nokolexbor 0.3.1 → 0.3.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6bfa3062e8718581cd8aabef75f0084a6626bd90301c9db4957ee975215edec1
-  data.tar.gz: 243f5c217c85750f63bfc0dcb8af575c019d3f2fd920f41769c0affc8a219802
+  metadata.gz: e83f343175997fd437b89893ae63b7cc6eaffa86711800a42e8ce64371c03c85
+  data.tar.gz: ea272e36f2a37cd5bc3e8e13b3fa8993098503abdaa85299bcf99f59ae484918
 SHA512:
-  metadata.gz: e66b390ef696ed5cafa480065e8ac8fc66ce98d86a7cb15f2f59d8acb7df21381942e5fe6f0244e3b9c7bf54787131c5241a49faeaba328eb49e65af0ac3216c
-  data.tar.gz: d266e2a7c49a202aec4112667807158b54878cc014b42fc32492fc0977c06db76b1a5ded827d9810f2e7966ba3e5068fe54ab89915a4c56e499181ddc190b6d7
+  metadata.gz: be0f026e8f2571705f1ae6225ddc4a3b7e78d7108e441a63d8b0988d4d0c91afa7af57ce8c6c318c1f2a5c38ebf00d9190ce29ec27de8e17505250273209d834
+  data.tar.gz: 0000f9c78d7743e789c8edc11aaeaf4bf8a75503d0b6631a0a25a9d3c571a56fb424fc69d3e4c2a737d38c4a84c23424cd79a87f4b6f967b749522b217bd1abd

data/ext/nokolexbor/libxml/xpathInternals.h

@@ -477,11 +477,10 @@ XMLPUBFUN xmlXPathParserContextPtr XMLCALL
 XMLPUBFUN void XMLCALL
 		nl_xmlXPathFreeParserContext	(xmlXPathParserContextPtr ctxt);
 
-/* TODO: remap to xmlXPathValuePop and Push. */
-static xmlXPathObjectPtr XMLCALL
-		valuePop			(xmlXPathParserContextPtr ctxt);
-static int XMLCALL
-		valuePush			(xmlXPathParserContextPtr ctxt,
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		nl_xmlXPathValuePop		(xmlXPathParserContextPtr ctxt);
+XMLPUBFUN int XMLCALL
+		nl_xmlXPathValuePush		(xmlXPathParserContextPtr ctxt,
 						 xmlXPathObjectPtr value);
 
 XMLPUBFUN xmlXPathObjectPtr XMLCALL

data/ext/nokolexbor/nl_xpath_context.c

@@ -15,12 +15,110 @@ VALUE cNokolexborXpathContext;
 VALUE mNokolexborXpath;
 VALUE cNokolexborXpathSyntaxError;
 
+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
 free_xml_xpath_context(xmlXPathContextPtr ctx)
 {
   nl_xmlXPathFreeContext(ctx);
 }
 
