From 57c3cda9f98e70a9f614a22eb6d518051cc60b19 Mon Sep 17 00:00:00 2001
From: Titus Wormer <tituswormer@gmail.com>
Date: Mon, 12 Sep 2022 17:18:30 +0200
Subject: Fix gfm email autolink literals overlapping
---
src/construct/gfm_autolink_literal.rs | 26 +++++++++++++-------------
src/event.rs | 2 +-
tests/fuzz.rs | 19 +++++++++++++++++--
3 files changed, 31 insertions(+), 16 deletions(-)
diff --git a/src/construct/gfm_autolink_literal.rs b/src/construct/gfm_autolink_literal.rs
index c25f04c..c4ef637 100644
--- a/src/construct/gfm_autolink_literal.rs
+++ b/src/construct/gfm_autolink_literal.rs
@@ -636,14 +636,14 @@ pub fn resolve(tokenizer: &mut Tokenizer) {
let mut replace = Vec::new();
let mut point = tokenizer.events[index - 1].point.clone();
let start_index = point.index;
- let mut start = 0;
+ let mut min = 0;
while byte_index < bytes.len() {
if bytes[byte_index] == b'@' {
let mut range = (0, 0, Name::GfmAutolinkLiteralEmail);
- if let Some(start) = peek_bytes_atext(bytes, byte_index) {
- let (start, kind) = peek_protocol(bytes, start);
+ if let Some(start) = peek_bytes_atext(bytes, min, byte_index) {
+ let (start, kind) = peek_protocol(bytes, min, start);
if let Some(end) = peek_bytes_email_domain(
bytes,
@@ -658,8 +658,8 @@ pub fn resolve(tokenizer: &mut Tokenizer) {
byte_index = range.1;
// If there is something between the last link
- // (or the start) and this link.
- if start != range.0 {
+ // (or `min`) and this link.
+ if min != range.0 {
replace.push(Event {
kind: Kind::Enter,
name: Name::Data,
@@ -691,7 +691,7 @@ pub fn resolve(tokenizer: &mut Tokenizer) {
point: point.clone(),
link: None,
});
- start = range.1;
+ min = range.1;
}
}
@@ -699,7 +699,7 @@ pub fn resolve(tokenizer: &mut Tokenizer) {
}
// If there was a link, and we have more bytes left.
- if start != 0 && start < bytes.len() {
+ if min != 0 && min < bytes.len() {
replace.push(Event {
kind: Kind::Enter,
name: Name::Data,
@@ -739,13 +739,13 @@ pub fn resolve(tokenizer: &mut Tokenizer) {
/// ^-- from
/// ^-- to
/// ```
-fn peek_bytes_atext(bytes: &[u8], end: usize) -> Option<usize> {
+fn peek_bytes_atext(bytes: &[u8], min: usize, end: usize) -> Option<usize> {
let mut index = end;
// Take simplified atext.
// See `email_atext` in `autolink.rs` for a similar algorithm.
// Source: <https://github.com/github/cmark-gfm/blob/ef1cfcb/extensions/autolink.c#L301>.
- while index > 0
+ while index > min
&& matches!(bytes[index - 1], b'+' | b'-' | b'.' | b'0'..=b'9' | b'A'..=b'Z' | b'_' | b'a'..=b'z')
{
index -= 1;
@@ -755,7 +755,7 @@ fn peek_bytes_atext(bytes: &[u8], end: usize) -> Option<usize> {
// The reference code is a bit weird, but that’s what it results in.
// Source: <https://github.com/github/cmark-gfm/blob/ef1cfcb/extensions/autolink.c#L307>.
// Other than slash, every preceding character is allowed.
- if index == end || (index > 0 && bytes[index - 1] == b'/') {
+ if index == end || (index > min && bytes[index - 1] == b'/') {
None
} else {
Some(index)
@@ -772,14 +772,14 @@ fn peek_bytes_atext(bytes: &[u8], end: usize) -> Option<usize> {
/// ^-- from
/// ^-- to
/// ```
-fn peek_protocol(bytes: &[u8], end: usize) -> (usize, Name) {
+fn peek_protocol(bytes: &[u8], min: usize, end: usize) -> (usize, Name) {
let mut index = end;
- if index > 0 && bytes[index - 1] == b':' {
+ if index > min && bytes[index - 1] == b':' {
index -= 1;
// Take alphanumerical.
- while index > 0 && matches!(bytes[index - 1], b'0'..=b'9' | b'A'..=b'Z' | b'a'..=b'z') {
+ while index > min && matches!(bytes[index - 1], b'0'..=b'9' | b'A'..=b'Z' | b'a'..=b'z') {
index -= 1;
}
diff --git a/src/event.rs b/src/event.rs
index 6ea52fb..de3f95f 100644
--- a/src/event.rs
+++ b/src/event.rs
@@ -3424,7 +3424,7 @@ impl Point {
/// point, to `index.`
pub fn shift_to(&self, bytes: &[u8], index: usize) -> Point {
let mut next = self.clone();
- debug_assert!(index > next.index, "expect");
+ debug_assert!(index > next.index, "expected to shift forward");
while next.index < index {
match bytes[next.index] {
diff --git a/tests/fuzz.rs b/tests/fuzz.rs
index 3cc1066..126032a 100644
--- a/tests/fuzz.rs
+++ b/tests/fuzz.rs
@@ -1,5 +1,5 @@
extern crate micromark;
-use micromark::micromark;
+use micromark::{micromark, micromark_with_options, Constructs, Options};
use pretty_assertions::assert_eq;
#[test]
@@ -7,7 +7,22 @@ fn fuzz() -> Result<(), String> {
assert_eq!(
micromark("[\n~\na\n-\n\n"),
"<p>[\n~\na</p>\n<ul>\n<li></li>\n</ul>\n",
- "1"
+ "1: label, blank lines, and code"
+ );
+
+ assert_eq!(
+ // The first link is stopped by the `+` (so it’s `a@b.c`), but the next
+ // link overlaps it (`b.c+d@e.f`).
+ micromark_with_options(
+ "a@b.c+d@e.f",
+ &Options {
+ constructs: Constructs::gfm(),
+ gfm_tagfilter: true,
+ ..Options::default()
+ }
+ )?,
+ "<p><a href=\"mailto:a@b.c\">a@b.c</a><a href=\"mailto:+d@e.f\">+d@e.f</a></p>",
+ "2: gfm: email autolink literals running into each other"
);
Ok(())
--
cgit