feat: open source the JavaScript extractor code

language/javascript/extractor/.eslintignore

@@ -0,0 +1,4 @@
+node_modules/
+test/
+dist/
+out/

language/javascript/extractor/.eslintrc.js

@@ -0,0 +1,10 @@
+// eslint-disable-next-line no-undef
+module.exports = {
+  root: true,
+  parser: '@typescript-eslint/parser',
+  plugins: ['@typescript-eslint'],
+  extends: ['eslint:recommended', 'plugin:@typescript-eslint/recommended'],
+  rules: {
+    '@typescript-eslint/no-empty-interface': 'off',
+  },
+};

language/javascript/extractor/.gitignore

@@ -0,0 +1,14 @@
+.DS_Store
+
+.vscode/
+.idea/
+
+node_modules/
+coverage/
+dist/
+out/
+
+bazel-*
+
+*.db
+*.db-journal

language/javascript/extractor/.prettierignore

@@ -0,0 +1,4 @@
+node_modules/
+test/
+dist/
+out/

language/javascript/extractor/.prettierrc.json

@@ -0,0 +1,15 @@
+{
+  "arrowParens": "avoid",
+  "bracketSpacing": true,
+  "endOfLine": "lf",
+  "bracketSameLine": false,
+  "jsxSingleQuote": false,
+  "printWidth": 80,
+  "proseWrap": "preserve",
+  "quoteProps": "as-needed",
+  "semi": true,
+  "singleQuote": true,
+  "tabWidth": 2,
+  "trailingComma": "all",
+  "useTabs": false
+}

language/javascript/extractor/README.md

@@ -0,0 +1,9 @@
+# Introduction
+coref-js-src-extractor is a data generation tool that extracts JavaScript / TypeScript source code into COREF format data.
+
+# Quick Start
+1. Environment requirements Node>=18.
+2. Install yarn. `npm install --global yarn` and then install dependencies with `yarn`.
+3. Build && package. `npm run pkg`. A coref-javascript-src-extractor executable file will be generated in the current directory.
+4. Run the extractor. `./coref-javascript-src-extractor extract -s {source_root}`.
+5. For more help. `./coref-javascript-src-extractor -h` or `./coref-javascript-src-extractor extract -h`.

language/javascript/extractor/README_cn.md

@@ -0,0 +1,9 @@
+# 介绍
+coref-js-src-extractor 是一个数据生成工具，该工具将 JavaScript / TypeScript 源码抽取为COREF格式的数据。
+
+# 快速开始
+1. 环境要求 Node>=18。
+2. 安装 yarn。 `npm install --global yarn` 并安装依赖 `yarn`。
+3. 构建 && 打包。  `npm run pkg`. 当前目录下会生成 `coref-javascript-src-extractor` 可执行文件。
+4. 运行抽取器。`./coref-javascript-src-extractor extract -s {source_root}`。
+5. 查看更多帮助。`./coref-javascript-src-extractor -h` 或 `./coref-javascript-src-extractor extract -h`。

