Migrating markdown scopes (#2489)

Fixes #2378

## Checklist

- [x] I have added
[tests](https://www.cursorless.org/docs/contributing/test-case-recorder/)
- [/] I have updated the
[docs](https://github.com/cursorless-dev/cursorless/tree/main/docs) and
[cheatsheet](https://github.com/cursorless-dev/cursorless/tree/main/cursorless-talon/src/cheatsheet)
- [/] I have not broken the cheatsheet

data/fixtures/scopes/markdown/section.scope

@@ -0,0 +1,66 @@
+# H1
+
+Sub 1
+
+## H2
+
+Sub 2
+
+# H1.2
+---
+
+[#1 Content] =
+[#1 Domain] = 0:0-6:5
+  >----
+0| # H1
+1|
+2| Sub 1
+3|
+4| ## H2
+5|
+6| Sub 2
+   -----<
+
+[#1 Removal] = 0:0-8:0
+  >----
+0| # H1
+1|
+2| Sub 1
+3|
+4| ## H2
+5|
+6| Sub 2
+7|
+8| # H1.2
+   <
+
+[#1 Insertion delimiter] = "\n\n"
+
+
+[#2 Content] =
+[#2 Domain] = 4:0-6:5
+  >-----
+4| ## H2
+5|
+6| Sub 2
+   -----<
+
+[#2 Removal] = 4:0-8:0
+  >-----
+4| ## H2
+5|
+6| Sub 2
+7|
+8| # H1.2
+   <
+
+[#2 Insertion delimiter] = "\n\n"
+
+
+[#3 Content] =
+[#3 Removal] =
+[#3 Domain] = 8:0-8:6
+  >------<
+8| # H1.2
+
+[#3 Insertion delimiter] = "\n\n"

packages/common/src/scopeSupportFacets/markdown.ts

@@ -9,5 +9,6 @@ const { supported, unsupported, notApplicable } = ScopeSupportFacetLevel;
 export const markdownScopeSupport: LanguageScopeSupportFacetMap = {
   "comment.line": supported,
   "comment.block": supported,
+  section: supported,
   notebookCell: supported,
 };

packages/common/src/scopeSupportFacets/scopeSupportFacetInfos.ts

@@ -38,6 +38,11 @@ export const scopeSupportFacetInfos: Record<
     scopeType: "environment",
   },
 
+  section: {
+    description: "A document section",
+    scopeType: "section",
+  },
+
   list: {
     description: "A list/array",
     scopeType: "list",

packages/common/src/scopeSupportFacets/scopeSupportFacets.types.ts

@@ -14,6 +14,8 @@ const scopeSupportFacets = [
 
   "environment",
 
+  "section",
+
   "list",
   "map",
   "ifStatement",
@@ -168,7 +170,6 @@ const scopeSupportFacets = [
   "notebookCell",
 
   // FIXME: Still in legacy
-  // section
   // selector
   // unit
   // collectionItem

packages/cursorless-engine/src/languages/LegacyLanguageId.ts

@@ -7,7 +7,6 @@ export const legacyLanguageIds = [
   "csharp",
   "css",
   "latex",
-  "markdown",
   "php",
   "ruby",
   "rust",

packages/cursorless-engine/src/languages/TreeSitterQuery/queryPredicateOperators.ts

@@ -168,10 +168,12 @@ class TrimEnd extends QueryPredicateOperator<TrimEnd> {
     const { document, range } = nodeInfo;
     const text = document.getText(range);
     const whitespaceLength = text.length - text.trimEnd().length;
-    nodeInfo.range = new Range(
-      range.start,
-      adjustPosition(document, range.end, -whitespaceLength),
-    );
+    if (whitespaceLength > 0) {
+      nodeInfo.range = new Range(
+        range.start,
+        adjustPosition(document, range.end, -whitespaceLength),
+      );
+    }
     return true;
   }
 }