+/* find a CSS class in an HTML element's `class` attribute */
+static const xmlChar *
+builtin_css_class(const xmlChar *str, const xmlChar *val)
+{
+  int val_len;
+
+  if (str == NULL) {
+    return (NULL);
+  }
+  if (val == NULL) {
+    return (NULL);
+  }
+
+  val_len = nl_xmlStrlen(val);
+  if (val_len == 0) {
+    return (str);
+  }
+
+  while (*str != 0) {
+    if ((*str == *val) && !nl_xmlStrncmp(str, val, val_len)) {
+      const xmlChar *next_byte = str + val_len;
+
+      /* only match if the next byte is whitespace or end of string */
+      if ((*next_byte == 0) || (IS_BLANK_CH(*next_byte))) {
+        return ((const xmlChar *)str);
+      }
+    }
+
+    /* advance str to whitespace */
+    while ((*str != 0) && !IS_BLANK_CH(*str)) {
+      str++;
+    }
+
+    /* advance str to start of next word or end of string */
+    while ((*str != 0) && IS_BLANK_CH(*str)) {
+      str++;
+    }
+  }
+
+  return (NULL);
+}
+
+/* xmlXPathFunction to wrap builtin_css_class() */
+static void
+xpath_builtin_css_class(xmlXPathParserContextPtr ctxt, int nargs)
+{
+  xmlXPathObjectPtr hay, needle;
+
+  CHECK_ARITY(2);
+
+  CAST_TO_STRING;
+  needle = nl_xmlXPathValuePop(ctxt);
+  if ((needle == NULL) || (needle->type != XPATH_STRING)) {
+    nl_xmlXPathFreeObject(needle);
+    XP_ERROR(XPATH_INVALID_TYPE);
+  }
+
+  CAST_TO_STRING;
+  hay = nl_xmlXPathValuePop(ctxt);
+  if ((hay == NULL) || (hay->type != XPATH_STRING)) {
+    nl_xmlXPathFreeObject(hay);
+    nl_xmlXPathFreeObject(needle);
+    XP_ERROR(XPATH_INVALID_TYPE);
+  }
+
+  if (builtin_css_class(hay->stringval, needle->stringval)) {
+    nl_xmlXPathValuePush(ctxt, nl_xmlXPathNewBoolean(1));
+  } else {
+    nl_xmlXPathValuePush(ctxt, nl_xmlXPathNewBoolean(0));
+  }
+
+  nl_xmlXPathFreeObject(hay);
+  nl_xmlXPathFreeObject(needle);
+}
+
+/* xmlXPathFunction to select nodes whose local name matches, for HTML5 CSS queries that should ignore namespaces */
+static void
+xpath_builtin_local_name_is(xmlXPathParserContextPtr ctxt, int nargs)
+{
+  xmlXPathObjectPtr element_name;
+  size_t tmp_len;
+
+  CHECK_ARITY(1);
+  CAST_TO_STRING;
+  CHECK_TYPE(XPATH_STRING);
+  element_name = nl_xmlXPathValuePop(ctxt);
+
+  const lxb_char_t *node_name = lxb_dom_node_name_qualified(ctxt->context->node, &tmp_len);
+  nl_xmlXPathValuePush(ctxt, nl_xmlXPathNewBoolean(nl_xmlStrEqual((xmlChar *)node_name, element_name->stringval)));
+
+  nl_xmlXPathFreeObject(element_name);
+}
+
 /*
  * call-seq:
  *  register_ns(prefix, uri)
@@ -228,6 +326,13 @@ nl_xpath_context_new(VALUE klass, VALUE rb_node)
   ctx = nl_xmlXPathNewContext(node->owner_document);
   ctx->node = node;
 
+  nl_xmlXPathRegisterNs(ctx, NOKOGIRI_PREFIX, NOKOGIRI_URI);
+  nl_xmlXPathRegisterNs(ctx, NOKOGIRI_BUILTIN_PREFIX, NOKOGIRI_BUILTIN_URI);
+  nl_xmlXPathRegisterFuncNS(ctx, (const xmlChar *)"css-class", NOKOGIRI_BUILTIN_URI,
+                         xpath_builtin_css_class);
+  nl_xmlXPathRegisterFuncNS(ctx, (const xmlChar *)"local-name-is", NOKOGIRI_BUILTIN_URI,
+                         xpath_builtin_local_name_is);
+
   self = Data_Wrap_Struct(klass, 0, free_xml_xpath_context, ctx);
   rb_iv_set(self, "@document", nl_rb_document_get(rb_node));
 

data/ext/nokolexbor/nokolexbor.c

@@ -54,6 +54,9 @@ void nl_raise_lexbor_error(lxb_status_t error)
 
 void Init_nokolexbor(void)
 {
+#ifndef NOKOLEXBOR_ASAN
+  lexbor_memory_setup(ruby_xmalloc, ruby_xrealloc, ruby_xcalloc, ruby_xfree);
+#endif
   mNokolexbor = rb_define_module("Nokolexbor");
   eLexborError = rb_define_class_under(mNokolexbor, "LexborError", rb_eStandardError);
   eLexborSyntaxError = rb_define_class_under(mNokolexbor, "LexborSyntaxError", eLexborError);

