X-Git-Url: http://jaekl.net/gitweb/?a=blobdiff_plain;f=book.rb;h=cd14ab690b6164c14c7f842b4755e85bb87d231e;hb=2122d68f0408b8471f9dedd9998ecf890a90f703;hp=aa61930bc357fb89dcc861085eedd38921b1f654;hpb=fffa1ed35ce07fdec65e1aa14a8f637fbdde9b10;p=quanlib.git

diff --git a/book.rb b/book.rb
index aa61930..cd14ab6 100644
--- a/book.rb
+++ b/book.rb
@@ -1,8 +1,10 @@
 
 require 'nokogiri'
+require 'rubygems'
 require 'zip'
 
 require 'author'
+require 'classification'
 require 'cover'
 require 'store'
 
@@ -11,6 +13,7 @@ class Book
 
   def initialize(store)
     @author = nil
+    @classification_id = nil
     @cover = nil
     @description = nil
     @path = nil
@@ -20,12 +23,12 @@ class Book
     @volume = nil
   end
 
-  def loadFromFile(fileName)
+  def load_from_file!(fileName)
     @path = fileName
-    parseFileName!(fileName)
+    parse_file_name!(fileName)
   end
 
-  def self.canHandle?(fileName)
+  def self.can_handle?(fileName)
     if nil == fileName
       return false
     end
@@ -37,6 +40,10 @@ class Book
       return true
     end
 
+    if lowerName.end_with?(".pdf")
+      return true
+    end
+
     return false
   end
 
@@ -48,6 +55,14 @@ class Book
     @author = value
   end
 
+  def classification_id
+    @classification_id
+  end
+
+  def classification_id=(value)
+    @classification_id = value
+  end
+
   def cover
     return @cover
   end
@@ -75,7 +90,7 @@ class Book
     if nil != @author
       result.push('<i>by ' + @author.reading_order + '</i>')
     end
-    
+
     seriesInfo = []
     series = @store.load_series(@series_id)
     if nil != series and nil != series.descr
@@ -88,6 +103,19 @@ class Book
       result.push(seriesInfo.join(' '))
     end
 
+    classification = nil
+    if nil != @classification_id
+      classification = @store.load_classification(@classification_id)
+    end
+    if nil != classification
+      if nil != classification.ddc
+        result.push('Dewey: ' + classification.ddc.to_s)
+      end
+      if nil != classification.lcc
+        result.push('LCC: ' + classification.lcc.to_s)
+      end
+    end
+
     return result.join('<br/>')
   end
 
@@ -125,7 +153,7 @@ class Book
   def series_id
     @series_id
   end
-  
+
   def series_id=(value)
     @series_id = value
   end
@@ -142,6 +170,14 @@ class Book
     @title = value
   end
 
+  def title_grouping
+    if nil == @path
+      return nil
+    end
+
+    return File.basename(@path, '.*')
+  end
+
   def volume
     @volume
   end
@@ -204,8 +240,11 @@ class Book
   end
 
   protected
-  def parseFileName!(fileName)
-    parts = fileName.split('/')
+  def parse_file_name!(file_name)
+    category = nil   # e.g., non-fiction, fan-fiction
+    grouping = ''
+
+    parts = file_name.split('/')
     (series_code, @volume, @title) = processTitle(parts[-1])
     if parts.length > 1
       grouping = parts[-2]
@@ -214,16 +253,32 @@ class Book
       @author = Author.new(grouping, reading_order, sort_order)
       @series_id = @store.get_series(grouping, series_code)
     end
+    if parts.length > 2
+      category = parts[-3]
+    end
+
+    lc_file_name = file_name.downcase
+    if lc_file_name.end_with?(".epub")
+      scanEpub!(file_name)
+    elsif lc_file_name.end_with?(".pdf")
+      scan_pdf!(file_name)
+    end
 
-    if fileName.downcase.end_with?(".epub")
-      scanEpub!(fileName)
+    @classification_id = @store.find_classification(@author.grouping, File.basename(file_name, '.*'))
+
+    # TODO:  Fix horrible hard-coded strings and paths
+    if ('01_nonfic' == category) && (nil == classification_id)
+      open(Store.unclassified_csv, 'a') do |fd|
+        fd.puts('"' + grouping.to_s + '","' + path + '"')
+      end
     end
   end
 
-  protected 
+  protected
   def scanEpub!(fileName)
     #puts 'Scanning "' + fileName.to_s + '"...'
     begin
+      Zip.warn_invalid_date = false
       Zip::File.open(fileName) do |zipfile|
         entry = zipfile.find_entry('META-INF/container.xml')
         if nil == entry
@@ -243,6 +298,25 @@ class Book
     end
   end
 
+  protected
+  def scan_pdf!(file_name)
+    #puts 'Scanning "' + file_name.to_s + '"...'
+
+    pdf_path = File.expand_path(file_name).to_s
+    if ! pdf_path.end_with?('.pdf')
+      puts 'Unexpected internal error:  path "' + file_name.to_s + '" does not end with ".pdf".'
+      return
+    end
+
+    jpeg_path = pdf_path[0..-5] + '.jpeg'
+    if File.file?(jpeg_path)
+      File.open(jpeg_path, 'r') do |is|
+        @cover = Cover.new(is, jpeg_path, 'image/jpeg')
+      end
+    end
+  end
+
+
   protected
   def scanOpf!(zipfile, opfPath)
     coverId = nil
@@ -288,7 +362,7 @@ class Book
 
     #---------------------------------------
     # Description
-    
+
     descrNodes = opfDoc.css('dc|description', 'dc' => @@DC_NS_URL)
     if (descrNodes.length > 0)
       descrNode = descrNodes[0]
@@ -339,9 +413,9 @@ class Book
         entry = zipfile.find_entry(href)
 
         if nil == entry
-          # Although the epub standard requires the path to be relative 
+          # Although the epub standard requires the path to be relative
           # to the base of the epub (zip), some books encountered in the
-          # wild have been found to use a bath relative to the location 
+          # wild have been found to use a bath relative to the location
           # of the opf file.
           parts = opfPath.split('/')
           opfBasePath = opfPath.split('/')[0..-2].join('/')
@@ -349,6 +423,14 @@ class Book
           entry = zipfile.find_entry(coverPath)
         end
 
+        unless entry
+          # Another case found in the wild:  cover image is at the root, but path is '../cover.jpeg'
+          if href.start_with? '../'
+            coverPath = href[3..-1]
+            entry = zipfile.find_entry(coverPath)
+          end
+        end
+
         if nil == entry
           puts 'WARNING!  Cover image "' + href + '" not found in file "' + @path + '".'
           return nil