POM Generation Pipeline (Java, JavaScript, TypeScript)

Code generation is model-driven and language-configurable. AUTOPOM generates output from PageModel artifacts, not directly from crawl events.

Pipeline

1. Build PageModel JSON from extraction stage.
2. Verify/heal selectors and update confidence.
3. Resolve target language via pom_language.
4. Render language-specific classes:
   • BasePage.<ext>
   • <PageName>.<ext>
5. Persist to output/<language>.

Supported output languages

Language	CLI value	Extension	Output folder
Java	java	.java	output/java
JavaScript	javascript	.js	output/javascript
TypeScript	typescript	.ts	output/typescript

Locator Storage Strategy

You can control how selectors are stored in the generated code using the locator_storage parameter:

• inline: Selectors are embedded directly in the class fields or methods. This is standard for most Playwright implementations.
• external: Selectors are stored in a sidecar metadata file, keeping the Page Object class focused purely on interaction logic.

Selection strategy

• Set --pom-language in CLI or pom_language in CrawlConfig.
• Keep language-specific style differences inside the generator layer.
• Reuse the same semantic schema and action translation logic for all languages.

Clean code rules

• Encapsulated locator fields in generated page classes.
• Public intent-level methods (login, searchProduct, submitOrder).
• Descriptive naming and minimal UI implementation leakage.
• Base page abstraction shared by all generated page classes.

Example generated actions

public LoginPage login(String username, String password) {
    usernameInput.fill(username);
    passwordInput.fill(password);
    signInButton.click();
    return this;
}

async login(username: string, password: string): Promise<LoginPage> {
  await this.usernameInput.fill(username)
  await this.passwordInput.fill(password)
  await this.signInButton.click()
  return this
}