html5ever 0.39.0

// Copyright 2014-2017 The html5ever Project Developers. See the
// COPYRIGHT file at the top-level directory of this distribution.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.

extern crate html5ever;
extern crate typed_arena;

use html5ever::interface::tree_builder::{ElementFlags, NodeOrText, QuirksMode, TreeSink};
use html5ever::tendril::{StrTendril, TendrilSink};
use html5ever::{parse_document, Attribute, QualName};
use std::borrow::Cow;
use std::cell::{Cell, RefCell};
use std::collections::HashSet;
use std::io::{self, Read};
use std::ptr;

/// By using our Sink type, the arena is filled with parsed HTML.
fn html5ever_parse_slice_into_arena<'a>(bytes: &[u8], arena: Arena<'a>) -> Ref<'a> {
    let sink = Sink {
        arena,
        document: arena.alloc(Node::new(NodeData::Document)),
        quirks_mode: Cell::new(QuirksMode::NoQuirks),
    };

    parse_document(sink, Default::default())
        .from_utf8()
        .one(bytes)
}

type Arena<'arena> = &'arena typed_arena::Arena<Node<'arena>>;
type Ref<'arena> = &'arena Node<'arena>;
type Link<'arena> = Cell<Option<Ref<'arena>>>;

/// Sink struct is responsible for handling how the data that comes out of the HTML parsing
/// unit (TreeBuilder in our case) is handled.
struct Sink<'arena> {
    arena: Arena<'arena>,
    document: Ref<'arena>,
    quirks_mode: Cell<QuirksMode>,
}

/// DOM node which contains links to other nodes in the tree.
pub struct Node<'arena> {
    parent: Link<'arena>,
    next_sibling: Link<'arena>,
    previous_sibling: Link<'arena>,
    first_child: Link<'arena>,
    last_child: Link<'arena>,
    data: NodeData<'arena>,
}

/// HTML node data which can be an element, a comment, a string, a DOCTYPE, etc...
#[derive(Clone)]
pub enum NodeData<'arena> {
    Document,
    Doctype {
        name: StrTendril,
        public_id: StrTendril,
        system_id: StrTendril,
    },
    Text {
        contents: RefCell<StrTendril>,
    },
    Comment {
        contents: StrTendril,
    },
    Element {
        name: QualName,
        attrs: RefCell<Vec<Attribute>>,
        template_contents: Option<Ref<'arena>>,
        mathml_annotation_xml_integration_point: bool,
    },
    ProcessingInstruction {
        target: StrTendril,
        contents: StrTendril,
    },
}

impl<'arena> Node<'arena> {
    fn new(data: NodeData<'arena>) -> Self {
        Node {
            parent: Cell::new(None),
            previous_sibling: Cell::new(None),
            next_sibling: Cell::new(None),
            first_child: Cell::new(None),
            last_child: Cell::new(None),
            data,
        }
    }

    fn detach(&self) {
        let parent = self.parent.take();
        let previous_sibling = self.previous_sibling.take();
        let next_sibling = self.next_sibling.take();

        if let Some(next_sibling) = next_sibling {
            next_sibling.previous_sibling.set(previous_sibling);
        } else if let Some(parent) = parent {
            parent.last_child.set(previous_sibling);
        }

        if let Some(previous_sibling) = previous_sibling {
            previous_sibling.next_sibling.set(next_sibling);
        } else if let Some(parent) = parent {
            parent.first_child.set(next_sibling);
        }
    }

    fn append(&'arena self, new_child: &'arena Self) {
        new_child.detach();
        new_child.parent.set(Some(self));
        if let Some(last_child) = self.last_child.take() {
            new_child.previous_sibling.set(Some(last_child));
            debug_assert!(last_child.next_sibling.get().is_none());
            last_child.next_sibling.set(Some(new_child));
        } else {
            debug_assert!(self.first_child.get().is_none());
            self.first_child.set(Some(new_child));
        }
        self.last_child.set(Some(new_child));
    }

