psych-pure 0.1.2 → 0.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0fb90297ca479adfe033c6d8db3544ec41c8e8083f23884e990de4ee2d8eb402
-  data.tar.gz: 8f895282ff5733903665ddcd0d1be2a28696dce0380242f8804ad42ce45123fe
+  metadata.gz: 62e40e29dab03fe116feadefc2a62967ec840deb52bee18bff89ae1318bf4d02
+  data.tar.gz: 3f5612a277682bbef3a698e03cc3bde6e9da02dc3f3024d9978aff34af8a624c
 SHA512:
-  metadata.gz: 4ad4470231f6587e40da0fe271ef65a1ba441f70ecc87bd7900f4fcc17a2da4be63396b73d15a131359977b62e53cfbd9d333969dbd920e984c6647741bfab34
-  data.tar.gz: 17965e31133a24c3f14c9d452884665f0d40e0c2e3edbf573fc59db5888471fb934caed61b9bed836287253061e549e540e4b79085360f9ec7a9a8082f4b7107
+  metadata.gz: 5be228176e34e4e908e1ea6af1cebbf41b59fa311de3e4a5abec21aefd9e6028f87473714b9e2c0ccf554bec1a379986b9323744c94ebdf9144cdc08d9116991
+  data.tar.gz: 4ba02fec03109e76b40883918d1d46125bdd8c1afb9d486a5af2decb868ebc6c967229dda3758e6281b5df709a89b2e6f704d747bad7f307d748861aba580b98

data/CHANGELOG.md

@@ -6,6 +6,12 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/) a
 
 ## [Unreleased]
 
+## [0.1.3] - 2025-10-24
+
+- Fix up roundtripping when using `<<` inside mappings.
+- Fix up roundtripping when using duplicate keys inside mappings.
+- Fix up comment handling when using duplicate keys inside mappings.
+
 ## [0.1.2] - 2025-03-04
 
 - Fix up comment dumping to not drift around objects.
@@ -25,7 +31,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/) a
 
 - 🎉 Initial release. 🎉
 
-[unreleased]: https://github.com/kddnewton/psych-pure/compare/v0.1.2...HEAD
+[unreleased]: https://github.com/kddnewton/psych-pure/compare/v0.1.3...HEAD
+[0.1.3]: https://github.com/kddnewton/psych-pure/compare/v0.1.2...v0.1.3
 [0.1.2]: https://github.com/kddnewton/psych-pure/compare/v0.1.1...v0.1.2
 [0.1.1]: https://github.com/kddnewton/psych-pure/compare/v0.1.0...v0.1.1
 [0.1.0]: https://github.com/kddnewton/psych-pure/compare/24de62...v0.1.0

data/lib/psych/pure/version.rb

@@ -2,6 +2,6 @@
 
 module Psych
   module Pure
-    VERSION = "0.1.2"
+    VERSION = "0.1.3"
   end
 end

data/lib/psych/pure.rb

@@ -7,6 +7,14 @@ require "strscan"
 require "stringio"
 
 module Psych
+  module Nodes
+    class Scalar
+      # The source of the scalar, as it was found in the input. This may be set
+      # in order to be reused when dumping the object.
+      attr_accessor :source
+    end
+  end
+
   # A YAML parser written in Ruby.
   module Pure
     # An internal exception is an exception that should not have occurred. It is
@@ -17,7 +25,7 @@ module Psych
       end
     end
 
-    # A source is wraps the input string and provides methods to access line and
+    # A source wraps the input string and provides methods to access line and
     # column information from a byte offset.
     class Source
       def initialize(string)
@@ -27,7 +35,13 @@ module Psych
         offset = 0
         string.each_line do |line|
           @line_offsets << offset
-          @trimmable_lines << line.match?(/\A(?: *#.*)?\n\z/)
+          @trimmable_lines <<
+            case line
+            when /\A *#.*\n\z/ then :comment
+            when /\A *\n\z/ then :blank
+            else false
+            end
+
           offset += line.bytesize
         end
 
