psych-pure 0.1.1 → 0.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d126f1c02cd1f7eba6b74a19f76dc938a6c828dc8dfea96a5663b3b4c8ac5bac
-  data.tar.gz: 6df58931a63090a6248a44d94986057fa60faa5c6ec2015b832da7a4d1fefe7a
+  metadata.gz: 62e40e29dab03fe116feadefc2a62967ec840deb52bee18bff89ae1318bf4d02
+  data.tar.gz: 3f5612a277682bbef3a698e03cc3bde6e9da02dc3f3024d9978aff34af8a624c
 SHA512:
-  metadata.gz: 00f44a5e4889ac4674d844f5dbfbff8a39fb1126f40b7135a33165539a3e6d27d5d5303b0289259471ba3d8c3dfd7e58e75f2fa8c5339258239a89b58e314c2d
-  data.tar.gz: feaacc73c47c767edf1f454de1ecf4db4c812557f10578d88f715508b1975c524edc9acf5c21b40f77c40e1fc9eb25e7e79ee6496d8322cfb49e0030d457be3c
+  metadata.gz: 5be228176e34e4e908e1ea6af1cebbf41b59fa311de3e4a5abec21aefd9e6028f87473714b9e2c0ccf554bec1a379986b9323744c94ebdf9144cdc08d9116991
+  data.tar.gz: 4ba02fec03109e76b40883918d1d46125bdd8c1afb9d486a5af2decb868ebc6c967229dda3758e6281b5df709a89b2e6f704d747bad7f307d748861aba580b98

data/CHANGELOG.md

@@ -6,6 +6,19 @@ 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.
+- Rely on source formatting when possible for scalar values.
+- Fix up multi-line literals with the "keep" flag (|+).
+- Fix up aliasing hashes that have comments loaded.
+
 ## [0.1.1] - 2025-02-13
 
 - Fix up comment handling to preserve within hashes.
@@ -18,6 +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/v0.1.1...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 = "0.1.1"
+    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
@@ -147,48 +178,89 @@ module Psych
       end
     end
 
-    # Wraps a Ruby object with its comments from the source input.
-    class CommentsObject < SimpleDelegator
-      attr_reader :psych_comments
+    # Wraps a Ruby object with its node from the source input.
+    class LoadedObject < SimpleDelegator
+      # The node associated with the object.
+      attr_reader :psych_node
+
+      # Whether or not this object has been modified. If it has, then we cannot
+      # rely on the source formatting, and need to format it ourselves.
+      attr_reader :dirty
 
-      def initialize(object, psych_comments)
-        @psych_comments = psych_comments
+      def initialize(object, psych_node, dirty = false)
         super(object)
+        @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 comments from the source input.
-    class CommentsHash < SimpleDelegator
-      attr_reader :psych_comments, :psych_key_comments
+    # Wraps a Ruby hash with its node from the source input.
+    class LoadedHash < SimpleDelegator
+      class PsychKey
+        attr_reader :key_node, :value_node
 
-      def initialize(object, psych_comments, psych_key_comments = {})
-        @psych_comments = psych_comments
-        @psych_key_comments = psych_key_comments
-        commentless = {}
+        def initialize(key_node, value_node)
+          @key_node = key_node
+          @value_node = value_node
+        end
 
-        object.each do |key, value|
-          if key.is_a?(CommentsObject)
-            @psych_key_comments[key.__getobj__] = key.psych_comments
-            commentless[key.__getobj__] = value
-          else
-            commentless[key] = value
-          end
+        def replace(value_node)
+          @value_node = value_node
         end
+      end
 
-        super(commentless)
+      # The node associated with the hash.
+      attr_reader :psych_node
+
+      def initialize(object, psych_node)
+        super(object)
+        @psych_node = psych_node
+        @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?(CommentsObject)
-            value = CommentsObject.new(value, previous.psych_comments)
-          elsif previous.is_a?(CommentsHash)
-            value = CommentsHash.new(value, previous.psych_comments, previous.psych_key_comments)
+        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
@@ -304,9 +376,18 @@ module Psych
       # Extend the document stream to be able to attach comments to the
       # document.
       module DocumentStream
+        def start_document(version, tag_directives, implicit)
+          node = Nodes::Document.new(version, tag_directives, implicit)
+          set_start_location(node)
+          push(node)
+        end
+
         def end_document(implicit_end = !streaming?)
           @last.implicit_end = implicit_end
-          @block.call(attach_comments(pop))
+          node = pop
+          set_end_location(node)
+          attach_comments(node)
+          @block.call(node)
         end
       end
 
