Action structure validation (#61)

* Validate action

* dependencies

* dependencies

* typo

Author: MatthewZMSU

actions/action.js

@@ -24,7 +24,7 @@ exports.action = async function action(page, request) {
         throw new exceptions.IncorrectArgumentError("Invalid action function.\n" +
             "Valid action function: \"async function action(page, request) " +
             "{ ... some actions with request and page in puppeteer " +
-            "syntax};\"");
+            "syntax};\".");
     }
 
     return {

actions/click.js

@@ -1,8 +1,9 @@
 const utils = require('../helpers/utils');
+const exceptions = require('../helpers/exceptions');
 
 const DEFAULT_TIMEOUT = 1000;  // 1 second
 
-/*
+/**
  * body = {
  *     "selector": "", // <string> A selector to search for element to click. If there are multiple elements satisfying the selector, the first will be clicked.
  *     "clickOptions": {
@@ -11,10 +12,19 @@ const DEFAULT_TIMEOUT = 1000;  // 1 second
  *         "delay"  // <number> Time to wait between mousedown and mouseup in milliseconds. Defaults to 0.
  *     },
  *     "waitOptions": {...},  // same as in goto action, defaults to 1s timeout
- *     "navigationOptions": {...}  // same as in goto action
+ *     "navigationOptions": {...},  // same as in goto action
  * }
  */
 exports.click = async function click(page, request) {
+    // Validation
+    if (!("selector" in request.body)) {
+        throw new exceptions.IncorrectArgumentError("No selector to click in request.");
+    }
+    if (typeof request.body.selector != "string") {
+        throw new exceptions.IncorrectArgumentError("Selector should be a string.");
+    }
+
+    // Evaluation
     await page.hover(request.body.selector);
     if (request.body.navigationOptions) {
         await Promise.all([

actions/compose.js

@@ -1,3 +1,5 @@
+const exceptions = require("../helpers/exceptions");
+
 endpoint2action = {
     action: require("./action").action,
     click: require("./click").click,
@@ -12,7 +14,27 @@ endpoint2action = {
     scroll: require("./scroll").scroll,
 }
 
+/**
+ * Compose is a sequence of other actions that are performed on the page.
+ * The function does not allow body to contain compose endpoint inside.
+ *
+ * body = {
+ *      "actions": [  An array of actions
+ *          {
+ *              "endpoint": <string>,  Action endpoint
+ *              "body": <object>,  A body that is processable by action on the endpoint
+ *          },
+ *          ...,
+ *     ],
+ * }
+ */
 async function compose(page, request) {
+    // Validation
+    if (!(request.body instanceof Object)) {
+        throw new exceptions.IncorrectArgumentError("Body of compose method should be an Object.");
+    }
+
+    // Evaluation
     const originalClosePage = request.query.closePage;
     const originalBody = structuredClone(request.body);
 

actions/fill_form.js

@@ -1,16 +1,23 @@
 const utils = require('../helpers/utils');
+const exceptions = require("../helpers/exceptions");
 
-/*
+/**
  * body = {
- *     "inputMapping": { A dictionary where each key is a CSS selector, and each value is another dictionary containing details about the input for that element:
- *         "selector": <string> The CSS selector for the input element (used as the key).
- *         "value": <string> The text to be inputted into the element.
- *         "delay": <number> A delay (in milliseconds) between each keystroke when inputting the text. Defaults to 0 if not provided.
+ *     "inputMapping": {  A dictionary where each key is a CSS selector, and each value is another dictionary containing details about the input for that element:
+ *         "selector": <string>,  The CSS selector for the input element (used as the key).
+ *         "value": <string>,  The text to be inputted into the element.
+ *         "delay": <number>,  A delay (in milliseconds) between each keystroke when inputting the text. Defaults to 0 if not provided.
  *     },
- *     "submitButton": <string> The CSS selector for the form's submit button. If provided, the button will be clicked after filling in the form.
+ *     "submitButton": <string>, The CSS selector for the form's submit button. If provided, the button will be clicked after filling in the form.
  * }
  */
 exports.fillForm = async function fillForm(page, request) {
+    // Validation
+    if (!("inputMapping" in request.body)) {
+        throw new exceptions.IncorrectArgumentError("No inputMapping provided in fill_form request.");
+    }
+
+    // Evaluation
     const inputMapping = request.body.inputMapping;
     const submitButton = request.body.submitButton;
 
