Implement parent_nodes + nth_child (#25)

Co-authored-by: spicychickensauce <[email protected]>
Co-authored-by: Jonatan Kłosko <[email protected]>

Author: 3 people

c_src/lazy_html.cpp

@@ -7,6 +7,7 @@
 #include <stdexcept>
 #include <string>
 #include <tuple>
+#include <unordered_set>
 #include <variant>
 
 #include <lexbor/html/html.h>
@@ -43,8 +44,10 @@ auto resource = fine::Atom("resource");
 
 struct DocumentRef {
   lxb_html_document_t *document;
+  bool is_fragment;
 
-  DocumentRef(lxb_html_document_t *document) : document(document) {}
+  DocumentRef(lxb_html_document_t *document, bool is_fragment)
+      : document(document), is_fragment(is_fragment) {}
 
   ~DocumentRef() { lxb_html_document_destroy(this->document); }
 };
@@ -97,7 +100,7 @@ ExLazyHTML from_document(ErlNifEnv *env, ErlNifBinary html) {
     throw std::runtime_error("failed to parse html document");
   }
 
-  auto document_ref = std::make_shared<DocumentRef>(document);
+  auto document_ref = std::make_shared<DocumentRef>(document, false);
   document_guard.deactivate();
 
   auto nodes = std::vector<lxb_dom_node_t *>();
@@ -129,7 +132,7 @@ ExLazyHTML from_fragment(ErlNifEnv *env, ErlNifBinary html) {
     throw std::runtime_error("failed to parse html fragment");
   }
 
-  auto document_ref = std::make_shared<DocumentRef>(document);
+  auto document_ref = std::make_shared<DocumentRef>(document, true);
   document_guard.deactivate();
 
   auto nodes = std::vector<lxb_dom_node_t *>();
@@ -522,7 +525,10 @@ ExLazyHTML from_tree(ErlNifEnv *env, std::vector<fine::Term> tree) {
     nodes.push_back(node);
   }
 
-  auto document_ref = std::make_shared<DocumentRef>(document);
+  bool is_fragment =
+      nodes.empty() || !lxb_html_tree_node_is(nodes.front(), LXB_TAG_HTML);
+
+  auto document_ref = std::make_shared<DocumentRef>(document, is_fragment);
   document_guard.deactivate();
 
   return ExLazyHTML(fine::make_resource<LazyHTML>(document_ref, nodes, false));
@@ -714,6 +720,57 @@ ExLazyHTML child_nodes(ErlNifEnv *env, ExLazyHTML ex_lazy_html) {
 
 FINE_NIF(child_nodes, 0);
 
+ExLazyHTML parent_node(ErlNifEnv *env, ExLazyHTML ex_lazy_html) {
+  bool is_document = !ex_lazy_html.resource->document_ref->is_fragment;
+  auto nodes = std::vector<lxb_dom_node_t *>();
+  auto inserted_nodes = std::unordered_set<lxb_dom_node_t *>();
+
+  for (auto node : ex_lazy_html.resource->nodes) {
+    auto parent = lxb_dom_node_parent(node);
+    if (parent != NULL && parent->type == LXB_DOM_NODE_TYPE_ELEMENT &&
+        (is_document || !lxb_html_tree_node_is(parent, LXB_TAG_HTML))) {
+      auto inserted_node = inserted_nodes.find(parent);
+      if (inserted_node == inserted_nodes.end()) {
+        inserted_nodes.insert(parent);
+        nodes.push_back(parent);
+      }
+    }
+  }
+  return ExLazyHTML(fine::make_resource<LazyHTML>(
+      ex_lazy_html.resource->document_ref, nodes, true));
+}
+FINE_NIF(parent_node, ERL_NIF_DIRTY_JOB_CPU_BOUND);
+
+std::vector<int64_t> nth_child(ErlNifEnv *env, ExLazyHTML ex_lazy_html) {
+  auto values = std::vector<int64_t>();
+  for (auto node : ex_lazy_html.resource->nodes) {
+    if (node->type != LXB_DOM_NODE_TYPE_ELEMENT) {
+      continue;
+    }
+
+    auto parent = lxb_dom_node_parent(node);
+    if (parent == NULL) {
+      // We're at the root, nth_child is 1
+      values.push_back(1);
+    } else {
+      int64_t i = 1;
+      for (auto child = lxb_dom_node_first_child(parent); child != NULL;
+           child = lxb_dom_node_next(child)) {
+        if (child == node) {
+          break;
+        }
+        if (child->type == LXB_DOM_NODE_TYPE_ELEMENT) {
+          i++;
+        }
+      }
+      values.push_back(i);
+    }
+  }
+
+  return values;
+}
+FINE_NIF(nth_child, ERL_NIF_DIRTY_JOB_CPU_BOUND);
+
 std::string text(ErlNifEnv *env, ExLazyHTML ex_lazy_html) {
   auto document = ex_lazy_html.resource->document_ref->document;
 

lib/lazy_html.ex

@@ -357,6 +357,48 @@ defmodule LazyHTML do
     LazyHTML.NIF.child_nodes(lazy_html)
   end
 
