Merge pull request gollum#166 from arr2036/mediawiki_compat

Gollum doesn't play nice with un-canonicalized filenames, and breaks offsite links when replacing mediawiki

lib/gollum/committer.rb

@@ -73,7 +73,7 @@ def parents
     #
     # dir    - The String subdirectory of the Gollum::Page without any
     #          prefix or suffix slashes (e.g. "foo/bar").
-    # name   - The String Gollum::Page name.
+    # name   - The String Gollum::Page filename_stripped.
     # format - The Symbol Gollum::Page format.
     # data   - The String wiki data to store in the tree map.
     # allow_same_ext - A Boolean determining if the tree map allows the same
@@ -111,7 +111,8 @@ def add_to_index(dir, name, format, data, allow_same_ext = false)
     # is a working directory present.
     #
     # dir    - The String directory in which the file lives.
-    # name   - The String name of the page (may be in human format).
+    # name   - The String name of the page or the stripped filename
+    #          (should be pre-canonicalized if required).
     # format - The Symbol format of the page.
     #
     # Returns nothing.

lib/gollum/page.rb

@@ -76,11 +76,20 @@ def self.format_for(filename)
     end
 
     # Reusable filter to turn a filename (without path) into a canonical name.
-    # Strips extension, converts spaces to dashes.
+    # Strips extension, converts dashes to spaces.
     #
     # Returns the filtered String.
     def self.canonicalize_filename(filename)
-      filename.split('.')[0..-2].join('.').gsub('-', ' ')
+      strip_filename(filename).gsub('-', ' ')
+    end
+
+    # Reusable filter to strip extension and path from filename
+    #
+    # filename - The string path or filename to strip
+    #
+    # Returns the stripped String. 
+    def self.strip_filename(filename)
+      ::File.basename(filename, ::File.extname(filename))
     end
 
     # Public: Initialize a page.
@@ -100,6 +109,13 @@ def filename
       @blob && @blob.name
     end
 
+    # Public: The on-disk filename of the page with extension stripped.
+    #
+    # Returns the String name.
+    def filename_stripped
+      self.class.strip_filename(filename)
+    end
+
     # Public: The canonical page name without extension, and dashes converted
     # to spaces.
     #
@@ -246,17 +262,22 @@ def historical?
 
     # Convert a human page name into a canonical page name.
     #
-    # name - The String human page name.
+    # name           - The String human page name.
+    # char_white_sub - Substitution for whitespace
+    # char_other_sub - Substitution for other special chars
     #
     # Examples
     #
     #   Page.cname("Bilbo Baggins")
     #   # => 'Bilbo-Baggins'
     #
+    #   Page.cname("Bilbo Baggins",'_')
+    #   # => 'Bilbo_Baggins'
+    #
     # Returns the String canonical name.
-    def self.cname(name)
-      name.respond_to?(:gsub)      ?
-        name.gsub(%r{[ /<>]}, '-') :
+    def self.cname(name, char_white_sub = '-', char_other_sub = '-')
+      name.respond_to?(:gsub) ?
+        name.gsub(%r{\s},char_white_sub).gsub(%r{[/<>+]}, char_other_sub) :
         ''
     end
 
@@ -370,10 +391,11 @@ def tree_path(treemap, tree)
     # Returns a Boolean.
     def page_match(name, filename)
       if match = self.class.valid_filename?(filename)
-        Page.cname(name).downcase == Page.cname(match).downcase
-      else
-        false
+        @wiki.ws_subs.each do |sub|
+          return true if Page.cname(name).downcase == Page.cname(match, sub).downcase
+        end
       end
+      false
     end
 
     # Loads a sub page.  Sub page nanes (footers) are prefixed with

lib/gollum/wiki.rb

@@ -20,7 +20,10 @@ class << self
 
       # Sets the default email for commits.
       attr_accessor :default_committer_email
-
+
+      # Array of chars to substitute whitespace for when trying to locate file in git repo.
+      attr_accessor :default_ws_subs
+
       # Sets sanitization options. Set to false to deactivate
       # sanitization altogether.
       attr_writer :sanitization
@@ -102,6 +105,8 @@ def history_sanitization
     self.default_ref = 'master'
     self.default_committer_name  = 'Anonymous'
     self.default_committer_email = 'anon@anon.com'
+
+    self.default_ws_subs = ['_','-']
 
     # The String base path to prefix to internal links. For example, when set
     # to "/wiki", the page "Hobbit" will be linked as "/wiki/Hobbit". Defaults
@@ -119,6 +124,9 @@ def history_sanitization
 
     # Gets the String directory in which all page files reside.
     attr_reader :page_file_dir
+
+    # Gets the Array of chars to sub for ws in filenames.
+    attr_reader :ws_subs
 
