moxml 0.5.107 → 0.5.109

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: abbc78c5ee5efd00d9513dbe54fa844d69518c656f2eaec2c16984230ef86e09
-  data.tar.gz: 2d45e9d308e8c5826d1dafdbeab32e1c0ef1f44dbbb51586283183436b6e42cf
+  metadata.gz: 04f63f1599959406386523aafe8289335b021ee5b729b0ac88dec032e08fac55
+  data.tar.gz: f4d6cbd0012ef39253c519edd2af087c8ff230b7689622bad6143138e21c9d7d
 SHA512:
-  metadata.gz: 3074fb3116eab8bc30697dd0fb75221c10be71f2d2cc884d1954911c7b054ca314db8407a31637fa9c59bd53c00d7bf9b648d436d87cfaec3bffe1102c7af366
-  data.tar.gz: 3a08135e70f29be37a8c97c75e01239386bd43a4d14c32837525b9395b68ff2778181a1a5499116884eb796bb6b08e30a40509c7d2a5ee25b4056d34ff188dcf
+  metadata.gz: f18cb5879bc64127726fc090c2c0f570edb813a032186e883d0c8ac3af7c96ec544f5c3fae7ca3d760b4fb5aba8ac5a1fb2a6a1422e27e7029f074543a036389
+  data.tar.gz: a22ade19383fc5da7052f713761fe7bd458e062c59de10e6c8ab00d07f747a00b9627d7503148092630c798359074cbcabb05b91512b1953fd237087beafc2b5

data/README.adoc

@@ -655,8 +655,48 @@ root.add_child(title)
 puts doc.to_xml(indent: 2)
 ----
 
+=== Performance versus nokogiri
+
+CPU-time medians (1,000-record catalog, 20 iterations) through the
+moxml adapter layer on both sides — moxml + leptris 1.9.307 vs
+moxml + nokogiri:
+
+[cols="2,1,4"]
+|===
+|Operation |leptris vs nokogiri |Notes
+
+|parse
+|3.27x
+|Engine pool-carved parse
+
+|serialize
+|3.38x
+|
+
+|XPath
+|50x
+|Native compiled-eval seams
+
+|walk + read (`.text`, `[]`)
+|1.90x
+|C klass-mint wrappers + wrapper read memos
+
+|warm re-read
+|2.10x
+|The memo family — zero string allocation per node
+
+|build (`add_element`)
+|4.24x
+|C-mint construction + bulk attributes
+
+|attribute pairs
+|1.73x
+|One C crossing, no Attribute objects
+|===
+
 === Fast paths and diagnostics
 
+
 Read-only attribute listing without Attribute node objects — the
 walk-hot shape when only names and values are needed (leptris
 answers in one C crossing; document order, duplicates included):

data/lib/moxml/element.rb

@@ -122,8 +122,14 @@ module Moxml
     # per-attribute objects. Names may be shared frozen strings —
     # dup before mutating. Mutation and per-attribute namespace
     # resolution stay on #attributes.
+    # One C crossing per read re-allocates the row set (moxml#329);
+    # memoized on the attribute-value clock — element-side value
+    # writes bump it (see invalidate_attribute_value_cache!).
     def attribute_pairs
-      adapter.attribute_pairs(@native)
+      generation = attribute_value_generation
+      @pairs = nil if @pairs_gen != generation
+      @pairs_gen = generation
+      @attribute_pairs ||= adapter.attribute_pairs(@native)
     end
 
     # Create an element with attributes and attach it under this
@@ -248,9 +254,18 @@ module Moxml
     alias add_namespace_definition add_namespace
 
     # it's NOT the same as namespaces.first
+    # Per-read namespace resolution pays an FFI fetch plus a wrapper
+    # mint on every call (moxml#329: a top attribution site in the
+    # fallback-path census). Memoized on the namespace-scope clock —
+    # every scope mutation bumps it (see invalidate_namespace_cache!).
     def namespace
-      ns = adapter.namespace(@native)
-      ns && Wrappers::Namespace.new(ns, context)
+      generation = context.namespace_scope_generation
+      @namespace = nil if @namespace_gen != generation
+      @namespace_gen = generation
+      @namespace ||= begin
+        ns = adapter.namespace(@native)
+        ns && Wrappers::Namespace.new(ns, context)
+      end
     end
 
     # add the prefix to the element name

data/lib/moxml/materializer.rb

@@ -72,7 +72,11 @@ module Moxml
       def group_flat(flat, stride)
         return EMPTY_ATTRIBUTES if flat.empty?
 
