rubygem-nokogiri/CVE-2020-26247.patch

From 9c87439d9afa14a365ff13e73adc809cb2c3d97b Mon Sep 17 00:00:00 2001
From: Mike Dalessio <mike.dalessio@gmail.com>
Date: Mon, 23 Nov 2020 00:47:02 -0500
Subject: [PATCH] feat: XML::Schema and RelaxNG creation accept optional
 ParseOptions

I'm trying out a new pattern, which is that the parsed object carries
around the ParseOptions it was created with, which should make some
testing a bit easier.

I'm also not implementing the "config block" pattern in use for
Documents, because I think the UX is weird and I'm hoping to change
everything to use kwargs in a 2.0 release, anyway.
---
 ext/nokogiri/xml_relax_ng.c       | 39 ++++++++++++++++++--------
 ext/nokogiri/xml_schema.c         | 46 +++++++++++++++++++++++--------
 lib/nokogiri/xml/parse_options.rb |  2 ++
 lib/nokogiri/xml/relax_ng.rb      |  4 +--
 lib/nokogiri/xml/schema.rb        | 10 ++++---
 5 files changed, 72 insertions(+), 29 deletions(-)

diff --git a/ext/nokogiri/xml_relax_ng.c b/ext/nokogiri/xml_relax_ng.c
index e17b11a..f361d27 100644
--- a/ext/nokogiri/xml_relax_ng.c
+++ b/ext/nokogiri/xml_relax_ng.c
@@ -53,16 +53,24 @@ static VALUE validate_document(VALUE self, VALUE document)
  *
  * Create a new RelaxNG from the contents of +string+
  */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE read_memory(int argc, VALUE *argv, VALUE klass)
 {
-  xmlRelaxNGParserCtxtPtr ctx = xmlRelaxNGNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE content;
+  VALUE parse_options;
+  xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
-  VALUE errors = rb_ary_new();
+  VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
+  ctx = xmlRelaxNGNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
   xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
@@ -90,6 +98,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -100,18 +109,25 @@ static VALUE read_memory(VALUE klass, VALUE content)
  *
  * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE from_document(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE document;
+  VALUE parse_options;
   xmlDocPtr doc;
   xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
   VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
 
   Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
   ctx = xmlRelaxNGNewDocParserCtxt(doc);
 
@@ -142,6 +158,7 @@ static VALUE from_document(VALUE klass, VALUE document)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -155,7 +172,7 @@ void init_xml_relax_ng()
 
   cNokogiriXmlRelaxNG = klass;
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  rb_define_singleton_method(klass, "read_memory", read_memory, -1);
+  rb_define_singleton_method(klass, "from_document", from_document, -1);
   rb_define_private_method(klass, "validate_document", validate_document, 1);
 }
diff --git a/ext/nokogiri/xml_schema.c b/ext/nokogiri/xml_schema.c
index 439f721..ea7c3d3 100644
--- a/ext/nokogiri/xml_schema.c
+++ b/ext/nokogiri/xml_schema.c
@@ -93,15 +93,26 @@ static VALUE validate_file(VALUE self, VALUE rb_filename)
  *
  * Create a new Schema from the contents of +string+
  */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE read_memory(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE content;
+  VALUE parse_options;
+  int parse_options_int;
+  xmlSchemaParserCtxtPtr ctx;
   xmlSchemaPtr schema;
-  xmlSchemaParserCtxtPtr ctx = xmlSchemaNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE errors;
   VALUE rb_schema;
-  VALUE errors = rb_ary_new();
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+  parse_options_int = (int)NUM2INT(rb_funcall(parse_options, rb_intern("to_i"), 0));
+
+  ctx = xmlSchemaNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
   xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
@@ -109,7 +120,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
     ctx,
     Nokogiri_error_array_pusher,
     (void *)errors
-  );
+    );
 #endif
 
    schema = xmlSchemaParse(ctx);
@@ -129,6 +140,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -164,18 +176,27 @@ static int has_blank_nodes_p(VALUE cache)
  *
  * Create a new Schema from the Nokogiri::XML::Document +doc+
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE from_document(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE document;
+  VALUE parse_options;
+  int parse_options_int;
   xmlDocPtr doc;
   xmlSchemaParserCtxtPtr ctx;
   xmlSchemaPtr schema;
   VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
 
   Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+  parse_options_int = (int)NUM2INT(rb_funcall(parse_options, rb_intern("to_i"), 0));
 
   if (has_blank_nodes_p(DOC_NODE_CACHE(doc))) {
     rb_raise(rb_eArgError, "Creating a schema from a document that has blank nodes exposed to Ruby is dangerous");
@@ -211,6 +232,7 @@ static VALUE from_document(VALUE klass, VALUE document)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 
@@ -226,8 +248,8 @@ void init_xml_schema()
 
   cNokogiriXmlSchema = klass;
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  rb_define_singleton_method(klass, "read_memory", read_memory, -1);
+  rb_define_singleton_method(klass, "from_document", from_document, -1);
 
   rb_define_private_method(klass, "validate_document", validate_document, 1);
   rb_define_private_method(klass, "validate_file",     validate_file, 1);
diff --git a/lib/nokogiri/xml/parse_options.rb b/lib/nokogiri/xml/parse_options.rb
index 8969578..c6d3d1c 100644
--- a/lib/nokogiri/xml/parse_options.rb
+++ b/lib/nokogiri/xml/parse_options.rb
@@ -72,6 +72,8 @@ module Nokogiri
       DEFAULT_XML  = RECOVER | NONET
       # the default options used for parsing HTML documents
       DEFAULT_HTML = RECOVER | NOERROR | NOWARNING | NONET
+      # the default options used for parsing XML schemas
+      DEFAULT_SCHEMA = NONET
 
       attr_accessor :options
       def initialize options = STRICT
diff --git a/lib/nokogiri/xml/relax_ng.rb b/lib/nokogiri/xml/relax_ng.rb
index 5a645a4..79bc30c 100644
--- a/lib/nokogiri/xml/relax_ng.rb
+++ b/lib/nokogiri/xml/relax_ng.rb
@@ -4,8 +4,8 @@ module Nokogiri
       ###
       # Create a new Nokogiri::XML::RelaxNG document from +string_or_io+.
       # See Nokogiri::XML::RelaxNG for an example.
-      def RelaxNG string_or_io
-        RelaxNG.new(string_or_io)
+      def RelaxNG(string_or_io, options = ParseOptions::DEFAULT_SCHEMA)
+        RelaxNG.new(string_or_io, options)
       end
     end
 
diff --git a/lib/nokogiri/xml/schema.rb b/lib/nokogiri/xml/schema.rb
index 65a7bcd..a88f69c 100644
--- a/lib/nokogiri/xml/schema.rb
+++ b/lib/nokogiri/xml/schema.rb
@@ -4,8 +4,8 @@ module Nokogiri
       ###
       # Create a new Nokogiri::XML::Schema object using a +string_or_io+
       # object.
-      def Schema string_or_io
-        Schema.new(string_or_io)
+      def Schema(string_or_io, options = ParseOptions::DEFAULT_SCHEMA)
+        Schema.new(string_or_io, options)
       end
     end
 
@@ -29,12 +29,14 @@ module Nokogiri
     class Schema
       # Errors while parsing the schema file
       attr_accessor :errors
+      # The Nokogiri::XML::ParseOptions used to parse the schema
+      attr_accessor :parse_options
 
       ###
       # Create a new Nokogiri::XML::Schema object using a +string_or_io+
       # object.
-      def self.new string_or_io
-        from_document Nokogiri::XML(string_or_io)
+      def self.new string_or_io, options = ParseOptions::DEFAULT_SCHEMA
+        from_document(Nokogiri::XML(string_or_io), options)
       end
 
       ###
-- 
2.23.0