language/javascript/extractor/coref-javascript-er.puml

@@ -0,0 +1,221 @@
+@startuml
+' https://plantuml.com/ie-diagram
+
+' Location table
+entity location {
+    oid: INTEGER <<PK>>
+    file_oid: INTEGER
+    start_line_number: INTEGER
+    start_column_number: INTEGER
+    end_line_number: INTEGER
+    end_column_number: INTEGER
+    text: TEXT
+}
+
+
+' Number of lines table
+entity number_of_lines {
+    location_oid: INTEGER <<PK>>
+    lines: INTEGER
+    code_lines: INTEGER
+    comment_lines: INTEGER
+}
+
+
+entity file {
+    oid: INTEGER <<PK>>
+    name: TEXT
+    extension: TEXT
+    relative_path: TEXT
+    location_oid: INTEGER
+}
+
+entity directory {
+    oid: INTEGER <<PK>>
+    name: TEXT
+    relative_path: TEXT
+    location_oid: INTEGER
+}
+
+' parent_oid: ref directory
+' child_oid: ref directory | file
+entity directory_hierarchy {
+    parent_oid: INTEGER
+    child_oid: INTEGER <<PK>>
+}
+
+' kind:
+' enum {
+'     script = 0
+'     inline_script = 1
+'     event_handler = 2
+'     javascript_url = 3
+'     template_top_level = 4
+' }
+entity top_level {
+    oid: INTEGER <<PK>>
+    kind: INTEGER
+    location_oid: INTEGER
+}
+
+entity node {
+    oid: INTEGER <<PK>>
+    kind: INTEGER
+    parent_oid: INTEGER
+    index: INTEGER
+    location_oid: INTEGER
+}
+
+entity literal {
+    oid: INTEGER <<PK>>
+    value: TEXT
+}
+
+entity binding_element_property_name {
+    oid: INTEGER <<PK>>
+    property_name_oid: INTEGER
+}
+
+entity binding_element_name {
+    oid: INTEGER <<PK>>
+    name_oid: INTEGER
+}
+
+entity binding_element_initializer {
+    oid: INTEGER <<PK>>
+    initializer_oid: INTEGER
+}
+
+' 
+' Class declaration or expression
+'
+entity class_like_declaration {
+    oid: INTEGER <<PK>>
+    kind: INTEGER
+    name: TEXT
+}
+
+entity function_like_declaration {
+    oid: INTEGER <<PK>>
+    kind: INTEGER
+    name: TEXT
+}
+
+entity function_enclosing_node {
+    node_oid: INTEGER <<PK>>
+    function_oid: INTEGER
+}
+
+entity modifier {
+    oid: INTEGER <<PK>>
+    index: INTEGER
+}
+
+' use `symbol_` instead of `symbol` to avoid the typescript error TS2457: Type alias name cannot be 'symbol'.
+entity symbol_ {
+    oid: INTEGER <<PK>>
+    name: TEXT
+    description: TEXT
+}
+
+entity node_symbol {
+    node_oid: INTEGER <<PK>>
+    symbol_oid: INTEGER
+}
+
+entity shorthand_assignment_value_symbol {
+    node_oid: INTEGER <<PK>>
+    symbol_oid: INTEGER
+}
+
+entity call_site {
+    invoke_expression_oid: INTEGER <<PK>>
+    callee_oid: INTEGER
+}
+
+entity cfg_entry_node {
+    oid: INTEGER <<PK>>
+    ast_node_oid: INTEGER
+}
+
+entity cfg_exit_node {
+    oid: INTEGER <<PK>>
+    ast_node_oid: INTEGER
+}
+
+
+' ----
+' Type
+' ----
+
+' ' Type table
+' '
+' entity type {
+'     oid: INTEGER <<PK>>
+'     kind: INTEGER
+'     name: TEXT
+' }
+
+' ' Type hierarchy table
+' '
+' ' child_oid: ref type
+' ' parent_oid: ref type
+' ' index: the index of child type in the parent type
+' entity type_hierarchy {
+'     oid: INTEGER <<PK>>
+'     parent_oid: INTEGER
+'     child_oid: INTEGER
+'     index: INTEGER
+' }
+
+' ' Type alias table
+' '
+' ' alias_type_oid: ref type
+' ' underlying_type_oid: ref type
+' entity type_alias {
+'     alias_type_oid: INTEGER <<PK>>
+'     underlying_type_oid: INTEGER
+' }
+
+' Comment table
+entity comment {
+    oid: INTEGER <<PK>>
+    kind: INTEGER
+    location_oid: INTEGER
+}
+
+' Node comment table
+entity node_comment {
+    oid: INTEGER <<PK>>
+    node_oid: INTEGER
+    comment_oid: INTEGER
+    type: INTEGER
+}
+
+' entity token {
+'     oid: INTEGER <<PK>>
+'     kind: INTEGER
+'     location_oid: INTEGER
+' }
+
+' JS parse error table
+entity js_parse_error {
+    oid: INTEGER <<PK>>
+    message: TEXT
+    line: TEXT
+}
+
+entity metadata {
+    oid: INTEGER <<PK>>
+    version: TEXT
+    created_time: DATETIME
+}
+
+entity ignored_path {
+    oid: INTEGER <<PK>>
+    path_kind: INTEGER
+    path: TEXT
+    ignore_kind: INTEGER
+}
+
+@enduml