#[macro_use] extern crate html5ever; #[macro_use] extern crate lazy_static; use html5ever::{driver as html, QualName}; use html5ever::rcdom::{Handle, NodeData, RcDom}; use html5ever::serialize::{serialize, SerializeOpts}; use pulldown_cmark_fork::{Parser, Options}; use std::collections::HashSet; use std::mem; use std::rc::{Rc, Weak}; use tendril::stream::TendrilSink; use regex::Regex; mod suite; #[inline(never)] pub fn test_markdown_html(input: &str, output: &str) { let mut s = String::new(); let mut opts = Options::empty(); opts.insert(Options::ENABLE_TABLES); opts.insert(Options::ENABLE_FOOTNOTES); opts.insert(Options::ENABLE_STRIKETHROUGH); opts.insert(Options::ENABLE_TASKLISTS); let p = Parser::new_ext(input, opts); pulldown_cmark_fork::html::push_html(&mut s, p); assert_eq!(normalize_html(output), normalize_html(&s)); } lazy_static! { static ref WHITESPACE_RE: Regex = Regex::new(r"\s+").unwrap(); static ref LEADING_WHITESPACE_RE: Regex = Regex::new(r"\A\s+").unwrap(); static ref TRAILING_WHITESPACE_RE: Regex = Regex::new(r"\s+\z").unwrap(); static ref BLOCK_TAGS: HashSet<&'static str> = [ "article", "header", "aside", "hgroup", "blockquote", "hr", "iframe", "body", "li", "map", "button", "object", "canvas", "ol", "caption", "output", "col", "p", "colgroup", "pre", "dd", "progress", "div", "section", "dl", "table", "td", "dt", "tbody", "embed", "textarea", "fieldset", "tfoot", "figcaption", "th", "figure", "thead", "footer", "tr", "form", "ul", "h1", "h2", "h3", "h4", "h5", "h6", "video", "script", "style" ].iter().cloned().collect(); static ref PRE_TAGS: HashSet<&'static str> = [ "pre", "code" ].iter().cloned().collect(); static ref TABLE_TAGS: HashSet<&'static str> = [ "table", "thead", "tbody", "tr", "td" ].iter().cloned().collect(); } fn make_html_parser() -> html::Parser { html::parse_fragment( RcDom::default(), html::ParseOpts::default(), QualName::new(None, ns!(html), local_name!("div")), vec![], ) } fn normalize_html(s: &str) -> String { let parser = make_html_parser(); let dom = parser.one(s); let body = normalize_dom(&dom); let opts = SerializeOpts::default(); let mut ret_val = Vec::new(); serialize(&mut ret_val, &body, opts) .expect("Writing to a string shouldn't fail (expect on OOM)"); String::from_utf8(ret_val) .expect("html5ever should always produce UTF8") } fn normalize_dom(dom: &RcDom) -> Handle { let body = { let children = dom.document.children.borrow(); children[0].clone() }; let mut current_level = Vec::new(); let mut next_level = Vec::new(); current_level.extend( body.children.borrow() .iter() .cloned() .rev() ); loop { while let Some(mut node) = current_level.pop() { let parent = node.parent.replace(None); node.parent.replace(parent.clone()); let parent = parent .expect("a node in the DOM will have a parent, except the root, which is not processed") .upgrade().expect("a node's parent will be pointed to by its parent (or the root pointer), and will not be dropped"); let retain = normalize_node(&parent, &mut node); if !retain { let mut siblings = parent.children.borrow_mut(); siblings.retain(|s| !Rc::ptr_eq(&node, s)); } else { next_level.extend( node.children.borrow() .iter() .cloned() .rev(), ); } } if next_level.is_empty() { break }; mem::swap(&mut next_level, &mut current_level); } body } // Returns false if node is an empty text node or an empty tbody. // Returns true otherwise. fn normalize_node(parent: &Handle, node: &mut Handle) -> bool { match node.data { NodeData::Comment { .. } | NodeData::Doctype { .. } | NodeData::Document | NodeData::ProcessingInstruction { .. } => true, NodeData::Text { ref contents, .. } => { let mut contents = contents.borrow_mut(); let is_pre = { let mut parent = parent.clone(); loop { let is_pre = if let NodeData::Element{ ref name, .. } = parent.data { PRE_TAGS.contains(&&*name.local.to_ascii_lowercase()) } else { false }; if is_pre { break true }; let parent_ = parent.parent.replace(None); parent.parent.replace(parent_.clone()); let parent_ = parent_.as_ref().and_then(Weak::upgrade); if let Some(parent_) = parent_ { parent = parent_ } else { break false }; } }; if !is_pre { let (is_first_in_block, is_last_in_block) = { let mut is_first_in_block = true; let mut is_last_in_block = true; let mut parent = parent.clone(); let mut node = node.clone(); loop { let reached_block = if let NodeData::Element{ ref name, .. } = parent.data { BLOCK_TAGS.contains(&&*name.local.to_ascii_lowercase()) } else { false }; let (is_first, is_last) = { let siblings = parent.children.borrow(); let n = &node; (siblings.get(0).map(|s| Rc::ptr_eq(s, n)).unwrap_or(false), siblings.len() > 0 && siblings.get(siblings.len() - 1).map(|s| Rc::ptr_eq(s, n)).unwrap_or(false)) }; is_first_in_block = is_first_in_block && is_first; is_last_in_block = is_last_in_block && is_last; if (is_first_in_block || is_last_in_block) && !reached_block { node = parent.clone(); let parent_ = parent.parent.replace(None); parent.parent.replace(parent_.clone()); let parent_ = parent_.as_ref().and_then(Weak::upgrade); if let Some(parent_) = parent_ { parent = parent_; } else { break (is_first_in_block, is_last_in_block) } } else { break (is_first_in_block, is_last_in_block) } } }; let is_preceeded_by_ws = { let mut parent = parent.clone(); let mut node = node.clone(); 'ascent: loop { let is_first = { let siblings = parent.children.borrow(); let n = &node; siblings.get(0).map(|s| Rc::ptr_eq(s, n)).unwrap_or(false) }; if is_first { node = parent.clone(); let parent_ = parent.parent.replace(None); parent.parent.replace(parent_.clone()); let parent_ = parent_.as_ref().and_then(Weak::upgrade); if let Some(parent_) = parent_ { parent = parent_; } else { break 'ascent false } } else { let siblings = parent.children.borrow(); let n = &node; let mut pos = !0; 'search: for (i, s) in siblings.iter().enumerate() { if Rc::ptr_eq(s, n) { pos = i; break 'search; } } assert!(pos != !0, "The list of node's parent's children shall contain node"); assert!(pos != 0, "If node is not first, then node's position shall not be zero"); let mut preceeding = siblings[pos-1].clone(); 'descent: loop { if let NodeData::Text { .. } = preceeding.data { break 'descent } preceeding = { let ch = preceeding.children.borrow(); if ch.len() == 0 { break 'descent } if let Some(preceeding_) = ch.get(ch.len() - 1) { preceeding_.clone() } else { break 'descent } }; } if let NodeData::Text { ref contents, .. } = preceeding.data { break 'ascent TRAILING_WHITESPACE_RE.is_match(&*contents.borrow()) } else { break 'ascent false } } } }; let is_in_table = if let NodeData::Element{ ref name, .. } = parent.data { TABLE_TAGS.contains(&&*name.local.to_ascii_lowercase()) } else { false }; let whitespace_replacement = if is_in_table { "" } else { " " }; *contents = WHITESPACE_RE.replace_all(&*contents, whitespace_replacement).as_ref().into(); if is_first_in_block || is_preceeded_by_ws { *contents = LEADING_WHITESPACE_RE.replace_all(&*contents, "").as_ref().into(); } if is_last_in_block { *contents = TRAILING_WHITESPACE_RE.replace_all(&*contents, "").as_ref().into(); } // TODO: collapse whitespace when adjacent to whitespace. // For example, the whitespace in the span should be collapsed in all of these cases: // // " q " // "q q" // "q q" // "q q" // "q q" } &**contents != "" } NodeData::Element { ref attrs, ref name, .. } => { let mut attrs = attrs.borrow_mut(); for a in attrs.iter_mut() { a.name.local = a.name.local.to_ascii_lowercase().into(); } attrs.sort_by(|a: &html5ever::Attribute, b: &html5ever::Attribute| { (&*a.name.local).cmp(&*b.name.local) }); let ascii_name = &*name.local.to_ascii_lowercase(); // drop empty tbody's ascii_name != "tbody" || node.children.borrow().len() > 1 || node.children.borrow().iter().next().map(|only_child| match only_child.data { NodeData::Text { ref contents, .. } => { !contents.borrow().chars().all(|c| c.is_whitespace()) } _ => { true } }).unwrap_or(false) } } } #[test] fn strip_div_newline() { assert_eq!("
", normalize_html("
\n
")); } #[test] fn strip_end_newline() { assert_eq!("test", normalize_html("test\n")); } #[test] fn strip_double_space() { assert_eq!("test mess", normalize_html("test mess")); } #[test] fn strip_inline_internal_text() { assert_eq!("a b c", normalize_html(" a b c ")) } #[test] fn strip_inline_block_internal_text() { assert_eq!("a b c", normalize_html(" a b c ")) } #[test] fn leaves_necessary_whitespace_alone() { assert_eq!("a b c", normalize_html("a b c")) } #[test] fn leaves_necessary_whitespace_alone_weird() { assert_eq!("a b c", normalize_html(" a b c")) } #[test] fn leaves_necessary_whitespace_all_nested() { assert_eq!("", normalize_html(" ")) } #[test] fn drops_empty_tbody() { assert_eq!( "
hi
", normalize_html("
hi
") ) } #[test] fn leaves_nonempty_tbody() { let input = "
hi
"; assert_eq!(input, normalize_html(input)) }