bookscan 0.1.3 → 0.2.0

data/Rakefile

@@ -16,7 +16,7 @@ Jeweler::Tasks.new do |gem|
   gem.homepage = "http://github.com/tumf/bookscan"
   gem.license = "MIT"
   gem.summary = %Q{BookScan Scraper}
-  gem.description = %Q{This is a scraper of BookScan (http://www.bookscan.co.jp/) Service.This is *NOT* a official software of BookScan.}
+  gem.description = %Q{This is a scraper of Bookscan (http://www.bookscan.co.jp) Service.This is *NOT* a official software of Bookscan.}
   gem.email = "y.takahara@gmail.com"
   gem.authors = ["Yoshihiro TAKAHARA"]
   # Include your dependencies below. Runtime dependencies are required when using your gem,

data/VERSION

@@ -1 +1 @@
-0.1.3
+0.2.0

data/bookscan.gemspec

@@ -5,13 +5,13 @@
 
 Gem::Specification.new do |s|
   s.name = %q{bookscan}
-  s.version = "0.1.3"
+  s.version = "0.2.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Yoshihiro TAKAHARA"]
   s.date = %q{2011-01-18}
   s.default_executable = %q{bookscan}
-  s.description = %q{This is a scraper of BookScan (http://www.bookscan.co.jp/) Service.This is *NOT* a official software of BookScan.}
+  s.description = %q{This is a scraper of Bookscan (http://www.bookscan.co.jp) Service.This is *NOT* a official software of Bookscan.}
   s.email = %q{y.takahara@gmail.com}
   s.executables = ["bookscan"]
   s.extra_rdoc_files = [

data/lib/bookscan.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 module Bookscan
   BSURL = "http://system.bookscan.co.jp"
-  TUNE_TYPES = ["ipad","iphone","kindle3","kindledx","android","sonyreader","nook","jpg"]
-  TUNED_PATTERN = /((iphone|ipad|kindle3|kindledx|android|sonyreader|nook|jpg)_([^>%]*\.pdf))/
+  TUNE_TYPES = ["ipad","iphone4","kindle3","kindledx","android","sonyreader","nook","jpg"]
+  TUNED_PATTERN = /((iphone4|ipad|kindle3|kindledx|android|sonyreader|nook|jpg)_([^>%]*\.pdf))/
 end

data/lib/bookscan/agent.rb

@@ -64,19 +64,35 @@ module Bookscan
       bs
     end
 
-    def tune book,type
-      getr(book.group_url)
-      page.search("a[@class=downloading]").each do |u|
-        if u.text == "変換" and
-            /_#{book.isbn}/ =~ u.attributes["href"].value.to_s
-          click(u)
-          page.forms.first["optimize_type"] = type;
-          page.forms.first["cover_flg"] = "1";
-          page.forms.first.submit
-          puts "tune %s " % [book.title]
-          return true
-        end
+    def tuning?(book,type)
+      @tuning = tuning unless @tuning
+      @tuning.each { |b|
+        return true if b.title == type+"_"+book.title
+      }
+      false
+    end
+
+    def tune(book,type,is_premium = true)
+      if is_premium
+        max_queue = 10
+      else
+        max_queue = 1
       end
+
+      @tuning = tuning unless @tuning
+      # チューニングいっぱい
+      raise "tune queue is full" if @tuning.length >= max_queue
+      # チューニング
+      return false if tuning?(book,type)
+      # tune
+      getr(book.tune_url)
+      page.forms.first["optimize_type"] = type;
+      page.forms.first["cover_flg"] = "1";
+      page.forms.first.submit
+      tuned = book.clone
+      tuned.title = type +"_"+book.title
+      @tuning << tuned
+      tuned
     end
 
     def groups
@@ -131,7 +147,7 @@ module Bookscan
           book.title = u.text.to_s
           book.url = u.attributes["href"].value.to_s
           book.group_url = url
-          books[book.isbn] = book
+          books[book.book_id] = book
         end
       end
       books

data/lib/bookscan/book.rb

@@ -1,6 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 require 'digest/md5'
+require 'uri'
 require 'rubygems'
 require 'mutter'
 
@@ -14,7 +15,7 @@ module Bookscan
       end
       
       each do |b|
-        table << [b.id,b.title]
+        table << [b.book_id,b.title]
       end
       table.to_s if length > 0
     end
@@ -22,20 +23,20 @@ module Bookscan
     def ids
       a = Array.new
       each do |b|
-        a << b.id
+        a << b.book_id
       end
       a
     end
 
     def by_id(book_id)
       each do |b|
-        return b if b.id == book_id
+        return b if b.book_id == book_id
       end
     end
 
     def has?(book_id)
       each do |b|
-        return true if b.id == book_id
+        return true if b.book_id == book_id
       end
       false
     end
@@ -43,6 +44,15 @@ module Bookscan
 
   class Book
     attr_accessor :url,:title,:group_url
+    def tune_url
+      "/bookoptimize.php?hash=%s&d=%s&filename=%s" % [hash,d,URI.encode(@title)]
+    end
+    def d
+      return $1 if /.*download.php\?d=([^&]+)/ =~ @url
+    end
+    def hash
+      return $1 if /.*bookdetail.php\?hash=(.*)/ =~ @group_url
+    end
 
     def to_s
       @title
@@ -51,7 +61,7 @@ module Bookscan
     def filename
       return @title if isbn
       if /(.*)\.pdf$/ =~ @title
-        return $1 + "_" + id + ".pdf"
+        return $1 + "_" + book_id + ".pdf"
       end
       raise "Can't make filename"
     end
@@ -68,7 +78,7 @@ module Bookscan
       return $1 if /_([0-9a-zA-Z]+)\.pdf$/ =~ @title
     end
 
