Upstream truffleruby patches needed when using system libraries

.github/workflows/ci.yml

@@ -279,6 +279,30 @@ jobs:
       - run: bundle exec rake test
       - run: bundle exec rake test:bench
 
+  truffleruby-head:
+    needs: ["basic"]
+    strategy:
+      fail-fast: false
+      matrix:
+        flags:
+          - "--disable-system-libraries --disable-static"
+          - "--disable-system-libraries --enable-static"
+          - "--enable-system-libraries"
+    runs-on: ubuntu-latest
+    container:
+      image: ghcr.io/sparklemotion/nokogiri-test:truffle-nightly
+    steps:
+      - uses: actions/checkout@v2
+        with:
+          submodules: true
+      - uses: actions/cache@v2
+        with:
+          path: ports/archives
+          key: tarballs-ubuntu-${{hashFiles('dependencies.yml', 'patches/**/*.patch')}}
+      - run: bundle install --local || bundle install
+      - run: bundle exec rake compile -- ${{matrix.flags}}
+      - run: bundle exec rake test
+
   bsd:
     continue-on-error: true # we're seeing VMs hang and fail the whole workflow
     needs: ["basic"]

ext/nokogiri/nokogiri.h

@@ -99,6 +99,9 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 #  endif
 #endif
 
+#if defined(TRUFFLERUBY) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
+#  define TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+#endif
 
 NOKOPUBVAR VALUE mNokogiri ;
 NOKOPUBVAR VALUE mNokogiriGumbo ;

ext/nokogiri/xml_sax_parser.c

@@ -203,10 +203,16 @@ warning_func(void *ctx, const char *msg, ...)
   VALUE doc = rb_iv_get(self, "@document");
   VALUE rb_message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  rb_message = rb_sprintf("warning_func: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
   rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
   rb_funcall(doc, id_warning, 1, rb_message);
 }
@@ -219,10 +225,16 @@ error_func(void *ctx, const char *msg, ...)
   VALUE doc = rb_iv_get(self, "@document");
   VALUE rb_message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  rb_message = rb_sprintf("error_func: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
   rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
   rb_funcall(doc, id_error, 1, rb_message);
 }

ext/nokogiri/xml_xpath_context.c

@@ -298,10 +298,16 @@ xpath_generic_exception_handler(void *ctx, const char *msg, ...)
 {
   VALUE rb_message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  rb_message = rb_sprintf("xpath_generic_exception_handler: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
   rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
   rb_exc_raise(rb_exc_new3(rb_eRuntimeError, rb_message));
 }
@@ -336,7 +342,12 @@ evaluate(int argc, VALUE *argv, VALUE self)
   }
 
   xmlResetLastError();
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  VALUE errors = rb_ary_new();
+  xmlSetStructuredErrorFunc(errors, Nokogiri_error_array_pusher);
+#else
   xmlSetStructuredErrorFunc(NULL, Nokogiri_error_raise);
+#endif
 
   /* For some reason, xmlXPathEvalExpression will blow up with a generic error */
   /* when there is a non existent function. */
@@ -346,6 +357,12 @@ evaluate(int argc, VALUE *argv, VALUE self)
   xmlSetStructuredErrorFunc(NULL, NULL);
   xmlSetGenericErrorFunc(NULL, NULL);
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  if (RARRAY_LEN(errors) > 0) {
+    rb_exc_raise(rb_ary_entry(errors, 0));
+  }
+#endif
+
   if (xpath == NULL) {
     xmlErrorPtr error = xmlGetLastError();
     rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));

ext/nokogiri/xslt_stylesheet.c

@@ -26,10 +26,16 @@ xslt_generic_error_handler(void *ctx, const char *msg, ...)
 {
   VALUE message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  message = rb_sprintf("xslt_generic_error_handler: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
   message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
   rb_str_concat((VALUE)ctx, message);
 }

test/helper.rb

@@ -119,6 +119,10 @@ def skip_unless_libxml2_patch(patch_name)
     def skip_unless_jruby(msg = "this test should only run with jruby")
       skip(msg) unless Nokogiri.jruby?
     end
+
+    def truffleruby_system_libraries?
+      RUBY_ENGINE == "truffleruby" && !Nokogiri::PACKAGED_LIBRARIES
+    end
   end
 
   class TestBenchmark < Minitest::BenchSpec
@@ -151,8 +155,12 @@ def initialize_nokogiri_test_gc_level
         "compact"
       elsif (ENV["NOKOGIRI_TEST_GC_LEVEL"] == "verify") && defined?(GC.verify_compaction_references)
         "verify"
+      elsif RUBY_ENGINE == "truffleruby"
+        "normal"
+      elsif defined?(GC.compact)
+        "compact"
       else
-        defined?(GC.compact) ? "compact" : "major"
+        "major"
       end
       if ["compact", "verify"].include?(@@gc_level)
         # the only way of detecting an unsupported platform is actually

test/test_xslt_transforms.rb

@@ -349,10 +349,17 @@ def test_non_html_xslt_transform
         exception = assert_raises(RuntimeError) do
           xslt.transform(doc)
         end
-        assert_match(
-          /xmlXPathCompOpEval: function decimal not found|java.lang.NoSuchMethodException.*decimal/,
-          exception.message,
-        )
+        if truffleruby_system_libraries?
+          assert_equal(
+            "xslt_generic_error_handler: xmlXPathCompOpEval: function %s not found\nxslt_generic_error_handler: %s",
+            exception.message
+          )
+        else
+          assert_match(
+            /xmlXPathCompOpEval: function decimal not found|java.lang.NoSuchMethodException.*decimal/,
+            exception.message
+          )
+        end
       end
 
       describe "DEFAULT_XSLT parse options" do

test/xml/sax/test_parser.rb

@@ -484,7 +484,11 @@ def call_parse_io_with_encoding(encoding)
           XML
           parser.parse(xml)
           refute_empty(parser.document.warnings)
-          assert_match(/URI .* is not absolute/, parser.document.warnings.first)
+          if truffleruby_system_libraries?
+            assert_equal("warning_func: %s", parser.document.warnings.first)
+          else
+            assert_match(/URI .* is not absolute/, parser.document.warnings.first)
+          end
         end
       end
     end

test/xml/test_entity_reference.rb

@@ -193,7 +193,11 @@ def setup
         html = File.read(xml_document)
         @parser.parse(html)
         refute_nil @parser.document.errors
-        assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+        if truffleruby_system_libraries?
+          assert_equal ["error_func: %s"], @parser.document.errors.map(&:to_s).map(&:strip)
+        else
+          assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+        end
       end
 
       test_relative_and_absolute_path :test_more_sax_entity_reference do
@@ -206,7 +210,11 @@ def setup
         ]
         @parser.parse(html)
         refute_nil @parser.document.errors
-        assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+        if truffleruby_system_libraries?
+          assert_equal ["error_func: %s"], @parser.document.errors.map(&:to_s).map(&:strip)
+        else
+          assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+        end
       end
     end