@@ -20,10 +27,9 @@ exports.fillForm = async function fillForm(page, request) {
         await page.type(selector, value, { delay });
     }
 
-    if (submitButton) {
+    if (submitButton) {  // Maybe we can use click action here?
         await page.click(submitButton);
     }
 
     return await utils.getContents(page);
-
 }

actions/goto.js

@@ -1,6 +1,7 @@
 const utils = require('../helpers/utils');
+const exceptions = require('../helpers/exceptions');
 
-/*
+/**
  * body = {
  *     "url": <string> URL to navigate page to. The url should include scheme, e.g. https://.
  *     "navigationOptions": { Navigation parameters which might have the following properties:
@@ -22,6 +23,11 @@ const utils = require('../helpers/utils');
  * }
  */
 exports.goto = async function goto(page, request) {
+    // Validation
+    if (!("url" in request.body)) {
+        throw new exceptions.IncorrectArgumentError("No URL provided in goto request.");
+    }
+
     await page.goto(request.body.url, request.body.navigationOptions);
     return await utils.getContents(page, request.body.waitOptions);
 }

actions/har.js

@@ -1,7 +1,10 @@
 const exceptions = require("../helpers/exceptions");
 
+/**
+ * Captures har of the page.
+ */
 exports.har = async function har(page, request) {
-    if (!(page.harWriter)){
+    if (!(page.harWriter)) {
         throw new exceptions.NoHarWriterError();
     }
 

actions/mhtml.js

@@ -1,5 +1,5 @@
-/*
- * Captures mhtml snapshot of a page
+/**
+ * Captures mhtml snapshot of the page.
  */
 exports.captureSnapshot = async function captureSnapshot(page, request) {
     const cdpSession = await page.target().createCDPSession();

actions/recaptcha_solver.js

@@ -1,37 +1,41 @@
 const utils = require('../helpers/utils')
+const exceptions = require("../helpers/exceptions");
 
 const DEFAULT_TIMEOUT = 1000;  // 1 second
 
-/*
- * This module introduces new ability to puppeteer-service.
- * It is capable of solving recaptchas on a given web-page.
- * If there is no recaptcha on the page nothing bad will happen.
- * If there is recaptcha it solves it and then inserts the special code
+/**
+ * The function solves recaptchas on the page.
+ * If there is no recaptcha on the page nothing will happen.
+ * If there is a recaptcha the function solves it and then inserts the special code
  * into the page automatically.
  *
  * Returns useful information about recaptcha_solving.
  * For more information about return value visit
  * https://github.com/berstend/puppeteer-extra/tree/master/packages/puppeteer-extra-plugin-recaptcha#result-object
  */
 exports.recaptchaSolver = async function recaptchaSolver(page, request) {
-    let recaptcha_data;
+    if (!("solve_recaptcha" in request.body)) {
+        throw new exceptions.IncorrectArgumentError("No solve_recaptcha parameter in request");
+    }
+
+    let recaptchaData;
 
     if (request.body.solve_recaptcha) {
-        recaptcha_data = await page.solveRecaptchas();
+        recaptchaData = await page.solveRecaptchas();
     } else {
-        recaptcha_data = await page.findRecaptchas();
+        recaptchaData = await page.findRecaptchas();
     }
 
     const waitOptions = request.body.waitOptions || { timeout: DEFAULT_TIMEOUT };
     const contents = await utils.getContents(page, waitOptions);
 
     if (request.query.closePage ||
-        (request.body.close_on_empty && recaptcha_data['captchas'].length === 0)) {
+        (request.body.close_on_empty && recaptchaData['captchas'].length === 0)) {
         await page.close();
     }
 
     return {
         ...contents,
-        recaptcha_data: recaptcha_data,
+        recaptcha_data: recaptchaData,
     }
 }

actions/screenshot.js

@@ -1,4 +1,4 @@
-/*
+/**
  * Method that returns screenshots of pages
  * more description of options you can see on GitHub:
  * https://github.com/GoogleChrome/puppeteer/blob/v1.19.0/docs/api.md#pagescreenshotoptions