-        flat.each_slice(stride).to_a
+        # Direct indexing over each_slice: no enumerator chain per
+        # record (moxml#329's slice-churn row — the enumerator
+        # allocation dominated the sub-array cost on deep trees).
+        count = flat.size / stride
+        Array.new(count) { |i| flat[i * stride, stride] }
       end
     end
 

data/lib/moxml/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Moxml
-  VERSION = "0.5.107"
+  VERSION = "0.5.109"
 end

data/lib/moxml/xpath/lexer.rb

@@ -155,6 +155,19 @@ module Moxml
 
       private
 
+      # After an E/e: a digit, or a sign followed by a digit. Anything
+      # else leaves the E as part of a name (1o vs 1e ambiguity guard).
+      def lookahead_after_e_is_digit_sign?
+        next_char = @expression[@position + 1]
+        return true if /\d/.match?(next_char)
+
+        if /[+-]/.match?(next_char)
+          return @expression[@position + 2]&.match?(/\d/) || false
+        end
+
+        false
+      end
+
       # Get current character
       #
       # @return [String, nil] Current character or nil if at end
@@ -259,6 +272,22 @@ module Moxml
           end
         end
 
+        # Exponent part (libxml2/nokogiri extension): digits [Ee] [+|-]
+        # digits. Without it, 1E+1 tokenized as 1E followed by + 1 and
+        # predicates raised "Expected ']' after predicate".
+        if @position < @length && current_char =~ /[Ee]/ &&
+            @position + 1 < @length && lookahead_after_e_is_digit_sign?
+          value += current_char
+          advance
+          value += current_char if /[+-]/.match?(current_char)
+          advance if /[+-]/.match?(current_char)
+
+          while @position < @length && current_char =~ /\d/
+            value += current_char
+            advance
+          end
+        end
+
         add_token(:number, value, start_pos)
       end
 

data/lib/moxml/xpath/parser.rb

@@ -451,8 +451,10 @@ module Moxml
         elsif match?(:number)
           value = current_value
           advance
-          # Convert string to actual numeric value
-          numeric_value = value.include?(".") ? value.to_f : value.to_i
+          # Convert string to actual numeric value. Exponent
+          # literals must go through to_f — to_i truncates at the E
+          # ("1E+1" became 1).
+          numeric_value = value.match?(/[.Ee]/) ? value.to_f : value.to_i
           return AST::Node.number(numeric_value)
         elsif match?(:dollar)
           advance

data/spec/moxml/xpath/number_literals_spec.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+# Signed exponent literals (libxml2/nokogiri extension): the lexer
+# consumed exponent digits but not their sign, so 1E+1 tokenized as
+# 1E + 1 and predicates raised "Expected ']'"; and the parser's
+# numeric conversion ran to_i on dotless literals, so 2E3 evaluated
+# as 2. Both fixed; the engine now matches nokogiri on every shape.
+RSpec.describe "XPath exponent number literals" do
+  %i[nokogiri ox].each do |adapter|
+    it "evaluates signed exponents on #{adapter}" do
+      ctx = Moxml.new(adapter)
+      doc = ctx.parse("<r><x v='2000'/><x v='5'/></r>")
+      expect(doc.xpath("//x[@v > 1E+1]").size).to eq(1)
+      expect(doc.xpath("//x[@v = 2E3]").size).to eq(1)
+      expect(doc.xpath("//x[@v > 5.2E-3]").size).to eq(2)
+      expect(doc.xpath("//x[@v < 1e1]").size).to eq(1)
+    end
+  end
+
+  it "tokenizes exponent literals as one number" do
+    tokens = Moxml::XPath::Lexer.new("1E+1 2E3 5.2E-3 1e2").tokenize
+    numbers = tokens.select { |t| t.first == :number }.map { |t| t[1] }
+    expect(numbers).to eq(["1E+1", "2E3", "5.2E-3", "1e2"])
+  end
+
+  it "parses exponent literals at full precision" do
+    ast = Moxml::XPath::Parser.parse("1E+1")
+    literal = ast
+    literal = literal.children.first while literal.children.any? { |c| c.respond_to?(:type) }
+    expect(literal.value).to eq(10.0)
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: moxml
 version: !ruby/object:Gem::Version
-  version: 0.5.107
+  version: 0.5.109
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-10-05 00:00:00.000000000 Z
+date: 2026-10-06 00:00:00.000000000 Z
 dependencies: []
 description: |
   Moxml is a unified XML manipulation library that provides a common API
@@ -557,6 +557,7 @@ files:
 - spec/moxml/xpath/functions/special_functions_spec.rb
 - spec/moxml/xpath/functions/string_functions_spec.rb
 - spec/moxml/xpath/lexer_spec.rb
+- spec/moxml/xpath/number_literals_spec.rb
 - spec/moxml/xpath/parser_integration_spec.rb
 - spec/moxml/xpath/parser_spec.rb
 - spec/moxml/xpath/ruby/generator_spec.rb