nokolexbor 0.5.4 → 0.6.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 86f8c3bb8682546901c14329e1baea4ca035923f1ade556d2bffd33a4e249788
-  data.tar.gz: b5b14d4c86663d12712fd5d96c59b034c226d0bd90608f07a46591b6ffb597dc
+  metadata.gz: bd05d369d77b5bcd7ba2e513ff1063cf2a72b9eb215172e4db12f71207c1de8e
+  data.tar.gz: 1e476b729e48e5e16753acd7c07306de2503659d3cd3b17f5f1d1674f073b058
 SHA512:
-  metadata.gz: 78aac523a6b2a2a9b843f921e3f59aab84dc8d0c8dce15795b91dfa92828a2feb49362b35d72d333a2f7df73c514d343762690335c5c718c1a1f637d67b54f91
-  data.tar.gz: 5a32ea2d5a01f3b09e6df403a40e0f10701e57ad669c1ee5f5660b1d01b5b4246936b52ca84679d405f6b41da0e4e09abf7c329542234c46ac1267cde3a3169e
+  metadata.gz: 7adf82735fc97339b9990999674c7a2afafa78e52cab6940edebbc3efe74e26f9e9231b9bc7e3ee3895cc5f938cba7891e482724f28f87be0d13256a8536db80
+  data.tar.gz: 599b41913bf994d633d4e3b86ec7013ce43db80f7c3bdb5d4560bfab4d2b5ae4e0fabed26d9807a0b8b493963ffd0b3bf1de16dda9758fe2d15de19df089c22d

data/ext/nokolexbor/extconf.rb

@@ -24,7 +24,20 @@ def which(cmd)
   return nil
 end
 
+if !find_executable('cmake')
+  abort "ERROR: CMake is required to build Lexbor."
+end
+
 cmake_flags = [ ENV["CMAKE_FLAGS"] ]
+
+cmake_version_full = `cmake --version`
+unless cmake_version_full.empty?
+  cmake_version = cmake_version_full[/cmake version ([0-9\.]+)/, 1]
+  if Gem::Version.new(cmake_version) >= Gem::Version.new('4.0')
+    cmake_flags << "-DCMAKE_POLICY_VERSION_MINIMUM=3.5"
+  end
+end
+
 cmake_flags << "-DCMAKE_TRY_COMPILE_TARGET_TYPE=STATIC_LIBRARY"
 # Set system name explicitly when cross-compiling
 cmake_flags << "-DCMAKE_SYSTEM_NAME=Windows -DWIN32=1" if windows?
@@ -64,14 +77,6 @@ end
 append_cflags("-DLEXBOR_STATIC")
 append_cflags("-DLIBXML_STATIC")
 
-def sys(cmd)
-  puts "-- #{cmd}"
-  unless ret = xsystem(cmd)
-    raise "ERROR: '#{cmd}' failed"
-  end
-  ret
-end
-
 # Thrown when we detect CMake is taking too long and we killed it
 class CMakeTimeout < StandardError
 end
@@ -124,10 +129,6 @@ LEXBOR_DIR = File.join(CWD, '..', '..', 'vendor', 'lexbor')
 EXT_DIR = File.join(CWD, '..', '..', 'ext', 'nokolexbor')
 INSTALL_DIR = File.join(LEXBOR_DIR, 'dist')
 
-if !find_executable('cmake')
-  abort "ERROR: CMake is required to build Lexbor."
-end
-
 Dir.chdir(LEXBOR_DIR) do
   # Patch lexbor
   Dir['../../patches/*-lexbor-*.patch'].each do |patch_file|
@@ -138,7 +139,7 @@ Dir.chdir(LEXBOR_DIR) do
 