    fn insert_before(&'arena self, new_sibling: &'arena Self) {
        new_sibling.detach();
        new_sibling.parent.set(self.parent.get());
        new_sibling.next_sibling.set(Some(self));
        if let Some(previous_sibling) = self.previous_sibling.take() {
            new_sibling.previous_sibling.set(Some(previous_sibling));
            debug_assert!(ptr::eq::<Node>(
                previous_sibling.next_sibling.get().unwrap(),
                self
            ));
            previous_sibling.next_sibling.set(Some(new_sibling));
        } else if let Some(parent) = self.parent.get() {
            debug_assert!(ptr::eq::<Node>(parent.first_child.get().unwrap(), self));
            parent.first_child.set(Some(new_sibling));
        }
        self.previous_sibling.set(Some(new_sibling));
    }
}

impl<'arena> Sink<'arena> {
    fn new_node(&self, data: NodeData<'arena>) -> Ref<'arena> {
        self.arena.alloc(Node::new(data))
    }

    fn append_common<P, A>(&self, child: NodeOrText<Ref<'arena>>, previous: P, append: A)
    where
        P: FnOnce() -> Option<Ref<'arena>>,
        A: FnOnce(Ref<'arena>),
    {
        let new_node = match child {
            NodeOrText::AppendText(text) => {
                // Append to an existing Text node if we have one.
                if let Some(&Node {
                    data: NodeData::Text { ref contents },
                    ..
                }) = previous()
                {
                    contents.borrow_mut().push_tendril(&text);
                    return;
                }
                self.new_node(NodeData::Text {
                    contents: RefCell::new(text),
                })
            },
            NodeOrText::AppendNode(node) => node,
        };

        append(new_node)
    }
}

/// By implementing the TreeSink trait we determine how the data from the tree building step
/// is processed. In our case, our data is allocated in the arena and added to the Node data
/// structure.
///
/// For deeper understating of each function go to the TreeSink declaration.
impl<'arena> TreeSink for Sink<'arena> {
    type Handle = Ref<'arena>;
    type Output = Ref<'arena>;
    type ElemName<'a>
        = &'a QualName
    where
        Self: 'a;

    fn finish(self) -> Ref<'arena> {
        self.document
    }

    fn parse_error(&self, _: Cow<'static, str>) {}

    fn get_document(&self) -> Ref<'arena> {
        self.document
    }

    fn set_quirks_mode(&self, mode: QuirksMode) {
        self.quirks_mode.set(mode);
    }

    fn same_node(&self, x: &Ref<'arena>, y: &Ref<'arena>) -> bool {
        ptr::eq::<Node>(*x, *y)
    }

    fn elem_name(&self, target: &Ref<'arena>) -> Self::ElemName<'_> {
        match target.data {
            NodeData::Element { ref name, .. } => name,
            _ => panic!("not an element!"),
        }
    }

    fn get_template_contents(&self, target: &Ref<'arena>) -> Ref<'arena> {
        if let NodeData::Element {
            template_contents: Some(contents),
            ..
        } = target.data
        {
            contents
        } else {
            panic!("not a template element!")
        }
    }

    fn is_mathml_annotation_xml_integration_point(&self, target: &Ref<'arena>) -> bool {
        if let NodeData::Element {
            mathml_annotation_xml_integration_point,
            ..
        } = target.data
        {
            mathml_annotation_xml_integration_point
        } else {
            panic!("not an element!")
        }
    }

    fn create_element(
        &self,
        name: QualName,
        attrs: Vec<Attribute>,
        flags: ElementFlags,
    ) -> Ref<'arena> {
        self.new_node(NodeData::Element {
            name,
            attrs: RefCell::new(attrs),
            template_contents: if flags.template {
                Some(self.new_node(NodeData::Document))
            } else {
                None
            },
            mathml_annotation_xml_integration_point: flags.mathml_annotation_xml_integration_point,
        })
    }