@@ -43,6 +57,14 @@ module Psych
         offset
       end
 
+      def trim_comments(offset)
+        while (l = line(offset)) != 0 && (offset == @line_offsets[l]) && @trimmable_lines[l - 1] == :comment
+          offset = @line_offsets[l - 1]
+        end
+
+        offset
+      end
+
       def line(offset)
         index = @line_offsets.bsearch_index { |line_offset| line_offset > offset }
         return @line_offsets.size - 1 if index.nil?
@@ -64,6 +86,10 @@ module Psych
         @pos_end = pos_end
       end
 
+      def range
+        @pos_start...@pos_end
+      end
+
       def start_line
         @source.line(@pos_start)
       end
@@ -89,6 +115,11 @@ module Psych
         Location.new(@source, @pos_start, @source.trim(@pos_end))
       end
 
+      # Trim trailing comments from this location.
+      def trim_comments
+        Location.new(@source, @pos_start, @source.trim_comments(@pos_end))
+      end
+
       def to_a
         [start_line, start_column, end_line, end_column]
       end
@@ -161,33 +192,75 @@ module Psych
         @psych_node = psych_node
         @dirty = dirty
       end
+
+      def replace(psych_node)
+        @psych_node = psych_node
+        @dirty = true
+      end
     end
 
     # Wraps a Ruby hash with its node from the source input.
     class LoadedHash < SimpleDelegator
+      class PsychKey
+        attr_reader :key_node, :value_node
+
+        def initialize(key_node, value_node)
+          @key_node = key_node
+          @value_node = value_node
+        end
+
+        def replace(value_node)
+          @value_node = value_node
+        end
+      end
+
       # The node associated with the hash.
       attr_reader :psych_node
 
-      # The nodes associated with the keys of the hash.
-      attr_reader :psych_node_keys
-
-      def initialize(object, psych_node, psych_node_keys = {})
+      def initialize(object, psych_node)
         super(object)
         @psych_node = psych_node
-        @psych_node_keys = psych_node_keys
+        @psych_keys = []
+      end
+
+      def psych_keys
+        @psych_keys.map do |psych_key|
+          [psych_key.key_node, psych_key.value_node]
+        end
       end
 
       def []=(key, value)
-        if (previous = self[key])
-          if previous.is_a?(LoadedObject)
-            value = LoadedObject.new(value, previous.psych_node, true)
-          elsif previous.is_a?(LoadedHash)
-            value = LoadedHash.new(value, previous.psych_node, previous.psych_node_keys)
+        if begin
+          @psych_keys.none? do |psych_key|
+            key_node = psych_key.key_node
+            key_node_inner =
+              if key_node.is_a?(LoadedHash) || key_node.is_a?(LoadedObject)
+                key_node.__getobj__
+              else
+                key_node
+              end
+
+            if key_node_inner.eql?(key)
+              psych_key.replace(value)
+              true
+            end
           end
+        end then
+          @psych_keys << PsychKey.new(key, value)
         end
 
         super(key, value)
       end
+
+      def set!(key_node, value_node)
+        @psych_keys << PsychKey.new(key_node, value_node)
+        __getobj__[key_node.__getobj__] = value_node
+      end
+
+      def join!(key_node, value_node)
+        @psych_keys << PsychKey.new(key_node, value_node)
+        merge!(value_node)
+      end
     end
 
     # This module contains all of the extensions to Psych that we need in order
@@ -349,8 +422,6 @@ module Psych
         def initialize(ss, class_loader, symbolize_names: false, freeze: false, comments: false)
           super(ss, class_loader, symbolize_names: symbolize_names, freeze: freeze)
           @comments = comments
-          @nodeless_objs = {}.compare_by_identity
-          @nodeless_keys = {}.compare_by_identity
         end
 
