nokogiri 1.13.0-aarch64-linux

data/ext/nokogiri/xslt_stylesheet.c

@@ -0,0 +1,264 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriXsltStylesheet ;
+
+static void
+mark(nokogiriXsltStylesheetTuple *wrapper)
+{
+  rb_gc_mark(wrapper->func_instances);
+}
+
+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);
+}
+
+static void
+xslt_generic_error_handler(void *ctx, const char *msg, ...)
+{
+  char *message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  rb_str_cat2((VALUE)ctx, message);
+
+  free(message);
+}
+
+VALUE
+Nokogiri_wrap_xslt_stylesheet(xsltStylesheetPtr ss)
+{
+  VALUE self;
+  nokogiriXsltStylesheetTuple *wrapper;
+
+  self = Data_Make_Struct(cNokogiriXsltStylesheet, nokogiriXsltStylesheetTuple,
+                          mark, dealloc, wrapper);
+
+  ss->_private = (void *)self;
+  wrapper->ss = ss;
+  wrapper->func_instances = rb_ary_new();
+
+  return self;
+}
+
+/*
+ * call-seq:
+ *   parse_stylesheet_doc(document)
+ *
+ * Parse a stylesheet from +document+.
+ */
+static VALUE
+parse_stylesheet_doc(VALUE klass, VALUE xmldocobj)
+{
+  xmlDocPtr xml, xml_cpy;
+  VALUE errstr, exception;
+  xsltStylesheetPtr ss ;
+  Data_Get_Struct(xmldocobj, xmlDoc, xml);
+
+  errstr = rb_str_new(0, 0);
+  xsltSetGenericErrorFunc((void *)errstr, xslt_generic_error_handler);
+
+  xml_cpy = xmlCopyDoc(xml, 1); /* 1 => recursive */
+  ss = xsltParseStylesheetDoc(xml_cpy);
+
+  xsltSetGenericErrorFunc(NULL, NULL);
+
+  if (!ss) {
+    xmlFreeDoc(xml_cpy);
+    exception = rb_exc_new3(rb_eRuntimeError, errstr);
+    rb_exc_raise(exception);
+  }
+
+  return Nokogiri_wrap_xslt_stylesheet(ss);
+}
+
+
+/*
+ * call-seq:
+ *   serialize(document)
+ *
+ * Serialize +document+ to an xml string.
+ */
+static VALUE
+serialize(VALUE self, VALUE xmlobj)
+{
+  xmlDocPtr xml ;
+  nokogiriXsltStylesheetTuple *wrapper;
+  xmlChar *doc_ptr ;
+  int doc_len ;
+  VALUE rval ;
+
+  Data_Get_Struct(xmlobj, xmlDoc, xml);
+  Data_Get_Struct(self, nokogiriXsltStylesheetTuple, wrapper);
+  xsltSaveResultToString(&doc_ptr, &doc_len, xml, wrapper->ss);
+  rval = NOKOGIRI_STR_NEW(doc_ptr, doc_len);
+  xmlFree(doc_ptr);
+  return rval ;
+}
+
+/*
+ *  call-seq:
+ *    transform(document, params = [])
+ *
+ *  Apply an XSLT stylesheet to an XML::Document.
+ *  +params+ is an array of strings used as XSLT parameters.
+ *  returns Nokogiri::XML::Document
+ *
+ *  Example:
+ *
+ *    doc   = Nokogiri::XML(File.read(ARGV[0]))
+ *    xslt  = Nokogiri::XSLT(File.read(ARGV[1]))
+ *    puts xslt.transform(doc, ['key', 'value'])
+ *
+ */
+static VALUE
+transform(int argc, VALUE *argv, VALUE self)
+{
+  VALUE xmldoc, paramobj, errstr, exception ;
+  xmlDocPtr xml ;
+  xmlDocPtr result ;
+  nokogiriXsltStylesheetTuple *wrapper;
+  const char **params ;
+  long param_len, j ;
+  int parse_error_occurred ;
+
+  rb_scan_args(argc, argv, "11", &xmldoc, &paramobj);
+  if (NIL_P(paramobj)) { paramobj = rb_ary_new2(0L) ; }
+  if (!rb_obj_is_kind_of(xmldoc, cNokogiriXmlDocument)) {
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::Document");
+  }
+
+  /* handle hashes as arguments. */
+  if (T_HASH == TYPE(paramobj)) {
+    paramobj = rb_funcall(paramobj, rb_intern("to_a"), 0);
+    paramobj = rb_funcall(paramobj, rb_intern("flatten"), 0);
+  }
+
+  Check_Type(paramobj, T_ARRAY);
+
+  Data_Get_Struct(xmldoc, xmlDoc, xml);
+  Data_Get_Struct(self, nokogiriXsltStylesheetTuple, wrapper);
+
+  param_len = RARRAY_LEN(paramobj);
