Fix JSON surrounding pairs (#370)

* Initial working version

* Fixed java surrounding pair

* Fix typescript

* Improve angles

* Add documentation

* Typescript template string tweaks

* Fix ci

* Remove unused import

src/languages/constants.ts

@@ -0,0 +1,15 @@
+export const supportedLanguageIds = [
+  "c",
+  "cpp",
+  "csharp",
+  "java",
+  "javascript",
+  "javascriptreact",
+  "json",
+  "jsonc",
+  "python",
+  "typescript",
+  "typescriptreact",
+] as const;
+
+export type SupportedLanguageId = typeof supportedLanguageIds[number];

src/languages/getNodeMatcher.ts

@@ -0,0 +1,99 @@
+import { SyntaxNode } from "web-tree-sitter";
+import { notSupported } from "../util/nodeMatchers";
+import { selectionWithEditorFromRange } from "../util/selectionUtils";
+import {
+  NodeMatcher,
+  NodeMatcherValue,
+  ScopeType,
+  SelectionWithEditor,
+} from "../typings/Types";
+import cpp from "./cpp";
+import csharp from "./csharp";
+import { patternMatchers as json } from "./json";
+import { patternMatchers as typescript } from "./typescript";
+import { patternMatchers as java } from "./java";
+import python from "./python";
+import { UnsupportedLanguageError } from "../errors";
+import { SupportedLanguageId } from "./constants";
+
+export function getNodeMatcher(
+  languageId: string,
+  scopeType: ScopeType,
+  includeSiblings: boolean
+): NodeMatcher {
+  const matchers = languageMatchers[languageId as SupportedLanguageId];
+
+  if (matchers == null) {
+    throw new UnsupportedLanguageError(languageId);
+  }
+
+  const matcher = matchers[scopeType];
+
+  if (matcher == null) {
+    return notSupported;
+  }
+
+  if (includeSiblings) {
+    return matcherIncludeSiblings(matcher);
+  }
+
+  return matcher;
+}
+
+const languageMatchers: Record<
+  SupportedLanguageId,
+  Record<ScopeType, NodeMatcher>
+> = {
+  c: cpp,
+  cpp: cpp,
+  csharp: csharp,
+  java,
+  javascript: typescript,
+  javascriptreact: typescript,
+  json,
+  jsonc: json,
+  python,
+  typescript,
+  typescriptreact: typescript,
+};
+
+function matcherIncludeSiblings(matcher: NodeMatcher): NodeMatcher {
+  return (
+    selection: SelectionWithEditor,
+    node: SyntaxNode
+  ): NodeMatcherValue[] | null => {
+    let matches = matcher(selection, node);
+    if (matches == null) {
+      return null;
+    }
+    matches = matches.flatMap((match) =>
+      iterateNearestIterableAncestor(
+        match.node,
+        selectionWithEditorFromRange(selection, match.selection.selection),
+        matcher
+      )
+    ) as NodeMatcherValue[];
+    if (matches.length > 0) {
+      return matches;
+    }
+    return null;
+  };
+}
+
+function iterateNearestIterableAncestor(
+  node: SyntaxNode,
+  selection: SelectionWithEditor,
+  nodeMatcher: NodeMatcher
+) {
+  let parent: SyntaxNode | null = node.parent;
+  while (parent != null) {
+    const matches = parent!.namedChildren
+      .flatMap((sibling) => nodeMatcher(selection, sibling))
+      .filter((match) => match != null) as NodeMatcherValue[];
+    if (matches.length > 0) {
+      return matches;
+    }
+    parent = parent.parent;
+  }
+  return [];
+}