         def accept(node)
@@ -360,31 +431,66 @@ module Psych
             case result
             when LoadedObject, LoadedHash
               # skip
-            when Hash
-              if !@nodeless_objs.key?(result)
-                nodeless_obj = {}
-                nodeless_key = {}
-
-                result.each do |key, value|
-                  if key.is_a?(LoadedObject) || key.is_a?(LoadedHash)
-                    nodeless_obj[key.__getobj__] = value
-                    nodeless_key[key.__getobj__] = key
-                  else
-                    nodeless_obj[key] = value
+            else
+              result = LoadedObject.new(result, node)
+            end
+          end
+
+          result
+        end
+
+        private
+
+        def revive_hash(hash, node, tagged = false)
+          return super unless @comments
+
+          revived = LoadedHash.new(hash, node)
+          node.children.each_slice(2) do |key_node, value_node|
+            key = accept(key_node)
+            value = accept(value_node)
+
+            if key == "<<" && key_node.tag != "tag:yaml.org,2002:str"
+              case value_node
+              when Nodes::Alias, Nodes::Mapping
+                begin
+                  # h1:
+                  #   <<: *h2
+                  #   <<: { k: v }
+                  revived.join!(key, value)
+                rescue TypeError
+                  # a: &a [1, 2, 3]
+                  # h: { <<: *a }
+                  revived.set!(key, value)
+                end
+              when Nodes::Sequence
+                # h1:
+                #   <<: [*h2, *h3]
+                begin
+                  temporary = {}
+                  value.reverse_each { |value| temporary.merge!(value) }
+                rescue TypeError
+                  revived.set!(key, value)
+                else
+                  value_node.children.zip(value).reverse_each do |(child_value_node, child_value)|
+                    revived.join!(key, child_value)
                   end
                 end
-
-                @nodeless_objs[result] = nodeless_obj
-                @nodeless_keys[result] = nodeless_key
+              else
+                # k: v
+                revived.set!(key, value)
               end
-
-              result = LoadedHash.new(@nodeless_objs.fetch(result), node, @nodeless_keys.fetch(result))
             else
-              result = LoadedObject.new(result, node)
+              if !tagged && @symbolize_names && key.is_a?(String)
+                key = key.to_sym
+              elsif !@freeze
+                key = deduplicate(key)
+              end
+
+              revived.set!(key, value)
             end
           end
 
-          result
+          revived
         end
       end
 
@@ -493,11 +599,12 @@ module Psych
     # A scalar event represents a single value in the YAML document. It can be
     # many different types.
     class Scalar
-      attr_reader :location, :value, :style
+      attr_reader :location, :source, :value, :style
       attr_accessor :anchor, :tag
 
-      def initialize(location, value, style)
+      def initialize(location, source, value, style)
         @location = location
+        @source = source
         @value = value
         @anchor = nil
         @tag = nil
@@ -506,14 +613,19 @@ module Psych
 
       def accept(handler)
         handler.event_location(*@location.trim)
-        handler.scalar(
-          @value,
-          @anchor,
-          @tag,
-          (!@tag || @tag == "!") && (@style == Nodes::Scalar::PLAIN),
-          (!@tag || @tag == "!") && (@style != Nodes::Scalar::PLAIN),
-          @style
-        )
+
+        event =
+          handler.scalar(
+            @value,
+            @anchor,
+            @tag,
+            (!@tag || @tag == "!") && (@style == Nodes::Scalar::PLAIN),
+            (!@tag || @tag == "!") && (@style != Nodes::Scalar::PLAIN),
+            @style
+          )
+
+        event.source = source if event.is_a?(Nodes::Scalar)
+        event
       end
     end
 
@@ -1453,7 +1565,7 @@ module Psych
       # e-scalar ::=
       #   <empty>
       def parse_e_scalar
-        events_push_flush_properties(Scalar.new(Location.point(@source, @scanner.pos), "", Nodes::Scalar::PLAIN))
+        events_push_flush_properties(Scalar.new(Location.point(@source, @scanner.pos), "", "", Nodes::Scalar::PLAIN))
         true
       end
 