+  params = calloc((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);
+    params[j] = ptr;
+  }
+  params[param_len] = 0 ;
+
+  errstr = rb_str_new(0, 0);
+  xsltSetGenericErrorFunc((void *)errstr, xslt_generic_error_handler);
+  xmlSetGenericErrorFunc((void *)errstr, xslt_generic_error_handler);
+
+  result = xsltApplyStylesheet(wrapper->ss, xml, params);
+  free(params);
+
+  xsltSetGenericErrorFunc(NULL, NULL);
+  xmlSetGenericErrorFunc(NULL, NULL);
+
+  parse_error_occurred = (Qfalse == rb_funcall(errstr, rb_intern("empty?"), 0));
+
+  if (parse_error_occurred) {
+    exception = rb_exc_new3(rb_eRuntimeError, errstr);
+    rb_exc_raise(exception);
+  }
+
+  return noko_xml_document_wrap((VALUE)0, result) ;
+}
+
+static void
+method_caller(xmlXPathParserContextPtr ctxt, int nargs)
+{
+  VALUE handler;
+  const char *function_name;
+  xsltTransformContextPtr transform;
+  const xmlChar *functionURI;
+
+  transform = xsltXPathGetTransformContext(ctxt);
+  functionURI = ctxt->context->functionURI;
+  handler = (VALUE)xsltGetExtData(transform, functionURI);
+  function_name = (const char *)(ctxt->context->function);
+
+  Nokogiri_marshal_xpath_funcall_and_return_values(ctxt, nargs, handler, (const char *)function_name);
+}
+
+static void *
+initFunc(xsltTransformContextPtr ctxt, const xmlChar *uri)
+{
+  VALUE modules = rb_iv_get(mNokogiriXslt, "@modules");
+  VALUE obj = rb_hash_aref(modules, rb_str_new2((const char *)uri));
+  VALUE args = { Qfalse };
+  VALUE methods = rb_funcall(obj, rb_intern("instance_methods"), 1, args);
+  VALUE inst;
+  nokogiriXsltStylesheetTuple *wrapper;
+  int i;
+
+  for (i = 0; i < RARRAY_LEN(methods); i++) {
+    VALUE method_name = rb_obj_as_string(rb_ary_entry(methods, i));
+    xsltRegisterExtFunction(ctxt,
+                            (unsigned char *)StringValueCStr(method_name), uri, method_caller);
+  }
+
+  Data_Get_Struct((VALUE)ctxt->style->_private, nokogiriXsltStylesheetTuple,
+                  wrapper);
+  inst = rb_class_new_instance(0, NULL, obj);
+  rb_ary_push(wrapper->func_instances, inst);
+
+  return (void *)inst;
+}
+
+static void
+shutdownFunc(xsltTransformContextPtr ctxt,
+             const xmlChar *uri, void *data)
+{
+  nokogiriXsltStylesheetTuple *wrapper;
+
+  Data_Get_Struct((VALUE)ctxt->style->_private, nokogiriXsltStylesheetTuple,
+                  wrapper);
+
+  rb_ary_clear(wrapper->func_instances);
+}
+
+/*
+ *  call-seq:
+ *    register(uri, custom_handler_class)
+ *
+ *  Register a class that implements custom XSLT transformation functions.
+ */
+static VALUE
+registr(VALUE self, VALUE uri, VALUE obj)
+{
+  VALUE modules = rb_iv_get(self, "@modules");
+  if (NIL_P(modules)) { rb_raise(rb_eRuntimeError, "wtf! @modules isn't set"); }
+
+  rb_hash_aset(modules, uri, obj);
+  xsltRegisterExtModule((unsigned char *)StringValueCStr(uri), initFunc, shutdownFunc);
+  return self;
+}
+
+void
+noko_init_xslt_stylesheet()
+{
+  rb_define_singleton_method(mNokogiriXslt, "register", registr, 2);
+  rb_iv_set(mNokogiriXslt, "@modules", rb_hash_new());
+
+  cNokogiriXsltStylesheet = rb_define_class_under(mNokogiriXslt, "Stylesheet", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriXsltStylesheet);
+
+  rb_define_singleton_method(cNokogiriXsltStylesheet, "parse_stylesheet_doc", parse_stylesheet_doc, 1);
+  rb_define_method(cNokogiriXsltStylesheet, "serialize", serialize, 1);
+  rb_define_method(cNokogiriXsltStylesheet, "transform", transform, -1);
+}

