nokolexbor 0.3.7 → 0.4.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 25a206e02eaf74545bf1e7abe7e43a53e38fd8626b337e8de7d6c207e352a686
-  data.tar.gz: 2b9f7c8a339b2618dbb2de5824b249f0a8fa690713028a9e22b8627a8bb0163a
+  metadata.gz: 0055757bc7b9a92e9a2d37235722ca6a6945fb438855880ce10055dfa10d0fcb
+  data.tar.gz: b42cb00bc57ac3db09e928ff0b4ae3b3f7fb950312ab64f2606436184785dcc0
 SHA512:
-  metadata.gz: 70e579aa1e64ae9cfcf71dd2d0bf3b77b634b69764d43c04aaa02b6db0c5b26cb992d0094ad4b2b9177ca2be46c6b32df38381224bd579bfb722c538075b436e
-  data.tar.gz: c89997e12178393f171bb5e3735346d99ee67047fb985ec02b08e9433aadc6f18d68278a5256c99ab7174ddd436f9bb49fb97a97808923205803097b083f7ed3
+  metadata.gz: 5527c86fea7d9efbbea6bcc1a2fcaaad506057ccf2d22d689f0de143b6c142f9bd1e8b33686ed2be1d01c37c1b02879c35e392a0ca81489f7f9cb125e569bf33
+  data.tar.gz: 4af797ac972b1ddfaafceec1ded5de2541daa92a366568f768f63f978bf3215e9a03295f598f4e05eea253aa2af8098ff8de33431119e7aeecad8a223de0c6f4

data/ext/nokolexbor/nl_node.c

@@ -802,6 +802,28 @@ nl_node_children(VALUE self)
   return nl_rb_node_set_create_with_data(array, nl_rb_document_get(self));
 }
 
+/**
+ * Get the element children of this node.
+ *
+ * @return [NodeSet] The set of this node's element children.
+ */
+static VALUE
+nl_node_element_children(VALUE self)
+{
+  lxb_dom_node_t *node = nl_rb_node_unwrap(self);
+  lxb_dom_node_t *child = node->first_child;
+  lexbor_array_t *array = lexbor_array_create();
+
+  while (child != NULL) {
+    if (child->type == LXB_DOM_NODE_TYPE_ELEMENT) {
+      lexbor_array_push(array, child);
+    }
+    child = child->next;
+  }
+
+  return nl_rb_node_set_create_with_data(array, nl_rb_document_get(self));
+}
+
 /**
  * Get the first child of this node.
  *
@@ -1031,20 +1053,8 @@ static VALUE
 nl_node_first_element_child(VALUE self)
 {
   lxb_dom_node_t *parent = nl_rb_node_unwrap(self);
-  lxb_dom_node_t *cur;
+  lxb_dom_node_t *cur = parent->first_child;
 
-  if (parent == NULL) {
-    return Qnil;
-  }
-  switch (parent->type) {
-  case LXB_DOM_NODE_TYPE_ELEMENT:
-  case LXB_DOM_NODE_TYPE_ENTITY:
-  case LXB_DOM_NODE_TYPE_DOCUMENT:
-    cur = parent->first_child;
-    break;
-  default:
-    return Qnil;
-  }
   while (cur != NULL) {
     if (cur->type == LXB_DOM_NODE_TYPE_ELEMENT) {
       return nl_rb_node_create(cur, nl_rb_document_get(self));
@@ -1061,20 +1071,8 @@ static VALUE
 nl_node_last_element_child(VALUE self)
 {
   lxb_dom_node_t *parent = nl_rb_node_unwrap(self);
-  lxb_dom_node_t *cur;
+  lxb_dom_node_t *cur = parent->last_child;
 
-  if (parent == NULL) {
-    return Qnil;
-  }
-  switch (parent->type) {
-  case LXB_DOM_NODE_TYPE_ELEMENT:
-  case LXB_DOM_NODE_TYPE_ENTITY:
-  case LXB_DOM_NODE_TYPE_DOCUMENT:
-    cur = parent->last_child;
-    break;
-  default:
-    return Qnil;
-  }
   while (cur != NULL) {
     if (cur->type == LXB_DOM_NODE_TYPE_ELEMENT) {
       return nl_rb_node_create(cur, nl_rb_document_get(self));
@@ -1133,6 +1131,16 @@ nl_node_inspect(int argc, VALUE *argv, VALUE self)
   return rb_call_super(argc, argv);
 }
 
