ph-api-scraper 3.4.0

package/example.js

@@ -0,0 +1,31 @@
+const pornhub = require('./src/index.js');
+
+async function demo() {
+    console.log('PornHub API Library - Demo');
+    console.log('===========================\n');
+    
+    console.log('This is a Node.js library for scraping PornHub.');
+    console.log('It provides the following main functions:\n');
+    
+    console.log('1. pornhub.page(url, keys) - Scrape a single video page');
+    console.log('2. pornhub.search(value, keys, options) - Search for videos');
+    console.log('3. pornhub.video(keys, options) - Search by category');
+    console.log('4. pornhub.model(name, keys) - Get model information\n');
+    
+    console.log('Example usage:');
+    console.log('---------------');
+    console.log(`
+const pornhub = require('@justalk/pornhub-api');
+
+// Scrape video info
+const video = await pornhub.page(url, ['title', 'description', 'views']);
+
+// Search for content
+const results = await pornhub.search('keyword', ['title', 'link', 'hd']);
+`);
+    
+    console.log('\nSee README.md for complete documentation.');
+    console.log('\nLibrary loaded successfully! All exports available.');
+}
+
+demo().catch(console.error);

package/package.json

@@ -0,0 +1,51 @@
+{
+        "name": "ph-api-scraper",
+        "version": "3.4.0",
+        "description": "Very complete scrapper for the famous porn website pornhub",
+        "license": "MIT",
+        "repository": {
+                "type": "git",
+                "url": "git+https://github.com/justalk/pornhub-api.git"
+        },
+        "author": "Justal Kevin <justal.kevin@gmail.com> (teamkd.online)",
+        "engines": {
+                "node": ">=12.0"
+        },
+        "scripts": {
+                "test": "xo --fix && nyc --reporter=html --reporter=text ava --verbose --timeout=20s",
+                "test-no": "ava tests/static/* --verbose --timeout=20s",
+                "test-s": "ava tests/static/*.js --verbose --timeout=20s",
+                "test-p": "ava tests/dynamic/tests_page.js --verbose --timeout=20s",
+                "coverage": "nyc report --reporter=text-lcov | coveralls",
+                "postinstall": "node src/packages/postinstall.js"
+        },
+        "keywords": [
+                "porn",
+                "pornhub",
+                "api",
+                "scraper",
+                "sex",
+                "download"
+        ],
+        "dependencies": {
+                "diff": "^4.0.2",
+                "ejs": "^3.1.5",
+                "got": "6.6.3",
+                "html-entities": "^1.3.1",
+                "jsdom": "^16.3.0",
+                "promise": "^8.1.0",
+                "uglify-js": "^3.10.1"
+        },
+        "devDependencies": {
+                "ava": "^2.4.0",
+                "coveralls": "^3.1.0",
+                "nock": "^13.0.4",
+                "nyc": "^15.1.0",
+                "xo": "*"
+        },
+        "bugs": {
+                "url": "https://github.com/justalk/pornhub-api/issues"
+        },
+        "homepage": "https://github.com/justalk/pornhub-api#readme",
+        "main": "src/index.js"
+}

package/replit.md

@@ -0,0 +1,48 @@
+# PornHub API
+
+A Node.js library for scraping PornHub website content.
+
+## Overview
+
+This is an npm package that provides a programmatic API for scraping PornHub. It is **not** a web application - it's a JavaScript library meant to be imported and used by other projects.
+
+## Project Structure
+
+- `src/` - Source code for the library
+  - `index.js` - Main entry point with exports
+  - `page.js` - Video page scraping
+  - `search.js` - Search functionality
+  - `model.js` - Model/pornstar page scraping
+  - `utils.js` - Utility functions
+  - `constants/` - Configuration constants
+  - `helpers/` - Helper utilities
+- `tests/` - Test files
+  - `static/` - Tests using mocked data
+  - `dynamic/` - Tests using live data
+  - `datas/` - Test fixture HTML files
+- `example.js` - Demo script showing library usage
+
+## Running the Project
+
+Run `node example.js` to see a demo of the library capabilities.
+
+Run `npm run test-s` to execute the static tests.
+
+## Main API Functions
+
+1. `pornhub.page(url, keys)` - Scrape a single video page
+2. `pornhub.search(value, keys, options)` - Search for videos
+3. `pornhub.video(keys, options)` - Search by category
+4. `pornhub.model(name, keys)` - Get model information
+
+## Dependencies
+
+- got - HTTP requests
+- jsdom - HTML parsing
+- ejs - Template engine
+- html-entities - Entity decoding
+- uglify-js - Code minification
+
+## Recent Changes
+
+- December 2025: Initial setup for Replit environment

