nokogiri 1.16.0-arm64-darwin → 1.16.1-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d5a49b88555da7599d4aa2beb0539d45e5f8b7c79a25cc6a3355a6b670b7046a
-  data.tar.gz: 8dcee519f9d75ddf5e3097128c9528f034b413f460bfd7da18bedb175ee0b078
+  metadata.gz: f267bea49d64a29a5ae0076f65389fe83c3b303723ae5641e31028e88a51df20
+  data.tar.gz: a905f6d9b6dee674a29659e311a8a61438b8a155b8ff79557907fa868ebc5281
 SHA512:
-  metadata.gz: bc52cebefe131ed7b8a30fa894b6113e1ef3913cbe71d34ef36519851128aee85b7303483345a21ecaa2e19cbd83e908ae1213a0512b258bffc0ce06f906466c
-  data.tar.gz: '009863ceea8574d646f85d81c7741c44e8d6458d638d366e7e25a66f81ba6a970ae95482eda3c1ad1b779a6e91abac708184676f6ae428e268981c039fd6db05'
+  metadata.gz: 9a6c7df9063244a83142dc3581a7337cb799457d563221d0dd7baeb840ea0f6b9970ce546039b1698afce1b870a57485fa016707cad247202f320eec17ce1070
+  data.tar.gz: 9201922886c52c2970435329c6b41b3baf062cf92f12e673d0df5894bee283ecd99533e0b33cd5d60a2625ee4dd618d000141de5a546555e7560630da2c12a0b

data/Gemfile

@@ -5,19 +5,25 @@ source "https://rubygems.org"
 gemspec
 
 group :development do
+  # ruby 3.4.0-dev removed some gems from the default set
+  #
+  # TODO: we should be able to remove these as our gem dependencies sort it out and we pull them in
+  # transitively.
+  gem "mutex_m"
+
   # bootstrapping
   gem "bundler", "~> 2.3"
   gem "rake", "13.1.0"
 
   # building extensions
-  gem "rake-compiler", "1.2.5"
+  gem "rake-compiler", "1.2.6"
   gem "rake-compiler-dock", "1.4.0"
 
   # parser generator
   gem "rexical", "= 1.0.7"
 
   # tests
-  gem "minitest", "5.20.0"
+  gem "minitest", "5.21.2"
   gem "minitest-parallel_fork", "2.0.0"
   gem "ruby_memcheck", "2.3.0"
   gem "rubyzip", "~> 2.3.2"
@@ -25,10 +31,10 @@ group :development do
 
   # rubocop
   if Gem::Requirement.new("~> 3.0").satisfied_by?(Gem::Version.new(RUBY_VERSION))
-    gem "rubocop", "1.59.0"
-    gem "rubocop-minitest", "0.34.2"
+    gem "rubocop", "1.60.2"
+    gem "rubocop-minitest", "0.34.5"
     gem "rubocop-packaging", "0.5.2"
-    gem "rubocop-performance", "1.20.1"
+    gem "rubocop-performance", "1.20.2"
     gem "rubocop-rake", "= 0.6.0"
     gem "rubocop-shopify", "2.14.0"
   end

data/dependencies.yml

@@ -1,8 +1,8 @@
 
 libxml2:
-  version: "2.12.3"
-  sha256: "8c8f1092340a89ff32bc44ad5c9693aff9bc8a7a3e161bb239666e5d15ac9aaa"
-  # sha-256 hash provided in https://download.gnome.org/sources/libxml2/2.12/libxml2-2.12.3.sha256sum
+  version: "2.12.4"
+  sha256: "497360e423cf0bd99eacdb7c6215dea92e6d6e89ee940393c2bae0e77cb9b7d0"
+  # sha-256 hash provided in https://download.gnome.org/sources/libxml2/2.12/libxml2-2.12.4.sha256sum
 
 libxslt:
   version: "1.1.39"