@@ -347,22 +428,77 @@ module Psych
           result = super
 
           if @comments
-            if result.is_a?(Hash)
-              result = CommentsHash.new(result, node.comments? ? node.comments : nil)
-            elsif node.comments?
-              result = CommentsObject.new(result, node.comments)
+            case result
+            when LoadedObject, LoadedHash
+              # skip
+            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
+              else
+                # k: v
+                revived.set!(key, value)
+              end
+            else
+              if !tagged && @symbolize_names && key.is_a?(String)
+                key = key.to_sym
+              elsif !@freeze
+                key = deduplicate(key)
+              end
+
+              revived.set!(key, value)
+            end
+          end
+
+          revived
+        end
       end
 
       # Extend the ToRuby singleton to be able to pass the comments option.
       module ToRubySingleton
         def create(symbolize_names: false, freeze: false, strict_integer: false, comments: false)
           class_loader = ClassLoader.new
-          scanner      = ScalarScanner.new(class_loader, strict_integer: strict_integer)
+          scanner = ScalarScanner.new(class_loader, strict_integer: strict_integer)
           new(scanner, class_loader, symbolize_names: symbolize_names, freeze: freeze, comments: comments)
         end
       end
@@ -463,11 +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
@@ -475,15 +612,20 @@ module Psych
       end
 
       def accept(handler)
-        handler.event_location(*@location)
-        handler.scalar(
-          @value,
-          @anchor,
-          @tag,
-          (!@tag || @tag == "!") && (@style == Nodes::Scalar::PLAIN),
-          (!@tag || @tag == "!") && (@style != Nodes::Scalar::PLAIN),
-          @style
-        )
+        handler.event_location(*@location.trim)
+
+        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
 
@@ -1423,7 +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
 
@@ -1527,7 +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
@@ -1671,7 +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
@@ -2211,10 +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
@@ -2440,9 +2585,7 @@ module Psych
           parse_l_literal_content(n + m, t)
         } then
           @in_scalar = false
-          lines = events_cache_pop
-          lines.pop if lines.length > 0 && lines.last.empty?
-          value = lines.map { |line| "#{line}\n" }.join
+          value = events_cache_pop.map { |line| "#{line}\n" }.join
 
           case t
           when :clip
@@ -2450,12 +2593,13 @@ module Psych
           when :strip
             value.sub!(/\n+\z/, "")
           when :keep
-            value.sub!(/\n(\n+)\z/) { $1 } if !value.match?(/\S/)
+            # nothing
           else
             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
@@ -2469,6 +2613,8 @@ module Psych
       #   l-empty(n,block-in)*
       #   s-indent(n) nb-char+
       def parse_l_nb_literal_text(n)
+        events_cache_size = @events_cache[-1].size
+
         try do
           if star { parse_l_empty(n, :block_in) } && parse_s_indent(n)
             pos_start = @scanner.pos
@@ -2477,6 +2623,12 @@ module Psych
               events_push(from(pos_start))
               true
             end
+          else
+            # When parsing all of the l_empty calls, we may have added a bunch
+            # of empty lines to the events cache. We need to clear those out
+            # here.
+            @events_cache[-1].slice!(events_cache_size..-1)
+            false
           end
         end
       end
@@ -2545,7 +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
@@ -2739,7 +2892,7 @@ module Psych
 
         if try { plus { try { parse_s_indent(n + m) && parse_ns_l_block_map_entry(n + m) } } }
           events_cache_flush
-          events_push_flush_properties(MappingEnd.new(Location.point(@source, @scanner.pos))) # TODO
+          events_push_flush_properties(MappingEnd.new(Location.point(@source, @scanner.pos)))
           true
         else
           events_cache_pop
@@ -2855,7 +3008,7 @@ module Psych
           star { try { parse_s_indent(n) && parse_ns_l_block_map_entry(n) } }
         } then
           events_cache_flush
-          events_push_flush_properties(MappingEnd.new(Location.point(@source, @scanner.pos))) # TODO
+          events_push_flush_properties(MappingEnd.new(Location.point(@source, @scanner.pos)))
           true
         else
           events_cache_pop
@@ -3042,12 +3195,12 @@ module Psych
       def parse_l_yaml_stream
         events_push_flush_properties(StreamStart.new(Location.point(@source, @scanner.pos)))
 
