtml : Myhtml::Parser)
+      chapters = [] of Array(String)
+      myhtml.css(".title a").each do |a|
+        link = a.attribute_by("href").to_s
+        matches = /\/chapter\/(\d+)/.match link
+        if matches
+          chapters.push [matches[1], a.inner_text]
+        end
+      end
+      chapters
+    end
   end
 end
diff --git a/src/thing.cr b/src/thing.cr
index 1787180..663b958 100644
--- a/src/thing.cr
+++ b/src/thing.cr
@@ -12,11 +12,14 @@ module Muse::Dl
     @summary_html : String
     @cover_url : String
     @thumbnail_url : String
+    @h : Myhtml::Parser
 
     getter :info, :title, :author, :date, :publisher, :summary, :summary_html, :cover_url, :thumbnail_url
 
+    private getter :h
+
     def initialize(html : String)
-      h = Myhtml::Parser.new html
+      @h = Myhtml::Parser.new html
       @info = InfoParser.infobox(h)
       @title = InfoParser.title(h)
       @author = InfoParser.author(h)
@@ -24,7 +27,6 @@ module Muse::Dl
       @publisher = InfoParser.publisher(h)
       @summary = InfoParser.summary(h)
       @summary_html = InfoParser.summary_html(h)
-
       @cover_url = "TODO"
       @thumbnail_url = "TODO"
     end