data/ext/nokogiri/include/libxml2/libxml/parser.h

@@ -859,10 +859,16 @@ XMLPUBFUN const char *const *__xmlParserVersion(void);
   XML_OP(xmlLoadExtDtdDefaultValue, int, XML_DEPRECATED) \
   XML_OP(xmlParserDebugEntities, int, XML_DEPRECATED) \
   XML_OP(xmlPedanticParserDefaultValue, int, XML_DEPRECATED) \
-  XML_OP(xmlSubstituteEntitiesDefaultValue, int, XML_DEPRECATED) \
-  XML_OP(xmlIndentTreeOutput, int, XML_NO_ATTR) \
-  XML_OP(xmlTreeIndentString, const char *, XML_NO_ATTR) \
-  XML_OP(xmlSaveNoEmptyTags, int, XML_NO_ATTR)
+  XML_OP(xmlSubstituteEntitiesDefaultValue, int, XML_DEPRECATED)
+
+#ifdef LIBXML_OUTPUT_ENABLED
+  #define XML_GLOBALS_PARSER_OUTPUT \
+    XML_OP(xmlIndentTreeOutput, int, XML_NO_ATTR) \
+    XML_OP(xmlTreeIndentString, const char *, XML_NO_ATTR) \
+    XML_OP(xmlSaveNoEmptyTags, int, XML_NO_ATTR)
+#else
+  #define XML_GLOBALS_PARSER_OUTPUT
+#endif
 
 #ifdef LIBXML_SAX1_ENABLED
   #define XML_GLOBALS_PARSER_SAX1 \
@@ -873,6 +879,7 @@ XMLPUBFUN const char *const *__xmlParserVersion(void);
 
 #define XML_GLOBALS_PARSER \
   XML_GLOBALS_PARSER_CORE \
+  XML_GLOBALS_PARSER_OUTPUT \
   XML_GLOBALS_PARSER_SAX1
 
 #define XML_OP XML_DECLARE_GLOBAL
@@ -896,9 +903,11 @@ XML_GLOBALS_PARSER
     XML_GLOBAL_MACRO(xmlPedanticParserDefaultValue)
   #define xmlSubstituteEntitiesDefaultValue \
     XML_GLOBAL_MACRO(xmlSubstituteEntitiesDefaultValue)
-  #define xmlIndentTreeOutput XML_GLOBAL_MACRO(xmlIndentTreeOutput)
-  #define xmlTreeIndentString XML_GLOBAL_MACRO(xmlTreeIndentString)
-  #define xmlSaveNoEmptyTags XML_GLOBAL_MACRO(xmlSaveNoEmptyTags)
+  #ifdef LIBXML_OUTPUT_ENABLED
+    #define xmlIndentTreeOutput XML_GLOBAL_MACRO(xmlIndentTreeOutput)
+    #define xmlTreeIndentString XML_GLOBAL_MACRO(xmlTreeIndentString)
+    #define xmlSaveNoEmptyTags XML_GLOBAL_MACRO(xmlSaveNoEmptyTags)
+  #endif
 #endif
 /** DOC_ENABLE */
 

data/ext/nokogiri/include/libxml2/libxml/xmlIO.h

@@ -218,7 +218,6 @@ XMLPUBFUN xmlParserInputBufferPtr
 XMLPUBFUN xmlParserInputBufferPtr
 	xmlParserInputBufferCreateMem		(const char *mem, int size,
 	                                         xmlCharEncoding enc);
-XML_DEPRECATED
 XMLPUBFUN xmlParserInputBufferPtr
 	xmlParserInputBufferCreateStatic	(const char *mem, int size,
 	                                         xmlCharEncoding enc);

data/ext/nokogiri/include/libxml2/libxml/xmlversion.h

@@ -29,21 +29,21 @@ XMLPUBFUN void xmlCheckVersion(int version);
  *
  * the version string like "1.2.3"
  */
-#define LIBXML_DOTTED_VERSION "2.12.3"
+#define LIBXML_DOTTED_VERSION "2.12.4"
 
 /**
  * LIBXML_VERSION:
  *
  * the version number: 1.2.3 value is 10203
  */
-#define LIBXML_VERSION 21203
+#define LIBXML_VERSION 21204
 
 /**
  * LIBXML_VERSION_STRING:
  *
  * the version number string, 1.2.3 value is "10203"
  */
-#define LIBXML_VERSION_STRING "21203"
+#define LIBXML_VERSION_STRING "21204"
 
 /**
  * LIBXML_VERSION_EXTRA:
@@ -58,7 +58,7 @@ XMLPUBFUN void xmlCheckVersion(int version);
  * Macro to check that the libxml version in use is compatible with
  * the version the software has been compiled against
  */
-#define LIBXML_TEST_VERSION xmlCheckVersion(21203);
+#define LIBXML_TEST_VERSION xmlCheckVersion(21204);
 
 #ifndef VMS
 #if 0

data/ext/nokogiri/xml_document.c

@@ -74,8 +74,10 @@ dealloc(void *data)
 
   ruby_xfree(doc->_private);
 