-        @document_start_event = DocumentStart.new(Location.point(@source, @scanner.pos))
-        @tag_directives = @document_start_event.tag_directives
-        @document_end_event = nil
-
         if try {
           if parse_l_document_prefix
+            @document_start_event = DocumentStart.new(Location.point(@source, @scanner.pos))
+            @tag_directives = @document_start_event.tag_directives
+            @document_end_event = nil
+
             parse_l_any_document
             star do
               try do
@@ -3105,11 +3258,11 @@ module Psych
       # aliases, since we may find that we need to add an anchor after the
       # object has already been flushed.
       class Node
-        attr_reader :value, :comments
+        attr_reader :value, :psych_node
 
-        def initialize(value, comments)
+        def initialize(value, psych_node)
           @value = value
-          @comments = comments
+          @psych_node = psych_node
           @anchor = nil
         end
 
@@ -3127,7 +3280,7 @@ module Psych
 
       # Represents an array of nodes.
       class ArrayNode < Node
-        attr_accessor :anchor
+        attr_accessor :anchor, :tag
 
         def accept(visitor)
           visitor.visit_array(self)
@@ -3136,7 +3289,7 @@ module Psych
 
       # Represents a hash of nodes.
       class HashNode < Node
-        attr_accessor :anchor
+        attr_accessor :anchor, :tag
 
         def accept(visitor)
           visitor.visit_hash(self)
@@ -3150,6 +3303,14 @@ module Psych
       # Represents a generic object that is not matched by any of the other node
       # types.
       class ObjectNode < 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_object(self)
         end
@@ -3175,6 +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
@@ -3195,14 +3364,10 @@ module Psych
         # Visit an ArrayNode.
         def visit_array(node)
           with_comments(node) do |value|
-            if (anchor = node.anchor)
-              @q.text("&#{anchor} ")
-            end
-
-            if value.empty?
-              @q.text("[]")
+            if value.empty? || ((psych_node = node.psych_node).is_a?(Nodes::Sequence) && psych_node.style == Nodes::Sequence::FLOW)
+              visit_array_contents_flow(node.anchor, node.tag, value)
             else
-              visit_array_contents(value)
+              visit_array_contents_block(node.anchor, node.tag, value)
             end
           end
         end
@@ -3210,16 +3375,10 @@ module Psych
         # Visit a HashNode.
         def visit_hash(node)
           with_comments(node) do |value|
-            if (anchor = node.anchor)
-              @q.text("&#{anchor}")
-            end
-
-            if value.empty?
-              @q.text(" ") if anchor
-              @q.text("{}")
+            if value.empty? || ((psych_node = node.psych_node).is_a?(Nodes::Mapping) && psych_node.style == Nodes::Mapping::FLOW)
+              visit_hash_contents_flow(node.anchor, node.tag, value)
             else
-              @q.breakable if anchor
-              visit_hash_contents(value)
+              visit_hash_contents_block(node.anchor, node.tag, value)
             end
           end
         end
@@ -3227,50 +3386,88 @@ module Psych
         # Visit an ObjectNode.
         def visit_object(node)
           with_comments(node) do |value|
-            @q.text(Psych.dump(value, indentation: @q.indent)[/\A--- (.+)\n\z/m, 1]) # TODO
+            if !node.dirty && (psych_node = node.psych_node)
+              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
         end
 
         # Visit an OmapNode.
         def visit_omap(node)
           with_comments(node) do |value|
-            if (anchor = node.anchor)
-              @q.text("&#{anchor} ")
-            end
-
-            @q.text("!!omap")
-            @q.breakable
-
-            visit_array_contents(value)
+            visit_array_contents_block(node.anchor, "!!omap", value)
           end
         end
 
         # Visit a SetNode.
         def visit_set(node)
           with_comments(node) do |value|
-            if (anchor = node.anchor)
-              @q.text("&#{anchor} ")
-            end
-
-            @q.text("!set")
-            @q.breakable
-
-            visit_hash_contents(node.value)
+            visit_hash_contents_block(node.anchor, "!set", value)
           end
         end
 
         # Visit a StringNode.
-        alias visit_string visit_object
+        def visit_string(node)
+          with_comments(node) do |value|
+            if !node.dirty && (psych_node = node.psych_node)
+              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
+        end
 
         private
 
+        # TODO: Certain objects require special formatting. Usually this
+        # involves scanning the object itself and determining what kind of YAML
+        # object it is, then dumping it back out. We rely on Psych itself to do
+        # this formatting for us.
+        #
+        # Note this is the one place where we indirectly rely on libyaml,
+        # because Psych delegates to libyaml to dump the object. This is less
+        # than ideal, because it means in some circumstances we have an indirect
+        # dependency. Ideally this would all be removed in favor of our own
+        # formatting.
+        def dump_object(value)
+          Psych.dump(value, indentation: @q.indent)[/\A--- (.+?)(?:\n\.\.\.)?\n\z/m, 1]
+        end
+
         # Shortcut to visit a node by passing this visitor to the accept method.
         def visit(node)
           node.accept(self)
         end
 