-    def id
+    def book_id
       return isbn if isbn
       title = @title
       if TUNED_PATTERN =~ title

data/lib/bookscan/cache.rb

@@ -22,9 +22,21 @@ module Bookscan
       ts
     end
 
-    def books(group)
-      groups.each do |g|
-        return g.books if g.hash == group.hash
+    def tuned?(book,type)
+      tuned.has?(book.book_id) and tuned.by_id(book.book_id).tune_type == type
+    end
+
+    def books(group = nil)
+      if group
+        groups.each do |g|
+          return g.books if g.hash == group.hash
+        end
+      else
+        bs = Books.new
+        groups.each do |g|
+          bs += g.books
+        end
+        bs
       end
     end
 

data/lib/bookscan/commands.rb

@@ -172,6 +172,7 @@ module Bookscan
       directory = "."
       hash = nil
       type = nil
+      dry_run = false
       opt.on('-d DIR','--directory=DIR', 'download directory') do |v|
         directory = v
       end
@@ -181,31 +182,59 @@ module Bookscan
       opt.on('-t TYPE','--tuned=TYPE', 'download tuned') do |v|
         type = v
       end
+      opt.on('--dry-run', 'dry-run mode') do |v|
+        dry_run = true
+      end
       opt.parse!(@command_options)
       book_id = @command_options.shift
-      if type
-        book = ask_tuned_book_id(book_id,type)
+
+      if book_id == "all"
+        if type
+          bs = @cache.tuned
+        else
+          bs = @cache.books
+        end
+        bs.each { |book|
+          if Dir.glob(directory + "/**/*" + book.book_id + "*.pdf").length == 0
+            path = directory + "/" +book.filename
+            puts "download: " + path
+            unless dry_run
+              start
+              @agent.download(book.url,path)
+            end
+          end
+        }
       else
-        book = ask_book_id(book_id,hash)
+        if type
+          book = ask_tuned_book_id(book_id,type)
+        else
+          book = ask_book_id(book_id,hash)
+        end
+        
+        # download
+        path = directory + "/" +book.filename
+        puts "download: " + path
+        unless dry_run
+          start
+          @agent.download(book.url,path)
+        end
       end
 
-      # download
-      start
-      path = directory + "/" +book.filename
-      puts "download: " + path
-      @agent.download(book.url,path)
     end
 
     def tune
       opt = OptionParser.new
       hash = nil
+      dry_run = false
       opt.on('-g HASH','--group=HASH', 'group hash') do |v|
         hash = v
       end
+      opt.on('--dry-run', 'dry-run mode') do |v|
+        dry_run = true
+      end
       opt.parse!(@command_options)
       book_id = @command_options.shift
       type =  @command_options.shift
-      book = ask_book_id(book_id,hash)
       unless type
         type = ask('Enter tune type: ',TUNE_TYPES) do |q|
           q.validate = /\w+/
@@ -213,10 +242,27 @@ module Bookscan
         end
       end
 
-      # tune
-      start
-      @agent.tuning
-      @agent.tune(book,type)
+      if book_id == "all"
+        tuned = @cache.tuned
+        bs = @cache.books
+        bs.each { |book|
+          unless @cache.tuned?(book,type)
+            # tune
+            unless dry_run
+              start
+            end
+            puts "tune for %s: %s" % [type, book.title] if dry_run or @agent.tune(book,type)
+          end
+        }
+      else
+        book = ask_book_id(book_id,hash)
+        # tune
+        puts "tune for %s: %s" % [type, book.title]
+        unless dry_run
+          start
+          @agent.tune(book,type)
+        end
+      end
     end
 
     def tuning

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: bookscan
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 23
   prerelease: 
   segments: 
   - 0
-  - 1
-  - 3
-  version: 0.1.3
+  - 2
+  - 0
+  version: 0.2.0
 platform: ruby
 authors: 
 - Yoshihiro TAKAHARA
@@ -138,7 +138,7 @@ dependencies:
   requirement: *id008
   prerelease: false
   name: highline
-description: This is a scraper of BookScan (http://www.bookscan.co.jp/) Service.This is *NOT* a official software of BookScan.
+description: This is a scraper of Bookscan (http://www.bookscan.co.jp) Service.This is *NOT* a official software of Bookscan.
 email: y.takahara@gmail.com
 executables: 
 - bookscan