perf(parser-adapter-yaml-1-2): use tree-sitter cursor for CST travers…

…al (#2955) Refs #691
swagger-api · Jul 15, 2023 · c46ae6a · c46ae6a
1 parent e6a3d80
commit c46ae6a
Show file tree

Hide file tree

Showing 12 changed files with 923 additions and 115 deletions.
diff --git a/packages/apidom-parser-adapter-yaml-1-2/package.json b/packages/apidom-parser-adapter-yaml-1-2/package.json
@@ -39,8 +39,8 @@
     "typescript:declaration": "tsc -p declaration.tsconfig.json && rollup -c config/rollup/types.dist.js",
     "test": "cross-env NODE_ENV=test BABEL_ENV=cjs mocha",
     "perf": "cross-env NODE_ENV=test BABEL_ENV=cjs node ./test/perf/index.cjs",
-    "perf:parse": "cross-env NODE_ENV=test BABEL_ENV=cjs node ./test/perf/parse.cjs",
     "perf:lexical-analysis": "cross-env NODE_ENV=test BABEL_ENV=cjs node ./test/perf/lexical-analysis.cjs",
+    "perf:parse-syntactic-analysis-indirect": "cross-env NODE_ENV=test BABEL_ENV=cjs node ./test/perf/parse-syntactic-analysis-indirect.cjs",
     "prepack": "copyfiles -u 3 ../../LICENSES/* LICENSES && copyfiles -u 2 ../../NOTICE .",
     "postpack": "rimraf NOTICE LICENSES"
   },

diff --git a/packages/apidom-parser-adapter-yaml-1-2/src/adapter-browser.ts b/packages/apidom-parser-adapter-yaml-1-2/src/adapter-browser.ts
@@ -1,7 +1,7 @@
 import { ParseResultElement } from '@swagger-api/apidom-core';
 
 import lexicalAnalysis from './lexical-analysis/browser';
-import syntacticAnalysis from './syntactic-analysis/index';
+import syntacticAnalysis from './syntactic-analysis/indirect/index';
 
 export { mediaTypes, namespace } from './adapter';
 export { lexicalAnalysis, syntacticAnalysis };

diff --git a/packages/apidom-parser-adapter-yaml-1-2/src/adapter-node.ts b/packages/apidom-parser-adapter-yaml-1-2/src/adapter-node.ts
@@ -1,7 +1,7 @@
 import { ParseResultElement } from '@swagger-api/apidom-core';
 
 import lexicalAnalysis from './lexical-analysis/node';
-import syntacticAnalysis from './syntactic-analysis/index';
+import syntacticAnalysis from './syntactic-analysis/indirect/index';
 
 export { mediaTypes, namespace } from './adapter';
 export { lexicalAnalysis, syntacticAnalysis };