data/ext/nokolexbor/xml_xpath.c

@@ -2897,6 +2897,10 @@ valuePop(xmlXPathParserContextPtr ctxt)
     ctxt->valueTab[ctxt->valueNr] = NULL;
     return (ret);
 }
+
+xmlXPathObjectPtr
+nl_xmlXPathValuePop(xmlXPathParserContextPtr ctxt) { return valuePop(ctxt); }
+
 /**
  * valuePush:
  * @ctxt:  an XPath evaluation context
@@ -2941,6 +2945,9 @@ valuePush(xmlXPathParserContextPtr ctxt, xmlXPathObjectPtr value)
     return (ctxt->valueNr++);
 }
 
+int
+nl_xmlXPathValuePush(xmlXPathParserContextPtr ctxt, xmlXPathObjectPtr value) { return valuePush(ctxt, value); }
+
 /**
  * nl_xmlXPathPopBoolean:
  * @ctxt:  an XPath parser context

data/lib/nokolexbor/document.rb

@@ -93,5 +93,9 @@ module Nokolexbor
         head.prepend_child(element)
       end
     end
+
+    private
+
+    IMPLIED_XPATH_CONTEXTS = ["//"].freeze # :nodoc:
   end
 end

data/lib/nokolexbor/node.rb

@@ -192,6 +192,16 @@ module Nokolexbor
       at_css_impl(args.join(', '))
     end
 
+    def nokogiri_css(*args)
+      rules, handler, ns, _ = extract_params(args)
+
+      nokogiri_css_internal(self, rules, handler, ns)
+    end
+
+    def nokogiri_at_css(*args)
+      nokogiri_css(*args).first
+    end
+
     def xpath(*args)
       paths, handler, ns, binds = extract_params(args)
 
@@ -299,6 +309,10 @@ module Nokolexbor
 
     private
 
+    def nokogiri_css_internal(node, rules, handler, ns)
+      xpath_internal(node, css_rules_to_xpath(rules, ns), handler, ns, nil)
+    end
+
     def xpath_internal(node, paths, handler, ns, binds)
       # document = node.document
       # return NodeSet.new(document) unless document
@@ -326,6 +340,34 @@ module Nokolexbor
       ctx.evaluate(path, handler)
     end
 
+    def css_rules_to_xpath(rules, ns)
+      rules.map { |rule| xpath_query_from_css_rule(rule, ns) }
+    end
+
+    def ensure_nokogiri
+      unless defined?(Nokogiri) && defined?(Nokogiri::CSS)
+        require 'nokogiri'
+      end
+    rescue LoadError
+      fail('nokogiri_css and nokogiri_at_css require Nokogiri to be installed')
+    end
+
+    def xpath_query_from_css_rule(rule, ns)
+      ensure_nokogiri
+      if defined? Nokogiri::CSS::XPathVisitor::BuiltinsConfig
+        visitor = Nokogiri::CSS::XPathVisitor.new(
+          builtins: Nokogiri::CSS::XPathVisitor::BuiltinsConfig::OPTIMAL,
+          doctype: :html4,
+        )
+      else
+        visitor = Nokogiri::CSS::XPathVisitorOptimallyUseBuiltins.new
+      end
+      self.class::IMPLIED_XPATH_CONTEXTS.map do |implied_xpath_context|
+        Nokogiri::CSS.xpath_for(rule.to_s, { prefix: implied_xpath_context, ns: ns,
+                                   visitor: visitor, })
+      end.join(" | ")
+    end
+
     def extract_params(params)
       handler = params.find do |param|
         ![Hash, String, Symbol].include?(param.class)
@@ -344,5 +386,7 @@ module Nokolexbor
 