   Dir.chdir("build") do
     run_cmake(10 * 60, ".. -DCMAKE_INSTALL_PREFIX:PATH=#{INSTALL_DIR} #{lexbor_cmake_flags.join(' ')}")
-    sys("#{MAKE} install")
+    system("#{MAKE}", "install")
   end
 end
 

data/ext/nokolexbor/libxml/tree.h

@@ -23,7 +23,7 @@ extern "C" {
 #endif
 
 static size_t tmp_len;
-#define NODE_NAME(node) lxb_dom_node_name_qualified((node), &tmp_len)
+#define NODE_NAME(node) lxb_dom_node_name_qualified((lxb_dom_node_t *)(node), &tmp_len)
 #define NODE_NS_HREF(node) ((node)->prefix ? lxb_ns_by_id((node)->owner_document->ns, (node)->ns, &tmp_len) : NULL)
 #define NODE_NS_PREFIX(node) lxb_ns_by_id((node)->owner_document->prefix, (node)->prefix, &tmp_len)
 

data/ext/nokolexbor/nl_attribute.c

@@ -141,7 +141,7 @@ nl_attribute_parent(VALUE self)
   if (attr->owner == NULL) {
     return Qnil;
   }
-  return nl_rb_node_create(attr->owner, nl_rb_document_get(self));
+  return nl_rb_node_create((lxb_dom_node_t *)attr->owner, nl_rb_document_get(self));
 }
 
 /**
@@ -158,7 +158,7 @@ nl_attribute_previous(VALUE self)
   if (attr->prev == NULL) {
     return Qnil;
   }
-  return nl_rb_node_create(attr->prev, nl_rb_document_get(self));
+  return nl_rb_node_create((lxb_dom_node_t *)attr->prev, nl_rb_document_get(self));
 }
 
 /**
@@ -175,7 +175,7 @@ nl_attribute_next(VALUE self)
   if (attr->next == NULL) {
     return Qnil;
   }
-  return nl_rb_node_create(attr->next, nl_rb_document_get(self));
+  return nl_rb_node_create((lxb_dom_node_t *)attr->next, nl_rb_document_get(self));
 }
 
 static VALUE
@@ -189,7 +189,7 @@ nl_attribute_inspect(VALUE self)
 
   return rb_sprintf("#<%" PRIsVALUE " %s=\"%s\">", c,
                     lxb_dom_attr_qualified_name(attr, &len),
-                    attr_value == NULL ? "" : attr_value);
+                    attr_value == NULL ? "" : (char *)attr_value);
 }
 
 void Init_nl_attribute(void)

data/ext/nokolexbor/nl_document.c

@@ -5,11 +5,6 @@ extern VALUE mNokolexbor;
 extern VALUE cNokolexborNode;
 VALUE cNokolexborDocument;
 
-#ifdef HAVE_PTHREAD_H
-#include <pthread.h>
-pthread_key_t p_key_parser;
-#endif
-
 static void
 free_nl_document(lxb_html_document_t *document)
 {
@@ -27,47 +22,25 @@ const rb_data_type_t nl_document_type = {
     RUBY_TYPED_FREE_IMMEDIATELY,
 };
 
-/**
- * call-seq:
- *   parse(string_or_io) -> Document
- *
- * Parse HTML into a {Document}.
- *
- * @param string_or_io [String, #read]
- *   The HTML to be parsed. It may be a String, or any object that
- *   responds to #read such as an IO, or StringIO.
- */
 static VALUE
-nl_document_parse(VALUE self, VALUE rb_string_or_io)
+nl_document_parse_native(VALUE self, VALUE rb_html)
 {
-  VALUE id_read = rb_intern("read");
-  VALUE rb_html;
-  if (rb_respond_to(rb_string_or_io, id_read)) {
-    rb_html = rb_funcall(rb_string_or_io, id_read, 0);
-  } else {
-    rb_html = rb_string_or_io;
-  }
   const char *html_c = StringValuePtr(rb_html);
   size_t html_len = RSTRING_LEN(rb_html);
 
-#ifdef HAVE_PTHREAD_H
-  lxb_html_parser_t *g_parser = (lxb_html_parser_t *)pthread_getspecific(p_key_parser);
-#else
-  lxb_html_parser_t *g_parser = NULL;
-#endif
-  if (g_parser == NULL) {
-    g_parser = lxb_html_parser_create();
-    lxb_status_t status = lxb_html_parser_init(g_parser);
+  static lxb_html_parser_t *html_parser = NULL;
+  if (html_parser == NULL) {
+    html_parser = lxb_html_parser_create();
+    lxb_status_t status = lxb_html_parser_init(html_parser);
     if (status != LXB_STATUS_OK) {
+      lxb_html_parser_destroy(html_parser);
+      html_parser = NULL;
       nl_raise_lexbor_error(status);
     }
-    g_parser->tree->scripting = true;
-#ifdef HAVE_PTHREAD_H
-    pthread_setspecific(p_key_parser, g_parser);
-#endif
+    html_parser->tree->scripting = true;
   }
 
