Merge pull request #65 from missive/max-parse-errors

Add :max_parse_errors argument to .parse, .get and .fragment with 0 as default value
rubys · Jan 25, 2018 · 938dd3f · 938dd3f
2 parents a3fc29b + c7397aa
commit 938dd3f
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 17 deletions.
diff --git a/ext/nokogumboc/nokogumbo.c b/ext/nokogumboc/nokogumbo.c
@@ -184,11 +184,14 @@ static xmlNodePtr walk_tree(xmlDocPtr document, GumboNode *node) {
 }
 
 // Parse a string using gumbo_parse into a Nokogiri document
-static VALUE parse(VALUE self, VALUE string) {
-  const GumboOptions *options = &kGumboDefaultOptions;
+static VALUE parse(VALUE self, VALUE string, VALUE max_parse_errors) {
+  GumboOptions options;
+  memcpy(&options, &kGumboDefaultOptions, sizeof options);
+  options.max_errors = NUM2INT(max_parse_errors);
+
   const char *input = RSTRING_PTR(string);
   size_t input_len = RSTRING_LEN(string);
-  GumboOutput *output = gumbo_parse_with_options(options, input, input_len);
+  GumboOutput *output = gumbo_parse_with_options(&options, input, input_len);
   xmlDocPtr doc = xmlNewDoc(CONST_CAST "1.0");
 #ifdef NGLIB
   doc->type = XML_HTML_DOCUMENT_NODE;
@@ -219,7 +222,7 @@ static VALUE parse(VALUE self, VALUE string) {
   // Add parse errors to rdoc.
   if (output->errors.length) {
     GumboVector *errors = &output->errors;
-    GumboParser parser = { ._options = options };
+    GumboParser parser = { ._options = &options };
     GumboStringBuffer msg;
     VALUE rerrors = rb_ary_new2(errors->length);
 
@@ -253,7 +256,7 @@ static VALUE parse(VALUE self, VALUE string) {
     gumbo_string_buffer_destroy(&parser, &msg);
   }
 
-  gumbo_destroy_output(options, output);
+  gumbo_destroy_output(&options, output);
 
   return rdoc;
 }
@@ -288,5 +291,5 @@ void Init_nokogumboc() {
 
   // define Nokogumbo class with a singleton parse method
   VALUE Gumbo = rb_define_class("Nokogumbo", rb_cObject);
-  rb_define_singleton_method(Gumbo, "parse", parse, 1);
+  rb_define_singleton_method(Gumbo, "parse", parse, 2);
 }
diff --git a/lib/nokogumbo.rb b/lib/nokogumbo.rb
@@ -4,14 +4,14 @@
 module Nokogiri
   # Parse an HTML document.  +string+ contains the document.  +string+
   # may also be an IO-like object.  Returns a +Nokogiri::HTML::Document+.
-  def self.HTML5(string)
-    Nokogiri::HTML5.parse(string)
+  def self.HTML5(*args)
+    Nokogiri::HTML5.parse(*args)
   end
 
   module HTML5
     # Parse an HTML document.  +string+ contains the document.  +string+
     # may also be an IO-like object.  Returns a +Nokogiri::HTML::Document+.
-    def self.parse(string)
+    def self.parse(string, options={})
       if string.respond_to? :read
         string = string.read
       end
@@ -21,7 +21,7 @@ def self.parse(string)
         string = reencode(string)
       end
 
-      Nokogumbo.parse(string.to_s)
+      Nokogumbo.parse(string.to_s, options[:max_parse_errors] || 0)
     end
 
     # Fetch and parse a HTML document from the web, following redirects,
@@ -67,7 +67,7 @@ def self.get(uri, options={})
 
       case response
       when Net::HTTPSuccess
-        doc = parse(reencode(response.body, response['content-type']))
+        doc = parse(reencode(response.body, response['content-type']), options)
         doc.instance_variable_set('@response', response)
         doc.class.send(:attr_reader, :response)
         doc
@@ -83,8 +83,8 @@ def self.get(uri, options={})
     # while fragment is on the Gumbo TODO list, simulate it by doing
     # a full document parse and ignoring the parent <html>, <head>, and <body>
     # tags, and collecting up the children of each.
-    def self.fragment(string)
-      doc = parse(string)
+    def self.fragment(*args)
+      doc = parse(*args)
       fragment = Nokogiri::HTML::DocumentFragment.new(doc)
 
       if doc.children.length != 1 or doc.children.first.name != 'html'

diff --git a/test-nokogumbo.rb b/test-nokogumbo.rb
@@ -78,7 +78,7 @@ def test_bogus_encoding
   end
 
   def test_html5_doctype
-    doc = Nokogumbo.parse("<!DOCTYPE html><html></html>")
+    doc = Nokogiri::HTML5.parse("<!DOCTYPE html><html></html>")
     assert_match /<!DOCTYPE html>/, doc.to_html
   end
 
@@ -126,17 +126,43 @@ def test_root_comments
   end
 
   def test_parse_errors
-    doc = Nokogiri::HTML5("<!DOCTYPE html><html><!-- -- --></a>")
+    doc = Nokogiri::HTML5("<!DOCTYPE html><html><!-- -- --></a>", max_parse_errors: 10)
     assert_equal doc.errors.length, 2
-    doc = Nokogiri::HTML5("<!DOCTYPE html><html>")
+    doc = Nokogiri::HTML5("<!DOCTYPE html><html>", max_parse_errors: 10)
     assert_empty doc.errors
   end
 
+  def test_max_parse_errors
+    # This document contains 2 parse errors, but we force limit to 1.
+    doc = Nokogiri::HTML5("<!DOCTYPE html><html><!-- -- --></a>", max_parse_errors: 1)
+    assert_equal 1, doc.errors.length
+    doc = Nokogiri::HTML5("<!DOCTYPE html><html>", max_parse_errors: 1)
+    assert_empty doc.errors
+  end
+
+  def test_default_max_parse_errors
+    # This document contains 200 parse errors, but default limit is 0.
+    doc = Nokogiri::HTML5("<!DOCTYPE html><html>" + "</p>" * 200)
+    assert_equal 0, doc.errors.length
+  end
+
   def test_parse_fragment_errors
-    doc = Nokogiri::HTML5.fragment("<\r\n")
+    doc = Nokogiri::HTML5.fragment("<\r\n", max_parse_errors: 10)
     refute_empty doc.errors
   end
 
+  def test_fragment_max_parse_errors
+    # This fragment contains 3 parse errors, but we force limit to 1.
+    doc = Nokogiri::HTML5.fragment("<!-- -- --></a>", max_parse_errors: 1)
+    assert_equal 1, doc.errors.length
+  end
+
+  def test_fragment_default_max_parse_errors
+    # This fragment contains 201 parse errors, but default limit is 0.
+    doc = Nokogiri::HTML5.fragment("</p>" * 200)
+    assert_equal 0, doc.errors.length
+  end
+
 private
 
   def buffer