       [params, handler, ns, binds]
     end
+
+    IMPLIED_XPATH_CONTEXTS = [".//"].freeze
   end
-end
+end

data/lib/nokolexbor/node_set.rb

@@ -126,6 +126,23 @@ module Nokolexbor
         end
       end
     end
+
+    def nokogiri_css(*args)
+      rules, handler, ns, _ = extract_params(args)
+      paths = css_rules_to_xpath(rules, ns)
+
+      NodeSet.new(@document) do |set|
+        each do |node|
+          node.send(:xpath_internal, node, paths, handler, ns, nil).each do |inner_node|
+            set << inner_node
+          end
+        end
+      end
+    end
+
+    private
+
+    IMPLIED_XPATH_CONTEXTS = [".//", "self::"].freeze # :nodoc:
       
   end
 end

data/lib/nokolexbor/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Nokolexbor
-  VERSION = '0.3.1'
+  VERSION = '0.3.2'
 end

data/patches/0001-lexbor-support-text-pseudo-element.patch

@@ -13,7 +13,7 @@ index 2b19f18..5e56181 100644
  lxb_css_selector_pseudo_element_id_t;
  
 diff --git i/source/lexbor/css/selectors/pseudo_res.h w/source/lexbor/css/selectors/pseudo_res.h
-index 398d1bb..b35bfab 100644
+index 398d1bb..ddd5f3e 100644
 --- i/source/lexbor/css/selectors/pseudo_res.h
 +++ w/source/lexbor/css/selectors/pseudo_res.h
 @@ -124,7 +124,8 @@ static const lxb_css_selectors_pseudo_data_t lxb_css_selectors_pseudo_data_pseud
@@ -36,7 +36,7 @@ index 398d1bb..b35bfab 100644
      {"grammar-error", (void *) &lxb_css_selectors_pseudo_data_pseudo_element[LXB_CSS_SELECTOR_PSEUDO_ELEMENT_GRAMMAR_ERROR], 13, 0}, 
      {"before", (void *) &lxb_css_selectors_pseudo_data_pseudo_element[LXB_CSS_SELECTOR_PSEUDO_ELEMENT_BEFORE], 6, 0}, 
 diff --git i/source/lexbor/selectors/selectors.c w/source/lexbor/selectors/selectors.c
-index c2d18ea..84b5b14 100644
+index fef05d3..1ceaa19 100644
 --- i/source/lexbor/selectors/selectors.c
 +++ w/source/lexbor/selectors/selectors.c
 @@ -156,6 +156,12 @@ lxb_selectors_descendant(lxb_selectors_t *selectors, lxb_selectors_entry_t *entr
@@ -91,16 +91,7 @@ index c2d18ea..84b5b14 100644
  
          node = node->next;
      }
-@@ -262,6 +286,8 @@ lxb_selectors_sibling(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
- {
-     node = node->next;
- 
-+    if (node == NULL) return NULL;
-+
-     do {
-         if (node->type == LXB_DOM_NODE_TYPE_ELEMENT) {
-             if (lxb_selectors_match(selectors, entry, selector, node)) {
-@@ -290,6 +316,12 @@ lxb_selectors_following(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry
+@@ -289,6 +313,12 @@ lxb_selectors_following(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry
          {
              return node;
          }
@@ -113,7 +104,7 @@ index c2d18ea..84b5b14 100644
  
          node = node->next;
      }
-@@ -1429,6 +1461,8 @@ lxb_selectors_pseudo_element(lxb_selectors_t *selectors,
+@@ -1428,6 +1458,8 @@ lxb_selectors_pseudo_element(lxb_selectors_t *selectors,
      lxb_css_selector_pseudo_t *pseudo = &selector->u.pseudo;
  
      switch (pseudo->type) {

data/patches/0002-lexbor-match-id-class-case-sensitive.patch

@@ -2,7 +2,7 @@ diff --git i/source/lexbor/selectors/selectors.c w/source/lexbor/selectors/selec
 index c2d18ea..84b5b14 100644
 --- i/source/lexbor/selectors/selectors.c
 +++ w/source/lexbor/selectors/selectors.c
-@@ -751,7 +783,7 @@ lxb_selectors_match(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
+@@ -750,7 +780,7 @@ lxb_selectors_match(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
              src = &selector->name;
  