-  lxb_html_document_t *document = lxb_html_parse(g_parser, (const lxb_char_t *)html_c, html_len);
+  lxb_html_document_t *document = lxb_html_parse(html_parser, (const lxb_char_t *)html_c, html_len);
 
   if (document == NULL) {
     rb_raise(rb_eRuntimeError, "Error parsing document");
@@ -84,7 +57,7 @@ nl_document_parse(VALUE self, VALUE rb_string_or_io)
 static VALUE
 nl_document_new(VALUE self)
 {
-  return nl_document_parse(self, rb_str_new("", 0));
+  return nl_document_parse_native(self, rb_str_new("", 0));
 }
 
 lxb_dom_document_t *
@@ -104,7 +77,7 @@ static VALUE
 nl_document_get_title(VALUE self)
 {
   size_t len;
-  lxb_char_t *str = lxb_html_document_title(nl_rb_document_unwrap(self), &len);
+  lxb_char_t *str = lxb_html_document_title((lxb_html_document_t *)nl_rb_document_unwrap(self), &len);
   return str == NULL ? rb_str_new("", 0) : rb_utf8_str_new(str, len);
 }
 
@@ -126,7 +99,7 @@ nl_document_set_title(VALUE self, VALUE rb_title)
 {
   const char *c_title = StringValuePtr(rb_title);
   size_t len = RSTRING_LEN(rb_title);
-  lxb_html_document_title_set(nl_rb_document_unwrap(self), (const lxb_char_t *)c_title, len);
+  lxb_html_document_title_set((lxb_html_document_t *)nl_rb_document_unwrap(self), (const lxb_char_t *)c_title, len);
   return rb_title;
 }
 
@@ -142,24 +115,11 @@ nl_document_root(VALUE self)
   return nl_rb_node_create(lxb_dom_document_root(doc), self);
 }
 
-static void
-free_parser(void *data)
-{
-  lxb_html_parser_t *g_parser = (lxb_html_parser_t *)data;
-  if (g_parser != NULL) {
-    g_parser = lxb_html_parser_destroy(g_parser);
-  }
-}
-
 void Init_nl_document(void)
 {
-#ifdef HAVE_PTHREAD_H
-  pthread_key_create(&p_key_parser, free_parser);
-#endif
-
   cNokolexborDocument = rb_define_class_under(mNokolexbor, "Document", cNokolexborNode);
   rb_define_singleton_method(cNokolexborDocument, "new", nl_document_new, 0);
-  rb_define_singleton_method(cNokolexborDocument, "parse", nl_document_parse, 1);
+  rb_define_singleton_method(cNokolexborDocument, "parse_native", nl_document_parse_native, 1);
   rb_define_method(cNokolexborDocument, "title", nl_document_get_title, 0);
   rb_define_method(cNokolexborDocument, "title=", nl_document_set_title, 1);
   rb_define_method(cNokolexborDocument, "root", nl_document_root, 0);

data/ext/nokolexbor/nl_node.c

@@ -159,7 +159,7 @@ nl_node_attribute(VALUE self, VALUE rb_name)
   if (attr->owner == NULL) {
     attr->owner = lxb_dom_interface_element(node);
   }
-  return nl_rb_node_create(attr, nl_rb_document_get(self));
+  return nl_rb_node_create((lxb_dom_node_t *)attr, nl_rb_document_get(self));
 }
 
 /**
@@ -185,7 +185,7 @@ nl_node_attribute_nodes(VALUE self)
     if (attr->owner == NULL) {
       attr->owner = lxb_dom_interface_element(node);
     }
-    rb_ary_push(ary, nl_rb_node_create(attr, rb_doc));
+    rb_ary_push(ary, nl_rb_node_create((lxb_dom_node_t *)attr, rb_doc));
     attr = attr->next;
   }
 
@@ -366,28 +366,32 @@ nl_node_find(VALUE self, VALUE selector, lxb_selectors_cb_f cb, void *ctx)
   lxb_dom_node_t *node = nl_rb_node_unwrap(self);
 
   lxb_status_t status;
-  lxb_css_parser_t *parser = NULL;
-  lxb_selectors_t *selectors = NULL;
+  static lxb_css_parser_t *css_parser = NULL;
+  static lxb_selectors_t *selectors = NULL;
   lxb_css_selector_list_t *list = NULL;
 