    fn create_comment(&self, text: StrTendril) -> Ref<'arena> {
        self.new_node(NodeData::Comment { contents: text })
    }

    fn create_pi(&self, target: StrTendril, data: StrTendril) -> Ref<'arena> {
        self.new_node(NodeData::ProcessingInstruction {
            target,
            contents: data,
        })
    }

    fn append(&self, parent: &Ref<'arena>, child: NodeOrText<Ref<'arena>>) {
        self.append_common(
            child,
            || parent.last_child.get(),
            |new_node| parent.append(new_node),
        )
    }

    fn append_before_sibling(&self, sibling: &Ref<'arena>, child: NodeOrText<Ref<'arena>>) {
        self.append_common(
            child,
            || sibling.previous_sibling.get(),
            |new_node| sibling.insert_before(new_node),
        )
    }

    fn append_based_on_parent_node(
        &self,
        element: &Ref<'arena>,
        prev_element: &Ref<'arena>,
        child: NodeOrText<Ref<'arena>>,
    ) {
        if element.parent.get().is_some() {
            self.append_before_sibling(element, child)
        } else {
            self.append(prev_element, child)
        }
    }

    fn append_doctype_to_document(
        &self,
        name: StrTendril,
        public_id: StrTendril,
        system_id: StrTendril,
    ) {
        self.document.append(self.new_node(NodeData::Doctype {
            name,
            public_id,
            system_id,
        }))
    }

    fn add_attrs_if_missing(&self, target: &Ref<'arena>, attrs: Vec<Attribute>) {
        let mut existing = if let NodeData::Element { ref attrs, .. } = target.data {
            attrs.borrow_mut()
        } else {
            panic!("not an element")
        };

        let existing_names = existing
            .iter()
            .map(|e| e.name.clone())
            .collect::<HashSet<_>>();
        existing.extend(
            attrs
                .into_iter()
                .filter(|attr| !existing_names.contains(&attr.name)),
        );
    }

    fn remove_from_parent(&self, target: &Ref<'arena>) {
        target.detach()
    }

    fn reparent_children(&self, node: &Ref<'arena>, new_parent: &Ref<'arena>) {
        let mut next_child = node.first_child.get();
        while let Some(child) = next_child {
            debug_assert!(ptr::eq::<Node>(child.parent.get().unwrap(), *node));
            next_child = child.next_sibling.get();
            new_parent.append(child)
        }
    }
}

/// In this example an "arena" is created and filled with the DOM nodes.
/// "Arena" is a type of allocation in which a block of memory is allocated
/// and later filled with data, DOM nodes in this case. When the arena is deallocated
/// it is destroyed with all of its items.
///
/// Further info about arena: https://docs.rs/typed-arena/latest/typed_arena/
fn main() {
    // Read HTML from the standard input
    let mut bytes = Vec::new();
    io::stdin().read_to_end(&mut bytes).unwrap();

    let arena = typed_arena::Arena::new();
    let dom = html5ever_parse_slice_into_arena(&bytes, &arena);

    // Print the DOM structure
    print_node(dom, 0);
}

fn print_node<'arena>(node: &Node<'arena>, depth: usize) {
    let indent = "  ".repeat(depth);

    match &node.data {
        NodeData::Document => println!("{}Document", indent),
        NodeData::Doctype { name, .. } => println!("{}<!DOCTYPE {}>", indent, name),
        NodeData::Text { contents } => {
            let text = contents.borrow();
            if !text.trim().is_empty() {
                println!("{}\"{}\"", indent, text.trim());
            }
        },
        NodeData::Comment { contents } => println!("{}<!-- {} -->", indent, contents),
        NodeData::Element { name, .. } => println!("{}<{}>", indent, name.local),
        NodeData::ProcessingInstruction { target, .. } => println!("{}<?{}>", indent, target),
    }

    // Print all children
    let mut child = node.first_child.get();
    while let Some(current_child) = child {
        print_node(current_child, depth + 1);
        child = current_child.next_sibling.get();
    }
}