src/languages/getTextFragmentExtractor.ts

@@ -0,0 +1,128 @@
+import { SyntaxNode } from "web-tree-sitter";
+import { SelectionWithEditor } from "../typings/Types";
+import { stringTextFragmentExtractor as jsonStringTextFragmentExtractor } from "./json";
+import { stringTextFragmentExtractor as javaStringTextFragmentExtractor } from "./java";
+import { stringTextFragmentExtractor as typescriptStringTextFragmentExtractor } from "./typescript";
+import { UnsupportedLanguageError } from "../errors";
+import { Range } from "vscode";
+import { SupportedLanguageId } from "./constants";
+import {
+  getNodeInternalRange,
+  getNodeRange,
+  makeRangeFromPositions,
+} from "../util/nodeSelectors";
+import { getNodeMatcher } from "./getNodeMatcher";
+import { notSupported } from "../util/nodeMatchers";
+
+export type TextFragmentExtractor = (
+  node: SyntaxNode,
+  selection: SelectionWithEditor
+) => Range | null;
+
+function constructDefaultTextFragmentExtractor(
+  languageId: SupportedLanguageId,
+  stringTextFragmentExtractor?: TextFragmentExtractor
+): TextFragmentExtractor {
+  const commentNodeMatcher = getNodeMatcher(languageId, "comment", false);
+  stringTextFragmentExtractor =
+    stringTextFragmentExtractor ??
+    constructDefaultStringTextFragmentExtractor(languageId);
+
+  return (node: SyntaxNode, selection: SelectionWithEditor) => {
+    const stringTextFragment = stringTextFragmentExtractor!(node, selection);
+
+    if (stringTextFragment != null) {
+      return stringTextFragment;
+    }
+
+    if (
+      commentNodeMatcher !== notSupported &&
+      commentNodeMatcher(selection, node) != null
+    ) {
+      return getNodeRange(node);
+    }
+
+    // Treat error nodes as raw text so that the surrounding pair matcher can
+    // still be useful when we have a bad parse tree
+    if (node.type === "ERROR") {
+      return getNodeRange(node);
+    }
+
+    return null;
+  };
+}
+
+function constructDefaultStringTextFragmentExtractor(
+  languageId: SupportedLanguageId
+): TextFragmentExtractor {
+  const stringNodeMatcher = getNodeMatcher(languageId, "string", false);
+
+  return (node: SyntaxNode, selection: SelectionWithEditor) => {
+    if (stringNodeMatcher(selection, node) != null) {
+      // Exclude starting and ending quotation marks
+      return getNodeInternalRange(node);
+    }
+
+    return null;
+  };
+}
+
+/**
+ * Returns a function which can be used to extract the range of a text fragment
+ * from within a parsed language. This function should only return a nominal
+ * range for fragments within the document that should be treated like raw text,
+ * such as comments strings or error nodes. In these cases we want our
+ * surrounding pair algorithm to fall back to a pure raw text-based approach.
+ * @param languageId The language for which to get the text fragment extractor
+ * for
+ * @returns The text fragment extractor for the given language
+ */
+export default function getTextFragmentExtractor(
+  languageId: string
+): TextFragmentExtractor {
+  const extractor = textFragmentExtractors[languageId as SupportedLanguageId];
+
+  if (extractor == null) {
+    throw new UnsupportedLanguageError(languageId);
+  }
+
+  return extractor;
+}
+
+const textFragmentExtractors: Record<
+  SupportedLanguageId,
+  TextFragmentExtractor
+> = {
+  c: constructDefaultTextFragmentExtractor("c"),
+  cpp: constructDefaultTextFragmentExtractor("cpp"),
+  csharp: constructDefaultTextFragmentExtractor("csharp"),
+  java: constructDefaultTextFragmentExtractor(
+    "java",
+    javaStringTextFragmentExtractor
+  ),
+  javascript: constructDefaultTextFragmentExtractor(
+    "javascript",
+    typescriptStringTextFragmentExtractor
+  ),
+  javascriptreact: constructDefaultTextFragmentExtractor(
+    "javascriptreact",
+    typescriptStringTextFragmentExtractor
+  ),
+  jsonc: constructDefaultTextFragmentExtractor(
+    "jsonc",
+    jsonStringTextFragmentExtractor
+  ),
+  json: constructDefaultTextFragmentExtractor(
+    "json",
+    jsonStringTextFragmentExtractor
+  ),
+  python: constructDefaultTextFragmentExtractor("python"),
+  typescript: constructDefaultTextFragmentExtractor(
+    "typescript",
+    typescriptStringTextFragmentExtractor
+  ),
+  typescriptreact: constructDefaultTextFragmentExtractor(
+    "typescriptreact",
+    typescriptStringTextFragmentExtractor
+  ),
+};

src/languages/index.ts

@@ -1,95 +1,7 @@
-import { SyntaxNode } from "web-tree-sitter";
-import { notSupported } from "../util/nodeMatchers";
-import { selectionWithEditorFromRange } from "../util/selectionUtils";
-import {
-  NodeMatcher,
-  NodeMatcherValue,
-  ScopeType,
-  SelectionWithEditor,
-} from "../typings/Types";
-import cpp from "./cpp";
-import csharp from "./csharp";
-import java from "./java";
-import json from "./json";
-import python from "./python";
-import typescript from "./typescript";
-import { UnsupportedLanguageError } from "../errors";
+import { SupportedLanguageId, supportedLanguageIds } from "./constants";
 