data/gumbo-parser/CHANGES.md

@@ -0,0 +1,63 @@
+## Gumbo 0.10.1 (2015-04-30)
+
+Same as 0.10.0, but with the version number bumped because the last version-number commit to v0.9.4 makes GitHub think that v0.9.4 is the latest version and so it's not highlighted on the webpage.
+
+## Gumbo 0.10.0 (2015-04-30)
+
+* Full support for `<template>` tag (kevinhendricks, nostrademons).
+* Some fixes for `<rtc>`/`<rt>` handling (kevinhendricks, vmg).
+* All html5lib-trunk tests pass now! (kevinhendricks, vmg, nostrademons)
+* Support for fragment parsing (vmg)
+* A couple additional example programs (kevinhendricks)
+* Performance improvements totaling an estimated 30-40% total improvement (vmg, nostrademons).
+
+## Gumbo 0.9.4 (2015-04-30)
+
+* Additional Visual Studio fixes (lowjoel, nostrademons)
+* Fixed some unused variable warnings.
+* Fix for glibtoolize vs. libtoolize build errors on Mac.
+* Fixed `CDATA` end tag handling.
+
+## Gumbo 0.9.3 (2015-02-17)
+
+* Bugfix for `&AElig;` entities (rgrove)
+* Fix `CDATA` handling; `CDATA` sections now generate a `GUMBO_NODE_CDATA` node rather
+than plain text.
+* Fix `get_title example` to handle whitespace nodes (gsnedders)
+* Visual Studio compilation fixes (fishioon)
+* Take the namespace into account when determining whether a node matches a
+certain tag (aroben)
+* Replace the varargs tag functions with a tagset bytevector, for a 20-30%
+speedup in overall parse time (kevinhendricks, vmg)
+* Add MacOS X support to Travis CI, and fix the deployment/DLL issues this
+uncovered (nostrademons, kevinhendricks, vmg)
+
+## Gumbo 0.9.2 (2014-09-21)
+
+* Performance improvements: Ragel-based char ref decoder and DFA-based UTF8
+decoder, totaling speedups of up to 300%.
+* Added benchmarking program and some sample data.
+* Fixed a compiler error under Visual Studio.
+* Fix an error in the ctypes bindings that could lead to memory corruption in
+the Python bindings.
+* Fix duplicate attributes when parsing `<isindex>` tags.
+* Don't leave semicolons behind when consuming entity references (rgrove)
+* Internally rename some functions in preparation for an amalgamation file
+(jdeng)
+* Add proper cflags for gyp builds (skabbes)
+
+## Gumbo 0.9.1 (2014-08-07)
+
+* First version listed on PyPi.
+* Autotools files excluded from GitHub and generated via autogen.sh. (endgame)
+* Numerous compiler warnings fixed. (bnoordhuis, craigbarnes)
+* Google security audit passed.
+* Gyp support (tfarina)
+* Naming convention for structs changed to avoid C reserved words.
+* Fix several integer and buffer overflows (Maxime2)
+* Some Visual Studio compiler support (bugparty)
+* Python3 compatibility for the ctypes bindings.
+
+## Gumbo 0.9.0 (2013-08-13)
+
+* Initial release open-sourced by Google.