-        # Visit the elements within an array.
-        def visit_array_contents(contents)
+        # Visit the elements within an array in the block format.
+        def visit_array_contents_block(anchor, tag, contents)
+          if anchor
+            @q.text("&#{anchor}")
+            tag ? @q.text(" ") : @q.breakable
+          end
+
+          if tag
+            @q.text(tag)
+            @q.breakable
+          end
+
           @q.seplist(contents, -> { @q.breakable }) do |element|
             @q.text("-")
             next if element.is_a?(NilNode)
@@ -3278,79 +3475,164 @@ module Psych
             @q.text(" ")
             @q.nest(2) { visit(element) }
           end
+
+          @q.current_group.break
         end
 
-        # Visit the key/value pairs within a hash.
-        def visit_hash_contents(contents)
-          @q.seplist(contents, -> { @q.breakable }) do |key, value|
-            inlined = false
+        # Visit the elements within an array in the flow format.
+        def visit_array_contents_flow(anchor, tag, contents)
+          @q.group do
+            @q.text("&#{anchor} ") if anchor
+            @q.text("#{tag} ") if tag
+            @q.text("[")
 
-            case key
-            when NilNode
-              @q.text("! ''")
-            when ArrayNode, HashNode, OmapNode, SetNode
-              if key.anchor.nil?
-                @q.text("? ")
-                @q.nest(2) { visit(key) }
-                @q.breakable
-                inlined = true
-              else
-                visit(key)
+            unless contents.empty?
+              @q.nest(2) do
+                @q.breakable("")
+                @q.seplist(contents, -> { @q.comma_breakable }) { |element| visit(element) }
               end
-            when AliasNode, ObjectNode
+              @q.breakable("")
+            end
+
+            @q.text("]")
+          end
+        end
+
+        # Visit a key value pair within a hash.
+        def visit_hash_key_value(key, value)
+          inlined = false
+
+          case key
+          when NilNode
+            @q.text("! ''")
+          when ArrayNode, HashNode, OmapNode, SetNode
+            if key.anchor.nil?
+              @q.text("? ")
+              @q.nest(2) { visit(key) }
+              @q.breakable
+              inlined = true
+            else
               visit(key)
-            when StringNode
-              if key.value.include?("\n")
-                @q.text("? ")
-                visit(key)
-                @q.breakable
-                inlined = true
-              else
-                visit(key)
-              end
             end
+          when AliasNode, ObjectNode
+            visit(key)
+          when StringNode
+            if key.value.include?("\n")
+              @q.text("? ")
+              visit(key)
+              @q.breakable
+              inlined = true
+            else
+              visit(key)
+            end
+          else
+            raise InternalException
+          end
 
-            @q.text(":")
+          @q.text(":")
 
