Skip to content

Latest commit

 

History

History
369 lines (320 loc) · 8.56 KB

README.md

File metadata and controls

369 lines (320 loc) · 8.56 KB
Raw

使用 AST 解析 React TypeScript Component 接口定义

背景

团队使用 TypeScript 进行 React 组件开发。开发组件的同时，需要为组件撰写文档（使用 Markdown 编写文档）。文档中需要对组件的 props 定义进行说明。

Antd 组件的 API 说明

在开发组件的时候，是编写组件 props 的接口定义。这时候就希望能够偷懒了，直接抽取组件源代码中的接口定义和注释来生成这部分文档。

分析

基于上面背景进行分析，可以发现核心需求是抽取组件接口定义和注释形成特定格式的文档。

import React from "react";

export interface IProps {
  // 文字
  text: string;
  // 点击事件
  onClick: () => void;
}

const Button = ({ text }) => {
  return <button>{text}</button>;
};

export default Button;

其实就是抽取 IProps 定义然后转换成其他格式文档。

思路

在人人都说 AST 的今天，我们肯定是要使用 AST 来完成。

  1. 将源代码转换成 AST
  2. 遍历 AST 抽取 interface 定义
  3. 继续遍历 interface 抽取各个字段定义 definition
code -> ast -> interface -> definitions

源代码 -> AST

站在巨人的肩膀上，使用 babel 解析源代码即可。

const parser = require("@babel/parser");

function transformCode2Ast(code) {
  return parser.parse(code, {
    sourceType: "module",
    plugins: [
      "jsx",
      "typescript",
      "asyncGenerators",
      "bigInt",
      "classProperties",
      "classPrivateProperties",
      "classPrivateMethods",
      ["decorators", { decoratorsBeforeExport: false }],
      "doExpressions",
      "dynamicImport",
      "exportDefaultFrom",
      "exportNamespaceFrom",
      "functionBind",
      "functionSent",
      "importMeta",
      "logicalAssignment",
      "nullishCoalescingOperator",
      "numericSeparator",
      "objectRestSpread",
      "optionalCatchBinding",
      "optionalChaining",
      ["pipelineOperator", { proposal: "minimal" }],
      "throwExpressions",
      "topLevelAwait",
      "estree",
    ],
  });
}

转换上面的 Button 组件，AST 大概长这样：

Node {
  type: 'File',
  start: 0,
  end: 207,
  loc:
   SourceLocation {
     start: Position { line: 1, column: 0 },
     end: Position { line: 16, column: 0 } },
  errors: [],
  program:
   Node {
     type: 'Program',
     start: 0,
     end: 207,
     loc: SourceLocation { start: [Position], end: [Position] },
     sourceType: 'module',
     interpreter: null,
     body: [ [Node], [Node], [Node], [Node] ] },
  comments:
   [ { type: 'CommentLine',
       value: ' 文字',
       start: 57,
       end: 62,
       loc: [SourceLocation] },
     { type: 'CommentLine',
       value: ' 点击事件',
       start: 81,
       end: 88,
       loc: [SourceLocation] } ] }

遍历 AST

同样站在巨人的肩膀上，使用 ast-types 对 ast 进行遍历。

const { visit } = require("ast-types");

function findInterface(ast) {
  let ret = Object.create(null);
  let currentInterface = null;

  visit(ast, {
    visitTSInterfaceDeclaration(nodePath) {
      currentInterface = nodePath.value.id.name;
      this.traverse(nodePath);
    },
    visitTSPropertySignature(nodePath) {
      ret[currentInterface] = ret[currentInterface] || [];
      ret[currentInterface].push(nodePath.value);
      return false;
    },
  });
  return ret;
}

对 AST 遍历抽取 Interface 后的结果大概长这样。

Node {
  type: 'TSPropertySignature',
  start: 65,
  end: 78,
  loc: [SourceLocation],
  key: [Node],
  computed: false,
  typeAnnotation: [Node],
  leadingComments: [Array],
  trailingComments: [Array] },
Node {
  type: 'TSPropertySignature',
  start: 91,
  end: 111,
  loc: [SourceLocation],
  key: [Node],
  computed: false,
  typeAnnotation: [Node],
  leadingComments: [Array] } ] }