+#if defined(__GNUC__) && __GNUC__ >= 5
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wdeprecated-declarations" // xmlDeregisterNodeDefault is deprecated as of libxml2 2.11.0
+#endif
   /*
    * libxml-ruby < 3.0.0 uses xmlDeregisterNodeDefault. If the user is using one of those older
    * versions, the registered callback from libxml-ruby will access the _private pointers set by
@@ -90,7 +92,9 @@ dealloc(void *data)
   if (xmlDeregisterNodeDefaultValue) {
     remove_private((xmlNodePtr)doc);
   }
+#if defined(__GNUC__) && __GNUC__ >= 5
 #pragma GCC diagnostic pop
+#endif
 
   xmlFreeDoc(doc);
 }

data/ext/nokogiri/xml_reader.c

@@ -5,8 +5,14 @@ VALUE cNokogiriXmlReader;
 static void
 xml_reader_deallocate(void *data)
 {
+  // free the document separately because we _may_ have triggered preservation by calling
+  // xmlTextReaderCurrentDoc during a read_more.
   xmlTextReaderPtr reader = data;
+  xmlDocPtr doc = xmlTextReaderCurrentDoc(reader);
   xmlFreeTextReader(reader);
+  if (doc) {
+    xmlFreeDoc(doc);
+  }
 }
 
 static const rb_data_type_t xml_reader_type = {
@@ -515,6 +521,7 @@ read_more(VALUE self)
   xmlErrorConstPtr error;
   VALUE error_list;
   int ret;
+  xmlDocPtr c_document;
 
   TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
 
@@ -524,6 +531,16 @@ read_more(VALUE self)
   ret = xmlTextReaderRead(reader);
   xmlSetStructuredErrorFunc(NULL, NULL);
 
+  c_document = xmlTextReaderCurrentDoc(reader);
+  if (c_document && c_document->encoding == NULL) {
+    VALUE constructor_encoding = rb_iv_get(self, "@encoding");
+    if (RTEST(constructor_encoding)) {
+      c_document->encoding = xmlStrdup(BAD_CAST StringValueCStr(constructor_encoding));
+    } else {
+      c_document->encoding = xmlStrdup(BAD_CAST "UTF-8");
+    }
+  }
+
   if (ret == 1) { return self; }
   if (ret == 0) { return Qnil; }
 
@@ -707,15 +724,18 @@ rb_xml_reader_encoding(VALUE rb_reader)
   const char *parser_encoding;
   VALUE constructor_encoding;
 
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
+  parser_encoding = (const char *)xmlTextReaderConstEncoding(c_reader);
+  if (parser_encoding) {
+    return NOKOGIRI_STR_NEW2(parser_encoding);
+  }
+
   constructor_encoding = rb_iv_get(rb_reader, "@encoding");
   if (RTEST(constructor_encoding)) {
     return constructor_encoding;
   }
 
-  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
-  parser_encoding = (const char *)xmlTextReaderConstEncoding(c_reader);
-  if (parser_encoding == NULL) { return Qnil; }
-  return NOKOGIRI_STR_NEW2(parser_encoding);
+  return Qnil;
 }
 
 void

data/ext/nokogiri/xslt_stylesheet.c

@@ -71,7 +71,12 @@ Nokogiri_wrap_xslt_stylesheet(xsltStylesheetPtr ss)
  * call-seq:
  *   parse_stylesheet_doc(document)
  *
- * Parse a stylesheet from +document+.
+ * Parse an XSLT::Stylesheet from +document+.
+ *
+ * [Parameters]
+ * - +document+ (Nokogiri::XML::Document) the document to be parsed.
+ *
+ * [Returns] Nokogiri::XSLT::Stylesheet
  */
 static VALUE
 parse_stylesheet_doc(VALUE klass, VALUE xmldocobj)
@@ -104,7 +109,7 @@ parse_stylesheet_doc(VALUE klass, VALUE xmldocobj)
  * call-seq:
  *   serialize(document)
  *
- * Serialize +document+ to an xml string.
+ * Serialize +document+ to an xml string, as specified by the +method+ parameter in the Stylesheet.
  */
 static VALUE
 rb_xslt_stylesheet_serialize(VALUE self, VALUE xmlobj)
@@ -133,7 +138,7 @@ rb_xslt_stylesheet_serialize(VALUE self, VALUE xmlobj)
  *   transform(document)
  *   transform(document, params = {})
  *
- * Apply an XSLT stylesheet to an XML::Document.
+ * Transform an XML::Document as defined by an XSLT::Stylesheet.
  *
  * [Parameters]
  * - +document+ (Nokogiri::XML::Document) the document to be transformed.

