Rollup merge of #93038 - GuillaumeGomez:block-doc-comments, r=notriddle

Fix star handling in block doc comments

Fixes #92872.

Some extra explanation about this PR and why #92357 created this regression: when we merge doc comment kinds for example in:

```rust
/// he
/**
* hello
*/
#[doc = "boom"]
```

We don't want to remove the empty lines between them. However, to correctly compute the "horizontal trim", we still need it, so instead, I put back a part of the "vertical trim" directly in the "horizontal trim" computation so it doesn't impact the output buffer but allows us to correctly handle the stars.

r? `@camelid`

Author: matthiaskrgr

compiler/rustc_ast/src/attr/mod.rs

@@ -242,6 +242,17 @@ impl Attribute {
         }
     }
 
+    pub fn doc_str_and_comment_kind(&self) -> Option<(Symbol, CommentKind)> {
+        match self.kind {
+            AttrKind::DocComment(kind, data) => Some((data, kind)),
+            AttrKind::Normal(ref item, _) if item.path == sym::doc => item
+                .meta_kind()
+                .and_then(|kind| kind.value_str())
+                .map(|data| (data, CommentKind::Line)),
+            _ => None,
+        }
+    }
+
     pub fn doc_str(&self) -> Option<Symbol> {
         match self.kind {
             AttrKind::DocComment(.., data) => Some(data),

compiler/rustc_ast/src/util/comments.rs

@@ -1,3 +1,4 @@
+use crate::token::CommentKind;
 use rustc_span::source_map::SourceMap;
 use rustc_span::{BytePos, CharPos, FileName, Pos, Symbol};
 
@@ -25,7 +26,7 @@ pub struct Comment {
 
 /// Makes a doc string more presentable to users.
 /// Used by rustdoc and perhaps other tools, but not by rustc.
-pub fn beautify_doc_string(data: Symbol) -> Symbol {
+pub fn beautify_doc_string(data: Symbol, kind: CommentKind) -> Symbol {
     fn get_vertical_trim(lines: &[&str]) -> Option<(usize, usize)> {
         let mut i = 0;
         let mut j = lines.len();
@@ -42,10 +43,28 @@ pub fn beautify_doc_string(data: Symbol) -> Symbol {
         if i != 0 || j != lines.len() { Some((i, j)) } else { None }
     }
 
-    fn get_horizontal_trim(lines: &[&str]) -> Option<usize> {
+    fn get_horizontal_trim(lines: &[&str], kind: CommentKind) -> Option<usize> {
         let mut i = usize::MAX;
         let mut first = true;
 
+        // In case we have doc comments like `/**` or `/*!`, we want to remove stars if they are
+        // present. However, we first need to strip the empty lines so they don't get in the middle
+        // when we try to compute the "horizontal trim".
+        let lines = if kind == CommentKind::Block {
+            let mut i = 0;
+            let mut j = lines.len();
+
+            while i < j && lines[i].trim().is_empty() {
+                i += 1;
+            }
+            while j > i && lines[j - 1].trim().is_empty() {
+                j -= 1;
+            }
+            &lines[i..j]
+        } else {
+            lines
+        };
+
         for line in lines {
             for (j, c) in line.chars().enumerate() {
                 if j > i || !"* \t".contains(c) {
@@ -79,11 +98,13 @@ pub fn beautify_doc_string(data: Symbol) -> Symbol {
         } else {
             &mut lines
         };
-        if let Some(horizontal) = get_horizontal_trim(&lines) {
+        if let Some(horizontal) = get_horizontal_trim(&lines, kind) {
             changes = true;
             // remove a "[ \t]*\*" block from each line, if possible
             for line in lines.iter_mut() {
-                *line = &line[horizontal + 1..];
+                if horizontal + 1 < line.len() {
+                    *line = &line[horizontal + 1..];
+                }
             }
         }
         if changes {