actions/scroll.js

@@ -2,7 +2,7 @@ const utils = require('../helpers/utils');
 
 const DEFAULT_TIMEOUT = 1000;  // 1 second
 
-/*
+/**
  * Method that scrolls page to a certain selector.
  * Example body:
  * body = {

helpers/middlewares/process_exception.js

@@ -20,7 +20,7 @@ exports.processExceptionMiddleware = async function processExceptionMiddleware(e
         res.status(400);
     } else if (err instanceof exceptions.NoHarWriterError) {
         res.status(400);
-    }else if (err instanceof exceptions.TooManyContextsError) {
+    } else if (err instanceof exceptions.TooManyContextsError) {
         res.status(429);  // Too Many Requests
     } else if (err.contextId) {  // there was a context, but something went wrong
         res.status(500);

helpers/utils.js

@@ -234,4 +234,3 @@ exports.performAction = async function performAction(request, action) {
         return response;
     });
 };
-

routes/click.js

@@ -2,15 +2,10 @@ const express = require('express');
 
 const {click} = require('../actions/click');
 const utils = require('../helpers/utils');
-const exceptions = require('../helpers/exceptions');
 
 const router = express.Router();
 
 router.post('/', async function (req, res, next) {
-    if (!("selector" in req.body)) {
-        throw new exceptions.IncorrectArgumentError("No selector to click in request");
-    }
-
     try {
         let response = await utils.performAction(req, click);
         res.header('scrapy-puppeteer-service-context-id', response.contextId);

routes/compose.js

@@ -1,16 +1,11 @@
 const express = require('express');
 
 const {compose} = require('../actions/compose');
-const exceptions = require('../helpers/exceptions');
 const utils = require("../helpers/utils");
 
 const router = express.Router();
 
 router.post('/', async function (req, res, next){
-    if (!(req.body instanceof Object)) {
-        throw new exceptions.IncorrectArgumentError("Body of compose method should be an Object");
-    }
-
     try {
         let response = await utils.performAction(req, compose);
         res.header('scrapy-puppeteer-service-context-id', response.contextId);

routes/fill_form.js

@@ -2,15 +2,10 @@ const express = require('express');
 
 const {fillForm} = require('../actions/fill_form');
 const utils = require('../helpers/utils');
-const exceptions = require('../helpers/exceptions');
 
 const router = express.Router();
 
 router.post('/', async function (req, res, next) {
-    if (!req.body.inputMapping) {
-        throw new exceptions.IncorrectArgumentError("No inputMapping provided in fill_form request");
-    }
-
     try {
         let response = await utils.performAction(req, fillForm);
         res.header('scrapy-puppeteer-service-context-id', response.contextId);

routes/goto.js

@@ -2,15 +2,10 @@ const express = require('express');
 
 const {goto} = require('../actions/goto');
 const utils = require('../helpers/utils');
-const exceptions = require('../helpers/exceptions');
 
 const router = express.Router();
 
 router.post('/', async function (req, res, next) {
-    if (!req.body.url) {
-        throw new exceptions.IncorrectArgumentError("No URL provided in goto request");
-    }
-
     try {
         let response = await utils.performAction(req, goto);
         res.header('scrapy-puppeteer-service-context-id', response.contextId);

routes/recaptcha_solver.js

@@ -6,21 +6,13 @@ const utils = require('../helpers/utils')
 const exceptions = require('../helpers/exceptions');
 
 router.post('/', async function (req, res, next) {
-    if (!req.query.contextId || !req.query.pageId) {
-        throw new exceptions.IncorrectArgumentError("No page in request");
-    }
-
     if (!process.env.TOKEN_2CAPTCHA) {
         res.status(501);
         res.send("TOKEN_2CAPTCHA is not provided!");
         next();
         return;
     }
 
-    if (!("solve_recaptcha" in req.body)) {
-        throw new exceptions.IncorrectArgumentError("No solve_recaptcha parameter in request");
-    }
-
     try {
         let response = await utils.performAction(req, recaptchaSolver);
         res.header('scrapy-puppeteer-service-context-id', response.contextId);