From 2be76355caa579d444c8e3c0d25563eb9778bfb2 Mon Sep 17 00:00:00 2001
From: Grzegorz Bizon <grzesiek.bizon@gmail.com>
Date: Tue, 12 Jan 2016 10:04:26 +0100
Subject: Support only valid UTF-8 paths in build artifacts browser

---
 lib/gitlab/ci/build/artifacts/metadata.rb      |  8 +++++---
 lib/gitlab/ci/build/artifacts/metadata/path.rb | 10 ++++++++--
 2 files changed, 13 insertions(+), 5 deletions(-)

(limited to 'lib')

diff --git a/lib/gitlab/ci/build/artifacts/metadata.rb b/lib/gitlab/ci/build/artifacts/metadata.rb
index 2b17712cdbe..d9c051be9f3 100644
--- a/lib/gitlab/ci/build/artifacts/metadata.rb
+++ b/lib/gitlab/ci/build/artifacts/metadata.rb
@@ -10,7 +10,8 @@ module Gitlab
           attr_reader :file, :path, :full_version
 
           def initialize(file, path)
-            @file, @path = file, path
+            @file = file
+            @path = path.force_encoding('ASCII-8BIT')
             @full_version = read_version
           end
 
@@ -42,7 +43,7 @@ module Gitlab
 
           def match_entries(gz)
             paths, metadata = [], []
-            match_pattern = %r{^#{Regexp.escape(@path)}[^/\s]*/?$}
+            match_pattern = %r{^#{Regexp.escape(@path)}[^/]*/?$}
             invalid_pattern = %r{(^\.?\.?/)|(/\.?\.?/)}
 
             until gz.eof? do
@@ -51,11 +52,12 @@ module Gitlab
                 meta = read_string(gz)
                
                 next unless path =~ match_pattern
+                next unless path.force_encoding('UTF-8').valid_encoding?
                 next if path =~ invalid_pattern
 
                 paths.push(path)
                 metadata.push(JSON.parse(meta.chomp, symbolize_names: true))
-              rescue JSON::ParserError
+              rescue JSON::ParserError, Encoding::CompatibilityError
                 next
               end
             end
diff --git a/lib/gitlab/ci/build/artifacts/metadata/path.rb b/lib/gitlab/ci/build/artifacts/metadata/path.rb
index 80ead335d57..6896aa936d5 100644
--- a/lib/gitlab/ci/build/artifacts/metadata/path.rb
+++ b/lib/gitlab/ci/build/artifacts/metadata/path.rb
@@ -8,18 +8,24 @@ module Gitlab
       # This is IO-operations safe class, that does similar job to
       # Ruby's Pathname but without the risk of accessing filesystem.
       #
+      # This class is working only with UTF-8 encoded paths.
+      #
       class Path
         attr_reader :path, :universe
         attr_accessor :name
 
         def initialize(path, universe, metadata = [])
-          @path = path
+          @path = path.force_encoding('UTF-8')
           @universe = universe
           @metadata = metadata
 
           if path.include?("\0")
             raise ArgumentError, 'Path contains zero byte character!'
           end
+
+          unless path.valid_encoding?
+            raise ArgumentError, 'Path contains non-UTF-8 byte sequence!'
+          end
         end
 
         def directory?
@@ -51,7 +57,7 @@ module Gitlab
           return [] unless directory?
           return @children if @children
 
-          child_pattern = %r{^#{Regexp.escape(@path)}[^/\s]+/?$}
+          child_pattern = %r{^#{Regexp.escape(@path)}[^/]+/?$}
           @children = select { |entry| entry =~ child_pattern }
         end
 
-- 
cgit v1.2.1