-const languageMatchers: Record<string, Record<ScopeType, NodeMatcher>> = {
-  c: cpp,
-  cpp: cpp,
-  csharp: csharp,
-  java,
-  javascript: typescript,
-  javascriptreact: typescript,
-  json,
-  jsonc: json,
-  python,
-  typescript,
-  typescriptreact: typescript,
-};
-
-export function getNodeMatcher(
-  languageId: string,
-  scopeType: ScopeType,
-  includeSiblings: boolean
-): NodeMatcher {
-  const matchers = languageMatchers[languageId];
-
-  if (matchers == null) {
-    throw new UnsupportedLanguageError(languageId);
-  }
-
-  const matcher = matchers[scopeType];
-
-  if (matcher == null) {
-    return notSupported;
-  }
-
-  if (includeSiblings) {
-    return matcherIncludeSiblings(matcher);
-  }
-
-  return matcher;
-}
-
-function matcherIncludeSiblings(matcher: NodeMatcher): NodeMatcher {
-  return (
-    selection: SelectionWithEditor,
-    node: SyntaxNode
-  ): NodeMatcherValue[] | null => {
-    let matches = matcher(selection, node);
-    if (matches == null) {
-      return null;
-    }
-    matches = matches.flatMap((match) =>
-      iterateNearestIterableAncestor(
-        match.node,
-        selectionWithEditorFromRange(selection, match.selection.selection),
-        matcher
-      )
-    ) as NodeMatcherValue[];
-    if (matches.length > 0) {
-      return matches;
-    }
-    return null;
-  };
-}
-
-function iterateNearestIterableAncestor(
-  node: SyntaxNode,
-  selection: SelectionWithEditor,
-  nodeMatcher: NodeMatcher
-) {
-  let parent: SyntaxNode | null = node.parent;
-  while (parent != null) {
-    const matches = parent!.namedChildren
-      .flatMap((sibling) => nodeMatcher(selection, sibling))
-      .filter((match) => match != null) as NodeMatcherValue[];
-    if (matches.length > 0) {
-      return matches;
-    }
-    parent = parent.parent;
-  }
-  return [];
+export function isLanguageSupported(
+  languageId: string
+): languageId is SupportedLanguageId {
+  return languageId in supportedLanguageIds;
 }

src/languages/java.ts

@@ -5,7 +5,13 @@ import {
   conditionMatcher,
   trailingMatcher,
 } from "../util/nodeMatchers";
-import { NodeMatcherAlternative, ScopeType } from "../typings/Types";
+import {
+  NodeMatcherAlternative,
+  ScopeType,
+  SelectionWithEditor,
+} from "../typings/Types";
+import { getNodeRange } from "../util/nodeSelectors";
+import { SyntaxNode } from "web-tree-sitter";
 
 // Generated by the following command:
 // > curl https://github.com/raw/tree-sitter/tree-sitter-java/master/src/node-types.json | jq '[.[] | select(.type == "statement" or .type == "declaration") | .subtypes[].type]'
@@ -70,4 +76,28 @@ const nodeMatchers: Partial<Record<ScopeType, NodeMatcherAlternative>> = {
   argumentOrParameter: argumentMatcher("formal_parameters", "argument_list"),
 };
 
-export default createPatternMatchers(nodeMatchers);
+export const patternMatchers = createPatternMatchers(nodeMatchers);
+
+/**
+ * Extracts string text fragments in java.
+ *
+ * This is a hack to deal with the fact that java doesn't have
+ * quotation mark tokens as children of the string. Rather than letting
+ * the parse tree handle the quotation marks in java, we instead just
+ * let the textual surround handle them by letting it see the quotation
+ * marks. In other languages we prefer to let the parser handle the
+ * quotation marks in case they are more than one character long.
+ * @param node The node which might be a string node
+ * @param selection The selection from which to expand
+ * @returns The range of the string text or null if the node is not a string
+ */
+export function stringTextFragmentExtractor(
+  node: SyntaxNode,
+  selection: SelectionWithEditor
+) {
+  if (node.type === "string_literal") {
+    return getNodeRange(node);
+  }
+
+  return null;
+}