@@ -1557,7 +1669,7 @@ module Psych
             end
           end
 
-          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), value, Nodes::Scalar::DOUBLE_QUOTED))
+          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), from(pos_start), value, Nodes::Scalar::DOUBLE_QUOTED))
           true
         end
       end
@@ -1701,7 +1813,7 @@ module Psych
           value.gsub!(/(?:[\ \t]*\r?\n[\ \t]*)/, "\n")
           value.gsub!(/\n(\n*)/) { $1.empty? ? " " : $1 }
           value.gsub!("''", "'")
-          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), value, Nodes::Scalar::SINGLE_QUOTED))
+          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), from(pos_start), value, Nodes::Scalar::SINGLE_QUOTED))
           true
         end
       end
@@ -2241,10 +2353,13 @@ module Psych
           end
 
         if result
-          value = from(pos_start)
+          source = from(pos_start)
+
+          value = source.dup
           value.gsub!(/(?:[\ \t]*\r?\n[\ \t]*)/, "\n")
           value.gsub!(/\n(\n*)/) { $1.empty? ? " " : $1 }
-          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), value, Nodes::Scalar::PLAIN))
+
+          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), source, value, Nodes::Scalar::PLAIN))
         end
 
         result
@@ -2470,8 +2585,7 @@ module Psych
           parse_l_literal_content(n + m, t)
         } then
           @in_scalar = false
-          lines = events_cache_pop
-          value = lines.map { |line| "#{line}\n" }.join
+          value = events_cache_pop.map { |line| "#{line}\n" }.join
 
           case t
           when :clip
@@ -2484,7 +2598,8 @@ module Psych
             raise InternalException, t.inspect
           end
 
-          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), value, Nodes::Scalar::LITERAL))
+          location = Location.new(@source, pos_start, @scanner.pos).trim_comments
+          events_push_flush_properties(Scalar.new(location, @scanner.string.byteslice(location.range).chomp, value, Nodes::Scalar::LITERAL))
           true
         else
           @in_scalar = false
@@ -2582,7 +2697,8 @@ module Psych
             raise InternalException, t.inspect
           end
 
-          events_push_flush_properties(Scalar.new(Location.new(@source, pos_start, @scanner.pos), value, Nodes::Scalar::FOLDED))
+          location = Location.new(@source, pos_start, @scanner.pos).trim_comments
+          events_push_flush_properties(Scalar.new(location, @scanner.string.byteslice(location.range).chomp, value, Nodes::Scalar::FOLDED))
           true
         else
           @in_scalar = false
@@ -3220,8 +3336,14 @@ module Psych
 
       # Represents a string object.
       class StringNode < Node
+        # The explicit tag associated with the object.
         attr_accessor :tag
 
+        # Whether or not this object was modified after being loaded. In this
+        # case we cannot rely on the source formatting, and need to instead
+        # format the value ourselves.
+        attr_accessor :dirty
+
         def accept(visitor)
           visitor.visit_string(self)
         end
@@ -3264,13 +3386,17 @@ module Psych
         # Visit an ObjectNode.
         def visit_object(node)
           with_comments(node) do |value|
-            if (tag = node.tag)
-              @q.text("#{tag} ")
-            end
-
             if !node.dirty && (psych_node = node.psych_node)
-              @q.text(psych_node.value)
+              if (tag = node.tag)
+                @q.text("#{tag} ")
+              end
+
+              @q.text(psych_node.source || psych_node.value)
             else
+              if (tag = node.tag) && tag != "tag:yaml.org,2002:binary"
+                @q.text("#{tag} ")
+              end
+
               @q.text(dump_object(value))
             end
           end
@@ -3293,11 +3419,19 @@ module Psych
         # Visit a StringNode.
         def visit_string(node)
           with_comments(node) do |value|
-            if (tag = node.tag)
-              @q.text("#{tag} ")
-            end
+            if !node.dirty && (psych_node = node.psych_node)
+              if (tag = node.tag)
+                @q.text("#{tag} ")
+              end
 
-            @q.text(dump_object(value))
+              @q.text(psych_node.source || psych_node.value)
+            else
+              if (tag = node.tag) && tag != "tag:yaml.org,2002:binary"
+                @q.text("#{tag} ")
+              end
+
+              @q.text(dump_object(value))
+            end
           end
         end
 