              if (trg->length == src->length
@@ -11,7 +11,7 @@ index c2d18ea..84b5b14 100644
              {
                  return true;
              }
-@@ -766,7 +798,7 @@ lxb_selectors_match(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
+@@ -765,7 +795,7 @@ lxb_selectors_match(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
              }
  
              return lxb_selectors_match_class(element->attr_class->value,

data/vendor/lexbor/source/lexbor/core/lexbor.h

@@ -13,6 +13,10 @@ extern "C" {
 
 #include "lexbor/core/def.h"
 
+typedef void *(*lexbor_memory_malloc_f)(size_t size);
+typedef void *(*lexbor_memory_realloc_f)(void *dst, size_t size);
+typedef void *(*lexbor_memory_calloc_f)(size_t num, size_t size);
+typedef void (*lexbor_memory_free_f)(void *dst);
 
 LXB_API void *
 lexbor_malloc(size_t size);
@@ -26,6 +30,10 @@ lexbor_calloc(size_t num, size_t size);
 LXB_API void *
 lexbor_free(void *dst);
 
+LXB_API lxb_status_t
+lexbor_memory_setup(lexbor_memory_malloc_f new_malloc, lexbor_memory_realloc_f new_realloc,
+                    lexbor_memory_calloc_f new_calloc, lexbor_memory_free_f new_free);
+
 
 #ifdef __cplusplus
 } /* extern "C" */

data/vendor/lexbor/source/lexbor/ports/posix/lexbor/core/memory.c

@@ -6,28 +6,48 @@
 
 #include "lexbor/core/base.h"
 
+static lexbor_memory_malloc_f lexbor_memory_malloc = malloc;
+static lexbor_memory_realloc_f lexbor_memory_realloc = realloc;
+static lexbor_memory_calloc_f lexbor_memory_calloc = calloc;
+static lexbor_memory_free_f lexbor_memory_free = free;
 
 void *
 lexbor_malloc(size_t size)
 {
-    return malloc(size);
+    return lexbor_memory_malloc(size);
 }
 
 void *
 lexbor_realloc(void *dst, size_t size)
 {
-    return realloc(dst, size);
+    return lexbor_memory_realloc(dst, size);
 }
 
 void *
 lexbor_calloc(size_t num, size_t size)
 {
-    return calloc(num, size);
+    return lexbor_memory_calloc(num, size);
 }
 
 void *
 lexbor_free(void *dst)
 {
-    free(dst);
+    lexbor_memory_free(dst);
     return NULL;
 }
+
+lxb_status_t
+lexbor_memory_setup(lexbor_memory_malloc_f new_malloc, lexbor_memory_realloc_f new_realloc,
+                    lexbor_memory_calloc_f new_calloc, lexbor_memory_free_f new_free)
+{
+    if (new_malloc == NULL || new_realloc == NULL || new_calloc == NULL || new_free == NULL) {
+        return LXB_STATUS_ERROR_OBJECT_IS_NULL;
+    }
+
+    lexbor_memory_malloc = new_malloc;
+    lexbor_memory_realloc = new_realloc;
+    lexbor_memory_calloc = new_calloc;
+    lexbor_memory_free = new_free;
+
+    return LXB_STATUS_OK;
+}

data/vendor/lexbor/source/lexbor/ports/windows_nt/lexbor/core/memory.c

@@ -6,28 +6,48 @@
 
 #include "lexbor/core/base.h"
 