+/**
+ * @return [Integer] The node's location at the source HTML. Returns 0 if the node is not parsed from a HTML string.
+ */
+static VALUE
+nl_node_source_location(VALUE self)
+{
+  lxb_dom_node_t *node = nl_rb_node_unwrap(self);
+  return ULONG2NUM(node->source_location);
+}
+
 void Init_nl_node(void)
 {
   cNokolexborNode = rb_define_class_under(mNokolexbor, "Node", rb_cObject);
@@ -1164,6 +1172,7 @@ void Init_nl_node(void)
   rb_define_method(cNokolexborNode, "next_element", nl_node_next_element, 0);
   rb_define_method(cNokolexborNode, "children", nl_node_children, 0);
   rb_define_method(cNokolexborNode, "child", nl_node_child, 0);
+  rb_define_method(cNokolexborNode, "element_children", nl_node_element_children, 0);
   rb_define_method(cNokolexborNode, "remove", nl_node_remove, 0);
   rb_define_method(cNokolexborNode, "destroy", nl_node_destroy, 0);
   rb_define_method(cNokolexborNode, "attrs", nl_node_attrs, 0);
@@ -1176,6 +1185,7 @@ void Init_nl_node(void)
   rb_define_method(cNokolexborNode, "last_element_child", nl_node_last_element_child, 0);
   rb_define_method(cNokolexborNode, "clone", nl_node_clone, 0);
   rb_define_method(cNokolexborNode, "inspect", nl_node_inspect, -1);
+  rb_define_method(cNokolexborNode, "source_location", nl_node_source_location, 0);
 
   rb_define_alias(cNokolexborNode, "attr", "[]");
   rb_define_alias(cNokolexborNode, "get_attribute", "[]");
@@ -1183,6 +1193,7 @@ void Init_nl_node(void)
   rb_define_alias(cNokolexborNode, "set_attribute", "[]=");
   rb_define_alias(cNokolexborNode, "has_attribute?", "key?");
   rb_define_alias(cNokolexborNode, "delete", "remove_attr");
+  rb_define_alias(cNokolexborNode, "elements", "element_children");
   rb_define_alias(cNokolexborNode, "remove_attribute", "remove_attr");
   rb_define_alias(cNokolexborNode, "text", "content");
   rb_define_alias(cNokolexborNode, "inner_text", "content");

data/ext/nokolexbor/nl_node_set.c

@@ -300,6 +300,64 @@ nl_node_set_union(VALUE self, VALUE other)
   return nl_rb_node_set_create_with_data(new_array, nl_rb_document_get(self));
 }
 
+/**
+ * @return [NodeSet] A new NodeSet with the common nodes only.
+ */
+static VALUE
+nl_node_set_intersection(VALUE self, VALUE other)
+{
+  if (!rb_obj_is_kind_of(other, cNokolexborNodeSet)) {
+    rb_raise(rb_eArgError, "Parameter must be a Nokolexbor::NodeSet");
+  }
+
+  lexbor_array_t *self_array = nl_rb_node_set_unwrap(self);
+  lexbor_array_t *other_array = nl_rb_node_set_unwrap(other);
+
+  lexbor_array_t *new_array = lexbor_array_create();
+
+  for (size_t i = 0; i < self_array->length; i++) {
+    for (size_t j = 0; j < other_array->length; j++) {
+      if (self_array->list[i] == other_array->list[j]) {
+        lexbor_array_push(new_array, self_array->list[i]);
+        break;
+      }
+    }
+  }
+
+  return nl_rb_node_set_create_with_data(new_array, nl_rb_document_get(self));
+}
+
+/**
+ * @return [NodeSet] A new NodeSet with the nodes in this NodeSet that aren't in +other+
+ */
+static VALUE
+nl_node_set_difference(VALUE self, VALUE other)
+{
+  if (!rb_obj_is_kind_of(other, cNokolexborNodeSet)) {
+    rb_raise(rb_eArgError, "Parameter must be a Nokolexbor::NodeSet");
+  }
+
+  lexbor_array_t *self_array = nl_rb_node_set_unwrap(self);
+  lexbor_array_t *other_array = nl_rb_node_set_unwrap(other);
+
+  lexbor_array_t *new_array = lexbor_array_create();
+
+  for (size_t i = 0; i < self_array->length; i++) {
+    bool found = false;
+    for (size_t j = 0; j < other_array->length; j++) {
+      if (self_array->list[i] == other_array->list[j]) {
+        found = true;
+        break;
+      }
+    }
+    if (!found) {
+      lexbor_array_push(new_array, self_array->list[i]);
+    }
+  }
+
+  return nl_rb_node_set_create_with_data(new_array, nl_rb_document_get(self));
+}
+
 static lxb_status_t
 nl_node_set_find(VALUE self, VALUE selector, lxb_selectors_cb_f cb, void *ctx)
 {
@@ -410,6 +468,8 @@ void Init_nl_node_set(void)
   rb_define_method(cNokolexborNodeSet, "[]", nl_node_set_slice, -1);
   rb_define_method(cNokolexborNodeSet, "push", nl_node_set_push, 1);
   rb_define_method(cNokolexborNodeSet, "|", nl_node_set_union, 1);
+  rb_define_method(cNokolexborNodeSet, "&", nl_node_set_intersection, 1);
+  rb_define_method(cNokolexborNodeSet, "-", nl_node_set_difference, 1);
   rb_define_method(cNokolexborNodeSet, "to_a", nl_node_set_to_array, 0);
   rb_define_method(cNokolexborNodeSet, "delete", nl_node_set_delete, 1);
   rb_define_method(cNokolexborNodeSet, "include?", nl_node_set_is_include, 1);

