chromium-dashboard/developer-documentation.md

14 KiB
Исходник Ответственный История

chromedash developer documentation

This doc covers some basic overview of the codebase to help developers navigate.

In summary, this web app is using Flask as the backend and uses Lit webcomponents in the front end. It uses Sign in with Google for authentication. Google Cloud Datastore is used as database.

Back end

In the Backend,

  • Flask is being used for:
    • All the request handlers (see basehandlers.py and all the code under api/ and pages/).
    • HTML templates (see FlaskHandler.render() in framework/basehandlers.py).

HISTORY:-

  • The app used to use a combination of Django plus Webapp2. However, now it uses Flask as mentioned above.
  • The app used to use DB Client Library for interacting with Google Cloud DataStore. It was later replaced by NDB Client Library. Now, it uses the Cloud NDB Library

Front end

Main site page rendering

All the pages are rendered in a combination of Jinja2 template (/templates) and front-end components (/client-src/elements).

  1. /templates/base.html and /templates/base_embed.html are the html skeleton.
  2. Templates in /templates (extend the _base.html or _embed_base.html) are the Jinja2 templates for each page.
    • The folder organization and template file names matches the router. (See template_path=os.path.join(path + '.html') in server.py)
    • lit-element components, css, js files are all imported/included in those templates.
    • We pass backend variables to js like this: const variableInJs = {{variable_in_template|safe}}.
  3. All Lit components are in /client-src/elements, and other Javascript files are in /client-src/js-src/.
  4. JavaScript is processed and code-split by Rollup, then output to /static/dist/ and included in templates.
  5. All *-css.js files used in client-side components are in /client-src/css/. The remaining css files still being included in templates are in /static/css/.

Adding an icon

Shoelace comes bundled with Bootstrap Icons, but we prefer to use Material Icons in most cases.

To add a new Bootstrap icon:

  1. Copy it from node_modules/@shoelace-style/shoelace/dist/assets/icons to static/shoelace/assets/icons.
  2. Reference it like <sl-icon name="icon-name">.

To add a new Material icon:

  1. Download the 24pt SVG file from https://fonts.google.com/icons?icon.set=Material+Icons
  2. Rename it to the icon name with underscores, and place it in static/shoelace/assets/material-icons.
  3. Reference it like <sl-icon library="material" name="icon_name">.

Creating a user with admin privileges

Creating or editing features normally requires a @google.com or @chromium.org account. To work around this when running locally, you can make a temporary change to the file framework/permissions.py to make function can_admin_site() return True. Once you restart the server and log in using any account, you will be able create or edit features.

To avoid needing to make this temporary change more than once, you can sign in and visit /admin/users/new to create a new registered account using the email address of any Google account that you own, such as an @gmail.com account.

Generating Diffs for sending emails to subscribers of a feature

  • When someone edits a feature, everyone who have subscribed to that feature will receive a email stating what fields were edited, the old values and the new values.
  • The body of this email (diffs) can be seen in the console logs. To see the logs, follow these steps:-
    1. Create a feature using one account.
    2. Now, signout and login with another account.
    3. Click on the star present in the feature box in the all features page.
    4. Now login again using the first account and edit a feature.
    5. On pressing submit after editing the feature, you will be able to see the diff in the console logs.

Local Development

  • When run locally, Datastore Emulator is used for storing all the entries. To reset local database, remove the local directory for storing data/config for the emulator. The default directory is <USER_CONFIG_DIR>/emulators/datastore. The value of <USER_CONFIG_DIR> can be found by running: $ gcloud info --format='get(config.paths.global_config_dir)' in the terminal. To learn more about using the Datastore Emulator CLI, execute $ gcloud beta emulators datastore --help.
  • Executing npm start or npm test automatically starts the Datastore Emulator and shuts it down afterwards.

Adding a new API

This section outlines the steps to consider when adding a new API.

Note: For all new APIs, please consider using OpenAPI. With OpenAPI, developers can write a specification for their API and have code generated for them on both the frontend and backend. This helps remove the burden of manually writing data models and data encoding and decoding for both sides. There is a tool installed as a devDependency called openapi-generator-cli to do the generation of the code.

The specification follows OpenAPI version 3 and is located at openapi/api.yaml.

Below are steps to help guide a developer along with a relatable example that follows the same steps.

Step 0: Additional Tools To Help

If using Visual Studio Code, install the following extensions. (These are pre-installed if using the devcontainer)

Step 1: Add the path to openapi/api.yaml

Before completing this step, read the Paths and Operations and Describing Parameters OpenAPI docs

  • Under paths, add the path. You'll add operations named after HTTP methods under this.
Example (click to expand)

openapi/api.yaml

paths:
  /features/{feature_id}:
    ...