     # Public: Initialize a new Gollum Repo.
     #
@@ -134,6 +142,7 @@ def history_sanitization
     #           :sanitization  - An instance of Sanitization.
     #           :page_file_dir - String the directory in which all page files reside
     #           :ref - String the repository ref to retrieve pages from
+    #           :ws_subs       - Array of chars to sub for ws in filenames.
     #
     # Returns a fresh Gollum::Repo.
     def initialize(path, options = {})
@@ -151,6 +160,8 @@ def initialize(path, options = {})
       @repo          = @access.repo
       @ref           = options[:ref] || self.class.default_ref
       @sanitization  = options[:sanitization] || self.class.sanitization
+      @ws_subs       = options[:ws_subs] ||
+        self.class.default_ws_subs
       @history_sanitization = options[:history_sanitization] ||
         self.class.history_sanitization
     end
@@ -228,12 +239,14 @@ def write_page(name, format, data, commit = {})
       else
         Committer.new(self, commit)
       end
-
-      committer.add_to_index('', name, format, data)
+
+      filename = Gollum::Page.cname(name)
+
+      committer.add_to_index('', filename, format, data)
 
       committer.after_commit do |index, sha|
         @access.refresh
-        index.update_working_dir('', name, format)
+        index.update_working_dir('', filename, format)
       end
 
       multi_commit ? committer : committer.commit
@@ -265,7 +278,10 @@ def update_page(page, name, format, data, commit = {})
       name   ||= page.name
       format ||= page.format
       dir      = ::File.dirname(page.path)
-      dir      = '' if dir == '.'
+      dir      = '' if dir == '.'   
+      filename = (rename = page.name != name) ?
+        Gollum::Page.cname(name) : page.filename_stripped
+
       multi_commit = false
 
       committer = if obj = commit[:committer]
@@ -274,18 +290,18 @@ def update_page(page, name, format, data, commit = {})
       else
         Committer.new(self, commit)
       end
-
-      if page.name == name && page.format == format
+      
+      if !rename && page.format == format
         committer.add(page.path, normalize(data))
       else
         committer.delete(page.path)
-        committer.add_to_index(dir, name, format, data, :allow_same_ext)
+        committer.add_to_index(dir, filename, format, data, :allow_same_ext)
       end
-
+      
       committer.after_commit do |index, sha|
         @access.refresh
-        index.update_working_dir(dir, page.name, page.format)
-        index.update_working_dir(dir, name, format)
+        index.update_working_dir(dir, page.filename_stripped, page.format)
+        index.update_working_dir(dir, filename, format)
       end
 
       multi_commit ? committer : committer.commit
@@ -324,7 +340,7 @@ def delete_page(page, commit)
         dir = '' if dir == '.'
 
         @access.refresh
-        index.update_working_dir(dir, page.name, page.format)
+        index.update_working_dir(dir, page.filename_stripped, page.format)
       end
 
       multi_commit ? committer : committer.commit
@@ -362,7 +378,7 @@ def revert_page(page, sha1, sha2 = nil, commit = {})
 
         files = []
         if page
-          files << [page.path, page.name, page.format]
+          files << [page.path, page.filename_stripped, page.format]
         else
           # Grit::Diff can't parse reverse diffs.... yet
           patch.each_line do |line|
@@ -522,13 +538,12 @@ def normalize(data)
 
     # Assemble a Page's filename from its name and format.
     #
-    # name   - The String name of the page (may be in human format).
+    # name   - The String name of the page (should be pre-canonicalized).
     # format - The Symbol format of the page.
     #
     # Returns the String filename.
     def page_file_name(name, format)
-      ext = @page_class.format_to_ext(format)
-      @page_class.cname(name) + '.' + ext
+      name + '.' + @page_class.format_to_ext(format)
     end
 
     # Fill an array with a list of pages.

test/examples/lotr.git/COMMIT_EDITMSG

@@ -1 +1 @@
-add sidebars
+Test out whitespace with Sam

test/examples/lotr.git/ORIG_HEAD

@@ -0,0 +1 @@
+a8ad3c09dd842a3517085bfadd37718856dee813

test/examples/lotr.git/logs/HEAD

@@ -1,2 +1,3 @@
 0000000000000000000000000000000000000000 60f12f4254f58801b9ee7db7bca5fa8aeefaa56b rick <technoweenie@gmail.com> 1291341857 -0800	clone: from /Users/rick/p/gollum/test/examples/lotr.git
 60f12f4254f58801b9ee7db7bca5fa8aeefaa56b a8ad3c09dd842a3517085bfadd37718856dee813 rick <technoweenie@gmail.com> 1291341922 -0800	commit: add sidebars
+a8ad3c09dd842a3517085bfadd37718856dee813 1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3 Arran Cudbard-Bell <a.cudbardb@freeradius.org> 1309107565 +0200	commit: Test out whitespace with Sam

