feat(node::Find): also change Node::find to return node::Find

Using `Node::find` to fetch only a few nodes should be _much_ faster as the descendant nodes are checked when required, instead of being checked upfront in the old implementation which returned a `Selection`. Only counting the nodes matching a predicate should also be faster than before due to no dynamic memory allocation in `Node::find` and `node::Find`. Fixes #19.
utkarshkukreti · Jul 20, 2016 · c75dba9 · c75dba9
1 parent 1cc1f66
commit c75dba9
Show file tree

Hide file tree

Showing 4 changed files with 84 additions and 26 deletions.
diff --git a/README.md b/README.md
@@ -29,26 +29,41 @@ pub fn main() {
 
     println!("# Top 5 Questions");
     for node in document.find(Class("question-summary")).iter().take(5) {
-        let question = node.find(Class("question-hyperlink")).first().unwrap();
-        let votes = node.find(Class("vote-count-post")).first().unwrap().text();
-        let answers = node.find(Class("status")).find(Name("strong")).first().unwrap().text();
-        let tags = node.find(Class("post-tag")).iter().map(|tag| tag.text()).collect::<Vec<_>>();
-        let asked_on = node.find(Class("relativetime")).first().unwrap().text();
-        let asker = node.find(Class("user-details")).find(Name("a")).first().unwrap().text();
+        let question = node.find(Class("question-hyperlink")).next().unwrap();
+        let votes = node.find(Class("vote-count-post")).next().unwrap().text();
+        let answers = node.find(Class("status"))
+            .into_selection()
+            .find(Name("strong"))
+            .first()
+            .unwrap()
+            .text();
+        let tags = node.find(Class("post-tag")).map(|tag| tag.text()).collect::<Vec<_>>();
+        let asked_on = node.find(Class("relativetime")).next().unwrap().text();
+        let asker = node.find(Class("user-details"))
+            .into_selection()
+            .find(Name("a"))
+            .first()
+            .unwrap()
+            .text();
         println!(" Question: {}", question.text());
         println!("  Answers: {}", answers);
         println!("    Votes: {}", votes);
         println!("   Tagged: {}", tags.join(", "));
         println!(" Asked on: {}", asked_on);
         println!("    Asker: {}", asker);
-        println!("Permalink: http://stackoverflow.com{}", question.attr("href").unwrap());
+        println!("Permalink: http://stackoverflow.com{}",
+                 question.attr("href").unwrap());
         println!("");
     }
 
     println!("# Top 10 Related Tags");
-    for node in document.find(Attr("id", "h-related-tags")).parent().find(Name("div")).iter().take(10) {
-        let tag = node.find(Name("a")).first().unwrap().text();
-        let count = node.find(Class("item-multiplier-count")).first().unwrap().text();
+    for node in document.find(Attr("id", "h-related-tags"))
+        .parent()
+        .find(Name("div"))
+        .iter()
+        .take(10) {
+        let tag = node.find(Name("a")).next().unwrap().text();
+        let count = node.find(Class("item-multiplier-count")).next().unwrap().text();
         println!("{} ({})", tag, count);
     }
 }

diff --git a/examples/stackoverflow.rs b/examples/stackoverflow.rs
@@ -16,26 +16,41 @@ pub fn main() {
 
