X-Git-Url: http://jaekl.net/gitweb/?p=quanlib.git;a=blobdiff_plain;f=walkdir.rb;h=b00645545fa984b5fbef17b09b45513957ab6f69;hp=fb23fcf56d44d1fd8bbbad76153aabd2debb6898;hb=HEAD;hpb=b8fe580f73b094e366643388ecd2d184b643616a

diff --git a/walkdir.rb b/walkdir.rb
deleted file mode 100644
index fb23fcf..0000000
--- a/walkdir.rb
+++ /dev/null
@@ -1,102 +0,0 @@
-# Walk the directory (and subdirectories), identifying books.
-#
-# Expected format:
-#   .../AuthorName/Title_of_the_Awesome_Book.ext
-#
-# Author is given as FirstLast.  For example, 
-# Robert Anson Heinlein is RoberHeinlein, and 
-# JKRowling is JoanneRowling.
-#
-# Book titles have spaces replaced with underscores,
-# and punctuation [,!?'] replaced with hyphens.
-#
-# If the book forms part of a series, then an all-capitals 
-# series designator, followed by a numeric volume number, 
-# followed by an underscore, is prefixed to the name.
-# For example, Hardy Boys' volume 1, The Tower Treasure, 
-# is rendered as .../FranklinDixon/HB001_The_Tower_Treasure.epub
-# and Mrs. Pollifax volume 6, On the China Station, is
-# .../DorothyGilman/P06_On_the_China_Station.epub.
-
-require 'book'
-require 'store'
-
-class WalkDir
-  def initialize(store, root)
-    @root = root
-    @store = store
-    @files = walk(@root)
-  end
-
-  def books
-    result = []
-    @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)
-        id = @store.store_book(book)
-        result.push(id)
-      end
-    end
-    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)
-    for child in children
-      fullName = (path.chomp("/")) + "/" + child
-      if (File.directory?(fullName)) and (child != ".") and (child != "..") and (!File.symlink?(fullName))
-        sub = walk(fullName)
-        if (sub != nil) and (sub.length > 0)
-          result.concat(sub)
-        end
-      elsif (! File.directory?(fullName))
-        result.push(fullName)
-      end
-    end
-    #puts result
-    return result
-  end
-end