Step 2: Add the operations

Operations = HTTP verbs. (e.g. GET, POST, PUT, etc)

  • Add the operation(s) under the path.
  • Ensure each operation has a summary, description and operationId
  • If your path has path parameters, describe the parameters now too.
    • Mark required parameters with required: true.
Example (click to expand)

openapi/api.yaml

paths:
  ...
  /features/{feature_id}:
    get:
      summary: Get a feature by ID.
      description: |
        Get a feature by ID. More details about this here.
        Also, can do more comments        
      operationId: getFeatureById
      parameters:
        - name: feature_id
          in: path
          description: Feature ID
          required: true
          schema:
            type: integer
    post:
      summary: Update a feature by ID.
      description: |
        Update a feature with the given ID.
        More details about this here.        
      operationId: updateFeatureById
      parameters:
        - name: feature_id
          in: path
          description: Feature ID
          required: true
          schema:
            type: integer

Step 3: Describe the request body

Before completing this step, read the Describing Request Body OpenAPI doc

Skip this step if there is no request body

  • Inside the operation (post in this example), add a requestBody.content.application/json.schema object.
  • Use a JSON Schema to define the request body.
  • You can re-use parts or all of the schema by writing $ref: '#/components/schemas/WellNamedObject' and defining the WellNamedObject under the top-level components.schemas object.
Example (click to expand)

openapi/api.yaml

paths:
  ...
  /features/{feature_id}:
    post:
      summary: Update a feature by ID.
      description: |
        Update a feature with the given ID.
        More details about this here.        
      operationId: updateFeatureById
      parameters:
        - name: feature_id
          in: path
          description: Feature ID
          required: true
          schema:
            type: integer
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Feature'
components:
  schemas:
    Feature:
      description: A feature
      type: object
      properties:
        id:
          type: integer
        name:
          type: string
        live:
          type: boolean
          description: Some optional field
      required:
        - id
        - name

For this example, only needed to describe a request body for the post operation.

Step 4: Describe the Responses

Before completing this step, read the Describing Responses OpenAPI doc

Skip this step if there is no response body

  • Add the appropriate response code(s)
    • Don't worry about describing global errors like unauthorized calls right now.
  • For each response code, describe the response object. As with the request body, you can refer to schemas defined in the components.schemas top-level object with $ref: '#/components/schemas/WellNamedObject'.
Example (click to expand)

openapi/api.yaml

paths:
  ...
  /features/{feature_id}:
    post:
      summary: Update a feature by ID.
      description: |
        Update a feature with the given ID.
        More details about this here.        
      operationId: updateFeatureById
      parameters:
        - name: feature_id
          in: path
          description: Feature ID
          required: true
          schema:
            type: integer
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Feature'
      responses:
        '200':
          description: An updated feature
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Feature'
components:
  schemas:
    Feature:
      description: A feature
      type: object
      ...

Step 5: Generate the Code

Validate that the linked schema objects are valid. There should be zero errors and zero warnings:

  • npm run openapi-validate

Generate the code:

  • npm run openapi

Step 6: Incorporate Into Backend

Currently, the repository is configured to use the generated Python data models for the backend. Once all routes are generated by OpenAPI, it would be wise to revisit using the controllers as well

  • Open main.py
  • Locate the api_routes variable.
  • Add a route.
    • In this example, it would be Route(f'{API_BASE}/features/<int:feature_id>', features_api.FeaturesAPI).
  • In the handler, the generated model classes can be imported from chromestatus_openapi.models.
  • Since we do not use the controllers, you will need to return a dictionary of the model class. Then, Flask can convert it appropriately to json. Each generated class has a to_dict() method to accomplish this.

Step 7: Incorporate Into Frontend

The frontend use @lit-labs/context to pass the client around. The benefits of it can be seen here and the advertised use cases here.

Your element needs to use a context consumer to retrieve the client that is provided by chromedash-app. Once you have the client, you can make an API call like normal.

import {ContextConsumer} from '@lit-labs/context';
import {chromestatusOpenApiContext} from '../contexts/openapi-context';
export class SomeElement extends LitElement {
  // Nice to have type hinting so that the IDE can auto complete the client and its functions.
  /** @type {ContextConsumer<import("../contexts/openapi-context").chromestatusOpenApiContext>} */
  _clientConsumer;

  constructor() {
    super();
    this._clientConsumer = new ContextConsumer(this, chromestatusOpenApiContext, undefined, true);
  }

  fetchData() {
    // Important to call .value to get the client from the context.
    this.clientConsumer.value.getFeature();
  }
  // other element stuff
}

Miscellaneous

Feature links provide users with easy access to additional information about URLs in the feature, without the need to navigate away from the site.

You can view all existing feature links and their associated statistics on admin page.

  1. Modify internals/link_helpers.py and internals/link_helpers_test.py to include support for parsing the new type of feature link.
  2. Update client-src/elements/feature-link.js to render the new type of feature link.
  3. Wait for the next cron job (which runs every Tuesday) to update the corresponding feature links, or trigger the job immediately by running /cron/update_all_feature_links

Feature links are automatically updated when a feature is created or edited. However, if youve performed a database cleanup, you can use a script to backfill the feature links.

  1. Run /scripts/backfill_feature_links to backfill the feature links without extracting information into the database. After running the script, you can verify the results by visiting admin page.
  2. Run /cron/update_all_feature_links?no_filter=true to update all existing feature links with latest information.