Make available the XPath path of the node where a SyntaxError occurre…

…d during Schema validation (#3316)

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

When validating an XML document with a schema it is difficult to
impossible to know where the error happened in the document. With the
path on the error, we can identify where in the document the error
occurred.

**Have you included adequate test coverage?**

I believe so. I updated the test in the schema to check to see if path
errors were being returned correctly.

**Does this change affect the behavior of either the C or the Java
implementations?**

This change the C implementation gently. Setting the path during error
creation.
I looked if xerces-j could do it, but I couldn't find anything on the
SAXException that would help.

---------

Co-authored-by: Mike Dalessio <mike.dalessio@gmail.com>

ext/nokogiri/xml_syntax_error.c

@@ -44,6 +44,7 @@ noko__error_raise(void *ctx, xmlErrorConstPtr error)
 VALUE
 noko_xml_syntax_error__wrap(xmlErrorConstPtr error)
 {
+  xmlChar *c_path ;
   VALUE msg, e, klass;
 
   klass = cNokogiriXmlSyntaxError;
@@ -61,16 +62,21 @@ noko_xml_syntax_error__wrap(xmlErrorConstPtr error)
       );
 
   if (error) {
+    c_path = xmlGetNodePath(error->node);
+
     rb_iv_set(e, "@domain", INT2NUM(error->domain));
     rb_iv_set(e, "@code", INT2NUM(error->code));
     rb_iv_set(e, "@level", INT2NUM((short)error->level));
     rb_iv_set(e, "@file", RBSTR_OR_QNIL(error->file));
     rb_iv_set(e, "@line", INT2NUM(error->line));
+    rb_iv_set(e, "@path", RBSTR_OR_QNIL(c_path));
     rb_iv_set(e, "@str1", RBSTR_OR_QNIL(error->str1));
     rb_iv_set(e, "@str2", RBSTR_OR_QNIL(error->str2));
     rb_iv_set(e, "@str3", RBSTR_OR_QNIL(error->str3));
     rb_iv_set(e, "@int1", INT2NUM(error->int1));
     rb_iv_set(e, "@column", INT2NUM(error->int2));
+
+    xmlFree(c_path);
   }
 
   return e;

lib/nokogiri/xml/syntax_error.rb

@@ -24,6 +24,15 @@ def aggregate(errors)
       attr_reader :level
       attr_reader :file
       attr_reader :line
+
+      # The XPath path of the node that caused the error when validating a `Nokogiri::XML::Document`.
+      #
+      # This attribute will only be non-nil when the error is emitted by `Schema#validate` on
+      # Document objects. It will return `nil` for DOM parsing errors and for errors emitted during
+      # Schema validation of files.
+      #
+      # ⚠ `#path` is not supported on JRuby, where it will always return `nil`.
+      attr_reader :path
       attr_reader :str1
       attr_reader :str2
       attr_reader :str3

test/html4/test_document.rb

@@ -813,6 +813,7 @@ def test_silencing_nonparse_errors_during_attribute_insertion_1262
                   Nokogiri::HTML4.parse(input, nil, nil, parse_options)
                 end
                 assert_match(/Parser without recover option encountered error or warning/, exception.to_s)
+                assert_nil(exception.path)
               end
             end
 
@@ -835,6 +836,7 @@ def test_silencing_nonparse_errors_during_attribute_insertion_1262
                   Nokogiri::HTML4.parse(input, nil, "UTF-8", parse_options)
                 end
                 assert_match(/Parser without recover option encountered error or warning/, exception.to_s)
+                assert_nil(exception.path)
               end
             end
 

test/xml/test_document.rb

@@ -1087,9 +1087,10 @@ def test_can_be_closed
               let(:parse_options) { xml_strict }
 
               it "raises exception on parse error" do
-                assert_raises Nokogiri::SyntaxError do
+                error = assert_raises Nokogiri::SyntaxError do
                   Nokogiri::XML.parse(input, nil, nil, parse_options)
                 end
+                assert_nil(error.path)
               end
             end
 

test/xml/test_schema.rb

@@ -159,6 +159,17 @@ class TestNokogiriXMLSchema < Nokogiri::TestCase
 
         assert(errors = xsd.validate(doc))
         assert_equal(2, errors.length)
+        if Nokogiri.uses_libxml?
+          assert_equal(
+            ["/purchaseOrder/billTo/state", "/purchaseOrder/shipTo/state"],
+            errors.map(&:path).sort,
+          )
+        else
+          assert_equal(
+            [nil, nil],
+            errors.map(&:path).sort,
+          )
+        end
       end
 
       it "validate_invalid_file" do
@@ -171,6 +182,10 @@ class TestNokogiriXMLSchema < Nokogiri::TestCase
 
         assert(errors = xsd.validate(tempfile.path))
         assert_equal(2, errors.length)
+        assert_equal(
+          [nil, nil],
+          errors.map(&:path).sort,
+        )
       end
 
       it "validate_non_document" do

test/xml/test_syntax_error.rb

@@ -58,5 +58,6 @@
       assert_nil error.column
       assert_nil error.level
     end
+    assert_nil error.path
   end
 end