path: root/src/lib.rs

use core::fmt;
pub use tag::{Tag, TagIterator, TagIteratorMut};

//mod query;
mod tag;

pub struct Html {
	pub nodes: Vec<Node>,
}

impl Html {
	pub fn parse<S: AsRef<str>>(raw: S) -> Self {
		let mut raw = raw.as_ref();

		let mut nodes = vec![];

		loop {
			let Consumed { node, remaining } = Self::parse_node(raw);
			nodes.push(node);

			match remaining {
				None => break Self { nodes },
				Some(rem) => raw = rem,
			}
		}
	}

	pub fn child_tags(&self) -> TagIterator {
		TagIterator {
			inner: self.nodes.iter(),
		}
	}

	pub fn child_tags_mut(&mut self) -> TagIteratorMut {
		TagIteratorMut {
			inner: self.nodes.iter_mut(),
		}
	}

	pub fn get_by_tag_name_mut(&mut self, looking: &str) -> Option<&mut Tag> {
		// depth first
		fn find_node<'a>(tag: &'a mut Tag, looking: &str) -> Option<&'a mut Tag> {
			if tag.name.as_str() == looking {
				return Some(tag);
			}

			for child in tag.child_tags_mut() {
				if let Some(tag) = find_node(child, looking) {
					return Some(tag);
				}
			}

			None
		}

		for child in self.child_tags_mut() {
			if let Some(tag) = find_node(child, looking) {
				return Some(tag);
			}
		}

		None
	}

	pub fn get_parent_that_contains_tag_name_mut(&mut self, looking: &str) -> Option<&mut Tag> {
		// depth first
		fn find_node<'a>(tag: &'a mut Tag, looking: &str) -> Option<&'a mut Tag> {
			if tag.has_tag(looking) {
				return Some(tag);
			}

			for child_tag in tag.child_tags_mut() {
				if let Some(tag) = find_node(child_tag, looking) {
					return Some(tag);
				}
			}

			None
		}

		for child_tag in self.child_tags_mut() {
			if let Some(tag) = find_node(child_tag, looking) {
				return Some(tag);
			}
		}

		None
	}

	pub fn get_by_id(&self, id: &str) -> Option<&Tag> {
		// depth first
		fn find_node<'a>(tag: &'a Tag, id: &str) -> Option<&'a Tag> {
			if tag.id().unwrap_or_default() == id {
				return Some(tag);
			}

			for child in tag.child_tags() {
				if let Some(tag) = find_node(child, id) {
					return Some(tag);
				}
			}

			None
		}

		for child in self.child_tags() {
			if let Some(tag) = find_node(child, id) {
				return Some(tag);
			}
		}

		None
	}

	fn parse_node(raw: &str) -> Consumed {
		match Self::is_tag(raw) {
			Some(_) => {
				if let Some(cmt) = Self::parse_comment(raw) {
					cmt
				} else {
					Self::parse_tag(raw)
				}
			}
			None => {
				let cons = Self::parse_text(raw);
				cons
			}
		}
	}

	fn parse_tag(raw: &str) -> Consumed {
		let (root_tag, mut rest) = Self::is_tag(raw).unwrap();
		let mut tag = if root_tag.body.is_empty() {
			Tag {
				name: root_tag.name.to_owned(),
				body: None,
				self_closing: root_tag.self_closing,
				children: vec![],
			}
		} else {
			Tag {
				name: root_tag.name.into(),
				body: Some(root_tag.body.to_owned()),
				self_closing: root_tag.self_closing,
				children: vec![],
			}
		};

		if root_tag.closing {
			panic!(
				"found closing tag when not expected! {:?}\n{raw}",
				root_tag.name
			)
		} else if root_tag.self_closing {
			return Consumed {
				node: Node::Tag(tag),
				remaining: rest,
			};
		}

		loop {
			// Special case <script> and <style>
			if root_tag.name == "script" && tag.get_attribute("src").is_none()
				|| root_tag.name == "style"
			{
				let special = Self::special_parse(rest.unwrap(), root_tag.name);

				match special {
					None => {
						panic!("found tag '{}' with no end", root_tag.name);
					}
					Some((text, remaining)) => {
						let remaining = if remaining.is_empty() {
							None
						} else {
							Some(remaining)
						};

						tag.children.push(text!(text));
						return Consumed {
							node: Node::Tag(tag),
							remaining,
						};
					}
				}
			}