+  @doc """
+  Returns the (unique) parent nodes of the root nodes in `lazy_html`.
+
+  ## Examples
+
+      iex> lazy_html = LazyHTML.from_fragment(~S|<div><span>Hello</span> <span>world</span></div>|)
+      iex> spans = LazyHTML.query(lazy_html, "span")
+      iex> LazyHTML.parent_node(spans)
+      #LazyHTML<
+        1 node (from selector)
+        #1
+        <div><span>Hello</span> <span>world</span></div>
+      >
+
+  """
+  @spec parent_node(t()) :: t()
+  def parent_node(lazy_html) do
+    LazyHTML.NIF.parent_node(lazy_html)
+  end
+
+  @doc """
+  Returns the position among its siblings for every root element in `lazy_html`.
+
+  The position numbering is 1-based and only considers siblings that
+  are elements, as to match the `:nth-child` CSS pseudo-class.
+
+  Note that if there are text or comment root nodes, they are ignored,
+  and they have no corresponding number in the result.
+
+  ## Examples
+
+      iex> lazy_html = LazyHTML.from_fragment(~S|<div><span>1</span><span>2</span></div>|)
+      iex> spans = LazyHTML.query(lazy_html, "span")
+      iex> LazyHTML.nth_child(spans)
+      [1, 2]
+
+  """
+  @spec nth_child(t()) :: list(integer())
+  def nth_child(lazy_html) do
+    LazyHTML.NIF.nth_child(lazy_html)
+  end
+
   @doc """
   Returns the text content of all nodes in `lazy_html`.
 

lib/lazy_html/nif.ex

@@ -21,6 +21,8 @@ defmodule LazyHTML.NIF do
   def filter(_lazy_html, _css_selector), do: err!()
   def query_by_id(_lazy_html, _id), do: err!()
   def child_nodes(_lazy_html), do: err!()
+  def parent_node(_lazy_html), do: err!()
+  def nth_child(_lazy_html), do: err!()
   def text(_lazy_html), do: err!()
   def attribute(_lazy_html, _name), do: err!()
   def attributes(_lazy_html), do: err!()

test/lazy_html_test.exs

@@ -250,6 +250,101 @@ defmodule LazyHTMLTest do
     end
   end
 
+  describe "parent_node/1" do
+    test "from selector of nodes on different levels" do
+      lazy_html =
+        LazyHTML.from_fragment("""
+        <div id="a">
+          <div id="b">
+            <span>Hello</span>
+          </div>
+          <span>world</span>
+        </div>
+        """)
+
+      spans = LazyHTML.query(lazy_html, "span")
+      parents = LazyHTML.parent_node(spans)
+      parent_ids = parents |> LazyHTML.attribute("id") |> Enum.sort()
+      assert parent_ids == ["a", "b"]
+
+      # parent of div#id="a" is null
+      grandparents = LazyHTML.parent_node(parents)
+      assert LazyHTML.tag(grandparents) == ["div"]
+
+      great_grandparents = LazyHTML.parent_node(grandparents)
+      assert great_grandparents |> Enum.count() == 0
+    end
+
+    test "from selector of nodes on same level" do
+      lazy_html =
+        LazyHTML.from_fragment("""
+        <div id="a">
+          <div id="b">
+            <span>Hello</span>
+          </div>
+          <div id="c">
+            <span>world</span>
+          </div>
+        </div>
+        """)
+
+      spans = LazyHTML.query(lazy_html, "span")
+      parents = LazyHTML.parent_node(spans)
+      parent_ids = parents |> LazyHTML.attribute("id") |> Enum.sort()
+      assert parent_ids == ["b", "c"]
+
+      # since they share the same parent, we now only have one node left
+      grandparent = LazyHTML.parent_node(parents)
+      assert LazyHTML.attribute(grandparent, "id") == ["a"]
+    end
+
+    defp ancestor_chain(node) do
+      parent = LazyHTML.parent_node(node)
+
+      if Enum.count(node) == 0 do
+        []
+      else
+        ancestor_chain(parent) ++ LazyHTML.tag(parent)
+      end
+    end
+
+    test "last parent node is <html> if instantiated via from_document and similar" do
+      lazy_html = LazyHTML.from_document("<html><body><div>root</div></body></html>")
+      assert lazy_html |> LazyHTML.query("div") |> ancestor_chain() == ["html", "body"]
+
+      lazy_html = LazyHTML.from_fragment("<div>root</div>")
+      assert lazy_html |> LazyHTML.query("div") |> ancestor_chain() == []
+
+      lazy_html = LazyHTML.from_tree([{"div", [], []}])
+      assert lazy_html |> LazyHTML.query("div") |> ancestor_chain() == []
+
+      lazy_html = LazyHTML.from_tree([{"html", [], [{"body", [], [{"div", [], []}]}]}])
+      assert lazy_html |> LazyHTML.query("div") |> ancestor_chain() == ["html", "body"]
+    end
+  end
+
+  describe "nth_child/1" do
+    test "nth_child gives position" do
+      lazy_html =
+        LazyHTML.from_fragment("""
+        <div>
+          Text isn't counted.
+          <span>1</span>
+          <!-- neither are comments -->
+          <span>2</span>
+        </div>
+        """)
+
+      assert LazyHTML.nth_child(lazy_html) == [1]
+      assert lazy_html["div"] |> LazyHTML.nth_child() == [1]
+      assert lazy_html["span"] |> LazyHTML.nth_child() == [1, 2]
+
+      # Verify numbering matches css selector
+      assert lazy_html["span:nth-child(1)"] |> LazyHTML.text() == "1"
+      assert lazy_html["span:nth-child(2)"] |> LazyHTML.text() == "2"
+    end
+  end
+
   describe "query_by_id/2" do
     test "raises when an empty id is given" do
       assert_raise ArgumentError, ~r/id cannot be empty/, fn ->