test/examples/lotr.git/logs/refs/heads/master

@@ -1,2 +1,3 @@
 0000000000000000000000000000000000000000 60f12f4254f58801b9ee7db7bca5fa8aeefaa56b rick <technoweenie@gmail.com> 1291341857 -0800	clone: from /Users/rick/p/gollum/test/examples/lotr.git
 60f12f4254f58801b9ee7db7bca5fa8aeefaa56b a8ad3c09dd842a3517085bfadd37718856dee813 rick <technoweenie@gmail.com> 1291341922 -0800	commit: add sidebars
+a8ad3c09dd842a3517085bfadd37718856dee813 1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3 Arran Cudbard-Bell <a.cudbardb@freeradius.org> 1309107565 +0200	commit: Test out whitespace with Sam

test/examples/lotr.git/objects/1d/b89ebba7e2c14d93b94ff98cfa3708a4f0d4e3

@@ -0,0 +1,2 @@
+x
��KN�0�a�9���Q�G�H#��p
'vh��t�����,�ͯ�n�pS~]Z���`��gK���q�X���]��X|�Y��c1Y����O	��$����1��k�|��C
+wyz���|�]^ڏ��,�~���3��m�S�G�5��;��'����p�������}3|�j�
K�Wo

test/examples/lotr.git/objects/a6/59b3763b822dd97544621fd0beef162ea37b14

@@ -0,0 +1,4 @@
+x
�RMK1���b�
+e���TPO��%�4;�6�Y�h-��;[mK�!���L�G�t�����)�c��H��]M�,�ǧ�p�(@�e�
#�4��SP�>�
+<v4ոG���'_�7<A[4o*�$����QG��
�
+�t�x�9��p��\�M,�ZUe���B\����J��?��k��:[6&?D��i/�6�n$q�8^�勜�Q�s��2Ӊuö
mY�s�`Љ�6�6=m�;���EG��!���\Q˶��_Ve�IX�瘌�h�s�J��}T3Y�He�RW��5��SŘK���T�>���O����ʳSh�A

test/examples/lotr.git/refs/heads/master

@@ -1 +1 @@
-a8ad3c09dd842a3517085bfadd37718856dee813
+1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3

test/test_committer.rb

@@ -50,7 +50,7 @@
   end
 
   test "parents with default master ref" do
-    ref = 'a8ad3c09dd842a3517085bfadd37718856dee813'
+    ref = '1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3'
     committer = Gollum::Committer.new(@wiki)
     assert_equal ref,  committer.parents.first.sha
   end

test/test_git_access.rb

@@ -18,9 +18,9 @@
     assert @access.ref_map.empty?
     assert @access.tree_map.empty?
     @access.tree 'master'
-    assert_equal({"master"=>"a8ad3c09dd842a3517085bfadd37718856dee813"}, @access.ref_map)
+    assert_equal({"master"=>"1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3"}, @access.ref_map)
 
-    map = @access.tree_map['a8ad3c09dd842a3517085bfadd37718856dee813']
+    map = @access.tree_map['1db89ebba7e2c14d93b94ff98cfa3708a4f0d4e3']
     assert_equal 'Bilbo-Baggins.md',        map[0].path
     assert_equal '',                        map[0].dir
     assert_equal map[0].path,               map[0].name

test/test_page.rb

@@ -25,7 +25,27 @@
   test "get existing page case insensitive" do
     assert_equal @wiki.page('Bilbo Baggins').path, @wiki.page('bilbo baggins').path
   end
+
+  test "get existing page with hyphen" do
+    assert_equal @wiki.page('Bilbo Baggins').path, @wiki.page('Bilbo-Baggins').path
+  end
 
+  test "get existing page with underscore" do
+    assert_nil @wiki.page('Bilbo_Baggins')
+  end
+
+  test "get existing page where filename contains whitespace, with hypen" do
+    assert_equal @wiki.page('Samwise Gamgee').path, @wiki.page('Samwise-Gamgee').path
+  end
+
+  test "get existing page where filename contains whitespace, with underscore" do
+    assert_equal @wiki.page('Samwise Gamgee').path, @wiki.page('Samwise_Gamgee').path
+  end
+
+  test "get existing page where filename contains whitespace, with whitespace" do
+    assert_equal @wiki.page('Samwise Gamgee').path, @wiki.page('Samwise Gamgee').path
+  end
+
   test "get nested page" do
     page = @wiki.page('Eye Of Sauron')
     assert_equal 'Mordor/Eye-Of-Sauron.md', page.path