@@ -3447,9 +3581,26 @@ module Psych
             @q.breakable
           end
 
+          current_line = nil
           ((0...children.length) % 2).each do |index|
-            @q.breakable if index != 0
-            visit_hash_key_value(children[index], children[index + 1])
+            key = children[index]
+            value = children[index + 1]
+
+            if index > 0
+              @q.breakable
+
+              if current_line && (psych_node = key.psych_node)
+                start_line = psych_node.start_line
+                if (leading = key.psych_node.comments.leading).any?
+                  start_line = leading.first.start_line
+                end
+
+                @q.breakable if start_line - current_line >= 2
+              end
+            end
+
+            current_line = (psych_node = value.psych_node) ? psych_node.end_line : nil
+            visit_hash_key_value(key, value)
           end
 
           @q.current_group.break
@@ -3576,6 +3727,7 @@ module Psych
         # Very rare circumstance here that there are leading comments attached
         # to the root object of a document that occur before the --- marker. In
         # this case we want to output them first here, then dump the object.
+        reload_comments = nil
         if (object.is_a?(LoadedObject) || object.is_a?(LoadedHash)) && (psych_node = object.psych_node).comments? && (leading = psych_node.comments.leading).any?
           leading = [*leading]
           line = psych_node.start_line - 1
@@ -3598,6 +3750,8 @@ module Psych
 
             line = comment.start_line
           end
+
+          reload_comments = leading.concat(psych_node.comments.leading)
         end
 
         @io << "---"
@@ -3620,6 +3774,12 @@ module Psych
 
           @io << q.output
         end
+
+        # If we initially split up the leading comments, then we need to reload
+        # them back to their original state here.
+        unless reload_comments.nil?
+          object.psych_node.comments.leading.replace(reload_comments)
+        end
       end
 
       private
@@ -3627,8 +3787,6 @@ module Psych
       # Dump the tag value for a given node.
       def dump_tag(value)
         case value
-        when nil, "tag:yaml.org,2002:binary"
-          nil
         when /\Atag:yaml.org,2002:(.+)\z/
           "!!#{$1}"
         else
@@ -3655,7 +3813,14 @@ module Psych
           end
 
           if @object_nodes.key?(object)
-            AliasNode.new(@object_nodes[object].anchor = (@object_anchors[object] ||= (@object_anchor += 1)), nil)
+            @object_anchors[object] ||=
+              if psych_node.is_a?(Nodes::Alias)
+                psych_node.anchor
+              else
+                @object_anchor += 1
+              end
+
+            AliasNode.new(@object_nodes[object].anchor = @object_anchors[object], psych_node)
           else
             case object
             when Psych::Omap
@@ -3670,8 +3835,7 @@ module Psych
             when Hash
               contents =
                 if base_object.is_a?(LoadedHash)
-                  psych_node_keys = base_object.psych_node_keys    
-                  object.flat_map { |key, value| [dump(psych_node_keys.fetch(key, key)), dump(value)] }
+                  base_object.psych_keys.flat_map { |(key, value)| [dump(key), dump(value)] }
                 else
                   object.flat_map { |key, value| [dump(key), dump(value)] }
                 end
@@ -3683,13 +3847,12 @@ module Psych
             when String
               dumped = StringNode.new(object, psych_node)
               dumped.tag = dump_tag(psych_node&.tag)
-
+              dumped.dirty = dirty
               dumped
             else
               dumped = ObjectNode.new(object, psych_node)
               dumped.tag = dump_tag(psych_node&.tag)
               dumped.dirty = dirty
-
               dumped
             end
           end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: psych-pure
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.3
 platform: ruby
 authors:
 - Kevin Newton
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-03-04 00:00:00.000000000 Z
+date: 2025-10-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: psych
@@ -113,7 +113,7 @@ licenses:
 - MIT
 metadata:
   bug_tracker_uri: https://github.com/kddnewton/psych-pure/issues
-  changelog_uri: https://github.com/kddnewton/psych-pure/blob/v0.1.2/CHANGELOG.md
+  changelog_uri: https://github.com/kddnewton/psych-pure/blob/v0.1.3/CHANGELOG.md
   source_code_uri: https://github.com/kddnewton/psych-pure
   rubygems_mfa_required: 'true'
 post_install_message: