Gumbo common options (#3205)

**What problem is this PR intended to solve?**

See discussion at #3199 

cc @stevecheckoway 

(Recreation of #3204 with additional commits)

ext/nokogiri/gumbo.c

@@ -316,6 +316,35 @@ parse_cleanup(VALUE parse_args)
   return Qnil;
 }
 
+// Scan the keyword arguments for options common to the document and fragment
+// parse.
+static GumboOptions
+common_options(VALUE kwargs) {
+  // The order of the keywords determines the order of the values below.
+  // If this order is changed, then setting the options below must change as
+  // well.
+  ID keywords[] = {
+    rb_intern_const("max_attributes"),
+    rb_intern_const("max_errors"),
+    rb_intern_const("max_tree_depth"),
+  };
+  VALUE values[sizeof keywords / sizeof keywords[0]];
+
+  // Extract the values coresponding to the required keywords. Raise an error
+  // if required arguments are missing.
+  rb_get_kwargs(kwargs, keywords, 3, 0, values);
+
+  GumboOptions options = kGumboDefaultOptions;
+  options.max_attributes = NUM2INT(values[0]);
+  options.max_errors = NUM2INT(values[1]);
+
+  // handle negative values
+  int depth = NUM2INT(values[2]);
+  options.max_tree_depth = depth < 0 ? UINT_MAX : (unsigned int)depth;
+
+  return options;
+}
+
 static VALUE parse_continue(VALUE parse_args);
 
 /*
@@ -331,10 +360,7 @@ rb_gumbo_s_parse(int argc, VALUE *argv, VALUE _self)
     kwargs = rb_hash_new();
   }
 
-  GumboOptions options = kGumboDefaultOptions;
-  options.max_attributes = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_attributes"))));
-  options.max_errors = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_errors"))));
-  options.max_tree_depth = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_tree_depth"))));
+  GumboOptions options = common_options(kwargs);
 
   GumboOutput *output = perform_parse(&options, input);
   ParseArgs args = {
@@ -440,6 +466,8 @@ rb_gumbo_s_fragment(int argc, VALUE *argv, VALUE _self)
     kwargs = rb_hash_new();
   }
 
+  GumboOptions options = common_options(kwargs);
+
   if (NIL_P(ctx)) {
     ctx_tag = "body";
     ctx_ns = GUMBO_NAMESPACE_HTML;
@@ -543,20 +571,15 @@ rb_gumbo_s_fragment(int argc, VALUE *argv, VALUE _self)
   }
 
   // Perform a fragment parse.
-  GumboOptions options = kGumboDefaultOptions;
-  options.max_attributes = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_attributes"))));
-  options.max_errors = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_errors"))));
-
-  // Add one to account for the HTML element.
-  int depth = NUM2INT(rb_hash_aref(kwargs, ID2SYM(rb_intern_const("max_tree_depth"))));
-  options.max_tree_depth = depth < 0 ? -1 : (depth + 1);
-
   options.fragment_context = ctx_tag;
   options.fragment_namespace = ctx_ns;
   options.fragment_encoding = encoding;
   options.quirks_mode = quirks_mode;
   options.fragment_context_has_form_ancestor = form;
 
+  // Add one to the max tree depth to account for the HTML element.
+  if (options.max_tree_depth < UINT_MAX) { options.max_tree_depth++; }
+
   GumboOutput *output = perform_parse(&options, tags);
   ParseArgs args = {
     .output = output,

lib/nokogiri/html5/document.rb

@@ -119,7 +119,7 @@ def do_parse(string_or_io, url, encoding, **options)
           string = HTML5.read_and_encode(string_or_io, encoding)
 
           options[:max_attributes] ||= Nokogiri::Gumbo::DEFAULT_MAX_ATTRIBUTES
-          options[:max_errors] ||= options[:max_parse_errors] || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
+          options[:max_errors] ||= options.delete(:max_parse_errors) || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
           options[:max_tree_depth] ||= Nokogiri::Gumbo::DEFAULT_MAX_TREE_DEPTH
 
           doc = Nokogiri::Gumbo.parse(string, url, self, **options)

lib/nokogiri/html5/document_fragment.rb

@@ -44,7 +44,7 @@ def initialize(doc, tags = nil, ctx = nil, options = {}) # rubocop:disable Lint/
         tags = Nokogiri::HTML5.read_and_encode(tags, nil)
 
         options[:max_attributes] ||= Nokogiri::Gumbo::DEFAULT_MAX_ATTRIBUTES
-        options[:max_errors] ||= options[:max_parse_errors] || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
+        options[:max_errors] ||= options.delete(:max_parse_errors) || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
         options[:max_tree_depth] ||= Nokogiri::Gumbo::DEFAULT_MAX_TREE_DEPTH
 
         Nokogiri::Gumbo.fragment(self, tags, ctx, **options)

test/html5/test_encoding.rb

@@ -189,7 +189,7 @@ def round_trip_through(str, enc)
     define_method("test_parse_encoded_#{enc[0]}".to_sym) do
       html = "<!DOCTYPE html><span>#{enc[1]}</span>"
       encoded_html = round_trip_through(html, enc[0])
-      doc = Nokogiri::HTML5(encoded_html, encoding: enc[0])
+      doc = Nokogiri::HTML5(encoded_html, enc[0])
       span = doc.at("/html/body/span")
       refute_nil span
       assert_equal enc[1], span.content
@@ -210,7 +210,7 @@ def round_trip_through(str, enc)
       skip "https://bugs.ruby-lang.org/issues/15033" if enc[0] == "ISO-2022-JP"
       round_trip_through(enc[1], enc[0])
       encoded = encodings_doc.serialize(encoding: enc[0])
-      doc = Nokogiri::HTML5(encoded, encoding: enc[0])
+      doc = Nokogiri::HTML5(encoded, enc[0])
       assert_equal encodings_html, doc.serialize
     end
   end

test/html5/test_nokogumbo.rb

@@ -268,6 +268,7 @@ def test_max_depth_parse
     end
 
     assert(Nokogiri::HTML5(html, max_tree_depth: depth))
+    assert(Nokogiri::HTML5(html, max_tree_depth: -1))
   end
 
   def test_max_depth_fragment
@@ -278,6 +279,7 @@ def test_max_depth_fragment
     end
 
     assert(Nokogiri::HTML5.fragment(html, max_tree_depth: depth))
+    assert(Nokogiri::HTML5.fragment(html, max_tree_depth: -1))
   end
 
   def test_document_encoding