-  /* Create CSS parser. */
-  parser = lxb_css_parser_create();
-  status = lxb_css_parser_init(parser, NULL, NULL);
-  if (status != LXB_STATUS_OK) {
-    goto cleanup;
+  /* CSS parser. */
+  if (css_parser == NULL) {
+    css_parser = lxb_css_parser_create();
+    status = lxb_css_parser_init(css_parser, NULL, NULL);
+    if (status != LXB_STATUS_OK) {
+      goto init_error;
+    }
   }
 
   /* Selectors. */
-  selectors = lxb_selectors_create();
-  status = lxb_selectors_init(selectors);
-  if (status != LXB_STATUS_OK) {
-    goto cleanup;
+  if (selectors == NULL) {
+    selectors = lxb_selectors_create();
+    status = lxb_selectors_init(selectors);
+    if (status != LXB_STATUS_OK) {
+      goto init_error;
+    }
   }
 
   /* Parse and get the log. */
-  list = lxb_css_selectors_parse_relative_list(parser, (const lxb_char_t *)selector_c, selector_len);
-  if (parser->status != LXB_STATUS_OK) {
-    status = parser->status;
+  list = lxb_css_selectors_parse_relative_list(css_parser, (const lxb_char_t *)selector_c, selector_len);
+  if (css_parser->status != LXB_STATUS_OK) {
+    status = css_parser->status;
     goto cleanup;
   }
 
@@ -398,11 +402,19 @@ nl_node_find(VALUE self, VALUE selector, lxb_selectors_cb_f cb, void *ctx)
   }
 
 cleanup:
+  /* Destroy all object for all CSS Selector List. */
+  lxb_css_selector_list_destroy_memory(list);
+
+  return status;
+
+init_error:
   /* Destroy Selectors object. */
-  (void)lxb_selectors_destroy(selectors, true);
+  lxb_selectors_destroy(selectors, true);
+  selectors = NULL;
 
   /* Destroy resources for CSS Parser. */
-  (void)lxb_css_parser_destroy(parser, true);
+  lxb_css_parser_destroy(css_parser, true);
+  css_parser = NULL;
 
   /* Destroy all object for all CSS Selector List. */
   lxb_css_selector_list_destroy_memory(list);
@@ -1014,9 +1026,9 @@ static VALUE
 nl_node_add_sibling(VALUE self, VALUE next_or_previous, VALUE new)
 {
   bool insert_after;
-  if (rb_eql(rb_String(next_or_previous), rb_str_new_literal("next"))) {
+  if (rb_str_cmp(rb_String(next_or_previous), rb_str_new_literal("next")) == 0) {
     insert_after = true;
-  } else if (rb_eql(rb_String(next_or_previous), rb_str_new_literal("previous"))) {
+  } else if (rb_str_cmp(rb_String(next_or_previous), rb_str_new_literal("previous")) == 0) {
     insert_after = false;
   } else {
     rb_raise(rb_eArgError, "Unsupported inserting position");

data/ext/nokolexbor/xml_tree.c

@@ -339,8 +339,8 @@ nl_xmlGetNodePath(const lxb_dom_node_t *node)
 
         } else if (cur->type == LXB_DOM_NODE_TYPE_ATTRIBUTE) {
             sep = "/@";
-            name = (const char *) lxb_dom_attr_qualified_name(cur, &tmp_len);
-            next = ((lxb_dom_attr_t_ptr)cur)->owner;
+            name = (const char *) lxb_dom_attr_qualified_name((lxb_dom_attr_t_ptr)cur, &tmp_len);
+            next = (lxb_dom_node_t *)((lxb_dom_attr_t_ptr)cur)->owner;
         } else {
             nl_xmlFree(buf);
             nl_xmlFree(buffer);

