Extract handling of relative file links to RelativeLinkFilter

4 files changed, 126 insertions, 145 deletions

diff --git a/app/helpers/gitlab_markdown_helper.rb b/app/helpers/gitlab_markdown_helper.rb
index 24263a0f619..c309f890d96 100644
--- a/app/helpers/gitlab_markdown_helper.rb
+++ b/app/helpers/gitlab_markdown_helper.rb
@@ -72,146 +72,6 @@ module GitlabMarkdownHelper
     end
   end
 
-  # TODO (rspeicher): This should be its own filter
-  def create_relative_links(text)
-    paths = extract_paths(text)
-
-    paths.uniq.each do |file_path|
-      # If project does not have repository
-      # its nothing to rebuild
-      #
-      # TODO: pass project variable to markdown helper instead of using
-      # instance variable. Right now it generates invalid path for pages out
-      # of project scope. Example: search results where can be rendered markdown
-      # from different projects
-      if @repository && @repository.exists? && !@repository.empty?
-        new_path = rebuild_path(file_path)
-        # Finds quoted path so we don't replace other mentions of the string
-        # eg. "doc/api" will be replaced and "/home/doc/api/text" won't
-        text.gsub!("\"#{file_path}\"", "\"/#{new_path}\"")
-      end
-    end
-
-    text
-  end
-
-  def extract_paths(text)
-    links = substitute_links(text)
-    image_links = substitute_image_links(text)
-    links + image_links
-  end
-
-  def substitute_links(text)
-    links = text.scan(/<a href=\"([^"]*)\">/)
-    relative_links = links.flatten.reject{ |link| link_to_ignore? link }
-    relative_links
-  end
-
-  def substitute_image_links(text)
-    links = text.scan(/<img src=\"([^"]*)\"/)
-    relative_links = links.flatten.reject{ |link| link_to_ignore? link }
-    relative_links
-  end
-
-  def link_to_ignore?(link)
-    if link =~ /\A\#\w+/
-      # ignore anchors like <a href="#my-header">
-      true
-    else
-      ignored_protocols.map{ |protocol| link.include?(protocol) }.any?
-    end
-  end
-
-  def ignored_protocols
-    ["http://","https://", "ftp://", "mailto:", "smb://"]
-  end
-
-  def rebuild_path(file_path)
-    file_path = file_path.dup
-    file_path.gsub!(/(#.*)/, "")
-    id = $1 || ""
-    file_path = relative_file_path(file_path)
-    file_path = sanitize_slashes(file_path)
-
-    [
-      Gitlab.config.gitlab.relative_url_root,
-      @project.path_with_namespace,
-      path_with_ref(file_path),
-      file_path
-    ].compact.join("/").gsub(/\A\/*|\/*\z/, '') + id
-  end
-
-  def sanitize_slashes(path)
-    path[0] = "" if path.start_with?("/")
-    path.chop if path.end_with?("/")
-    path
-  end
-
-  def relative_file_path(path)
-    requested_path = @path
-    nested_path = build_nested_path(path, requested_path)
-    return nested_path if file_exists?(nested_path)
-    path
-  end
-
-  # Covering a special case, when the link is referencing file in the same directory eg:
-  # If we are at doc/api/README.md and the README.md contains relative links like [Users](users.md)
-  # this takes the request path(doc/api/README.md), and replaces the README.md with users.md so the path looks like doc/api/users.md
-  # If we are at doc/api and the README.md shown in below the tree view
-  # this takes the request path(doc/api) and adds users.md so the path looks like doc/api/users.md
-  def build_nested_path(path, request_path)
-    return request_path if path == ""
-    return path unless request_path
-    if local_path(request_path) == "tree"
-      base = request_path.split("/").push(path)
-      base.join("/")
-    else
-      base = request_path.split("/")
-      base.pop
-      base.push(path).join("/")
-    end
-  end
-
-  # Checks if the path exists in the repo
-  # eg. checks if doc/README.md exists, if not then link to blob
-  def path_with_ref(path)
-    if file_exists?(path)
-      "#{local_path(path)}/#{correct_ref}"
-    else
-      "blob/#{correct_ref}"
-    end
-  end
-
-  def file_exists?(path)
-    return false if path.nil?
-    @repository.blob_at(current_sha, path).present? || @repository.tree(current_sha, path).entries.any?
-  end
-
-  # Check if the path is pointing to a directory(tree) or a file(blob)
-  # eg. doc/api is directory and doc/README.md is file
-  def local_path(path)
-    return "tree" if @repository.tree(current_sha, path).entries.any?
-    return "raw" if @repository.blob_at(current_sha, path).image?
-    "blob"
-  end
-
-  def current_sha
-    if @commit
-      @commit.id
-    elsif @repository && !@repository.empty?
-      if @ref
-        @repository.commit(@ref).try(:sha)
-      else
-        @repository.head_commit.sha
-      end
-    end
-  end
-
-  # We will assume that if no ref exists we can point to master
-  def correct_ref
-    @ref ? @ref : "master"
-  end
-
   private
 
