in_tail: Expand glob capability for square brackets and one character…

… matcher (fluent#4401)

Signed-off-by: Hiroshi Hatake <[email protected]>

lib/fluent/plugin/in_tail.rb

@@ -65,6 +65,8 @@ def initialize
     config_param :path, :string
     desc 'path delimiter used for spliting path config'
     config_param :path_delimiter, :string, default: ','
+    desc 'Choose using glob patterns. Adding capabilities to handle [] and ?, and {}.'
+    config_param :glob_policy, :enum, list: [:backward_compatible, :extended, :always], default: :backward_compatible
     desc 'The tag of the event.'
     config_param :tag, :string
     desc 'The paths to exclude the files from watcher list.'
@@ -141,6 +143,14 @@ def configure(conf)
         raise Fluent::ConfigError, "either of enable_watch_timer or enable_stat_watcher must be true"
       end
 
+      if @glob_policy == :always && @path_delimiter == ','
+        raise Fluent::ConfigError, "cannot use glob_policy as always with the default path_delimitor: `,\""
+      end
+
+      if @glob_policy == :extended && /\{.*,.*\}/.match(@path) && extended_glob_pattern(@path)
+        raise Fluent::ConfigError, "cannot include curly braces with glob patterns in `#{@path}\". Use glob_policy always instead."
+      end
+
       if RESERVED_CHARS.include?(@path_delimiter)
         rc = RESERVED_CHARS.join(', ')
         raise Fluent::ConfigError, "#{rc} are reserved words: #{@path_delimiter}"
@@ -288,6 +298,28 @@ def have_read_capability?
         @capability.have_capability?(:effective, :dac_override)
     end
 
+    def extended_glob_pattern(path)
+      path.include?('*') || path.include?('?') || /\[.*\]/.match(path)
+    end
+
+    # Curly braces is not supported with default path_delimiter
+    # because the default delimiter of path is ",".
+    # This should be collided for wildcard pattern for curly braces and
+    # be handled as an error on #configure.
+    def use_glob?(path)
+      if @glob_policy == :always
+        # For future extensions, we decided to use `always' term to handle
+        # regular expressions as much as possible.
+        # This is because not using `true' as a returning value
+        # when choosing :always here.
+        extended_glob_pattern(path) || /\{.*,.*\}/.match(path)
+      elsif @glob_policy == :extended
+        extended_glob_pattern(path)
+      elsif @glob_policy == :backward_compatible
+        path.include?('*')
+      end
+    end
+
     def expand_paths
       date = Fluent::EventTime.now
       paths = []
@@ -297,7 +329,7 @@ def expand_paths
                else
                  date.to_time.strftime(path)
                end
-        if path.include?('*')
+        if use_glob?(path)
           paths += Dir.glob(path).select { |p|
             begin
               is_file = !File.directory?(p)
@@ -332,7 +364,7 @@ def expand_paths
                else
                  date.to_time.strftime(path)
                end
-        path.include?('*') ? Dir.glob(path) : path
+        use_glob?(path) ? Dir.glob(path) : path
       }.flatten.uniq
       # filter out non existing files, so in case pattern is without '*' we don't do unnecessary work
       hash = {}

test/plugin/test_in_tail.rb

@@ -1538,6 +1538,147 @@ def test_expand_paths
       assert_equal(ex_paths - [ex_paths.last], plugin.expand_paths.values.sort_by { |path_ino| path_ino.path })
     end
 