data/gumbo-parser/Makefile

@@ -0,0 +1,101 @@
+.PHONY: all clean check coverage
+
+gumbo_objs := $(patsubst %.c,build/%.o,$(wildcard src/*.c))
+test_objs := $(patsubst %.cc,build/%.o,$(wildcard test/*.cc))
+gtest_lib := googletest/make/gtest_main.a
+
+# make SANITIZEFLAGS='-fsanitize=undefined -fsanitize=address'
+SANITIZEFLAGS :=
+CPPFLAGS := -Isrc
+CFLAGS := -std=c99 -Os -Wall
+CXXFLAGS := -isystem googletest/include -std=c++11 -Os -Wall
+LDFLAGS := -pthread
+
+all: check
+
+src/%.c: src/%.rl
+	ragel -F1 -o $@ $<
+
+build/src:
+	mkdir -p $@
+
+build/test:
+	mkdir -p $@
+
+build/src/%.o: src/%.c build/src/flags | build/src
+	$(CC) -MMD $(CPPFLAGS) $(CFLAGS) $(SANITIZEFLAGS) -c -o $@ $<
+
+build/test/%.o: test/%.cc build/test/flags | build/test
+	$(CXX) -MMD $(CPPFLAGS) $(CXXFLAGS) $(SANITIZEFLAGS) -c -o $@ $<
+
+build/run_tests: $(gumbo_objs) $(test_objs) $(gtest_lib)
+	$(CXX) -o $@ $+ $(LDFLAGS) $(SANITIZEFLAGS)
+
+check: build/run_tests
+	./build/run_tests
+
+coverage:
+	$(RM) build/{src,test}/*.gcda
+	$(RM) build/*.info
+	$(MAKE) CPPFLAGS='-Isrc -DNDEBUG=1' \
+		CFLAGS='-std=c99 --coverage -g -O0' \
+		CXXFLAGS='-isystem googletest/include -std=c++11 --coverage -g -O0' \
+		LDFLAGS='--coverage' \
+		build/run_tests
+	lcov	--no-external \
+		--initial \
+		--capture \
+		--base-directory . \
+		--directory build \
+		--output-file build/coverage-pre.info
+	awk	-F '[:,]' \
+		'/^SF:/ { delete defs } /^FN:/ { defs[$$2]=1 } /^DA:/ { if ($$3 == 0 && $$2 in defs) next } { print }' \
+		build/coverage-pre.info > build/coverage-initial.info
+	./build/run_tests
+	lcov	--no-external \
+		--capture \
+		--base-directory . \
+		--directory build \
+		--rc lcov_branch_coverage=1 \
+		--output-file build/coverage-test.info
+	lcov	--add-tracefile build/coverage-initial.info \
+		--add-tracefile build/coverage-test.info \
+		--rc lcov_branch_coverage=1 \
+		--output-file build/coverage.info
+	lcov	--remove build/coverage.info '$(CURDIR)/googletest/*' \
+		--rc lcov_branch_coverage=1 \
+		--output-file build/coverage.info
+	genhtml --branch-coverage \
+		--output-directory build/coverage \
+		build/coverage.info
+
+clean:
+	$(RM) -r build
+
+build/src/flags: | build/src
+	@echo 'old_CC := $(CC)' > $@
+	@echo 'old_CPPFLAGS := $(CPPFLAGS)' >> $@
+	@echo 'old_CFLAGS := $(CFLAGS)' >>$@
+	@echo 'old_SANITIZEFLAGS := $(SANITIZEFLAGS)' >> $@
+	@echo 'old_LDFLAGS := $(LDFLAGS)' >> $@
+
+build/test/flags: | build/test
+	@echo 'old_CXX := $(CXX)' > $@
+	@echo 'old_CPPFLAGS := $(CPPFLAGS)' >> $@
+	@echo 'old_CXXFLAGS := $(CXXFLAGS)' >> $@
+	@echo 'old_SANITIZEFLAGS := $(SANITIZEFLAGS)' >> $@
+	@echo 'old_LDFLAGS := $(LDFLAGS)' >> $@
+
+ifeq (,$(filter clean coverage,$(MAKECMDGOALS)))
+# Ensure that the flags are up to date.
+-include build/src/flags build/test/flags
+ifneq ($(old_CC) | $(old_CPPFLAGS) | $(old_CFLAGS) | $(old_SANITIZEFLAGS) | $(old_LDFLAGS),$(CC) | $(CPPFLAGS) | $(CFLAGS) | $(SANITIZEFLAGS) | $(LDFLAGS))
+.PHONY: build/src/flags
+endif
+ifneq ($(old_CXX) | $(old_CPPFLAGS) | $(old_CXXFLAGS) | $(old_SANITIZEFLAGS) | $(old_LDFLAGS),$(CXX) | $(CPPFLAGS) | $(CXXFLAGS) | $(SANITIZEFLAGS) | $(LDFLAGS))
+.PHONY: build/test/flags
+endif
+
+# Include dependencies.
+-include $(test_objs:.o=.d) $(gumbo_objs:.o=.d)
+endif

