X-Git-Url: http://jaekl.net/gitweb/?a=blobdiff_plain;ds=sidebyside;f=walkdir.rb;h=b00645545fa984b5fbef17b09b45513957ab6f69;hb=872d620121706ad345b7e667521be1c7326c2e00;hp=236e3feaf7244852b4b58e780e9fdce456da6fc2;hpb=fcaeedd4d1c128ff84371c0a7db5d0af6751492a;p=quanlib.git

diff --git a/walkdir.rb b/walkdir.rb
index 236e3fe..b006455 100644
--- a/walkdir.rb
+++ b/walkdir.rb
@@ -4,7 +4,7 @@
 #   .../AuthorName/Title_of_the_Awesome_Book.ext
 #
 # Author is given as FirstLast.  For example, 
-# Robert Anson Heinlein is RoberHeinlein, and 
+# Robert Anson Heinlein is RobertHeinlein, and 
 # JKRowling is JoanneRowling.
 #
 # Book titles have spaces replaced with underscores,
@@ -30,10 +30,11 @@ class WalkDir
 
   def books
     result = []
-    for file in @files.sort
+    @files = remove_duplicates(@files)
+    for file in @files.sort()
       if Book.can_handle?(file) && (!is_duplicate?(file))
         book = Book.new(@store)
-        book.load_from_file(file)
+        book.load_from_file!(file)
         id = @store.store_book(book)
         result.push(id)
       end
@@ -57,6 +58,30 @@ class WalkDir
     return false
   end
 
+  def remove_duplicates(files)
+    unique = {}
+    for file in files
+      if Book.can_handle?(file)
+        key = File.dirname(file) + '/' + File.basename(file, '.*')
+        if unique.has_key?(key)
+          new_ext = File.extname(file)
+          old_ext = File.extname(unique[key])
+          if ('.pdf' == old_ext) && ('.epub' == new_ext)
+            # Prefer EPUB over PDF
+            puts 'REPLACED ' + unique[key].to_s + ' with ' + file.to_s
+            unique[key] = file
+          else
+            puts 'DROPPED ' + file.to_s + " because it's superceded by " + unique[key].to_s
+          end
+        else
+          unique[key] = file
+        end
+      end
+    end
+
+    return unique.values
+  end
+
   def walk(path)
     result = []
     children = Dir.entries(path)
@@ -71,7 +96,6 @@ class WalkDir
         result.push(fullName)
       end
     end
-    #puts result
     return result
   end
 end