nokogiri 1.13.9-java → 1.14.0.rc1-java

data/ext/nokogiri/xml_sax_parser_context.c

@@ -7,13 +7,8 @@ static ID id_read;
 static void
 deallocate(xmlParserCtxtPtr ctxt)
 {
-  NOKOGIRI_DEBUG_START(ctxt);
-
   ctxt->sax = NULL;
-
   xmlFreeParserCtxt(ctxt);
-
-  NOKOGIRI_DEBUG_END(ctxt);
 }
 
 /*

data/ext/nokogiri/xml_sax_push_parser.c

@@ -5,12 +5,10 @@ VALUE cNokogiriXmlSaxPushParser ;
 static void
 deallocate(xmlParserCtxtPtr ctx)
 {
-  NOKOGIRI_DEBUG_START(ctx);
   if (ctx != NULL) {
     NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
     xmlFreeParserCtxt(ctx);
   }
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
 static VALUE

data/ext/nokogiri/xml_schema.c

@@ -5,9 +5,7 @@ VALUE cNokogiriXmlSchema;
 static void
 dealloc(xmlSchemaPtr schema)
 {
-  NOKOGIRI_DEBUG_START(schema);
   xmlSchemaFree(schema);
-  NOKOGIRI_DEBUG_END(schema);
 }
 
 /*

data/ext/nokogiri/xml_xpath_context.c

@@ -6,15 +6,15 @@ VALUE cNokogiriXmlXpathContext;
  * these constants have matching declarations in
  * ext/java/nokogiri/internals/NokogiriNamespaceContext.java
  */
+static const xmlChar *NOKOGIRI_PREFIX = (const xmlChar *)"nokogiri";
+static const xmlChar *NOKOGIRI_URI = (const xmlChar *)"http://www.nokogiri.org/default_ns/ruby/extensions_functions";
 static const xmlChar *NOKOGIRI_BUILTIN_PREFIX = (const xmlChar *)"nokogiri-builtin";
 static const xmlChar *NOKOGIRI_BUILTIN_URI = (const xmlChar *)"https://www.nokogiri.org/default_ns/ruby/builtins";
 
 static void
-deallocate(xmlXPathContextPtr ctx)
+xml_xpath_context_deallocate(xmlXPathContextPtr ctx)
 {
-  NOKOGIRI_DEBUG_START(ctx);
   xmlXPathFreeContext(ctx);
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
 /* find a CSS class in an HTML element's `class` attribute */
@@ -113,7 +113,7 @@ xpath_builtin_local_name_is(xmlXPathParserContextPtr ctxt, int nargs)
  * Register the namespace with +prefix+ and +uri+.
  */
 static VALUE
-register_ns(VALUE self, VALUE prefix, VALUE uri)
+rb_xml_xpath_context_register_ns(VALUE self, VALUE prefix, VALUE uri)
 {
   xmlXPathContextPtr ctx;
   Data_Get_Struct(self, xmlXPathContext, ctx);
@@ -132,7 +132,7 @@ register_ns(VALUE self, VALUE prefix, VALUE uri)
  * Register the variable +name+ with +value+.
  */
 static VALUE
-register_variable(VALUE self, VALUE name, VALUE value)
+rb_xml_xpath_context_register_variable(VALUE self, VALUE name, VALUE value)
 {
   xmlXPathContextPtr ctx;
   xmlXPathObjectPtr xmlValue;
@@ -154,28 +154,28 @@ register_variable(VALUE self, VALUE name, VALUE value)
  *  returns Qundef if no conversion was possible.
  */
 static VALUE
-xpath2ruby(xmlXPathObjectPtr xobj, xmlXPathContextPtr xctx)
+xpath2ruby(xmlXPathObjectPtr c_xpath_object, xmlXPathContextPtr ctx)
 {
-  VALUE retval;
+  VALUE rb_retval;
 
