Housekeeping: remove commented-out code.

[quanlib.git] / walkdir.rb

diff --git a/walkdir.rb b/walkdir.rb
index 1dadd2e35b95974f81ad43366698c46b93dac7b9..bbb56a263fbb70381875182d88bcf21fcb05c1b3 100644 (file)
--- 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,
@@ -18,8 +18,8 @@
 # and Mrs. Pollifax volume 6, On the China Station, is
 # .../DorothyGilman/P06_On_the_China_Station.epub.
 
-require 'book'
-require 'store'
+require_relative 'book'
+require_relative 'store'
 
 class WalkDir
   def initialize(store, root)
@@ -30,10 +30,11 @@ class WalkDir
 
   def books
     result = []
-    for file in @files.sort
-      if Book.canHandle?(file)
+    @files = remove_duplicates(@files)
+    for file in @files.sort()
+      if Book.can_handle?(file) && (!is_duplicate?(file))
         book = Book.new(@store)
-        book.loadFromFile(file)
+        book.load_from_file!(file)
         id = @store.store_book(book)
         result.push(id)
       end
@@ -41,6 +42,46 @@ class WalkDir
     return result
   end
 
+  # Duplicate versions of a text are named 
+  #   xxx_suffix.ext
+  # Where suffix is one of bis, ter, quater, quinquies
+  # for the 2nd, 3rd, 4th or 5th variant respectively.
+  def is_duplicate?(file)
+    s = file.to_s
+    suffix = ['_bis.', '_ter.', '_quater.', '_quinquies.']
+    suffix.each do |pat|
+      if s.include?(pat)
+        return true
+      end
+    end
+    
+    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+            puts 'DROPPED ' + file.to_s + " because it's superceded by " + unique[key].to_s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)
@@ -55,7 +96,6 @@ class WalkDir
         result.push(fullName)
       end
     end
-    #puts result
     return result
   end
 end