			// Find the closing end of out root_tag
			if let Some((parsed, remaining)) = Self::is_tag(rest.unwrap()) {
				if parsed.closing && parsed.name == root_tag.name {
					break Consumed {
						node: Node::Tag(tag),
						remaining,
					};
				}
			}

			// Not our closing root? parse and push
			let cons = Self::parse_node(rest.unwrap());
			rest = cons.remaining;
			tag.children.push(cons.node);
		}
	}

	fn special_parse<'a>(raw: &'a str, looking_for_name: &str) -> Option<(&'a str, &'a str)> {
		let close = format!("</{looking_for_name}>");

		let mut offset = 0;
		loop {
			match raw[offset..].find('\n') {
				None => return None,
				Some(nl_idx) => {
					offset += nl_idx + 1;
					match raw[offset..].find(|c: char| !c.is_ascii_whitespace()) {
						None => return None,
						Some(whole_idx) => {
							let whole_start = &raw[offset + whole_idx..];
							if let Some(stripped) = whole_start.strip_prefix(&close) {
								return Some((&raw[..offset + whole_idx], stripped));
							}
						}
					}
				}
			}
		}
	}

	fn parse_comment(raw: &str) -> Option<Consumed> {
		if let Some(after_start) = raw.strip_prefix("<!--") {
			after_start.find("-->").map(|end| Consumed {
				node: Node::Comment(after_start[..end].into()),
				remaining: after_start.get(end + 3..),
			})
		} else {
			None
		}
	}

	/// check if the start of the string is a valid tag
	#[rustfmt::skip]
	fn is_tag(raw: &str) -> Option<(ParsedTag,  Option<&str>)> {
		// Starts '<' and any non-whitespace character
		let starts_right = raw.starts_with('<')
			&& raw.chars().nth(1).map(|c| !c.is_ascii_whitespace()).unwrap_or(false);

		if !starts_right {
			return None;
		}

		match raw.find('>') {
			// not a tag if there's no close
			None => None,
			Some(idx) => {
				let rest = match raw.get(idx+1..) {
					None => None,
					Some("") => None,
					Some(txt) => Some(txt)
				};
				let tag_innards = &raw[1..idx];

				// Tag is closing. Don't check for self-close
				// as they cannot be on the same tag
				if let Some(closing) = tag_innards.strip_prefix('/') {
					match closing.find(' ') {
						None => return Some((ParsedTag {
							closing: true,
							self_closing: false,
							name: closing,
							body: ""
						}, rest)),
						Some(idx) => {
							let name = &closing[..idx];
							let body = &closing[idx..];

							return Some((ParsedTag{
								closing: true,
								self_closing: false,
								name,
								body
							}, rest))
						}
					}
				}

				if let Some(closing) = tag_innards.strip_suffix('/') {
					match closing.find(' ') {
						None => return Some((ParsedTag {
							closing: false,
							self_closing: true,
							name: closing,
							body: ""
						}, rest)),
						Some(idx) => {
							let name = &closing[..idx];
							let body = &closing[idx+1..];

							return Some((ParsedTag{
								closing: false,
								self_closing: true,
								name,
								body
							}, rest))
						}
					}
				}

				let (name, body) = match tag_innards.find(' '){
					None => {
						(tag_innards, "")
					},
					Some(idx) => {
						(&tag_innards[..idx], &tag_innards[idx+1..])
					}
				};