     println!("# Top 5 Questions");
     for node in document.find(Class("question-summary")).iter().take(5) {
-        let question = node.find(Class("question-hyperlink")).first().unwrap();
-        let votes = node.find(Class("vote-count-post")).first().unwrap().text();
-        let answers = node.find(Class("status")).find(Name("strong")).first().unwrap().text();
-        let tags = node.find(Class("post-tag")).iter().map(|tag| tag.text()).collect::<Vec<_>>();
-        let asked_on = node.find(Class("relativetime")).first().unwrap().text();
-        let asker = node.find(Class("user-details")).find(Name("a")).first().unwrap().text();
+        let question = node.find(Class("question-hyperlink")).next().unwrap();
+        let votes = node.find(Class("vote-count-post")).next().unwrap().text();
+        let answers = node.find(Class("status"))
+            .into_selection()
+            .find(Name("strong"))
+            .first()
+            .unwrap()
+            .text();
+        let tags = node.find(Class("post-tag")).map(|tag| tag.text()).collect::<Vec<_>>();
+        let asked_on = node.find(Class("relativetime")).next().unwrap().text();
+        let asker = node.find(Class("user-details"))
+            .into_selection()
+            .find(Name("a"))
+            .first()
+            .unwrap()
+            .text();
         println!(" Question: {}", question.text());
         println!("  Answers: {}", answers);
         println!("    Votes: {}", votes);
         println!("   Tagged: {}", tags.join(", "));
         println!(" Asked on: {}", asked_on);
         println!("    Asker: {}", asker);
-        println!("Permalink: http://stackoverflow.com{}", question.attr("href").unwrap());
+        println!("Permalink: http://stackoverflow.com{}",
+                 question.attr("href").unwrap());
         println!("");
     }
 
     println!("# Top 10 Related Tags");
-    for node in document.find(Attr("id", "h-related-tags")).parent().find(Name("div")).iter().take(10) {
-        let tag = node.find(Name("a")).first().unwrap().text();
-        let count = node.find(Class("item-multiplier-count")).first().unwrap().text();
+    for node in document.find(Attr("id", "h-related-tags"))
+        .parent()
+        .find(Name("div"))
+        .iter()
+        .take(10) {
+        let tag = node.find(Name("a")).next().unwrap().text();
+        let count = node.find(Class("item-multiplier-count")).next().unwrap().text();
         println!("{} ({})", tag, count);
     }
 }
diff --git a/src/node.rs b/src/node.rs
@@ -144,8 +144,12 @@ impl<'a> Node<'a> {
         String::from_utf8(buf).unwrap()
     }
 
-    pub fn find<P: Predicate>(&self, p: P) -> Selection<'a> {
-        Selection::new(self.document, [self.index].iter().cloned().collect()).find(p)
+    pub fn find<P: Predicate>(&self, predicate: P) -> Find<P> {
+        Find {
+            document: self.document,
+            descendants: self.descendants(),
+            predicate: predicate,
+        }
     }
 
     pub fn is<P: Predicate>(&self, p: P) -> bool {
@@ -267,3 +271,28 @@ impl<'a> Iterator for Descendants<'a> {
         Some(self.current)
     }
 }
+
+pub struct Find<'a, P: Predicate> {
+    document: &'a Document,
+    descendants: Descendants<'a>,
+    predicate: P,
+}
+
+impl<'a, P: Predicate> Find<'a, P> {
+    pub fn into_selection(self) -> Selection<'a> {
+        Selection::new(self.document, self.map(|node| node.index()).collect())
+    }
+}
+
+impl<'a, P: Predicate> Iterator for Find<'a, P> {
+    type Item = Node<'a>;
+
+    fn next(&mut self) -> Option<Node<'a>> {
+        for node in &mut self.descendants {
+            if self.predicate.matches(&node) {
+                return Some(node);
+            }
+        }
+        None
+    }
+}
diff --git a/tests/node_tests.rs b/tests/node_tests.rs
@@ -104,8 +104,8 @@ speculate! {
                 let main = document.find(Attr("id", "main"));
                 let main = main.iter().next().unwrap();
 
-                assert_eq!(main.find(Name("span")).len(), 1785);
-                assert_eq!(main.find(Name("div")).len(), 204);
+                assert_eq!(main.find(Name("span")).count(), 1785);
+                assert_eq!(main.find(Name("div")).count(), 204);
             };
         }
 
@@ -175,8 +175,7 @@ speculate! {
             for i in 0..document.nodes.len() {
                 let node = document.nth(i).unwrap();
                 let actual = node.descendants().map(|node| node.index()).collect::<Vec<_>>();
-                let expected = node.find(Any);
-                let expected = expected.iter().map(|node| node.index()).collect::<Vec<_>>();
+                let expected = node.find(Any).map(|node| node.index()).collect::<Vec<_>>();
                 assert_eq!(actual, expected);
             }
         }