packages/cursorless-engine/src/languages/branchMatcher.ts

@@ -1,33 +0,0 @@
-import { NodeMatcher } from "../typings/Types";
-import { patternFinder } from "../util/nodeFinders";
-import {
-  cascadingMatcher,
-  matcher,
-  patternMatcher,
-} from "../util/nodeMatchers";
-import { childRangeSelector } from "../util/nodeSelectors";
-
-/**
- * Constructs a branch matcher for constructs that have a primary branch
- * followed by zero or more optional branches, such as `if` statements or `try`
- * statements
- * @param statementType The top-level statement type for this construct, eg
- * "if_statement" or "try_statement"
- * @param optionalBranchTypes The optional branch type names that can be
- * children of the top-level statement, eg "else_clause" or "except_clause"
- * @returns A node matcher capable of matching this type of branch
- */
-export function branchMatcher(
-  statementType: string,
-  optionalBranchTypes: string[],
-): NodeMatcher {
-  return cascadingMatcher(
-    patternMatcher(...optionalBranchTypes),
-    matcher(
-      patternFinder(statementType),
-      childRangeSelector(optionalBranchTypes, [], {
-        includeUnnamedChildren: true,
-      }),
-    ),
-  );
-}

packages/cursorless-engine/src/languages/getNodeMatcher.ts

@@ -12,7 +12,6 @@ import clojure from "./clojure";
 import { LegacyLanguageId } from "./LegacyLanguageId";
 import csharp from "./csharp";
 import latex from "./latex";
-import markdown from "./markdown";
 import php from "./php";
 import { patternMatchers as ruby } from "./ruby";
 import rust from "./rust";
@@ -51,7 +50,6 @@ export const languageMatchers: Record<
   csharp,
   css: scss,
   latex,
-  markdown,
   php,
   ruby,
   rust,

packages/cursorless-engine/src/languages/markdown.ts

@@ -1,76 +0,0 @@
-import { SimpleScopeTypeType, TextEditor } from "@cursorless/common";
-import type { SyntaxNode } from "web-tree-sitter";
-import { NodeFinder, NodeMatcherAlternative } from "../typings/Types";
-import { leadingSiblingNodeFinder, patternFinder } from "../util/nodeFinders";
-import { createPatternMatchers, matcher } from "../util/nodeMatchers";
-import { extendUntilNextMatchingSiblingOrLast } from "../util/nodeSelectors";
-import { shrinkRangeToFitContent } from "../util/selectionUtils";
-
-const HEADING_MARKER_TYPES = [
-  "atx_h1_marker",
-  "atx_h2_marker",
-  "atx_h3_marker",
-  "atx_h4_marker",
-  "atx_h5_marker",
-  "atx_h6_marker",
-] as const;
-type HeadingMarkerType = (typeof HEADING_MARKER_TYPES)[number];
-
-/**
- * Returns a node finder that will only accept nodes of heading level at least
- * as high as the given heading level type
- * @param headingType The heading type of the node we'll be starting at
- * @returns A node finder that will return the next node that is of the same
- * marker level or higher than the original type
- */
-function makeMinimumHeadingLevelFinder(
-  headingType: HeadingMarkerType,
-): NodeFinder {
-  const markerIndex = HEADING_MARKER_TYPES.indexOf(headingType);
-  return (node: SyntaxNode) => {
-    return node.type === "atx_heading" &&
-      HEADING_MARKER_TYPES.indexOf(
-        node.firstNamedChild?.type as HeadingMarkerType,
-      ) <= markerIndex
-      ? node
-      : null;
-  };
-}
-
-function sectionExtractor(editor: TextEditor, node: SyntaxNode) {
-  const finder = makeMinimumHeadingLevelFinder(
-    node.firstNamedChild?.type as HeadingMarkerType,
-  );
-
-  const { context, selection } = extendUntilNextMatchingSiblingOrLast(
-    editor,
-    node,
-    finder,
-  );
-  return {
-    context,
-    selection: shrinkRangeToFitContent(editor, selection).toSelection(
-      selection.isReversed,
-    ),
-  };
-}
-
-function sectionMatcher(...patterns: string[]) {
-  const finder = patternFinder(...patterns);
-
-  return matcher(leadingSiblingNodeFinder(finder), sectionExtractor);
-}
-
-const nodeMatchers: Partial<
-  Record<SimpleScopeTypeType, NodeMatcherAlternative>
-> = {
-  section: sectionMatcher("atx_heading"),
-  sectionLevelOne: sectionMatcher("atx_heading.atx_h1_marker"),
-  sectionLevelTwo: sectionMatcher("atx_heading.atx_h2_marker"),
-  sectionLevelThree: sectionMatcher("atx_heading.atx_h3_marker"),
-  sectionLevelFour: sectionMatcher("atx_heading.atx_h4_marker"),
-  sectionLevelFive: sectionMatcher("atx_heading.atx_h5_marker"),
-  sectionLevelSix: sectionMatcher("atx_heading.atx_h6_marker"),
-};
-
-export default createPatternMatchers(nodeMatchers);