data/lib/nokolexbor/node_set.rb

@@ -54,6 +54,16 @@ module Nokolexbor
       length == 0
     end
 
+    # Insert +node+ before the first Node in this NodeSet
+    def before(node)
+      first.before(node)
+    end
+
+    # Insert +node+ after the last Node in this NodeSet
+    def after(node)
+      last.after(node)
+    end
+
     # @return [Integer] The index of the first node in this NodeSet that is equal to +node+ or meets the given block. Returns nil if no match is found.
     def index(node = nil)
       if node
@@ -169,6 +179,75 @@ module Nokolexbor
       self
     end
 
+    # Add the class attribute +name+ to all containing nodes.
+    #
+    # @see Node#add_class
+    def add_class(name)
+      each do |el|
+        el.add_class(name)
+      end
+      self
+    end
+
+    # Append the class attribute +name+ to all containing nodes.
+    #
+    # @see Node#append_class
+    def append_class(name)
+      each do |el|
+        el.append_class(name)
+      end
+      self
+    end
+
+    # Remove the class attribute +name+ from all containing nodes.
+    #
+    # @see Node#remove_class
+    def remove_class(name = nil)
+      each do |el|
+        el.remove_class(name)
+      end
+      self
+    end
+
+    # Remove the attributed named +name+ from all containing nodes.
+    #
+    # @see Node#remove_attr
+    def remove_attr(name)
+      each { |el| el.delete(name) }
+      self
+    end
+    alias_method :remove_attribute, :remove_attr
+
+    # Set attributes on each Node in the NodeSet, or get an
+    # attribute from the first Node in the NodeSet.
+    #
+    # @example Get an attribute from the first Node in a NodeSet.
+    #   node_set.attr("href")
+    #
+    # @example Set attributes on each node.
+    #   node_set.attr("href" => "http://example.com", "class" => "a")
+    #   node_set.attr("href", "http://example.com")
+    #   node_set.attr("href") { |node| "http://example.com" }
+    #
+    # @return [NodeSet] +self+, to support chaining of calls.
+    def attr(key, value = nil, &block)
+      unless key.is_a?(Hash) || (key && (value || block))
+        return first&.attribute(key)
+      end
+
+      hash = key.is_a?(Hash) ? key : { key => value }
+
+      hash.each do |k, v|
+        each do |node|
+          node[k] = v || yield(node)
+        end
+      end
+
+      self
+    end
+    alias_method :set, :attr
+    alias_method :attribute, :attr
+
     # (see Node#xpath)
     def xpath(*args)
       paths, handler, ns, binds = extract_params(args)

data/lib/nokolexbor/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Nokolexbor
-  VERSION = '0.3.7'
+  VERSION = '0.4.1'
 end

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

@@ -0,0 +1,24 @@
+diff --git a/source/lexbor/dom/interfaces/document.c b/source/lexbor/dom/interfaces/document.c
+index a2153f4..8a9c69f 100755
+--- a/source/lexbor/dom/interfaces/document.c
++++ b/source/lexbor/dom/interfaces/document.c
+@@ -12,6 +12,7 @@
+ #include "lexbor/dom/interfaces/cdata_section.h"
+ #include "lexbor/dom/interfaces/cdata_section.h"
+ #include "lexbor/dom/interfaces/processing_instruction.h"
++#include "lexbor/html/interfaces/template_element.h"
+ 
+ 
+ lxb_dom_document_t *
+@@ -449,6 +450,11 @@ lxb_dom_document_import_node(lxb_dom_document_t *doc, lxb_dom_node_t *node,
+             return NULL;
+         }
+ 
++        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_dom_node_insert_child(curr, cnode);
+ 
+         if (node->first_child != NULL) {

