Sync with cmark-gfm-0.29.0.gfm.11

README.md

@@ -6,7 +6,7 @@ Collective](https://opencollective.com/comrak/all/badge.svg?label=financial+cont
 [![crates.io version](https://img.shields.io/crates/v/comrak.svg)](https://crates.io/crates/comrak)
 [![docs.rs](https://docs.rs/comrak/badge.svg)](https://docs.rs/comrak)
 
-Rust port of [github's `cmark-gfm`](https://github.com/github/cmark).
+Rust port of [github's `cmark-gfm`](https://github.com/github/cmark). *Currently synced with release `0.29.0.gfm.11`*.
 
 - [Installation](#installation)
 - [Usage](#usage)

fuzz/fuzz_targets/quadratic.rs

@@ -2,7 +2,8 @@
 #![feature(int_roundings)]
 #![no_main]
 use comrak::{
-    markdown_to_html, ExtensionOptions, Options, ParseOptions,
+    markdown_to_html, markdown_to_commonmark, markdown_to_commonmark_xml,
+    ExtensionOptions, Options, ParseOptions,
     RenderOptions, ListStyleType,
 };
 use libfuzzer_sys::arbitrary::{self, Arbitrary};
@@ -273,20 +274,22 @@ fn fuzz_one_input(input: &Input, num_bytes: usize) -> (usize, Duration, f64) {
     let now = Instant::now();
     {
         let _ = markdown_to_html(&markdown, &input.options.to_options());
+        let _ = markdown_to_commonmark(&markdown, &input.options.to_options());
+        let _ = markdown_to_commonmark_xml(&markdown, &input.options.to_options());
     }
+
     let duration = now.elapsed();
+    let byte_length = markdown.len() * 3;
+    let duration_per_byte = duration.as_secs_f64() / (byte_length as f64);
 
     if DEBUG {
         println!(
             "do_one: {} bytes, duration = {:?}",
-            markdown.len(),
+            byte_length,
             duration
         );
     }
 
-    let byte_length = markdown.len();
-    let duration_per_byte = duration.as_secs_f64() / (markdown.len() as f64);
-
     (
         byte_length,
         duration,

src/cm.rs

@@ -308,13 +308,23 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
         self.node = node;
         let allow_wrap = self.options.render.width > 0 && !self.options.render.hardbreaks;
 
-        if !(matches!(
-            node.data.borrow().value,
-            NodeValue::Item(..) | NodeValue::TaskItem(..)
-        ) && node.previous_sibling().is_none()
-            && entering)
-        {
-            self.in_tight_list_item = self.get_in_tight_list_item(node);
+        let parent_node = node.parent();
+        if entering {
+            if parent_node.is_some()
+                && matches!(
+                    parent_node.unwrap().data.borrow().value,
+                    NodeValue::Item(..) | NodeValue::TaskItem(..)
+                )
+            {
+                self.in_tight_list_item = self.get_in_tight_list_item(node);
+            }
+        } else if matches!(node.data.borrow().value, NodeValue::List(..)) {
+            self.in_tight_list_item = parent_node.is_some()
+                && matches!(
+                    parent_node.unwrap().data.borrow().value,
+                    NodeValue::Item(..) | NodeValue::TaskItem(..)
+                )
+                && self.get_in_tight_list_item(node);
         }
 
         match node.data.borrow().value {
@@ -343,7 +353,13 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
             NodeValue::HtmlInline(ref literal) => {
                 self.format_html_inline(literal.as_bytes(), entering)
             }
-            NodeValue::Strong => self.format_strong(),
+            NodeValue::Strong => {
+                if parent_node.is_none()
+                    || !matches!(parent_node.unwrap().data.borrow().value, NodeValue::Strong)
+                {
+                    self.format_strong();
+                }
+            }
             NodeValue::Emph => self.format_emph(node),
             NodeValue::TaskItem(symbol) => self.format_task_item(symbol, node, entering),
             NodeValue::Strikethrough => self.format_strikethrough(),
@@ -410,13 +426,12 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
         let marker_width = if parent.list_type == ListType::Bullet {
             2
         } else {
-            let mut list_number = parent.start;
+            let list_number = match node.data.borrow().value {
+                NodeValue::Item(ref ni) => ni.start,
+                NodeValue::TaskItem(_) => parent.start,
+                _ => unreachable!(),
+            };
             let list_delim = parent.delimiter;
-            let mut tmpch = node;
-            while let Some(tmp) = tmpch.previous_sibling() {
-                tmpch = tmp;
-                list_number += 1;
-            }
             write!(
                 listmarker,
                 "{}{}{}",

src/html.rs

@@ -760,12 +760,17 @@ impl<'o> HtmlFormatter<'o> {
                 }
             }
             NodeValue::Strong => {
-                if entering {
-                    self.output.write_all(b"<strong")?;
-                    self.render_sourcepos(node)?;
-                    self.output.write_all(b">")?;
-                } else {
-                    self.output.write_all(b"</strong>")?;
+                let parent_node = node.parent();
+                if parent_node.is_none()
+                    || !matches!(parent_node.unwrap().data.borrow().value, NodeValue::Strong)
+                {
+                    if entering {
+                        self.output.write_all(b"<strong")?;
+                        self.render_sourcepos(node)?;
+                        self.output.write_all(b">")?;
+                    } else {
+                        self.output.write_all(b"</strong>")?;
+                    }
                 }
             }
             NodeValue::Emph => {
@@ -977,7 +982,7 @@ impl<'o> HtmlFormatter<'o> {
                     self.output.write_all(b">")?;
                     write!(
                         self.output,
-                        "<input type=\"checkbox\" disabled=\"\" {}/> ",
+                        "<input type=\"checkbox\" {}disabled=\"\" /> ",
                         if symbol.is_some() {
                             "checked=\"\" "
                         } else {

src/parser/mod.rs

@@ -31,6 +31,12 @@ use self::inlines::RefMap;
 const TAB_STOP: usize = 4;
 const CODE_INDENT: usize = 4;
 