diff --git a/packages/apidom-parser-adapter-yaml-1-2/src/syntactic-analysis/TreeCursorIterator.ts b/packages/apidom-parser-adapter-yaml-1-2/src/syntactic-analysis/TreeCursorIterator.ts
@@ -0,0 +1,121 @@
+import { TreeCursor as NodeTreeCursor } from 'tree-sitter';
+import { TreeCursor as WebTreeCursor } from 'web-tree-sitter';
+
+import TreeCursorSyntaxNode from './TreeCursorSyntaxNode';
+
+class TreeCursorIterator {
+  protected readonly cursor;
+
+  constructor(cursor: NodeTreeCursor | WebTreeCursor) {
+    this.cursor = cursor;
+  }
+
+  stream() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  yaml_directive() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  tag_directive() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  reserved_directive() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  document() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  block_node() {
+    return new TreeCursorSyntaxNode(this.cursor).setFieldName(this.cursor);
+  }
+
+  flow_node() {
+    return new TreeCursorSyntaxNode(this.cursor).setFieldName(this.cursor);
+  }
+
+  block_mapping() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  block_mapping_pair() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  flow_mapping() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  flow_pair() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  block_sequence() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  block_sequence_item() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  flow_sequence() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  plain_scalar() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  single_quote_scalar() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  double_quote_scalar() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  block_scalar() {
+    return new TreeCursorSyntaxNode(this.cursor);
+  }
+
+  ERROR() {
+    return new TreeCursorSyntaxNode(this.cursor).setHasError(this.cursor);
+  }
+
+  public *[Symbol.iterator]() {
+    let node: TreeCursorSyntaxNode;
+
+    if (this.cursor.nodeType in this) {
+      // @ts-ignore
+      node = this[this.cursor.nodeType]() as TreeCursorSyntaxNode;
+    } else {
+      node = new TreeCursorSyntaxNode(this.cursor);
+    }
+
+    if (this.cursor.gotoFirstChild()) {
+      const [firstChild] = new TreeCursorIterator(this.cursor);
+
+      node.pushChildren(firstChild);
+
+      while (this.cursor.gotoNextSibling()) {
+        const firstChildSiblings = Array.from(new TreeCursorIterator(this.cursor));
+        node.pushChildren(...firstChildSiblings);
+      }
+
+      node.children.reduce((previousNode: TreeCursorSyntaxNode | undefined, currentNode) => {
+        currentNode.setPreviousSibling(previousNode);
+        return currentNode;
+      }, undefined);
+
+      this.cursor.gotoParent();
+    }
+
+    yield node;
+  }
+}
+
+export default TreeCursorIterator;
diff --git a/packages/apidom-parser-adapter-yaml-1-2/src/syntactic-analysis/TreeCursorSyntaxNode.ts b/packages/apidom-parser-adapter-yaml-1-2/src/syntactic-analysis/TreeCursorSyntaxNode.ts
@@ -0,0 +1,105 @@
+import { TreeCursor as NodeTreeCursor, Point as NodePoint } from 'tree-sitter';
+import { TreeCursor as WebTreeCursor, Point as WebPoint } from 'web-tree-sitter';
+
+class TreeCursorSyntaxNode {
+  public readonly type: string;
+
+  public readonly startPosition: NodePoint | WebPoint;
+
+  public readonly endPosition: NodePoint | WebPoint;
+
+  public readonly startIndex: number;
+
+  public readonly endIndex: number;
+
+  public readonly text: string;
+
+  public readonly isNamed: boolean;
+
+  public readonly isMissing: boolean;
+
+  public fieldName: string | undefined;
+
+  public hasError = false;
+
+  public readonly children: TreeCursorSyntaxNode[] = [];
+
+  public previousSibling: TreeCursorSyntaxNode | undefined;
+
+  constructor(cursor: NodeTreeCursor | WebTreeCursor) {
+    this.type = cursor.nodeType;
+    this.startPosition = cursor.startPosition;
+    this.endPosition = cursor.endPosition;
+    this.startIndex = cursor.startIndex;
+    this.endIndex = cursor.endIndex;
+    this.text = cursor.nodeText;
+    this.isNamed = cursor.nodeIsNamed;
+    this.isMissing = cursor.nodeIsMissing;
+  }
+
+  get keyNode(): TreeCursorSyntaxNode | undefined {
+    if (this.type === 'flow_pair' || this.type === 'block_mapping_pair') {
+      return this.children.find((node) => node.fieldName === 'key');
+    }
+    return undefined;
+  }
+
+  get valueNode(): TreeCursorSyntaxNode | undefined {
+    if (this.type === 'flow_pair' || this.type === 'block_mapping_pair') {
+      return this.children.find((node) => node.fieldName === 'value');
+    }
+    return undefined;
+  }
+
+  get tag(): TreeCursorSyntaxNode | undefined {
+    let { previousSibling } = this;
+
+    while (typeof previousSibling !== 'undefined' && previousSibling.type !== 'tag') {
+      ({ previousSibling } = previousSibling);
+    }
+
+    return previousSibling;
+  }
+
+  get anchor(): TreeCursorSyntaxNode | undefined {
+    let { previousSibling } = this;
+
+    while (typeof previousSibling !== 'undefined' && previousSibling.type !== 'anchor') {
+      ({ previousSibling } = previousSibling);
+    }
+
+    return previousSibling;
+  }
+
+  get firstNamedChild(): TreeCursorSyntaxNode | undefined {
+    return this.children.find((node) => node.isNamed);
+  }
+
+  setFieldName(cursor: NodeTreeCursor | WebTreeCursor) {
+    if (typeof cursor.currentFieldName === 'function') {
+      this.fieldName = cursor.currentFieldName();
+    } else {
+      this.fieldName = cursor.currentFieldName;
+    }
+    return this;
+  }
+
+  setHasError(cursor: NodeTreeCursor | WebTreeCursor) {
+    if (typeof cursor.currentNode === 'function') {
+      this.hasError = cursor.currentNode().hasError();
+    } else {
+      this.hasError = cursor.currentNode.hasError();
+    }
+    return this;
+  }
+
+  setPreviousSibling(previousSibling: TreeCursorSyntaxNode | undefined) {
+    this.previousSibling = previousSibling;
+  }
+
+  pushChildren(...children: TreeCursorSyntaxNode[]) {
+    this.children.push(...children);
+  }
+}
+
+export default TreeCursorSyntaxNode;
diff --git a/...-yaml-1-2/src/syntactic-analysis/index.ts → .../src/syntactic-analysis/indirect/index.ts b/...-yaml-1-2/src/syntactic-analysis/index.ts → .../src/syntactic-analysis/indirect/index.ts
@@ -9,6 +9,8 @@ import YamlAstVisitor, {
   isNode as isAstNode,
   getNodeType as getAstNodeType,
 } from './visitors/YamlAstVisitor';
+import TreeCursorIterator from '../TreeCursorIterator';
+import TreeCursorSyntaxNode from '../TreeCursorSyntaxNode';
 
 type Tree = WebTree | NodeTree;
 
@@ -18,11 +20,14 @@ type Tree = WebTree | NodeTree;
  * Two traversals passes are needed to get from CST to ApiDOM.
  */
 const analyze = (cst: Tree, { sourceMap = false } = {}): ParseResultElement => {
+  const cursor = cst.walk();
+  const iterator = new TreeCursorIterator(cursor);
+  const rootNode = [...iterator].at(0) as TreeCursorSyntaxNode;
   const cstVisitor = CstVisitor();
   const astVisitor = YamlAstVisitor();
   const schema = JsonSchema();
 
-  const yamlAst = visit(cst.rootNode, cstVisitor, {
+  const yamlAst = visit(rootNode, cstVisitor, {
     // @ts-ignore
     keyMap: cstKeyMap,
     nodePredicate: isCstNode,
@@ -35,9 +40,7 @@ const analyze = (cst: Tree, { sourceMap = false } = {}): ParseResultElement => {
   return visit(yamlAst.rootNode, astVisitor, {
     // @ts-ignore
     keyMap: astKeyMap,
-    // @ts-ignore
     nodeTypeGetter: getAstNodeType,
-    // @ts-ignore
     nodePredicate: isAstNode,
     state: {
       sourceMap,