🐛 fix(vscode): Rust 连着写的行注释合成一块注释，和别的语言走同一套拆块

vscode/src/codeReader/languages/rust.ts

@@ -1,7 +1,8 @@
-import type { Access, ClassBlock, ClassFacts, CodeBlock, EnumBlock, FunctionBlock, ImplBlock, ImportBlock, ImportItem, LanguageAdapter, ModuleBlock, Node, ParameterFacts, TokenRole, TraitBlock, VariableBlock } from '../contract'
+import type { Access, ClassFacts, CodeBlock, EnumBlock, FunctionBlock, ImplBlock, ImportBlock, ImportItem, LanguageAdapter, Node, ParameterFacts, TokenRole, TraitBlock } from '../contract'
+import type { CommentRules } from './shared'
 import * as path from 'node:path'
 import { leadingAttributes, visibilityFacts } from '../rules/rustTree'
-import { compoundShape, extendHeader, headerSpan, leafBlock, leafShape, named } from './shared'
+import { compoundShape, extendHeader, headerSpan, leafBlock, leafShape, listBlocks, named } from './shared'
 
 const STORAGE = new Set(['pub', 'async', 'const', 'static', 'mut', 'let'])
 
@@ -10,7 +11,7 @@ export function createRustAdapter(wasmDir: string): LanguageAdapter {
     id: 'rust',
     languageIds: ['rust'],
     grammarPath: () => path.join(wasmDir, 'tree-sitter-rust.wasm'),
-    toBlocks: root => topBlocks(named(root)),
+    toBlocks: root => topBlocks(named(root), -1),
     tokenRole: rustTokenRole,
   }
 }
@@ -23,37 +24,28 @@ function unwrapExtern(item: Node): Node[] {
   return body ? named(body).flatMap(unwrapExtern) : []
 }
 
-function topBlocks(items: Node[]): CodeBlock[] {
-  const list = items.flatMap(unwrapExtern)
-  const blocks: CodeBlock[] = []
-  let index = 0
-  while (index < list.length) {
-    const item = list[index]
-    if (isDoc(item)) {
-      const docs = [item]
-      let next = index + 1
-      while (next < list.length && isDoc(list[next]) && list[next].startPosition.row <= docs[docs.length - 1].endPosition.row + 1)
-        docs.push(list[next++])
-      while (next < list.length && list[next].type === 'attribute_item')
-        next++
-      const target = list[next]
-      if (target && target.type !== 'line_comment' && target.type !== 'block_comment' && target.type !== 'empty_statement') {
-        blocks.push(withPreamble(itemBlock(target), docs))
-        index = next + 1
-        continue
-      }
-    }
-    if (item.type === 'attribute_item' || item.type === 'empty_statement') {
-      index++
-      continue
-    }
-    if (item.type === 'line_comment' || item.type === 'block_comment')
-      blocks.push(leafBlock('comment', item))
-    else
-      blocks.push(withPreamble(itemBlock(item), []))
-    index++
-  }
-  return blocks
+/**
+ * `//` 普通注释、讲所在模块的 `//!`、讲下一项的 `///` 各是一种写法，只和同一种连成一块。
+ * `///` 和 `/**` 开头的块注释是外部文档注释，讲紧跟着的那一项（中间可以隔着 `#[…]`）。
+ */
+const COMMENTS: CommentRules = {
+  lineStyle: comment => comment.type === 'line_comment' ? `//${docMarker(comment)}` : undefined,
+  isDoc: comment => named(comment).some(child => child.type === 'outer_doc_comment_marker'),
+  documents: new Set(['function', 'class', 'trait', 'impl', 'module', 'enum', 'const', 'variable']),
+}
+
+/** 文档注释记号：外部的 `/`（块注释是 `*`）、内部的 `!`；普通注释没有。 */
+function docMarker(comment: Node): string {
+  return named(comment).find(child => child.type === 'outer_doc_comment_marker' || child.type === 'inner_doc_comment_marker')?.text ?? ''
+}
+
+/** 一层的各项（文件顶层、模块身体、impl 身体）→ 块；afterRow 同 listBlocks。`#[…]` 和多出来的空 `;` 不出块。 */
+function topBlocks(items: Node[], afterRow: number): CodeBlock[] {
+  return listBlocks(items.flatMap(unwrapExtern), afterRow, (item) => {
+    if (item.type === 'attribute_item' || item.type === 'empty_statement')
+      return []
+    return [withAttributes(itemBlock(item))]
+  }, COMMENTS)
 }
 
 function itemBlock(item: Node): CodeBlock {
@@ -90,7 +82,7 @@ function moduleBlock(node: Node): CodeBlock {
   const { access, attributes } = visibilityFacts(node)
   return {
     kind: 'module',
-    ...compoundShape(node, body.startIndex, named(body), items => topBlocks(items)),
+    ...compoundShape(node, body.startIndex, named(body), topBlocks),
     ...(attributes && { attributes }),
     facts: classFacts(node, access),
   }
@@ -112,25 +104,10 @@ function valueBlock(node: Node, kind: 'const' | 'variable'): CodeBlock {
   }
 }
 
-/** 文档注释和写在前面的属性算进头部。文档注释的第一句是说明。 */
-function withPreamble(block: CodeBlock, docs: Node[]): CodeBlock {
-  const [doc] = docs
-  const attribute = leadingAttributes(block.node)[0]
-  let next = block
-  if (doc && isDocumentedKind(next))
-    next = { ...extendHeader(next, doc), facts: { ...next.facts, doc } } as typeof next
-  const from = doc ?? attribute
-  if (from && next.header.startIndex > from.startIndex)
-    next = extendHeader(next, from)
-  return next
-}
-
-function isDocumentedKind(block: CodeBlock): block is FunctionBlock | ClassBlock | TraitBlock | ImplBlock | ModuleBlock | EnumBlock | VariableBlock {
-  return block.kind === 'function' || block.kind === 'class' || block.kind === 'trait' || block.kind === 'impl' || block.kind === 'module' || block.kind === 'enum' || block.kind === 'const' || block.kind === 'variable'
-}
-
-function isDoc(node: Node): boolean {
-  return (node.type === 'line_comment' || node.type === 'block_comment') && named(node).some(child => child.type === 'outer_doc_comment_marker')
+/** 写在前面的 `#[…]` 算进头部。 */
+function withAttributes(block: CodeBlock): CodeBlock {
+  const [attribute] = leadingAttributes(block.node)
+  return attribute && block.header.startIndex > attribute.startIndex ? extendHeader(block, attribute) : block
 }
 
 function useBlock(node: Node): ImportBlock {
@@ -224,10 +201,9 @@ function traitBlock(node: Node): TraitBlock {
 function implBlock(node: Node): ImplBlock {
   const { access, attributes } = visibilityFacts(node)
   const body = node.childForFieldName('body')
-  const children = body ? topBlocks(named(body)) : []
   return {
     kind: 'impl',
-    ...compoundShape(node, body?.startIndex ?? node.endIndex, [], () => children),
+    ...compoundShape(node, body?.startIndex ?? node.endIndex, named(body), topBlocks),
     ...(attributes && { attributes }),
     facts: classFacts(node, access, implType(node)),
   }