Limited encoding (#368)

* WIP

* Format and test limited encoding

* Format and test limited encoding

* Prepare release

* Fix rubocop complaints

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All changes to the Ox gem are documented here. Releases follow semantic versioning.
 
+## [2.14.20] - 2025-01-12
+
+### Fixed
+
+- The instruction encoding attribute is now used to set the encoding in the `:limited` mode.
+
 ## [2.14.19] - 2024-12-25
 
 ### Fixed

ext/ox/parse.c

@@ -342,6 +342,15 @@ static void read_instruction(PInfo pi) {
         } else {
             pi->pcb->instruct(pi, target, attrs.head, content_ptr);
         }
+    } else {
+        for (Attr a = attrs.head; a < attrs.tail; a++) {
+            if (0 == strcasecmp(a->name, "encoding")) {
+                strncpy(pi->options->encoding, a->value, sizeof(pi->options->encoding) - 1);
+                pi->options->encoding[sizeof(pi->options->encoding) - 1] = '\0';
+                pi->options->rb_enc                                      = rb_enc_find(a->value);
+                break;
+            }
+        }
     }
     attr_stack_cleanup(&attrs);
     if (content_ptr != content) {

lib/ox/version.rb

@@ -1,4 +1,4 @@
 module Ox
   # Current version of the module.
-  VERSION = '2.14.19'
+  VERSION = '2.14.20'
 end

test/perf_mars.rb

@@ -13,10 +13,10 @@
 require 'optparse'
 require 'ox'
 
-it = 5000
+iter = 5000
 
 opts = OptionParser.new
-opts.on('-i', '--iterations [Int]', Integer, 'iterations')    { |i| it = i }
+opts.on('-i', '--iterations [Int]', Integer, 'iterations')    { |i| iter = i }
 opts.on('-h', '--help', 'Show this display')                  { puts opts; Process.exit!(0) }
 files = opts.parse(ARGV)
 
@@ -80,28 +80,28 @@ def initialize(v=[])
   # pp a
   # puts xml
   start = Time.now
-  (1..it).each do
+  (1..iter).each do
     obj = Ox.load(xml, mode: :object)
     # pp obj
   end
   ox_load_time = Time.now - start
 
   m = Marshal.dump(a)
   start = Time.now
-  (1..it).each do
+  (1..iter).each do
     obj = Marshal.load(m)
   end
   mars_load_time = Time.now - start
 
   obj = Ox.load(xml, mode: :object)
   start = Time.now
-  (1..it).each do
+  (1..iter).each do
     xml = Ox.dump(a, indent: -1)
   end
   ox_dump_time = Time.now - start
 
   start = Time.now
-  (1..it).each do
+  (1..iter).each do
     m = Marshal.dump(a)
   end
   mars_dump_time = Time.now - start

test/tests.rb

@@ -1932,6 +1932,13 @@ def dump_and_load(obj, trace=false, circular=false)
     end
     loaded
   end
+
+  def test_limit_encoding
+    Ox.default_options = $ox_object_options
+    xml = '<?xml version="1.0" encoding="UTF-8"?><doc><name>Martin</name></doc>'.encode('ASCII-8BIT')
+    enc = Ox.load(xml, mode: :limited).locate('name').first.text.encoding
+    assert_equal('UTF-8', enc.to_s)
+  end
 end
 
 class Bag