package/src/constants/consts_global.js

@@ -0,0 +1,95 @@
+const js_type = {
+	STRING: 'STRING',
+	URL: 'URL',
+	URL_PORNHUB: 'URL_PORNHUB',
+	NUMBER: 'NUMBER',
+	NUMBER_KM: 'NUMBER_KM',
+	NUMBER_SECONDS: 'NUMBER_SECONDS',
+	ARRAY: 'ARRAY',
+	DATE: 'DATE',
+	OBJECT: 'OBJECT',
+	BOOLEAN: 'BOOLEAN'
+};
+
+module.exports = {
+	NO_DATA: 'No Data',
+	js_type,
+	keys: {
+		COMMENTS: 'COMMENTS',
+		RELATED_VIDEOS: 'RELATED_VIDEOS',
+		DOWNLOAD_URLS: 'DOWNLOAD_URLS',
+		RESULTS: 'RESULTS'
+	},
+	type: {
+		TITLE: js_type.STRING,
+		VIEWS: js_type.NUMBER_KM,
+		UP_VOTES: js_type.NUMBER,
+		DOWN_VOTES: js_type.NUMBER,
+		PERCENT: js_type.NUMBER,
+		AUTHOR: js_type.STRING,
+		AUTHOR_SUBSCRIBER: js_type.NUMBER,
+		CATEGORIES: js_type.ARRAY,
+		TAGS: js_type.ARRAY,
+		PRODUCTION: js_type.STRING,
+		DESCRIPTION: js_type.STRING,
+		DURATION: js_type.NUMBER,
+		RANK_MODEL: js_type.NUMBER,
+		RANK_WEEK_MODEL: js_type.NUMBER,
+		RANK_MONTH_MODEL: js_type.NUMBER,
+		RANK_LAST_MONTH_MODEL: js_type.NUMBER,
+		RANK_YEAR_MODEL: js_type.NUMBER,
+		UPLOAD_DATE: js_type.DATE,
+		PORNSTARS: js_type.ARRAY,
+		DOWNLOAD_URLS: js_type.OBJECT,
+		LINK: js_type.URL_PORNHUB,
+		THUMBNAIL_URL: js_type.URL,
+		HD: js_type.BOOLEAN,
+		NUMBER_OF_COMMENT: js_type.NUMBER,
+		COMMENTS: js_type.OBJECT,
+		AVATAR: js_type.STRING,
+		USERNAME: js_type.STRING,
+		DATE: js_type.STRING,
+		MESSAGE: js_type.STRING,
+		TOTAL_VOTE: js_type.NUMBER,
+		PREMIUM: js_type.BOOLEAN,
+		RATING: js_type.NUMBER,
+		RELATED_SEARCH: js_type.ARRAY,
+		RELATED_PORNSTARS: js_type.ARRAY,
+		RELATED_VIDEOS: js_type.OBJECT,
+		RESULTS: js_type.OBJECT,
+		ACTOR: js_type.STRING,
+		VIDEO_NUMBER: js_type.NUMBER,
+		VIEW_NUMBER: js_type.STRING,
+		RANK: js_type.NUMBER,
+		LINK_MP4: js_type.URL,
+		LINK_WEBM: js_type.URL,
+		RELATIONSHIP_STATUS: js_type.STRING,
+		INTERESTED_IN: js_type.STRING,
+		GENDER: js_type.STRING,
+		BIRTHDAY: js_type.DATE,
+		AGE: js_type.NUMBER,
+		HEIGHT: js_type.STRING,
+		WEIGHT: js_type.STRING,
+		ETHNICITY: js_type.STRING,
+		VIDEO_VIEWS: js_type.NUMBER,
+		PROFILE_VIEWS: js_type.NUMBER,
+		VIDEOS_WATCHED: js_type.NUMBER,
+		JOINED: js_type.STRING
+	},
+	queries: {
+		PAGE_: 'PAGE'
+	},
+	links: {
+		BASE_URL: 'https://www.pornhub.com',
+		SEARCH: 'search',
+		VIDEO: 'video',
+		MODEL: 'model/'
+	},
+	types: {
+		MODEL: 'model',
+		PORNSTAR: 'pornstar'
+	},
+	errors: {
+		DEFAULT: 'An error occured'
+	}
+};

