Rollup merge of #76689 - jyn514:update-pulldown, r=GuillaumeGomez

Upgrade to pulldown-cmark 0.8.0

Thanks to marcusklaas' hard work in https://github.com/raphlinus/pulldown-cmark/pull/469, this fixes a lot of rustdoc bugs!

- Get rid of unnecessary `RefCell`
- Fix duplicate warnings for broken implicit reference link
- Remove unnecessary copy of links

Closes https://github.com/rust-lang/rust/issues/73264, closes https://github.com/rust-lang/rust/issues/76687.
r? @euclio

I'm not sure if the switch away from `locate` fixes any open bugs - euclio mentioned some in https://github.com/raphlinus/pulldown-cmark/issues/165, but I didn't see any related issues open for rustdoc. Let me know if I missed one.

Cargo.lock

@@ -534,7 +534,7 @@ dependencies = [
  "if_chain",
  "itertools 0.9.0",
  "lazy_static",
- "pulldown-cmark",
+ "pulldown-cmark 0.7.2",
  "quine-mc_cluskey",
  "quote",
  "regex-syntax",
@@ -1853,7 +1853,7 @@ dependencies = [
  "log",
  "memchr",
  "open",
- "pulldown-cmark",
+ "pulldown-cmark 0.7.2",
  "regex",
  "serde",
  "serde_derive",
@@ -2511,6 +2511,17 @@ dependencies = [
  "unicase",
 ]
 
+[[package]]
+name = "pulldown-cmark"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ffade02495f22453cd593159ea2f59827aae7f53fa8323f756799b670881dcf8"
+dependencies = [
+ "bitflags",
+ "memchr",
+ "unicase",
+]
+
 [[package]]
 name = "punycode"
 version = "0.4.1"
@@ -4122,7 +4133,7 @@ dependencies = [
  "expect-test",
  "itertools 0.9.0",
  "minifier",
- "pulldown-cmark",
+ "pulldown-cmark 0.8.0",
  "rustc-rayon",
  "serde",
  "serde_json",

src/librustdoc/Cargo.toml

@@ -8,7 +8,7 @@ edition = "2018"
 path = "lib.rs"
 
 [dependencies]
-pulldown-cmark = { version = "0.7", default-features = false }
+pulldown-cmark = { version = "0.8", default-features = false }
 minifier = "0.0.33"
 rayon = { version = "0.3.0", package = "rustc-rayon" }
 serde = { version = "1.0", features = ["derive"] }

src/librustdoc/html/markdown.rs

@@ -27,7 +27,6 @@
 use rustc_span::edition::Edition;
 use rustc_span::Span;
 use std::borrow::Cow;
-use std::cell::RefCell;
 use std::collections::VecDeque;
 use std::default::Default;
 use std::fmt::Write;
@@ -39,7 +38,7 @@
 use crate::html::highlight;
 use crate::html::toc::TocBuilder;
 
-use pulldown_cmark::{html, CodeBlockKind, CowStr, Event, Options, Parser, Tag};
+use pulldown_cmark::{html, BrokenLink, CodeBlockKind, CowStr, Event, Options, Parser, Tag};
 
 #[cfg(test)]
 mod tests;
@@ -931,15 +930,17 @@ pub fn into_string(self) -> String {
         if md.is_empty() {
             return String::new();
         }
-        let replacer = |_: &str, s: &str| {
+        let mut replacer = |broken_link: BrokenLink<'_>| {
-            if let Some(link) = links.iter().find(|link| &*link.original_text == s) {
+            if let Some(link) =
-                Some((link.href.clone(), link.new_text.clone()))
+                links.iter().find(|link| &*link.original_text == broken_link.reference)
+            {
+                Some((link.href.as_str().into(), link.new_text.as_str().into()))
             } else {
                 None
             }
         };
 
-        let p = Parser::new_with_broken_link_callback(md, opts(), Some(&replacer));
+        let p = Parser::new_with_broken_link_callback(md, opts(), Some(&mut replacer));
 
         let mut s = String::with_capacity(md.len() * 3 / 2);
 
@@ -1009,9 +1010,11 @@ pub fn into_string(self) -> String {
             return String::new();
         }
 
-        let replacer = |_: &str, s: &str| {
+        let mut replacer = |broken_link: BrokenLink<'_>| {
-            if let Some(link) = links.iter().find(|link| &*link.original_text == s) {
+            if let Some(link) =
-                Some((link.href.clone(), link.new_text.clone()))
+                links.iter().find(|link| &*link.original_text == broken_link.reference)
+            {
+                Some((link.href.as_str().into(), link.new_text.as_str().into()))
             } else {
                 None
             }
@@ -1020,7 +1023,7 @@ pub fn into_string(self) -> String {
         let p = Parser::new_with_broken_link_callback(
             md,
             Options::ENABLE_STRIKETHROUGH,
-            Some(&replacer),
+            Some(&mut replacer),
         );
 
         let mut s = String::new();
@@ -1067,7 +1070,7 @@ pub fn markdown_links(md: &str) -> Vec<(String, Option<Range<usize>>)> {
     }
 
     let mut links = vec![];
-    let shortcut_links = RefCell::new(vec![]);
+    let mut shortcut_links = vec![];
 
     {
         let locate = |s: &str| unsafe {
@@ -1084,11 +1087,13 @@ pub fn markdown_links(md: &str) -> Vec<(String, Option<Range<usize>>)> {
             }
         };
 
-        let push = |_: &str, s: &str| {
+        let mut push = |link: BrokenLink<'_>| {
-            shortcut_links.borrow_mut().push((s.to_owned(), locate(s)));
+            // FIXME: use `link.span` instead of `locate`
+            // (doing it now includes the `[]` as well as the text)
+            shortcut_links.push((link.reference.to_owned(), locate(link.reference)));
             None
         };
-        let p = Parser::new_with_broken_link_callback(md, opts(), Some(&push));
+        let p = Parser::new_with_broken_link_callback(md, opts(), Some(&mut push));
 
         // There's no need to thread an IdMap through to here because
         // the IDs generated aren't going to be emitted anywhere.
@@ -1106,8 +1111,7 @@ pub fn markdown_links(md: &str) -> Vec<(String, Option<Range<usize>>)> {
         }
     }
 
-    let mut shortcut_links = shortcut_links.into_inner();
+    links.append(&mut shortcut_links);
-    links.extend(shortcut_links.drain(..));
 
     links
 }

src/test/rustdoc-ui/intra-link-double-anchor.rs

@@ -0,0 +1,7 @@
+// check-pass
+
+// regression test for #73264
+// should only give one error
+/// docs [label][with#anchor#error]
+//~^ WARNING multiple anchors
+pub struct S;

src/test/rustdoc-ui/intra-link-double-anchor.stderr

@@ -0,0 +1,10 @@
+warning: `with#anchor#error` contains multiple anchors
+  --> $DIR/intra-link-double-anchor.rs:5:18
+   |
+LL | /// docs [label][with#anchor#error]
+   |                  ^^^^^^^^^^^^^^^^^ contains invalid anchor
+   |
+   = note: `#[warn(broken_intra_doc_links)]` on by default
+
+warning: 1 warning emitted
+