+// Very deeply nested lists can cause quadratic performance issues.
+// This constant is used in open_new_blocks() to limit the nesting
+// depth. It is unlikely that a non-contrived markdown document will
+// be nested this deeply.
+const MAX_LIST_DEPTH: usize = 100;
+
 macro_rules! node_matches {
     ($node:expr, $( $pat:pat )|+) => {{
         matches!(
@@ -222,7 +228,7 @@ pub struct ExtensionOptions {
     /// options.extension.tasklist = true;
     /// options.render.unsafe_ = true;
     /// assert_eq!(markdown_to_html("* [x] Done\n* [ ] Not done\n", &options),
-    ///            "<ul>\n<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Done</li>\n\
+    ///            "<ul>\n<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Done</li>\n\
     ///            <li><input type=\"checkbox\" disabled=\"\" /> Not done</li>\n</ul>\n");
     /// ```
     pub tasklist: bool,
@@ -954,11 +960,13 @@ impl<'a, 'o, 'c> Parser<'a, 'o, 'c> {
         let mut nl: NodeList = NodeList::default();
         let mut sc: scanners::SetextChar = scanners::SetextChar::Equals;
         let mut maybe_lazy = node_matches!(self.current, NodeValue::Paragraph);
+        let mut depth = 0;
 
         while !node_matches!(
             container,
             NodeValue::CodeBlock(..) | NodeValue::HtmlBlock(..)
         ) {
+            depth += 1;
             self.find_first_nonspace(line);
             let indented = self.indent >= CODE_INDENT;
 
@@ -1112,6 +1120,7 @@ impl<'a, 'o, 'c> Parser<'a, 'o, 'c> {
                 }
             } else if (!indented || node_matches!(container, NodeValue::List(..)))
                 && self.indent < 4
+                && depth < MAX_LIST_DEPTH
                 && unwrap_into_2(
                     parse_list_marker(
                         line,

src/tests.rs

@@ -6,6 +6,7 @@ use std::panic;
 
 mod api;
 mod autolink;
+mod commonmark;
 mod core;
 mod description_lists;
 mod footnotes;

src/tests/commonmark.rs

@@ -0,0 +1,11 @@
+use super::*;
+
+#[test]
+fn commonmark_removes_redundant_strong() {
+    let options = ComrakOptions::default();
+
+    let input = "This is **something **even** better**";
+    let output = "This is **something even better**\n";
+
+    commonmark(input, output, Some(&options));
+}

src/tests/fuzz.rs

@@ -10,7 +10,7 @@ fn tasklist() {
     html_opts!(
         [extension.tasklist, parse.relaxed_tasklist_matching],
         "* [*]",
-        "<ul>\n<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> </li>\n</ul>\n",
+        "<ul>\n<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> </li>\n</ul>\n",
     );
 }
 

src/tests/tasklist.rs

@@ -25,21 +25,21 @@ fn tasklist() {
         concat!(
             "<ul>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> Red</li>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Green</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Green</li>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> Blue</li>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Papayawhip</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Papayawhip</li>\n",
             "</ul>\n",
             "<!-- end list -->\n",
             "<ol>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> Bird</li>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> McHale</li>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Parish</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Parish</li>\n",
             "</ol>\n",
             "<!-- end list -->\n",
             "<ul>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> Red\n",
             "<ul>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Green\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Green\n",
             "<ul>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> Blue</li>\n",
             "</ul>\n",
@@ -58,7 +58,7 @@ fn tasklist_relaxed_regression() {
         "* [!] Red\n",
         concat!(
             "<ul>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Red</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Red</li>\n",
             "</ul>\n"
         ),
     );
@@ -74,7 +74,7 @@ fn tasklist_relaxed_regression() {
         "* [!] Red\n",
         concat!(
             "<ul>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> Red</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> Red</li>\n",
             "</ul>\n"
         ),
     );
@@ -93,7 +93,7 @@ fn tasklist_32() {
             "<ul>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> List item 1</li>\n",
             "<li><input type=\"checkbox\" disabled=\"\" /> This list item is <strong>bold</strong></li>\n",
-            "<li><input type=\"checkbox\" disabled=\"\" checked=\"\" /> There is some <code>code</code> here</li>\n",
+            "<li><input type=\"checkbox\" checked=\"\" disabled=\"\" /> There is some <code>code</code> here</li>\n",
             "</ul>\n"
         ),
     );

src/xml.rs

@@ -1,10 +1,13 @@
 use crate::nodes::{AstNode, ListType, NodeCode, NodeValue};
 use crate::parser::{Options, Plugins};
 use once_cell::sync::Lazy;
+use std::cmp;
 use std::io::{self, Write};
 
 use crate::nodes::NodeHtmlBlock;
 
+const MAX_INDENT: u32 = 40;
+
 /// Formats an AST as HTML, modified by the given options.
 pub fn format_document<'a>(
     root: &'a AstNode<'a>,
@@ -121,7 +124,7 @@ impl<'o> XmlFormatter<'o> {
     }
 
     fn indent(&mut self) -> io::Result<()> {
-        for _ in 0..self.indent {
+        for _ in 0..(cmp::min(self.indent, MAX_INDENT)) {
             self.output.write_all(b" ")?;
         }
         Ok(())