data/patches/0005-lexbor-add-source-location-to-node.patch

@@ -0,0 +1,65 @@
+diff --git a/source/lexbor/dom/interfaces/node.h b/source/lexbor/dom/interfaces/node.h
+index acd0c1c..f436257 100755
+--- a/source/lexbor/dom/interfaces/node.h
++++ b/source/lexbor/dom/interfaces/node.h
+@@ -46,6 +46,8 @@ struct lxb_dom_node {
+     uintptr_t              prefix;     /* lowercase: lalala */
+     uintptr_t              ns;         /* namespace */
+ 
++    size_t                 source_location; /* char offset to the source HTML */
++
+     lxb_dom_document_t     *owner_document;
+ 
+     lxb_dom_node_t         *next;
+diff --git a/source/lexbor/html/tokenizer.c b/source/lexbor/html/tokenizer.c
+index 741bced..6343c65 100755
+--- a/source/lexbor/html/tokenizer.c
++++ b/source/lexbor/html/tokenizer.c
+@@ -309,6 +309,7 @@ lxb_html_tokenizer_chunk(lxb_html_tokenizer_t *tkz, const lxb_char_t *data,
+ 
+     tkz->is_eof = false;
+     tkz->status = LXB_STATUS_OK;
++    tkz->first = data;
+     tkz->last = end;
+ 
+     while (data < end) {
+diff --git a/source/lexbor/html/tokenizer.h b/source/lexbor/html/tokenizer.h
+index ba9602f..08d0d9a 100755
+--- a/source/lexbor/html/tokenizer.h
++++ b/source/lexbor/html/tokenizer.h
+@@ -72,6 +72,7 @@ struct lxb_html_tokenizer {
+     lxb_char_t                       *pos;
+     const lxb_char_t                 *end;
+     const lxb_char_t                 *begin;
++    const lxb_char_t                 *first;
+     const lxb_char_t                 *last;
+ 
+     /* Entities */
+diff --git a/source/lexbor/html/tree.c b/source/lexbor/html/tree.c
+index 8c42990..28c97cc 100755
+--- a/source/lexbor/html/tree.c
++++ b/source/lexbor/html/tree.c
+@@ -484,6 +484,7 @@ lxb_html_tree_append_attributes(lxb_html_tree_t *tree,
+ 
+         attr->node.local_name = token_attr->name->attr_id;
+         attr->node.ns = ns;
++        attr->node.source_location = token_attr->name_begin - tree->tkz_ref->first;
+ 
+         /* Fix for adjust MathML/SVG attributes */
+         if (tree->before_append_attr != NULL) {
+diff --git a/source/lexbor/html/tree.h b/source/lexbor/html/tree.h
+index 231239d..bc0249e 100755
+--- a/source/lexbor/html/tree.h
++++ b/source/lexbor/html/tree.h
+@@ -266,8 +266,10 @@ lxb_inline lxb_dom_node_t *
+ lxb_html_tree_create_node(lxb_html_tree_t *tree,
+                           lxb_tag_id_t tag_id, lxb_ns_id_t ns)
+ {
+-    return (lxb_dom_node_t *) lxb_html_interface_create(tree->document,
++    lxb_dom_node_t *node = (lxb_dom_node_t *) lxb_html_interface_create(tree->document,
+                                                         tag_id, ns);
++    node->source_location = (tree->tkz_ref->token->begin > tree->tkz_ref->first ? tree->tkz_ref->token->begin : tree->tkz_ref->begin) - tree->tkz_ref->first;
++    return node;
+ }
+ 
+ lxb_inline bool

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: nokolexbor
 version: !ruby/object:Gem::Version
-  version: 0.3.7
+  version: 0.4.1
 platform: ruby
 authors:
 - Yicheng Zhou
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-02-12 00:00:00.000000000 Z
+date: 2023-04-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler
@@ -130,6 +130,8 @@ files:
 - patches/0001-lexbor-support-text-pseudo-element.patch
 - patches/0002-lexbor-match-id-class-case-sensitive.patch
 - patches/0003-lexbor-attach-template-content-to-self.patch
+- patches/0004-lexbor-fix-template-clone.patch
+- patches/0005-lexbor-add-source-location-to-node.patch
 - vendor/lexbor/CMakeLists.txt
 - vendor/lexbor/config.cmake
 - vendor/lexbor/feature.cmake