Merge pull request #1604 from middleman/frontmatter

Refactor frontmatter parsing

middleman-core/lib/middleman-core/util/data.rb

@@ -1,153 +1,82 @@
-# Core Pathname library used for traversal
+require 'yaml'
+require 'json'
 require 'pathname'
-
+require 'middleman-core/util'
-# DbC
 require 'middleman-core/contracts'
 
-# Shared util methods
+module Middleman::Util::Data
-require 'middleman-core/util'
+  include Contracts
 
-# Parsing YAML data
+  module_function
-require 'yaml'
 
-# Parsing JSON data
+  # Get the frontmatter and plain content from a file
-require 'json'
+  # @param [String] path
+  # @return [Array<Hash, String>]
+  Contract Pathname, Maybe[Symbol] => [Hash, Maybe[String]]
+  def parse(full_path, known_type=nil)
+    return [{}, nil] if Middleman::Util.binary?(full_path)
 
-module Middleman
+    # Avoid weird race condition when a file is renamed
-  module Util
+    begin
-    module Data
+      content = File.read(full_path)
-      include Contracts
+    rescue EOFError, IOError, Errno::ENOENT
+      return [{}, nil]
+    end
 
-      module_function
+    case known_type
+    when :yaml
+      return [parse_yaml(content, full_path), nil]
+    when :json
+      return [parse_json(content, full_path), nil]
+    end
 
-      YAML_ERRORS = [StandardError]
+    /
+      (?<start>^[-;]{3})[ ]*\r?\n
+      (?<frontmatter>.*?)[ ]*\r?\n
+      (?<stop>^[-.;]{3})[ ]*\r?\n?
+      (?<additional_content>.*)
+    /mx =~ content
 
-      # https://github.com/tenderlove/psych/issues/23
+    case [start, stop]
-      if defined?(Psych) && defined?(Psych::SyntaxError)
+    when %w[--- ---], %w[--- ...]
-        YAML_ERRORS << Psych::SyntaxError
+      [parse_yaml(frontmatter, full_path), additional_content]
-      end
+    when %w[;;; ;;;]
+      [parse_json(frontmatter, full_path), additional_content]
+    else
+      [{}, content]
+    end
+  end
 
-      # Get the frontmatter and plain content from a file
+  # Parse YAML frontmatter out of a string
-      # @param [String] path
+  # @param [String] content
-      # @return [Array<Hash, String>]
+  # @return [Array<Hash, String>]
-      Contract Pathname, Maybe[Symbol] => [Hash, Maybe[String]]
+  Contract String, Pathname, Bool => Hash
-      def parse(full_path, known_type=nil)
+  def parse_yaml(content, full_path)
-        data = {}
+    symbolize_recursive(YAML.load(content))
+  rescue StandardError, Psych::SyntaxError => error
+    warn "YAML Exception parsing #{full_path}: #{error.message}"
+    {}
+  end
 
-        return [data, nil] if ::Middleman::Util.binary?(full_path)
+  # Parse JSON frontmatter out of a string
+  # @param [String] content
+  # @return [Array<Hash, String>]
+  Contract String, Pathname => Hash
+  def parse_json(content, full_path)
+    symbolize_recursive(JSON.parse(content))
+  rescue StandardError => error
+    warn "JSON Exception parsing #{full_path}: #{error.message}"
+    {}
+  end
 
-        # Avoid weird race condition when a file is renamed.
+  def symbolize_recursive(value)
-        content = begin
+    case value
-          File.read(full_path)
+    when Hash
-        rescue ::EOFError
+      value.map { |k, v| [k.to_sym, symbolize_recursive(v)] }.to_h
-        rescue ::IOError
+    when Array
-        rescue ::Errno::ENOENT
+      value.map { |v| symbolize_recursive(v) }
-          ''
+    else
-        end
+      value
-
-        begin
-          if content =~ /\A.*coding:/
-            lines = content.split(/\n/)
-            lines.shift
-            content = lines.join("\n")
-          end
-
-          if known_type
-            if known_type == :yaml
-              result = parse_yaml(content, full_path, true)
-            elsif known_type == :json
-              result = parse_json(content, full_path)
-            end
-          else
-            result = parse_yaml(content, full_path, false)
-          end
-
-          return result if result
-        rescue
-          # Probably a binary file, move on
-        end
-
-        [data, content]
-      end
-
-      # Parse YAML frontmatter out of a string
-      # @param [String] content
-      # @return [Array<Hash, String>]
-      Contract String, Pathname, Bool => Maybe[[Hash, String]]
-      def parse_yaml(content, full_path, require_yaml=false)
-        total_delims = content.scan(/^(?:---|\.\.\.)\s*(?:\n|$)/).length
-        has_first_line_delim = !content.match(/\A(---\s*(?:\n|$))/).nil?
-        # has_closing_delim = (total_delims > 1 && has_first_line_delim) || (!has_first_line_delim && total_delims == 1)
-
-        parts = content.split(/^(?:---|\.\.\.)\s*(?:\n|$)/)
-        parts.shift if parts[0].empty?
-
-        yaml_string = nil
-        additional_content = nil
-
-        if require_yaml
-          yaml_string = parts[0]
-          additional_content = parts[1]
-        else
-          if total_delims > 1
-            if has_first_line_delim
-              yaml_string = parts[0]
-              additional_content = parts[1]
-            else
-              additional_content = content
-            end
-          else
-            additional_content = parts[0]
-          end
-        end
-
-        return [{}, additional_content] if yaml_string.nil?
-
-        begin
-          data = map_value(::YAML.load(yaml_string) || {})
-        rescue *YAML_ERRORS => e
-          $stderr.puts "YAML Exception parsing #{full_path}: #{e.message}"
-          return nil
-        end
-
-        [data, additional_content]
-      rescue
-        [{}, additional_content]
-      end
-
-      # Parse JSON frontmatter out of a string
-      # @param [String] content
-      # @return [Array<Hash, String>]
-      Contract String, Pathname => Maybe[[Hash, String]]
-      def parse_json(content, full_path)
-        begin
-          data = map_value(::JSON.parse(content))
-        rescue => e
-          $stderr.puts "JSON Exception parsing #{full_path}: #{e.message}"
-          return nil
-        end
-
-        [data, nil]
-      rescue
-        [{}, nil]
-      end
-
-      def symbolize_recursive(hash)
-        {}.tap do |h|
-          hash.each { |key, value| h[key.to_sym] = map_value(value) }
-        end
-      end
-
-      def map_value(thing)
-        case thing
-        when Hash
-          symbolize_recursive(thing)
-        when Array
-          thing.map { |v| map_value(v) }
-        else
-          thing
-        end
-      end
     end
   end
 end