-            case value
-            when NilNode
-              # skip
-            when OmapNode, SetNode
+          case value
+          when NilNode
+            # skip
+          when OmapNode, SetNode
+            @q.text(" ")
+            @q.nest(2) { visit(value) }
+          when ArrayNode
+            if ((psych_node = value.psych_node).is_a?(Nodes::Sequence) && psych_node.style == Nodes::Sequence::FLOW) || value.value.empty?
+              @q.text(" ")
+              visit(value)
+            elsif inlined || value.anchor || value.tag || value.value.empty?
               @q.text(" ")
               @q.nest(2) { visit(value) }
-            when ArrayNode
-              if value.value.empty?
-                @q.text(" []")
-              elsif inlined || value.anchor
-                @q.text(" ")
-                @q.nest(2) { visit(value) }
-              else
+            else
+              @q.breakable
+              visit(value)
+            end
+          when HashNode
+            if ((psych_node = value.psych_node).is_a?(Nodes::Mapping) && psych_node.style == Nodes::Mapping::FLOW) || value.value.empty?
+              @q.text(" ")
+              visit(value)
+            elsif inlined || value.anchor || value.tag
+              @q.text(" ")
+              @q.nest(2) { visit(value) }
+            else
+              @q.nest(2) do
                 @q.breakable
                 visit(value)
               end
-            when HashNode
-              if value.value.empty?
-                @q.text(" {}")
-              elsif inlined || value.anchor
-                @q.text(" ")
-                @q.nest(2) { visit(value) }
-              else
-                @q.nest(2) do
-                  @q.breakable
-                  visit(value)
+            end
+          when AliasNode, ObjectNode, StringNode
+            @q.text(" ")
+            @q.nest(2) { visit(value) }
+          else
+            raise InternalException
+          end
+        end
+
+        # Visit the key/value pairs within a hash in the block format.
+        def visit_hash_contents_block(anchor, tag, children)
+          if anchor
+            @q.text("&#{anchor}")
+            tag ? @q.text(" ") : @q.breakable
+          end
+
+          if tag
+            @q.text(tag)
+            @q.breakable
+          end
+
+          current_line = nil
+          ((0...children.length) % 2).each do |index|
+            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
-            when AliasNode, ObjectNode, StringNode
-              @q.text(" ")
-              @q.nest(2) { visit(value) }
             end
+
+            current_line = (psych_node = value.psych_node) ? psych_node.end_line : nil
+            visit_hash_key_value(key, value)
+          end
+
+          @q.current_group.break
+        end
+
+        # Visit the key/value pairs within a hash in the flow format.
+        def visit_hash_contents_flow(anchor, tag, children)
+          @q.group do
+            @q.text("&#{anchor} ") if anchor
+            @q.text("#{tag} ") if tag
+            @q.text("{")
+
+            unless children.empty?
+              @q.nest(2) do
+                @q.breakable
+
+                ((0...children.length) % 2).each do |index|
+                  @q.comma_breakable if index != 0
+                  visit_hash_key_value(children[index], children[index + 1])
+                end
+              end
+              @q.breakable
+            end
+
+            @q.text("}")
           end
         end
 
         # Print out the leading and trailing comments of a node, as well as
         # yielding the value of the node to the block.
         def with_comments(node)
-          if (comments = node.comments) && (leading = comments.leading).any?
+          if (comments = node.psych_node&.comments) && (leading = comments.leading).any?
             line = nil
 
             leading.each do |comment|
@@ -3435,14 +3717,45 @@ module Psych
       # This is the main entrypoint into this object. It is responsible for
       # pushing a new object onto the emitter, which is then represented as a
       # YAML document.
-      def <<(object)
+      def emit(object)
         if @started
-          @io << "...\n---"
+          @io << "...\n"
         else
-          @io << "---"
           @started = true
         end
 
+        # 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
+
+          while leading.any? && leading.last.start_line == line
+            leading.pop
+            line -= 1
+          end
+
+          psych_node.comments.leading.slice!(0, leading.length)
+          line = nil
+
+          leading.each do |comment|
+            if line && (line < comment.start_line)
+              @io << "\n" * (comment.start_line - line - 1)
+            end
+
+            @io << comment.value
+            @io << "\n"
+
+            line = comment.start_line
+          end
+
+          reload_comments = leading.concat(psych_node.comments.leading)
+        end
+
+        @io << "---"
+
         if (node = dump(object)).is_a?(NilNode)
           @io << "\n"
         else
@@ -3461,44 +3774,87 @@ 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
 
-      # Walk through the given object and convert it into a tree of nodes.
-      def dump(base_object, comments = nil)
-        object = base_object
-
-        if base_object.is_a?(CommentsObject) || base_object.is_a?(CommentsHash)
-          object = base_object.__getobj__
-          comments = base_object.psych_comments
+      # Dump the tag value for a given node.
+      def dump_tag(value)
+        case value
+        when /\Atag:yaml.org,2002:(.+)\z/
+          "!!#{$1}"
+        else
+          value
         end
+      end
 