+static lexbor_memory_malloc_f lexbor_memory_malloc = malloc;
+static lexbor_memory_realloc_f lexbor_memory_realloc = realloc;
+static lexbor_memory_calloc_f lexbor_memory_calloc = calloc;
+static lexbor_memory_free_f lexbor_memory_free = free;
 
 void *
 lexbor_malloc(size_t size)
 {
-    return malloc(size);
+    return lexbor_memory_malloc(size);
 }
 
 void *
 lexbor_realloc(void *dst, size_t size)
 {
-    return realloc(dst, size);
+    return lexbor_memory_realloc(dst, size);
 }
 
 void *
 lexbor_calloc(size_t num, size_t size)
 {
-    return calloc(num, size);
+    return lexbor_memory_calloc(num, size);
 }
 
 void *
 lexbor_free(void *dst)
 {
-    free(dst);
+    lexbor_memory_free(dst);
     return NULL;
 }
+
+lxb_status_t
+lexbor_memory_setup(lexbor_memory_malloc_f new_malloc, lexbor_memory_realloc_f new_realloc,
+                    lexbor_memory_calloc_f new_calloc, lexbor_memory_free_f new_free)
+{
+    if (new_malloc == NULL || new_realloc == NULL || new_calloc == NULL || new_free == NULL) {
+        return LXB_STATUS_ERROR_OBJECT_IS_NULL;
+    }
+
+    lexbor_memory_malloc = new_malloc;
+    lexbor_memory_realloc = new_realloc;
+    lexbor_memory_calloc = new_calloc;
+    lexbor_memory_free = new_free;
+
+    return LXB_STATUS_OK;
+}

data/vendor/lexbor/source/lexbor/selectors/selectors.c

@@ -262,7 +262,7 @@ lxb_selectors_sibling(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
 {
     node = node->next;
 
-    do {
+    while (node != NULL) {
         if (node->type == LXB_DOM_NODE_TYPE_ELEMENT) {
             if (lxb_selectors_match(selectors, entry, selector, node)) {
                 return node;
@@ -273,7 +273,6 @@ lxb_selectors_sibling(lxb_selectors_t *selectors, lxb_selectors_entry_t *entry,
 
         node = node->next;
     }
-    while (node != NULL);
 
     return NULL;
 }

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: nokolexbor
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.3.2
 platform: ruby
 authors:
 - Yicheng Zhou
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-12-31 00:00:00.000000000 Z
+date: 2023-01-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler
@@ -78,7 +78,6 @@ files:
 - ext/nokolexbor/libxml/xpath.h
 - ext/nokolexbor/libxml/xpathInternals.h
 - ext/nokolexbor/libxml/xpointer.h
-- ext/nokolexbor/memory.c
 - ext/nokolexbor/nl_cdata.c
 - ext/nokolexbor/nl_comment.c
 - ext/nokolexbor/nl_document.c

data/ext/nokolexbor/memory.c

@@ -1,46 +0,0 @@
-/*
- * This file overrides the default `lexbor_malloc` and other
- * related functions to reduce memory bloat and allow Ruby to
- * GC more often.
- * By calling `ruby_xmalloc` instead of `malloc`, Ruby can
- * count the memory usage of the C extension and run GC
- * whenever `malloc_increase_byte` exceeds the limit. Similar
- * to Nokogiri's solution by calling `xmlMemSetup`.
- * The downside will be the downgrade of performance because
- * of more frequent GC.
- */
-
-#include "lexbor/core/base.h"
-#include <ruby.h>
-
-// Disable using ruby memory functions when ASAN is enabled,
-// otherwise memory leak info will be all about ruby which
-// is useless.
-#ifndef NOKOLEXBOR_ASAN
-
-void *
-lexbor_malloc(size_t size)
-{
-  return ruby_xmalloc(size);
-}
-
-void *
-lexbor_realloc(void *dst, size_t size)
-{
-  return ruby_xrealloc(dst, size);
-}
-
-void *
-lexbor_calloc(size_t num, size_t size)
-{
-  return ruby_xcalloc(num, size);
-}
-
-void *
-lexbor_free(void *dst)
-{
-  ruby_xfree(dst);
-  return NULL;
-}
-
-#endif