lutaml-lml 0.3.3 → 0.3.5

data/lib/lutaml/lml/format/adapter/standard_adapter.rb

@@ -10,14 +10,7 @@ module Lutaml
         class StandardAdapter < Document
           TYPE_KEY = "__type__"
 
-          # The grammar's `word` rule is [A-Za-z0-9_] only (lml.parg) and
-          # `variable = quoted_string / 1*word` — a bare value is a run of
-          # word characters and nothing else. Anything else (a dash, a
-          # space, a digit-leading date...) commits the value parser to
-          # `number` or fails re-parsing, so it must be emitted quoted.
-          # This predicate is the single source of truth for the emit
-          # side; the parse side is pinned by the grammar itself.
-          BARE_WORD = /\A\w+\z/
+          include Lutaml::Lml::Quoting
 
           def self.parse(data, _options = {})
             return data if data.is_a?(Hash)
@@ -78,13 +71,7 @@ module Lutaml
           end
 
           def format_key(key)
-            key = key.to_s
-            return key if key.match?(BARE_WORD)
-
-            # Same escaping as quote_value: a key may contain any
-            # character, and the emitted form must re-parse (issue #78).
-            escaped = key.gsub('\\') { |bs| bs * 2 }.gsub('"') { |q| "\\#{q}" }
-            "\"#{escaped}\""
+            quote_scalar(key)
           end
 
           def format_value(prefix, key, value, indent)
@@ -96,7 +83,9 @@ module Lutaml
               format_array(prefix, key, value, indent)
             when Hash
               format_nested(prefix, key, value, indent)
-            when TrueClass, FalseClass, Integer, Float
+            when Float
+              "#{prefix}#{key} = #{number_scalar(value)}"
+            when TrueClass, FalseClass, Integer
               "#{prefix}#{key} = #{value}"
             else
               "#{prefix}#{key} = #{quote_value(value)}"
@@ -144,17 +133,27 @@ module Lutaml
           end
 
           def quote_value(val)
+            return number_scalar(val) if val.is_a?(Float)
             return val.to_s if val.is_a?(Numeric) || val.is_a?(TrueClass) || val.is_a?(FalseClass)
 
+            # RS 3001 §String values: escape backslash and double quote;
+            # every value is representable in double quotes. A bare word
+            # is only safe when the parser reads it back as the same
+            # string — `null`, booleans, and digit-leading tokens are
+            # data literals (and digit-led words do not re-parse).
             str = val.to_s
-            return str if str.match?(BARE_WORD)
+            return str if safe_bare_string?(str)
 
-            # RS 3001 §String values: escape backslash and double quote;
-            # every value is representable in double quotes.
-            # gsub replacement strings re-interpret backslash pairs, so
-            # the backslash doubling uses the block form.
-            escaped = str.gsub('\\') { |bs| bs * 2 }.gsub('"') { |q| "\\#{q}" }
-            "\"#{escaped}\""
+            quote_always(str)
+          end
+
+          private
+
+          def safe_bare_string?(str)
+            return false unless str.match?(Lutaml::Lml::Quoting::BARE_WORD)
+            return false if str.match?(/\A(?:true|false|null)\z/)
+
+            !str.match?(/\A\d/)
           end
         end
       end