会发现各个定义在 typeAnnotation 中，这时候对它进行解析即可。

解析 TypeAnnotation

通过分析 typeAnnotation 很容易写出。

const get = require("lodash/get");

function parseTSTypeReference(typeName) {
  const type = get(typeName, "type");
  switch (type) {
    case "TSQualifiedName":
      return `${get(typeName, "left.name")}.${get(typeName, "right.name")}`;
    default:
      return `Unknown ReferenceType`;
  }
}

function parseTSFunctionType(parameters, typeAnnotation) {
  const parseTSFunctionParameters = (parameters) => {
    if (!parameters || !parameters.length) {
      return `()`;
    }
    let args = parameters.map((parameter) => {
      return `${get(parameter, "name")}: ${parseTypeAnnotation(
        get(parameter, "typeAnnotation.typeAnnotation")
      )}`;
    });
    return "( " + args.join(", ") + ")";
  };
  const parseTSFunctionReturn = (typeAnnotation) => {
    const type = get(typeAnnotation, "type");
    switch (type) {
      case "TSVoidKeyword":
        return "void";
      case "TSTypeReference":
        return parseTSTypeReference(get(typeAnnotation, "typeName"));
      default:
        return `Unknown FunctionType`;
    }
  };
  return `${parseTSFunctionParameters(parameters)} => ${parseTSFunctionReturn(
    typeAnnotation
  )}`;
}

function parseTSTypeLiteral(members) {
  const ret = parseInterfaceDefinitions(members);
  let args = ret.map((t) => `${t.name}: ${t.type}`);
  return "{ " + args.join(", ") + " }";
}

function parseTypeAnnotation(typeAnnotation) {
  const type = get(typeAnnotation, "type");
  switch (type) {
    case "TSNumberKeyword":
    case "TSStringKeyword":
    case "TSBoleanKeyword":
    case "TSNullKeyword":
    case "TSUndefinedKeyword":
    case "TSSymbolKeyword":
    case "TSAnyKeyword":
      return type.match(/TS(\w+)Keyword/)[1].toLowerCase();
    case "TSUnionType":
      return get(typeAnnotation, "types", [])
        .map((type) => get(type, "literal.value"))
        .join(" | ");
    case "TSFunctionType":
      return parseTSFunctionType(
        get(typeAnnotation, "parameters"),
        get(typeAnnotation, "typeAnnotation.typeAnnotation")
      );
    case "TSTypeReference":
      return parseTSTypeReference(get(typeAnnotation, "typeName"));
    case "TSTypeLiteral":
      return parseTSTypeLiteral(get(typeAnnotation, "members"));
    default:
      return "UnKnowType";
  }
}

function parseInterfaceDefinitions(nodePaths) {
  const parseInterfaceDefinitionsNode = (nodePath) => {
    const name = get(nodePath, "key.name");
    const comments = get(nodePath, "leadingComments.0.value", "")
      .trim()
      .split(/[\r\n]/)
      .map((str) => str.trim().replace(/^\*/g, "").trim())
      .filter(Boolean);
    const typeAnnotation = get(nodePath, "typeAnnotation.typeAnnotation");
    const type = parseTypeAnnotation(typeAnnotation);
    return { name, type, comments };
  };
  return nodePaths.map(parseInterfaceDefinitionsNode);
}

至此，可以得到 Button 的接口定义。

[
  [
    {
      name: "text",
      type: "string",
      comments: ["文字"],
    },
    {
      name: "onClick",
      type: "() => void",
      comments: ["点击事件"],
    },
  ],
];

接下来只要将解析后的结果转成想要的格式即可。

测试

function parseTypeScriptComponentInterface(code) {
  let ast = transformCode2Ast(code);
  let interfaces = findInterfaces(ast);
  let definitions = Object.keys(interfaces).reduce((a, c) => {
    a[c] = a[c] || [];
    a[c].push(parseInterfaceDefinitions(interfaces[c]));
    return a;
  }, Object.create(null));
  return definitions;
}

const code = `
import React from 'react';

export interface IProps {
  /**
   * button 
   * 显示文字
   */
  text: string;
  // 点击事件
  onClick: () => void;
  // 属性 3
  props3: (arg: any) => void;
  // 属性 4
  props4: (arg: { name: string, age: number }) => React.Node
}

const Button = ({ text }) => {
  return <button>{text}</button>;
};

export default Button;
`;

let ret = parseTypeScriptComponentInterface(code);

输出：

{
  "IProps": [
    [
      {
        "name": "text",
        "type": "string",
        "comments": [
          "button",
          "显示文字"
        ]
      },
      {
        "name": "onClick",
        "type": "() => void",
        "comments": [
          "点击事件"
        ]
      },
      {
        "name": "props3",
        "type": "( arg: any) => void",
        "comments": [
          "属性 3"
        ]
      },
      {
        "name": "props4",
        "type": "( arg: { name: string, age: number }) => React.Node",
        "comments": [
          "属性 4"
        ]
      }
    ]
  ]
}