data/ext/nokolexbor/xml_xpath.c

@@ -6389,7 +6389,7 @@ xmlXPathNodeValHash(lxb_dom_node_t_ptr node) {
 		return(0);
 	    return(string[0] + (string[1] << 8));
 	case LXB_DOM_NODE_TYPE_ATTRIBUTE:
-	    string = lxb_dom_attr_value(node, &tmp_len);
+	    string = lxb_dom_attr_value((lxb_dom_attr_t_ptr)node, &tmp_len);
 	    if (string == NULL)
 		return(0);
 	    if (string[0] == 0)
@@ -8452,9 +8452,9 @@ nl_xmlXPathNextAttribute(xmlXPathParserContextPtr ctxt, lxb_dom_node_t_ptr cur)
     if (cur == NULL) {
         if (ctxt->context->node == (lxb_dom_node_t_ptr) ctxt->context->doc)
 	    return(NULL);
-        return lxb_dom_element_first_attribute(ctxt->context->node);
+        return (lxb_dom_node_t_ptr)lxb_dom_element_first_attribute((lxb_dom_element_t *)ctxt->context->node);
     }
-    return(((lxb_dom_attr_t *)cur)->next);
+    return (lxb_dom_node_t_ptr)(((lxb_dom_attr_t *)cur)->next);
 }
 
 /************************************************************************

data/lib/nokolexbor/document.rb

@@ -141,6 +141,26 @@ module Nokolexbor
       end
     end
 
+    # Parse HTML into a {Document}.
+    #
+    # @param string_or_io [String, #read]
+    #   The HTML to be parsed. It may be a String, or any object that
+    #   responds to #read such as an IO, or StringIO.
+    #
+    # @return [Document]
+    def self.parse(string_or_io)
+      html = string_or_io
+      if string_or_io.respond_to?(:read)
+        html = string_or_io.read
+      end
+
+      if html.respond_to?(:encoding) && html.encoding != Encoding::UTF_8
+        html = html.encode(Encoding::UTF_8, invalid: :replace, undef: :replace)
+      end
+
+      parse_native(html)
+    end
+
     private
 
     IMPLIED_XPATH_CONTEXTS = ["//"].freeze # :nodoc:

data/lib/nokolexbor/node.rb

@@ -717,18 +717,38 @@ module Nokolexbor
 
     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,
-        )
+
+      unless defined?(Gem)
+        require 'rubygems'
+      end
+
+      v_1_17_0 = Gem::Version.new("1.17.0")
+      current_version = Gem::Version.new(Nokogiri::VERSION)
+
+      if current_version < v_1_17_0
+        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(" | ")
       else
-        visitor = Nokogiri::CSS::XPathVisitorOptimallyUseBuiltins.new
+        self.class::IMPLIED_XPATH_CONTEXTS.map do |implied_xpath_context|
+          visitor = Nokogiri::CSS::XPathVisitor.new(
+            builtins: Nokogiri::CSS::XPathVisitor::BuiltinsConfig::OPTIMAL,
+            doctype: :html4,
+            prefix: implied_xpath_context,
+            namespaces: ns,
+          )
+          Nokogiri::CSS.xpath_for(rule.to_s, visitor: visitor)
+        end.join(" | ")
       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)

data/lib/nokolexbor/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Nokolexbor
-  VERSION = '0.5.4'
+  VERSION = '0.6.1'
 end

data/patches/0004-lexbor-fix-template-clone.patch

@@ -16,7 +16,7 @@ index a2153f4..8a9c69f 100755
  
 +        if (curr->local_name == LXB_TAG_TEMPLATE && curr->first_child != NULL && cnode->type == LXB_DOM_NODE_TYPE_DOCUMENT_FRAGMENT) {
 +            lxb_dom_node_remove(curr->first_child);
-+            lxb_html_interface_template(curr)->content = cnode;
++            lxb_html_interface_template(curr)->content = (lxb_dom_document_fragment_t *)cnode;
 +        }
 +
          lxb_dom_node_insert_child(curr, cnode);

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: nokolexbor
 version: !ruby/object:Gem::Version
-  version: 0.5.4
+  version: 0.6.1
 platform: ruby
 authors:
 - Yicheng Zhou
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-04-22 00:00:00.000000000 Z
+date: 2025-05-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler
@@ -523,7 +523,7 @@ licenses:
 - MIT
 metadata:
   msys2_mingw_dependencies: cmake
-post_install_message: 
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -539,7 +539,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubygems_version: 3.0.3.1
-signing_key: 
+signing_key:
 specification_version: 4
 summary: High-performance HTML5 parser, with support for both CSS selectors and XPath.
 test_files: []