packages/cursorless-engine/src/languages/ternaryBranchMatcher.ts

@@ -1,40 +0,0 @@
-import type { SyntaxNode } from "web-tree-sitter";
-import { NodeMatcher } from "../typings/Types";
-import { matcher } from "../util/nodeMatchers";
-
-/**
- * Constructs a matcher for matching ternary branches.  Branches are expected to
- * be named children at particular indices of a ternary node.
- *
- * NB: We can't just use the "foo[0]" syntax of our pattern language because
- * that just uses `foo` for the finder; the `[0]` is just part of the extractor,
- * so if we had `foo[0]` and `foo[1]`, they would both match for either branch.
- * @param ternaryTypename The type name for ternary expressions
- * @param acceptableNamedChildIndices Which named children, by index, of the
- * ternary node correspond to branches
- * @returns A matcher that can match ternary branches
- */
-export function ternaryBranchMatcher(
-  ternaryTypename: string,
-  acceptableNamedChildIndices: number[],
-): NodeMatcher {
-  function finder(node: SyntaxNode) {
-    const parent = node.parent;
-    if (parent == null) {
-      return null;
-    }
-
-    if (
-      parent.type === ternaryTypename &&
-      acceptableNamedChildIndices.some((index) =>
-        parent.namedChild(index)!.equals(node),
-      )
-    ) {
-      return node;
-    }
-
-    return null;
-  }
-
-  return matcher(finder);
-}

queries/markdown.scm

@@ -77,3 +77,61 @@
   (#trim-end! @notebookCell)
   (#insertion-delimiter! @notebookCell "\n\n")
 )
+
+;;!! # H1
+;;!! ## H2
+(
+  (section) @section @_.removal
+  (#trim-end! @section)
+)
+
+;;!! # H1
+(
+  (section
+    (atx_heading
+      (atx_h1_marker)
+    )
+  ) @sectionLevelOne @_.removal
+  (#trim-end! @sectionLevelOne)
+)
+;;!! ## H2
+(
+  (section
+    (atx_heading
+      (atx_h2_marker)
+    )
+  ) @sectionLevelTwo @_.removal
+  (#trim-end! @sectionLevelTwo)
+)
+(
+  (section
+    (atx_heading
+      (atx_h3_marker)
+    )
+  ) @sectionLevelThree @_.removal
+  (#trim-end! @sectionLevelThree)
+)
+(
+  (section
+    (atx_heading
+      (atx_h4_marker)
+    )
+  ) @sectionLevelFour @_.removal
+  (#trim-end! @sectionLevelFour)
+)
+(
+  (section
+    (atx_heading
+      (atx_h5_marker)
+    )
+  ) @sectionLevelFive @_.removal
+  (#trim-end! @sectionLevelFive)
+)
+(
+  (section
+    (atx_heading
+      (atx_h6_marker)
+    )
+  ) @sectionLevelSix @_.removal
+  (#trim-end! @sectionLevelSix)
+)