-  assert(xctx->doc);
-  assert(DOC_RUBY_OBJECT_TEST(xctx->doc));
+  assert(ctx->doc);
+  assert(DOC_RUBY_OBJECT_TEST(ctx->doc));
 
-  switch (xobj->type) {
+  switch (c_xpath_object->type) {
     case XPATH_STRING:
-      retval = NOKOGIRI_STR_NEW2(xobj->stringval);
-      xmlFree(xobj->stringval);
-      return retval;
+      rb_retval = NOKOGIRI_STR_NEW2(c_xpath_object->stringval);
+      xmlFree(c_xpath_object->stringval);
+      return rb_retval;
 
     case XPATH_NODESET:
-      return noko_xml_node_set_wrap(xobj->nodesetval,
-                                    DOC_RUBY_OBJECT(xctx->doc));
+      return noko_xml_node_set_wrap(c_xpath_object->nodesetval,
+                                    DOC_RUBY_OBJECT(ctx->doc));
 
     case XPATH_NUMBER:
-      return rb_float_new(xobj->floatval);
+      return rb_float_new(c_xpath_object->floatval);
 
     case XPATH_BOOLEAN:
-      return (xobj->boolval == 1) ? Qtrue : Qfalse;
+      return (c_xpath_object->boolval == 1) ? Qtrue : Qfalse;
 
     default:
       return Qundef;
@@ -183,52 +183,51 @@ xpath2ruby(xmlXPathObjectPtr xobj, xmlXPathContextPtr xctx)
 }
 
 void
-Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler,
-    const char *function_name)
+Nokogiri_marshal_xpath_funcall_and_return_values(
+  xmlXPathParserContextPtr ctx,
+  int argc,
+  VALUE rb_xpath_handler,
+  const char *method_name
+)
 {
-  VALUE result, doc;
+  VALUE rb_retval;
   VALUE *argv;
-  VALUE node_set = Qnil;
-  xmlNodeSetPtr xml_node_set = NULL;
-  xmlXPathObjectPtr obj;
+  VALUE rb_node_set = Qnil;
+  xmlNodeSetPtr c_node_set = NULL;
+  xmlXPathObjectPtr c_xpath_object;
 
   assert(ctx->context->doc);
   assert(DOC_RUBY_OBJECT_TEST(ctx->context->doc));
 
-  argv = (VALUE *)calloc((size_t)nargs, sizeof(VALUE));
-  for (int j = 0 ; j < nargs ; ++j) {
+  argv = (VALUE *)ruby_xcalloc((size_t)argc, sizeof(VALUE));
+  for (int j = 0 ; j < argc ; ++j) {
     rb_gc_register_address(&argv[j]);
   }
 
-  doc = DOC_RUBY_OBJECT(ctx->context->doc);
-
-  for (int j = nargs - 1 ; j >= 0 ; --j) {
-    obj = valuePop(ctx);
-    argv[j] = xpath2ruby(obj, ctx->context);
+  for (int j = argc - 1 ; j >= 0 ; --j) {
+    c_xpath_object = valuePop(ctx);
+    argv[j] = xpath2ruby(c_xpath_object, ctx->context);
     if (argv[j] == Qundef) {
-      argv[j] = NOKOGIRI_STR_NEW2(xmlXPathCastToString(obj));
+      argv[j] = NOKOGIRI_STR_NEW2(xmlXPathCastToString(c_xpath_object));
     }
-    xmlXPathFreeNodeSetList(obj);
+    xmlXPathFreeNodeSetList(c_xpath_object);
   }
 
-  result = rb_funcall2(handler, rb_intern((const char *)function_name), nargs, argv);
+  rb_retval = rb_funcall2(rb_xpath_handler, rb_intern((const char *)method_name), argc, argv);
 
-  for (int j = 0 ; j < nargs ; ++j) {
+  for (int j = 0 ; j < argc ; ++j) {
     rb_gc_unregister_address(&argv[j]);
   }
-  free(argv);
+  ruby_xfree(argv);
 
