X-Git-Url: http://jaekl.net/gitweb/?a=blobdiff_plain;f=book.rb;h=cd14ab690b6164c14c7f842b4755e85bb87d231e;hb=2122d68f0408b8471f9dedd9998ecf890a90f703;hp=b0a1bbff66de2851b9bab47b8ad6eb0d16cc65f3;hpb=b8fe580f73b094e366643388ecd2d184b643616a;p=quanlib.git

diff --git a/book.rb b/book.rb
index b0a1bbf..cd14ab6 100644
--- a/book.rb
+++ b/book.rb
@@ -1,5 +1,6 @@
 
 require 'nokogiri'
+require 'rubygems'
 require 'zip'
 
 require 'author'
@@ -89,7 +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
@@ -152,7 +153,7 @@ class Book
   def series_id
     @series_id
   end
-  
+
   def series_id=(value)
     @series_id = value
   end
@@ -240,6 +241,9 @@ class Book
 
   protected
   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
@@ -249,6 +253,9 @@ 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")
@@ -258,12 +265,20 @@ class Book
     end
 
     @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
@@ -347,7 +362,7 @@ class Book
 
     #---------------------------------------
     # Description
-    
+
     descrNodes = opfDoc.css('dc|description', 'dc' => @@DC_NS_URL)
     if (descrNodes.length > 0)
       descrNode = descrNodes[0]
@@ -398,9 +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('/')
@@ -408,6 +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