+    sub_test_case "expand_paths with glob" do |data|
+      sub_test_case "extended_glob" do
+        data("curly braces"          => [true, "always", "test/plugin/data/log_numeric/{0,1}*.log"],
+             "square brackets"       => [true, "always", "test/plugin/data/log_numeric/[0-1][2-4].log"],
+             "asterisk"              => [true, "always", "test/plugin/data/log/*.log"],
+             "one character matcher" => [true, "always", "test/plugin/data/log/tes?.log"],
+            )
+        def test_expand_paths_with_use_glob_p_and_almost_set_of_patterns
+          result, option, path = data
+          config = config_element("", "", {
+                                    "tag" => "tail",
+                                    "path" => path,
+                                    "format" => "none",
+                                    "pos_file" => "#{@tmp_dir}/tail.pos",
+                                    "read_from_head" => true,
+                                    "refresh_interval" => 30,
+                                    "glob_policy" => option,
+                                    "path_delimiter" => "|",
+                                    "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                                    "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                                  })
+          plugin = create_driver(config, false).instance
+          assert_equal(result, !!plugin.use_glob?(path))
+        end
+
+        data("curly braces"          => [true, false, "extended", "test/plugin/data/log_numeric/{0,1}*.log"],
+             "square brackets"       => [false, true, "extended", "test/plugin/data/log_numeric/[0-1][2-4].log"],
+             "asterisk"              => [false, true, "extended", "test/plugin/data/log/*.log"],
+             "one character matcher" => [false, true, "extended", "test/plugin/data/log/tes?.log"],
+            )
+        def test_expand_paths_with_use_glob_p
+          emit_exception_p, result, option, path = data
+          config = config_element("", "", {
+                                    "tag" => "tail",
+                                    "path" => path,
+                                    "format" => "none",
+                                    "pos_file" => "#{@tmp_dir}/tail.pos",
+                                    "read_from_head" => true,
+                                    "refresh_interval" => 30,
+                                    "glob_policy" => option,
+                                    "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                                    "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                                  })
+          if emit_exception_p
+            assert_raise(Fluent::ConfigError) do
+              plugin = create_driver(config, false).instance
+            end
+          else
+            plugin = create_driver(config, false).instance
+            assert_equal(result, !!plugin.use_glob?(path))
+          end
+        end
+      end
+
+      sub_test_case "only_use_backward_compatible" do
+        data("square brackets"       => [false, "backward_compatible", "test/plugin/data/log_numeric/[0-1][2-4].log"],
+             "asterisk"              => [true,  "backward_compatible", "test/plugin/data/log/*.log"],
+             "one character matcher" => [false, "backward_compatible", "test/plugin/data/log/tes?.log"],
+            )
+        def test_expand_paths_with_use_glob_p
+          result, option, path = data
+          config = config_element("", "", {
+                                    "tag" => "tail",
+                                    "path" => path,
+                                    "format" => "none",
+                                    "pos_file" => "#{@tmp_dir}/tail.pos",
+                                    "read_from_head" => true,
+                                    "refresh_interval" => 30,
+                                    "glob_policy" => option,
+                                    "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                                    "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                                  })
+          plugin = create_driver(config, false).instance
+          assert_equal(result, !!plugin.use_glob?(path))
+        end
+      end
+    end
+
+    def ex_config_with_brackets
+      config_element("", "", {
+                     "tag" => "tail",
+                     "path" => "test/plugin/data/log_numeric/[0-1][2-4].log",
+                     "format" => "none",
+                     "pos_file" => "#{@tmp_dir}/tail.pos",
+                     "read_from_head" => true,
+                     "refresh_interval" => 30,
+                     "glob_policy" => "extended",
+                     "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                     "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                   })
+    end
+
+    def test_config_with_always_with_default_delimiter
+      assert_raise(Fluent::ConfigError) do
+        config = config_element("", "", {
+                         "tag" => "tail",
+                         "path" => "test/plugin/data/log_numeric/[0-1][2-4].log",
+                         "format" => "none",
+                         "pos_file" => "#{@tmp_dir}/tail.pos",
+                         "read_from_head" => true,
+                         "refresh_interval" => 30,
+                         "glob_policy" => "always",
+                         "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                         "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                       })
+
+        create_driver(config, false).instance
+      end
+    end
+
+    def test_config_with_always_with_custom_delimiter
+      assert_nothing_raised do
+        config = config_element("", "", {
+                         "tag" => "tail",
+                         "path" => "test/plugin/data/log_numeric/[0-1][2-4].log",
+                         "format" => "none",
+                         "pos_file" => "#{@tmp_dir}/tail.pos",
+                         "read_from_head" => true,
+                         "refresh_interval" => 30,
+                         "glob_policy" => "always",
+                         "path_delimiter" => "|",
+                         "rotate_wait" => "#{EX_ROTATE_WAIT}s",
+                         "follow_inodes" => "#{EX_FOLLOW_INODES}",
+                       })
+
+        create_driver(config, false).instance
+      end
+    end
+
+    def test_expand_paths_with_brackets
+      expanded_paths = [
+        create_target_info('test/plugin/data/log_numeric/01.log'),
+        create_target_info('test/plugin/data/log_numeric/02.log'),
+        create_target_info('test/plugin/data/log_numeric/12.log'),
+        create_target_info('test/plugin/data/log_numeric/14.log'),
+      ]
+
+      plugin = create_driver(ex_config_with_brackets, false).instance
+      assert_equal(expanded_paths - [expanded_paths.first], plugin.expand_paths.values.sort_by { |path_ino| path_ino.path })
+    end
+
     def test_expand_paths_with_duplicate_configuration
       expanded_paths = [
         create_target_info('test/plugin/data/log/foo/bar.log'),