-  switch (TYPE(result)) {
+  switch (TYPE(rb_retval)) {
     case T_FLOAT:
     case T_BIGNUM:
     case T_FIXNUM:
-      xmlXPathReturnNumber(ctx, NUM2DBL(result));
+      xmlXPathReturnNumber(ctx, NUM2DBL(rb_retval));
       break;
     case T_STRING:
-      xmlXPathReturnString(
-        ctx,
-        xmlCharStrdup(StringValueCStr(result))
-      );
+      xmlXPathReturnString(ctx, xmlCharStrdup(StringValueCStr(rb_retval)));
       break;
     case T_TRUE:
       xmlXPathReturnTrue(ctx);
@@ -239,19 +238,17 @@ Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, i
     case T_NIL:
       break;
     case T_ARRAY: {
-      VALUE args[2];
-      args[0] = doc;
-      args[1] = result;
-      node_set = rb_class_new_instance(2, args, cNokogiriXmlNodeSet);
-      Data_Get_Struct(node_set, xmlNodeSet, xml_node_set);
-      xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
+      VALUE construct_args[2] = { DOC_RUBY_OBJECT(ctx->context->doc), rb_retval };
+      rb_node_set = rb_class_new_instance(2, construct_args, cNokogiriXmlNodeSet);
+      Data_Get_Struct(rb_node_set, xmlNodeSet, c_node_set);
+      xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, c_node_set));
     }
     break;
     case T_DATA:
-      if (rb_obj_is_kind_of(result, cNokogiriXmlNodeSet)) {
-        Data_Get_Struct(result, xmlNodeSet, xml_node_set);
+      if (rb_obj_is_kind_of(rb_retval, cNokogiriXmlNodeSet)) {
+        Data_Get_Struct(rb_retval, xmlNodeSet, c_node_set);
         /* Copy the node set, otherwise it will get GC'd. */
-        xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
+        xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, c_node_set));
         break;
       }
     default:
@@ -260,47 +257,56 @@ Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, i
 }
 
 static void
-ruby_funcall(xmlXPathParserContextPtr ctx, int nargs)
+method_caller(xmlXPathParserContextPtr ctx, int argc)
 {
-  VALUE handler = Qnil;
-  const char *function = NULL ;
+  VALUE rb_xpath_handler = Qnil;
+  const char *method_name = NULL ;
 
   assert(ctx);
   assert(ctx->context);
   assert(ctx->context->userData);
   assert(ctx->context->function);
 
-  handler = (VALUE)(ctx->context->userData);
-  function = (const char *)(ctx->context->function);
+  rb_xpath_handler = (VALUE)(ctx->context->userData);
+  method_name = (const char *)(ctx->context->function);
 
-  Nokogiri_marshal_xpath_funcall_and_return_values(ctx, nargs, handler, function);
+  Nokogiri_marshal_xpath_funcall_and_return_values(ctx, argc, rb_xpath_handler, method_name);
 }
 
 static xmlXPathFunction
-lookup(void *ctx,
-       const xmlChar *name,
-       const xmlChar *ns_uri)
+handler_lookup(void *ctx, const xmlChar *c_name, const xmlChar *c_ns_uri)
 {
-  VALUE xpath_handler = (VALUE)ctx;
-  if (rb_respond_to(xpath_handler, rb_intern((const char *)name))) {
-    return ruby_funcall;
+  VALUE rb_xpath_handler = (VALUE)ctx;
+  if (rb_respond_to(rb_xpath_handler, rb_intern((const char *)c_name))) {
+    return method_caller;
   }
 
   return NULL;
 }
 
-NORETURN(static void xpath_generic_exception_handler(void *ctx, const char *msg, ...));
+PRINTFLIKE_DECL(2, 3)
 static void
-xpath_generic_exception_handler(void *ctx, const char *msg, ...)
+generic_exception_pusher(void *ctx, const char *msg, ...)
 {
-  char *message;
+  VALUE rb_errors = (VALUE)ctx;
+  VALUE rb_message;
+  VALUE rb_exception;
+
+  Check_Type(rb_errors, T_ARRAY);
 