data/gumbo-parser/Makefile

@@ -13,6 +13,9 @@ LDFLAGS := -pthread
 
 all: check
 
+oss-fuzz:
+	./fuzzer/build-ossfuzz.sh
+
 fuzzers: fuzzer-normal fuzzer-asan fuzzer-ubsan fuzzer-msan
 
 fuzzer-normal:

data/lib/nokogiri/version/constant.rb

@@ -2,5 +2,5 @@
 
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = "1.16.0"
+  VERSION = "1.16.1"
 end

data/lib/nokogiri/xml/reader.rb

@@ -3,9 +3,11 @@
 module Nokogiri
   module XML
     ###
-    # Nokogiri::XML::Reader parses an XML document similar to the way a cursor
-    # would move.  The Reader is given an XML document, and yields nodes
-    # to an each block.
+    # Nokogiri::XML::Reader parses an XML document similar to the way a cursor would move. The
+    # Reader is given an XML document, and yields nodes to an each block.
+    #
+    # The Reader parser might be good for when you need the speed and low memory usage of the SAX
+    # parser, but do not want to write a Document handler.
     #
     # Here is an example of usage:
     #
@@ -22,13 +24,12 @@ module Nokogiri
     #
     #     end
     #
-    # Note that Nokogiri::XML::Reader#each can only be called once!!  Once
-    # the cursor moves through the entire document, you must parse the
-    # document again.  So make sure that you capture any information you
-    # need during the first iteration.
+    # ⚠ Nokogiri::XML::Reader#each can only be called once! Once the cursor moves through the entire
+    # document, you must parse the document again. It may be better to capture all information you
+    # need during a single iteration.
     #
-    # The Reader parser is good for when you need the speed of a SAX parser,
-    # but do not want to write a Document handler.
+    # ⚠ libxml2 does not support error recovery in the Reader parser. The `RECOVER` ParseOption is
+    # ignored. If a syntax error is encountered during parsing, an exception will be raised.
     class Reader
       include Enumerable
 

data/lib/nokogiri/xslt/stylesheet.rb

@@ -10,15 +10,37 @@ module Nokogiri
     #   doc   = Nokogiri::XML(File.read('some_file.xml'))
     #   xslt  = Nokogiri::XSLT(File.read('some_transformer.xslt'))
     #
-    #   puts xslt.transform(doc)
+    #   xslt.transform(doc) # => Nokogiri::XML::Document
     #
-    # See Nokogiri::XSLT::Stylesheet#transform for more transformation
-    # information.
+    # Many XSLT transformations include serialization behavior to emit a non-XML document. For these
+    # cases, please take care to invoke the #serialize method on the result of the transformation:
+    #
+    #   doc   = Nokogiri::XML(File.read('some_file.xml'))
+    #   xslt  = Nokogiri::XSLT(File.read('some_transformer.xslt'))
+    #   xslt.serialize(xslt.transform(doc)) # => String
+    #
+    # or use the #apply_to method, which is a shortcut for `serialize(transform(document))`:
+    #
+    #   doc   = Nokogiri::XML(File.read('some_file.xml'))
+    #   xslt  = Nokogiri::XSLT(File.read('some_transformer.xslt'))
+    #   xslt.apply_to(doc) # => String
+    #
+    # See Nokogiri::XSLT::Stylesheet#transform for more information and examples.
     class Stylesheet
-      ###
-      # Apply an XSLT stylesheet to an XML::Document.
-      # +params+ is an array of strings used as XSLT parameters.
-      # returns serialized document
+      # :call-seq:
+      #   apply_to(document, params = []) -> String
+      #
+      # Apply an XSLT stylesheet to an XML::Document and serialize it properly. This method is
+      # equivalent to calling #serialize on the result of #transform.
+      #
+      # [Parameters]
+      # - +document+ is an instance of XML::Document to transform
+      # - +params+ is an array of strings used as XSLT parameters, passed into #transform
+      #
+      # [Returns]
+      #   A string containing the serialized result of the transformation.
+      #
+      # See Nokogiri::XSLT::Stylesheet#transform for more information and examples.
       def apply_to(document, params = [])
         serialize(transform(document, params))
       end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: nokogiri
 version: !ruby/object:Gem::Version
-  version: 1.16.0
+  version: 1.16.1
 platform: arm64-darwin
 authors:
 - Mike Dalessio
@@ -20,7 +20,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-12-27 00:00:00.000000000 Z
+date: 2024-02-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: racc