-        if object.nil?
-          NilNode.new(object, comments)
-        elsif @object_nodes.key?(object)
-          AliasNode.new(@object_nodes[object].anchor = (@object_anchors[object] ||= (@object_anchor += 1)), comments)
+      # Walk through the given object and convert it into a tree of nodes.
+      def dump(base_object)
+        if base_object.nil?
+          NilNode.new(nil, nil)
         else
-          case object
-          when Psych::Omap
-            @object_nodes[object] = OmapNode.new(object.map { |(key, value)| HashNode.new({ dump(key) => dump(value) }, nil) }, comments)
-          when Psych::Set
-            @object_nodes[object] = SetNode.new(object.to_h { |key, value| [dump(key), dump(value)] }, comments)
-          when Array
-            @object_nodes[object] = ArrayNode.new(object.map { |element| dump(element) }, comments)
-          when Hash
-            dumped =
-              if base_object.is_a?(CommentsHash)
-                object.to_h { |key, value| [dump(key, base_object.psych_key_comments[key]), dump(value)] }
+          object = base_object
+          psych_node = nil
+          dirty = false
+
+          if base_object.is_a?(LoadedObject)
+            object = base_object.__getobj__
+            psych_node = base_object.psych_node
+            dirty = base_object.dirty
+          elsif base_object.is_a?(LoadedHash)
+            object = base_object.__getobj__
+            psych_node = base_object.psych_node
+          end
+
+          if @object_nodes.key?(object)
+            @object_anchors[object] ||=
+              if psych_node.is_a?(Nodes::Alias)
+                psych_node.anchor
               else
-                object.to_h { |key, value| [dump(key), dump(value)] }
+                @object_anchor += 1
               end
 
-            @object_nodes[object] = HashNode.new(dumped, comments)
-          when String
-            StringNode.new(object, comments)
+            AliasNode.new(@object_nodes[object].anchor = @object_anchors[object], psych_node)
           else
-            ObjectNode.new(object, comments)
+            case object
+            when Psych::Omap
+              @object_nodes[object] = OmapNode.new(object.map { |(key, value)| HashNode.new([dump(key), dump(value)], nil) }, psych_node)
+            when Psych::Set
+              @object_nodes[object] = SetNode.new(object.flat_map { |key, value| [dump(key), dump(value)] }, psych_node)
+            when Array
+              dumped = ArrayNode.new(object.map { |element| dump(element) }, psych_node)
+              dumped.tag = dump_tag(psych_node&.tag)
+
+              @object_nodes[object] = dumped
+            when Hash
+              contents =
+                if base_object.is_a?(LoadedHash)
+                  base_object.psych_keys.flat_map { |(key, value)| [dump(key), dump(value)] }
+                else
+                  object.flat_map { |key, value| [dump(key), dump(value)] }
+                end
+
+              dumped = HashNode.new(contents, psych_node)
+              dumped.tag = dump_tag(psych_node&.tag)
+
+              @object_nodes[object] = dumped
+            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
         end
       end
@@ -3538,10 +3894,10 @@ module Psych
       private
 
       # Dump the given object, ensuring that it is a permitted object.
-      def dump(base_object, comments = nil)
+      def dump(base_object)
         object = base_object
 
-        if base_object.is_a?(CommentsObject) || base_object.is_a?(CommentsHash)
+        if base_object.is_a?(LoadedObject) || base_object.is_a?(LoadedHash)
           object = base_object.__getobj__
         end
 
@@ -3677,7 +4033,7 @@ module Psych
 
       real_io = io || StringIO.new
       emitter = Emitter.new(real_io, options)
-      emitter << o
+      emitter.emit(o)
       io || real_io.string
     end
 
@@ -3691,7 +4047,7 @@ module Psych
 
       real_io = io || StringIO.new
       emitter = SafeEmitter.new(real_io, options)
-      emitter << o
+      emitter.emit(o)
       io || real_io.string
     end
 
@@ -3699,7 +4055,7 @@ module Psych
     def self.dump_stream(*objects)
       real_io = io || StringIO.new
       emitter = Emitter.new(real_io, {})
-      objects.each { |object| emitter << object }
+      objects.each { |object| emitter.emit(object) }
       io || real_io.string
     end
   end

data/psych-pure.gemspec

@@ -40,4 +40,5 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "bundler"
   spec.add_development_dependency "minitest"
   spec.add_development_dependency "rake"
+  spec.add_development_dependency "simplecov"
 end

metadata

@@ -1,13 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: psych-pure
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.1.3
 platform: ruby
 authors:
 - Kevin Newton
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-02-13 00:00:00.000000000 Z
+date: 2025-10-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: psych
@@ -79,6 +80,21 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
+- !ruby/object:Gem::Dependency
+  name: simplecov
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+description:
 email:
 - kddnewton@gmail.com
 executables: []
@@ -97,9 +113,10 @@ licenses:
 - MIT
 metadata:
   bug_tracker_uri: https://github.com/kddnewton/psych-pure/issues
-  changelog_uri: https://github.com/kddnewton/psych-pure/blob/v0.1.1/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:
 rdoc_options: []
 require_paths:
 - lib
@@ -114,7 +131,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.2
+rubygems_version: 3.5.16
+signing_key:
 specification_version: 4
 summary: A YAML parser written in Ruby
 test_files: []