data/gumbo-parser/THANKS

@@ -0,0 +1,27 @@
+Gumbo HTML parser THANKS file
+
+Gumbo was originally written by Jonathan Tang, but many people helped out through suggestions, question-answering, code reviews, bugfixes, and organizational support.  Here is a list of these people.  Help me keep it complete and exempt of errors.
+
+Adam Barth
+Adam Roben
+Ben Noordhuis
+Bowen Han
+Constantinos Michael
+Craig Barnes
+Geoffrey Sneddon
+Ian Hickson
+Jack Deng
+Joel Low
+Jonathan Shneier
+Kevin Hendricks
+Mason Tang
+Maxim Zakharov
+Michal Zalewski
+Neal Norwitz
+Othar Hansson
+Ryan Grove
+Stefan Haustein
+Steffen Meschkat
+Steven Kabbes
+Thiago Farina
+Vicent Marti

data/lib/nokogiri/class_resolver.rb

@@ -0,0 +1,67 @@
+# coding: utf-8
+# frozen_string_literal: true
+
+require "set"
+
+module Nokogiri
+  #
+  # Some classes in Nokogiri are namespaced as a group, for example
+  # Document, DocumentFragment, and Builder.
+  #
+  # It's sometimes necessary to look up the related class, e.g.:
+  #
+  #   XML::Builder → XML::Document
+  #   HTML4::Builder → HTML4::Document
+  #   HTML5::Document → HTML5::DocumentFragment
+  #
+  # This module is included into those key classes who need to do this.
+  #
+  module ClassResolver
+    # #related_class restricts matching namespaces to those matching this set.
+    VALID_NAMESPACES = Set.new(["HTML", "HTML4", "HTML5", "XML"])
+
+    # :call-seq:
+    #   related_class(class_name) → Class
+    #
+    # Find a class constant within the
+    #
+    # Some examples:
+    #
+    #   Nokogiri::XML::Document.new.related_class("DocumentFragment")
+    #   # => Nokogiri::XML::DocumentFragment
+    #   Nokogiri::HTML4::Document.new.related_class("DocumentFragment")
+    #   # => Nokogiri::HTML4::DocumentFragment
+    #
+    # Note this will also work for subclasses that follow the same convention, e.g.:
+    #
+    #   Loofah::HTML::Document.new.related_class("DocumentFragment")
+    #   # => Loofah::HTML::DocumentFragment
+    #
+    # And even if it's a subclass, this will iterate through the superclasses:
+    #
+    #   class ThisIsATopLevelClass < Nokogiri::HTML4::Builder ; end
+    #   ThisIsATopLevelClass.new.related_class("Document")
+    #   # => Nokogiri::HTML4::Document
+    #
+    def related_class(class_name)
+      klass = nil
+      inspecting = self.class
+
+      while inspecting
+        namespace_path = inspecting.name.split("::")[0..-2]
+        inspecting = inspecting.superclass
+
+        next unless VALID_NAMESPACES.include?(namespace_path.last)
+
+        related_class_name = (namespace_path << class_name).join("::")
+        klass = begin
+          Object.const_get(related_class_name)
+        rescue NameError
+          nil
+        end
+        break if klass
+      end
+      klass
+    end
+  end
+end

data/lib/nokogiri/css/node.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+module Nokogiri
+  module CSS
+    class Node # :nodoc:
+      ALLOW_COMBINATOR_ON_SELF = [:DIRECT_ADJACENT_SELECTOR, :FOLLOWING_SELECTOR, :CHILD_SELECTOR]
+
+      # Get the type of this node
+      attr_accessor :type
+      # Get the value of this node
+      attr_accessor :value
+
+      # Create a new Node with +type+ and +value+
+      def initialize(type, value)
+        @type = type
+        @value = value
+      end
+
+      # Accept +visitor+
+      def accept(visitor)
+        visitor.send(:"visit_#{type.to_s.downcase}", self)
+      end
+
+      ###
+      # Convert this CSS node to xpath with +prefix+ using +visitor+
+      def to_xpath(prefix, visitor)
+        prefix = "." if ALLOW_COMBINATOR_ON_SELF.include?(type) && value.first.nil?
+        prefix + visitor.accept(self)
+      end
+
+      # Find a node by type using +types+
+      def find_by_type(types)
+        matches = []
+        matches << self if to_type == types
+        @value.each do |v|
+          matches += v.find_by_type(types) if v.respond_to?(:find_by_type)
+        end
+        matches
+      end
+
+      # Convert to_type
+      def to_type
+        [@type] + @value.map do |n|
+          n.to_type if n.respond_to?(:to_type)
+        end.compact
+      end
+
+      # Convert to array
+      def to_a
+        [@type] + @value.map { |n| n.respond_to?(:to_a) ? n.to_a : [n] }
+      end
+    end
+  end
+end