   # Return +text+, truncated to +max_chars+ characters, excluding any HTML
diff --git a/lib/gitlab/markdown.rb b/lib/gitlab/markdown.rb
index 63294aa54c0..dadd0e0d64b 100644
--- a/lib/gitlab/markdown.rb
+++ b/lib/gitlab/markdown.rb
@@ -15,6 +15,7 @@ module Gitlab
     autoload :IssueReferenceFilter,         'gitlab/markdown/issue_reference_filter'
     autoload :LabelReferenceFilter,         'gitlab/markdown/label_reference_filter'
     autoload :MergeRequestReferenceFilter,  'gitlab/markdown/merge_request_reference_filter'
+    autoload :RelativeLinkFilter,           'gitlab/markdown/relative_link_filter'
     autoload :SanitizationFilter,           'gitlab/markdown/sanitization_filter'
     autoload :SnippetReferenceFilter,       'gitlab/markdown/snippet_reference_filter'
     autoload :TableOfContentsFilter,        'gitlab/markdown/table_of_contents_filter'
@@ -64,7 +65,12 @@ module Gitlab
         current_user:    current_user,
         only_path:       options[:reference_only_path],
         project:         project,
-        reference_class: html_options[:class]
+        reference_class: html_options[:class],
+
+        # RelativeLinkFilter
+        ref: @ref,
+        requested_path: @path,
+        project_wiki: @project_wiki
       }
 
       result = pipeline.call(text, context)
@@ -91,6 +97,8 @@ module Gitlab
       [
         Gitlab::Markdown::SanitizationFilter,
 
+        Gitlab::Markdown::RelativeLinkFilter,
+
         Gitlab::Markdown::EmojiFilter,
         Gitlab::Markdown::TableOfContentsFilter,
         Gitlab::Markdown::AutolinkFilter,
diff --git a/lib/gitlab/markdown/relative_link_filter.rb b/lib/gitlab/markdown/relative_link_filter.rb
new file mode 100644
index 00000000000..084f84b25f7
--- /dev/null
+++ b/lib/gitlab/markdown/relative_link_filter.rb
@@ -0,0 +1,117 @@
+require 'uri'
+
+module Gitlab
+  module Markdown
+    # HTML filter that "fixes" relative links to files in a repository.
+    #
+    # Context options:
+    #   :commit
+    #   :project
+    #   :project_wiki
+    #   :requested_path
+    #   :ref
+    class RelativeLinkFilter < HTML::Pipeline::Filter
+
+      def call
+        if !project_wiki && repository.try(:exists?) && !repository.empty?
+          doc.search('a').each do |el|
+            process_link_attr el.attribute('href')
+          end
+
+          doc.search('img').each do |el|
+            process_link_attr el.attribute('src')
+          end
+        end
+
+        doc
+      end
+
+      protected
+
+      def process_link_attr(html_attr)
+        return if html_attr.blank?
+
+        uri = URI(html_attr.value)
+        if uri.relative? && uri.path.present?
+          html_attr.value = rebuild_relative_uri(uri).to_s
+        end
+      end
+
+      def rebuild_relative_uri(uri)
+        file_path = relative_file_path(uri.path)
+
+        uri.path = [
+          relative_url_root,
+          project.path_with_namespace,
+          path_type(file_path),
+          ref || 'master',  # assume that if no ref exists we can point to master
+          file_path
+        ].compact.join('/').squeeze('/').chomp('/')
+
+        uri
+      end
+
+      def relative_file_path(path)
+        nested_path = build_nested_path(path, requested_path)
+        file_exists?(nested_path) ? nested_path : path
+      end
+
+      # Covering a special case, when the link is referencing file in the same
+      # directory.
+      # If we are at doc/api/README.md and the README.md contains relative
+      # links like [Users](users.md), this takes the request
+      # path(doc/api/README.md) and replaces the README.md with users.md so the
+      # path looks like doc/api/users.md.
+      # If we are at doc/api and the README.md shown in below the tree view
+      # this takes the request path(doc/api) and adds users.md so the path
+      # looks like doc/api/users.md
+      def build_nested_path(path, request_path)
+        return request_path if path.empty?
+        return path unless request_path
+
+        parts = request_path.split('/')
+        parts.pop if path_type(request_path) != 'tree'
+        parts.push(path).join('/')
+      end
+
+      def file_exists?(path)
+        return false if path.nil?
+        repository.blob_at(current_sha, path).present? ||
+          repository.tree(current_sha, path).entries.any?
+      end
+
+      # Check if the path is pointing to a directory(tree) or a file(blob)
+      # eg. doc/api is directory and doc/README.md is file.
+      def path_type(path)
+        return 'tree' if repository.tree(current_sha, path).entries.any?
+        return 'raw' if repository.blob_at(current_sha, path).try(:image?)
+        'blob'
+      end
+
+      def current_sha
+        if commit
+          commit.id
+        elsif ref
+          repository.commit(ref).try(:sha)
+        else
+          repository.head_commit.sha
+        end
+      end
+
+      def relative_url_root
+        Gitlab.config.gitlab.relative_url_root.presence || '/'
+      end
+
+      [:commit, :project, :project_wiki, :requested_path, :ref].each do |name|
+        define_method(name) do
+          context[name]
+        end
+      end
+
+      def repository
+        return if project.nil?
+        project.repository
+      end
+    end
+  end
+end
diff --git a/lib/redcarpet/render/gitlab_html.rb b/lib/redcarpet/render/gitlab_html.rb
index bea66e6cdc1..15eb6effe08 100644
--- a/lib/redcarpet/render/gitlab_html.rb
+++ b/lib/redcarpet/render/gitlab_html.rb
@@ -36,10 +36,6 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
   end
 
   def postprocess(full_document)
-    unless @template.instance_variable_get("@project_wiki") || @project.nil?
-      full_document = h.create_relative_links(full_document)
-    end
-
     h.gfm_with_options(full_document, @options)
   end
 end