				Some((ParsedTag{
					closing: false,
					self_closing: false,
					name,
					body
				}, rest))
			}
		}
	}

	fn parse_text(raw: &str) -> Consumed {
		let mut end_idx = 0;
		let mut search_from = raw;

		loop {
			match search_from.find('<') {
				// if we ever run out of <'s, the entire string was text
				None => {
					break Consumed {
						node: Node::Text(raw.to_owned()),
						remaining: None,
					}
				}
				Some(idx) => {
					end_idx += idx;

					if Self::is_tag(&search_from[idx..]).is_some() {
						// we've found a new tag, this text node is done
						break Consumed {
							node: Node::Text(raw[..end_idx].to_owned()),
							remaining: Some(&raw[end_idx..]),
						};
					} else {
						// step over the <
						end_idx += 1;
						search_from = &raw[end_idx..];
					}
				}
			}
		}
	}
}

impl fmt::Display for Html {
	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
		for node in &self.nodes {
			write!(f, "{node}")?;
		}
		Ok(())
	}
}

struct Consumed<'a> {
	node: Node,
	remaining: Option<&'a str>,
}

struct ParsedTag<'a> {
	closing: bool,
	name: &'a str,
	// a tag's body is what exists between the end of the name and the end of
	// the tag (including a self-close that may be there and any whitespace)
	body: &'a str,
	self_closing: bool,
}

#[derive(Clone, Debug, PartialEq)]
pub enum Node {
	Text(String),
	Tag(Tag),
	Comment(String),
}

impl fmt::Display for Node {
	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
		match self {
			Self::Text(txt) => write!(f, "{txt}"),
			Self::Tag(tag) => write!(f, "{tag}"),
			Self::Comment(cmt) => write!(f, "<!--{cmt}-->"),
		}
	}
}

#[macro_export]
macro_rules! tag {
	($name:expr) => {
		$crate::Node::Tag($crate::Tag {
			name: String::from($name),
			body: None,
			self_closing: false,
			children: vec![],
		})
	};

	($name:expr, [$($children:expr),+]) => {
		$crate::Node::Tag($crate::Tag {
			name: String::from($name),
			body: None,
			self_closing: false,
			children: vec![$($children),+],
		})
	};

	($name:expr, $body:expr) => {
		$crate::Node::Tag($crate::Tag {
			name: String::from($name),
			body: Some(String::from($body)),
			self_closing: false,
			children: vec![],
		})
	};

	($name:expr, $body:expr, [$($children:expr),+]) => {
		$crate::Node::Tag($crate::Tag {
			name: String::from($name),
			body: Some(String::from($body)),
			self_closing: false,
			children: vec![$($children),+],
		})
	};
}

#[macro_export]
macro_rules! text {
	($text:expr) => {
		$crate::Node::Text(String::from($text))
	};
}

#[macro_export]
macro_rules! comment {
	($text:expr) => {
		$crate::Node::Comment(String::from($text))
	};
}

#[cfg(test)]
mod test {
	use crate::{Html, Node};

	macro_rules! text {
		($text:expr) => {
			Node::Text(String::from($text))
		};
	}

	#[test]
	fn parse_text_finds_start_of_tag() {
		let no_tag = "Hello, World!";
		let starts_tag = "<p>Hello, World!";
		let ends_tag = "Hello, World!</p>";

		let no_tag_res = Html::parse_text(no_tag);
		assert_eq!(no_tag_res.node, text!("Hello, World!"));
		assert!(no_tag_res.remaining.is_none());

		let starts_tag_res = Html::parse_text(starts_tag);
		assert_eq!(starts_tag_res.node, text!(""));
		assert_eq!(starts_tag_res.remaining, Some(starts_tag));

		let ends_tag_res = Html::parse_text(ends_tag);
		assert_eq!(ends_tag_res.node, text!("Hello, World!"));
		assert_eq!(ends_tag_res.remaining, Some("</p>"));
	}