+#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("generic_exception_pusher: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
-  vasprintf(&message, msg, args);
+  rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
-  rb_raise(rb_eRuntimeError, "%s", message);
+  rb_exception = rb_exc_new_str(cNokogiriXmlXpathSyntaxError, rb_message);
+  rb_ary_push(rb_errors, rb_exception);
 }
 
 /*
@@ -310,13 +316,14 @@ xpath_generic_exception_handler(void *ctx, const char *msg, ...)
  * Evaluate the +search_path+ returning an XML::XPath object.
  */
 static VALUE
-evaluate(int argc, VALUE *argv, VALUE self)
+rb_xml_xpath_context_evaluate(int argc, VALUE *argv, VALUE self)
 {
   VALUE search_path, xpath_handler;
   VALUE retval = Qnil;
   xmlXPathContextPtr ctx;
   xmlXPathObjectPtr xpath;
   xmlChar *query;
+  VALUE errors = rb_ary_new();
 
   Data_Get_Struct(self, xmlXPathContext, ctx);
 
@@ -329,23 +336,19 @@ evaluate(int argc, VALUE *argv, VALUE self)
   if (Qnil != xpath_handler) {
     /* FIXME: not sure if this is the correct place to shove private data. */
     ctx->userData = (void *)xpath_handler;
-    xmlXPathRegisterFuncLookup(ctx, lookup, (void *)xpath_handler);
+    xmlXPathRegisterFuncLookup(ctx, handler_lookup, (void *)xpath_handler);
   }
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc(NULL, Nokogiri_error_raise);
-
-  /* For some reason, xmlXPathEvalExpression will blow up with a generic error */
-  /* when there is a non existent function. */
-  xmlSetGenericErrorFunc(NULL, xpath_generic_exception_handler);
+  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+  xmlSetGenericErrorFunc((void *)errors, generic_exception_pusher);
 
   xpath = xmlXPathEvalExpression(query, ctx);
+
   xmlSetStructuredErrorFunc(NULL, NULL);
   xmlSetGenericErrorFunc(NULL, NULL);
 
   if (xpath == NULL) {
-    xmlErrorPtr error = xmlGetLastError();
-    rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    rb_exc_raise(rb_ary_entry(errors, 0));
   }
 
   retval = xpath2ruby(xpath, ctx);
@@ -365,7 +368,7 @@ evaluate(int argc, VALUE *argv, VALUE self)
  * Create a new XPathContext with +node+ as the reference point.
  */
 static VALUE
-new (VALUE klass, VALUE nodeobj)
+rb_xml_xpath_context_new(VALUE klass, VALUE nodeobj)
 {
   xmlNodePtr node;
   xmlXPathContextPtr ctx;
@@ -381,13 +384,14 @@ new (VALUE klass, VALUE nodeobj)
   ctx = xmlXPathNewContext(node->doc);
   ctx->node = node;
 
+  xmlXPathRegisterNs(ctx, NOKOGIRI_PREFIX, NOKOGIRI_URI);
   xmlXPathRegisterNs(ctx, NOKOGIRI_BUILTIN_PREFIX, NOKOGIRI_BUILTIN_URI);
   xmlXPathRegisterFuncNS(ctx, (const xmlChar *)"css-class", NOKOGIRI_BUILTIN_URI,
                          xpath_builtin_css_class);
   xmlXPathRegisterFuncNS(ctx, (const xmlChar *)"local-name-is", NOKOGIRI_BUILTIN_URI,
                          xpath_builtin_local_name_is);
 