compiler/rustc_ast/src/util/comments/tests.rs

@@ -5,7 +5,7 @@ use rustc_span::create_default_session_globals_then;
 fn test_block_doc_comment_1() {
     create_default_session_globals_then(|| {
         let comment = "\n * Test \n **  Test\n *   Test\n";
-        let stripped = beautify_doc_string(Symbol::intern(comment));
+        let stripped = beautify_doc_string(Symbol::intern(comment), CommentKind::Block);
         assert_eq!(stripped.as_str(), " Test \n*  Test\n   Test");
     })
 }
@@ -14,7 +14,7 @@ fn test_block_doc_comment_1() {
 fn test_block_doc_comment_2() {
     create_default_session_globals_then(|| {
         let comment = "\n * Test\n *  Test\n";
-        let stripped = beautify_doc_string(Symbol::intern(comment));
+        let stripped = beautify_doc_string(Symbol::intern(comment), CommentKind::Block);
         assert_eq!(stripped.as_str(), " Test\n  Test");
     })
 }
@@ -23,21 +23,21 @@ fn test_block_doc_comment_2() {
 fn test_block_doc_comment_3() {
     create_default_session_globals_then(|| {
         let comment = "\n let a: *i32;\n *a = 5;\n";
-        let stripped = beautify_doc_string(Symbol::intern(comment));
+        let stripped = beautify_doc_string(Symbol::intern(comment), CommentKind::Block);
         assert_eq!(stripped.as_str(), " let a: *i32;\n *a = 5;");
     })
 }
 
 #[test]
 fn test_line_doc_comment() {
     create_default_session_globals_then(|| {
-        let stripped = beautify_doc_string(Symbol::intern(" test"));
+        let stripped = beautify_doc_string(Symbol::intern(" test"), CommentKind::Line);
         assert_eq!(stripped.as_str(), " test");
-        let stripped = beautify_doc_string(Symbol::intern("! test"));
+        let stripped = beautify_doc_string(Symbol::intern("! test"), CommentKind::Line);
         assert_eq!(stripped.as_str(), "! test");
-        let stripped = beautify_doc_string(Symbol::intern("test"));
+        let stripped = beautify_doc_string(Symbol::intern("test"), CommentKind::Line);
         assert_eq!(stripped.as_str(), "test");
-        let stripped = beautify_doc_string(Symbol::intern("!test"));
+        let stripped = beautify_doc_string(Symbol::intern("!test"), CommentKind::Line);
         assert_eq!(stripped.as_str(), "!test");
     })
 }

compiler/rustc_save_analysis/src/lib.rs

@@ -821,9 +821,9 @@ impl<'tcx> SaveContext<'tcx> {
         let mut result = String::new();
 
         for attr in attrs {
-            if let Some(val) = attr.doc_str() {
+            if let Some((val, kind)) = attr.doc_str_and_comment_kind() {
                 // FIXME: Should save-analysis beautify doc strings itself or leave it to users?
-                result.push_str(beautify_doc_string(val).as_str());
+                result.push_str(beautify_doc_string(val, kind).as_str());
                 result.push('\n');
             }
         }

src/librustdoc/clean/types.rs

@@ -1040,9 +1040,9 @@ impl Attributes {
     ) -> Attributes {
         let mut doc_strings: Vec<DocFragment> = vec![];
         let clean_attr = |(attr, parent_module): (&ast::Attribute, Option<DefId>)| {
-            if let Some(value) = attr.doc_str() {
+            if let Some((value, kind)) = attr.doc_str_and_comment_kind() {
                 trace!("got doc_str={:?}", value);
-                let value = beautify_doc_string(value);
+                let value = beautify_doc_string(value, kind);
                 let kind = if attr.is_doc_comment() {
                     DocFragmentKind::SugaredDoc
                 } else {

src/test/rustdoc/strip-block-doc-comments-stars.docblock.html

@@ -0,0 +1,2 @@
+<div class="docblock"><p>a</p>
+</div>

src/test/rustdoc/strip-block-doc-comments-stars.rs

@@ -0,0 +1,11 @@
+#![crate_name = "foo"]
+
+// The goal of this test is to answer that it won't be generated as a list because
+// block doc comments can have their lines starting with a star.
+
+// @has foo/fn.foo.html
+// @snapshot docblock - '//*[@class="rustdoc-toggle top-doc"]//*[@class="docblock"]'
+/**
+ *     a
+ */
+pub fn foo() {}