	#[test]
	fn parse_text_correctly_ignores_nontags() {
		let sentence = "The condition 2 < 1 should be 1 > 2";
		let weird = "Hello, < p>";
		let no_close = "Hello <p my name is ";

		let sentence_res = Html::parse_text(sentence);
		assert_eq!(sentence_res.node, text!(sentence));
		assert!(sentence_res.remaining.is_none());

		let weird_res = Html::parse_text(weird);
		assert_eq!(weird_res.node, text!(weird));
		assert!(weird_res.remaining.is_none());

		let no_close_res = Html::parse_text(no_close);
		assert_eq!(no_close_res.node, text!(no_close));
		assert!(no_close_res.remaining.is_none());
	}

	#[test]
	fn special_parse_find_tag_end() {
		let basic = "words words\n</script>";
		let special = Html::special_parse(basic, "script");
		assert_eq!(special.unwrap().0, "words words\n");
		assert!(special.unwrap().1.is_empty());
	}

	#[test]
	fn special_parse_correctly_ignore_non_start() {
		let nonstart = "first_line\nlet end = '</script>';\n";
		let special = Html::special_parse(nonstart, "script");
		assert!(special.is_none());
	}

	#[test]
	fn special_parse_correctly_handles_leading_whitespace() {
		let white = "words words\n  \t\t</script>";
		let special = Html::special_parse(white, "script");
		assert_eq!(special.unwrap().0, "words words\n  \t\t");
	}

	#[test]
	fn parse_node_parses_comment() {
		let cmt = "<!-- Comment! -->";

		let node = Html::parse_node(cmt);
		assert_eq!(node.node, comment!(" Comment! "));
	}

	#[test]
	fn parse_node_parses_tag() {
		let basic = "<p>Hello!</p>";

		let hh = Html::parse_node(basic);
		assert_eq!(hh.node, tag!("p", [text!("Hello!")]))
	}

	#[test]
	fn parse_node_parses_nested_tags() {
		let nested = "<p><p>Hello!</p></p>";

		let hh = Html::parse_node(nested);
		assert_eq!(hh.node, tag!("p", [tag!("p", [text!("Hello!")])]))
	}

	#[test]
	fn parse_multiple_toplevel() {
		let nested = "<p>Hello </p><p>World!</p>";

		let hh = Html::parse(nested);
		assert_eq!(
			hh.nodes,
			vec![tag!("p", [text!("Hello ")]), tag!("p", [text!("World!")])]
		)
	}

	#[test]
	fn parse_script() {
		let raw = "<head>\n\t<script>let k=\"v\";\n\t</script>\n</head>";

		let hh = Html::parse(raw);
		assert_eq!(
			hh.nodes,
			vec![tag!(
				"head",
				[
					text!("\n\t"),
					tag!("script", [text!("let k=\"v\";\n\t")]),
					text!("\n")
				]
			)]
		)
	}

	#[test]
	fn parse_external_script() {
		let raw = "<head>\n\t<script src=\"script.js\"></script>\n</head>";

		let hh = Html::parse(raw);
		assert_eq!(
			hh.nodes,
			vec![tag!(
				"head",
				[
					text!("\n\t"),
					tag!("script", "src=\"script.js\""),
					text!("\n")
				]
			)]
		)
	}

	fn test_roundtrip(raw: &str) {
		let html = Html::parse(raw);
		let string = html.to_string();

		for (raw, html) in raw.lines().zip(string.lines()) {
			assert_eq!(raw, html)
		}
	}

	#[test]
	fn round_trip_simple() {
		test_roundtrip("<p>Hello!</p>")
	}

	#[test]
	fn round_trip_complex() {
		test_roundtrip(
			r#"
<html>
	<head>
		<link rel="style.css"/>
		<title>Title!</title>
		<script>
			alert("hello!");
		</script>
	</head>
	<body>
		<p>Hello, <i>World!</i></p>
	</body>
</html>"#,
		)
	}
}