-  self = Data_Wrap_Struct(klass, 0, deallocate, ctx);
+  self = Data_Wrap_Struct(klass, 0, xml_xpath_context_deallocate, ctx);
   return self;
 }
 
@@ -401,9 +405,9 @@ noko_init_xml_xpath_context(void)
 
   rb_undef_alloc_func(cNokogiriXmlXpathContext);
 
-  rb_define_singleton_method(cNokogiriXmlXpathContext, "new", new, 1);
+  rb_define_singleton_method(cNokogiriXmlXpathContext, "new", rb_xml_xpath_context_new, 1);
 
-  rb_define_method(cNokogiriXmlXpathContext, "evaluate", evaluate, -1);
-  rb_define_method(cNokogiriXmlXpathContext, "register_variable", register_variable, 2);
-  rb_define_method(cNokogiriXmlXpathContext, "register_ns", register_ns, 2);
+  rb_define_method(cNokogiriXmlXpathContext, "evaluate", rb_xml_xpath_context_evaluate, -1);
+  rb_define_method(cNokogiriXmlXpathContext, "register_variable", rb_xml_xpath_context_register_variable, 2);
+  rb_define_method(cNokogiriXmlXpathContext, "register_ns", rb_xml_xpath_context_register_ns, 2);
 }

data/ext/nokogiri/xslt_stylesheet.c

@@ -12,27 +12,28 @@ static void
 dealloc(nokogiriXsltStylesheetTuple *wrapper)
 {
   xsltStylesheetPtr doc = wrapper->ss;
-
-  NOKOGIRI_DEBUG_START(doc);
-  xsltFreeStylesheet(doc); /* commented out for now. */
-  NOKOGIRI_DEBUG_END(doc);
-
-  free(wrapper);
+  xsltFreeStylesheet(doc);
+  ruby_xfree(wrapper);
 }
 
+PRINTFLIKE_DECL(2, 3)
 static void
 xslt_generic_error_handler(void *ctx, const char *msg, ...)
 {
-  char *message;
+  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);
-  vasprintf(&message, msg, args);
+  message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
-  rb_str_cat2((VALUE)ctx, message);
-
-  free(message);
+  rb_str_concat((VALUE)ctx, message);
 }
 
 VALUE
@@ -248,7 +249,7 @@ transform(int argc, VALUE *argv, VALUE self)
   Data_Get_Struct(self, nokogiriXsltStylesheetTuple, wrapper);
 
   param_len = RARRAY_LEN(paramobj);
-  params = calloc((size_t)param_len + 1, sizeof(char *));
+  params = ruby_xcalloc((size_t)param_len + 1, sizeof(char *));
   for (j = 0 ; j < param_len ; j++) {
     VALUE entry = rb_ary_entry(paramobj, j);
     const char *ptr = StringValueCStr(entry);
@@ -261,7 +262,7 @@ transform(int argc, VALUE *argv, VALUE self)
   xmlSetGenericErrorFunc((void *)errstr, xslt_generic_error_handler);
 
   result = xsltApplyStylesheet(wrapper->ss, xml, params);
-  free(params);
+  ruby_xfree(params);
 
   xsltSetGenericErrorFunc(NULL, NULL);
   xmlSetGenericErrorFunc(NULL, NULL);

data/gumbo-parser/Makefile

@@ -13,8 +13,18 @@ LDFLAGS := -pthread
 
 all: check
 
+# don't try to regenerate ragel or gperf files in CI, that should be a development-only action and
+# the generated files should be committed to SCM
+ifneq ($(CI),true)
+src/foreign_attrs.c: src/foreign_attrs.gperf
+	gperf -m100 -n $< | ./gperf-filter.sed > $@
+
+src/%.c: src/%.gperf
+	gperf -m100 $< | ./gperf-filter.sed > $@
+
 src/%.c: src/%.rl
 	ragel -F1 -o $@ $<
+endif
 
 build/src:
 	mkdir -p $@

data/lib/nokogiri/css/node.rb

@@ -40,9 +40,9 @@ module Nokogiri
 
       # Convert to_type
       def to_type
-        [@type] + @value.map do |n|
+        [@type] + @value.filter_map do |n|
           n.to_type if n.respond_to?(:to_type)
-        end.compact
+        end
       end
 
       # Convert to array

data/lib/nokogiri/css/xpath_visitor.rb

@@ -278,7 +278,9 @@ module Nokogiri
       end
 
       def nth(node, options = {})
-        raise ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}" unless node.value.size == 4
+        unless node.value.size == 4
+          raise(ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}")
+        end
 
         a, b = read_a_and_positive_b(node.value)
         position = if options[:child]

data/lib/nokogiri/css.rb

@@ -40,9 +40,15 @@ module Nokogiri
       # 💡 Note that translated queries are cached for performance concerns.
       #
       def xpath_for(selector, options = {})
+        raise TypeError, "no implicit conversion of #{selector.inspect} to String" unless selector.respond_to?(:to_str)
+
+        selector = selector.to_str
+        raise Nokogiri::CSS::SyntaxError, "empty CSS selector" if selector.empty?
+
         prefix = options.fetch(:prefix, Nokogiri::XML::XPath::GLOBAL_SEARCH_PREFIX)
         visitor = options.fetch(:visitor) { Nokogiri::CSS::XPathVisitor.new }
         ns = options.fetch(:ns, {})
+
         Parser.new(ns).xpath_for(selector, prefix, visitor)
       end
     end

data/lib/nokogiri/encoding_handler.rb

@@ -0,0 +1,57 @@
+# encoding: utf-8
+# frozen_string_literal: true
+
+module Nokogiri
+  class EncodingHandler
+    # Popular encoding aliases not known by all iconv implementations that Nokogiri should support.
+    USEFUL_ALIASES = {
+      # alias_name => true_name
+      "NOKOGIRI-SENTINEL" => "UTF-8", # indicating the Nokogiri has installed aliases
+      "Windows-31J" => "CP932", # Windows-31J is the IANA registered name of CP932.
+      "UTF-8" => "UTF-8", # for JRuby tests, this is a no-op in CRuby
+    }
+
+    class << self
+      def install_default_aliases
+        USEFUL_ALIASES.each do |alias_name, name|
+          EncodingHandler.alias(name, alias_name) if EncodingHandler[alias_name].nil?
+        end
+      end
+    end
+
+    # :stopdoc:
+    if Nokogiri.jruby?
+      class << self
+        def [](name)
+          storage.key?(name) ? new(storage[name]) : nil
+        end
+
+        def alias(name, alias_name)
+          storage[alias_name] = name
+        end
+
+        def delete(name)
+          storage.delete(name)
+        end
+
+        def clear_aliases!
+          storage.clear
+        end
+
+        private
+
+        def storage
+          @storage ||= {}
+        end
+      end
+
+      def initialize(name)
+        @name = name
+      end
+
+      attr_reader :name
+    end
+  end
+end
+
+Nokogiri::EncodingHandler.install_default_aliases

data/lib/nokogiri/extension.rb

@@ -3,13 +3,14 @@
 # load the C or Java extension
 begin
   # native precompiled gems package shared libraries in <gem_dir>/lib/nokogiri/<ruby_version>
-  ::RUBY_VERSION =~ /(\d+\.\d+)/
+  RUBY_VERSION =~ /(\d+\.\d+)/
   require_relative "#{Regexp.last_match(1)}/nokogiri"
 rescue LoadError => e
   if /GLIBC/.match?(e.message)
     warn(<<~EOM)
 
-      ERROR: It looks like you're trying to use Nokogiri as a precompiled native gem on a system with glibc < 2.17:
+      ERROR: It looks like you're trying to use Nokogiri as a precompiled native gem on a system
+             with an unsupported version of glibc.
 
         #{e.message}