package/src/constants/consts_model.js

@@ -0,0 +1,23 @@
+module.exports = {
+	PROFIL_INFOS_LIST: '.infoPiece',
+	model_selectors: {
+		TITLE: 'h1',
+		RANK_MODEL: '.rankingInfo div:nth-child(1) .big',
+		RANK_WEEK_MODEL: '.rankingInfo div:nth-child(2) .big',
+		RANK_MONTH_MODEL: '.rankingInfo div:nth-child(3) .big',
+		RANK_LAST_MONTH_MODEL: '.rankingInfo div:nth-child(4) .big',
+		RANK_YEAR_MODEL: '.rankingInfo div:nth-child(5) .big',
+		DESCRIPTION: '.model-details .aboutMeSection div:nth-child(2)',
+		VIDEO_NUMBER: '.pornstarVideosCounter'
+	},
+	model_element_attributs: {
+		TITLE: 'innerHTML',
+		DESCRIPTION: 'innerHTML',
+		RANK_MODEL: 'textContent',
+		RANK_WEEK_MODEL: 'textContent',
+		RANK_MONTH_MODEL: 'textContent',
+		RANK_LAST_MONTH_MODEL: 'textContent',
+		RANK_YEAR_MODEL: 'textContent',
+		VIDEO_NUMBER: 'textContent'
+	}
+};

package/src/constants/consts_page.js

@@ -0,0 +1,83 @@
+module.exports = {
+	COMMENTS_LIST: '#cmtContent .topCommentBlock',
+	RELATED_VIDEOS_LIST: '#relatedVideosCenter li',
+	page_element_attributs: {
+		TITLE: 'textContent',
+		VIEWS: 'textContent',
+		UP_VOTES: 'textContent',
+		DOWN_VOTES: 'textContent',
+		PERCENT: 'textContent',
+		AUTHOR: 'textContent',
+		AUTHOR_SUBSCRIBER: 'textContent',
+		PORNSTARS: 'multi_textContent',
+		CATEGORIES: 'multi_textContent',
+		TAGS: 'multi_textContent',
+		PRODUCTION: 'textContent',
+		DURATION: 'dataContent',
+		NUMBER_OF_COMMENT: 'textContent',
+		AVATAR: 'data-src',
+		USERNAME: 'innerHTML',
+		DATE: 'innerHTML',
+		MESSAGE: 'innerHTML',
+		TOTAL_VOTE: 'innerHTML',
+		UPLOAD_DATE: 'javascript',
+		DESCRIPTION: 'javascript',
+		THUMBNAIL_URL: 'javascript'
+	},
+	page_related_videos_element_attributs: {
+		LINK: 'href',
+		TITLE: 'title',
+		HD: null,
+		DURATION: 'innerHTML',
+		VIEWS: 'innerHTML',
+		PREMIUM: null,
+		AUTHOR: 'innerHTML',
+		RATINGS: 'innerHTML',
+		RELATED_SEARCH: 'multi_textContent',
+		RELATED_PORNSTARS: 'multi_textContent',
+		ACTOR: 'innerHTML',
+		VIDEO_NUMBER: 'innerHTML',
+		VIEW_NUMBER: 'innerHTML',
+		RANK: 'innerHTML'
+	},
+	page_selectors: {
+		TITLE: '.title-container .title .inlineFree',
+		VIEWS: '.count',
+		UP_VOTES: '.votesUp',
+		DOWN_VOTES: '.votesDown',
+		PERCENT: '.percent',
+		AUTHOR: '.video-detailed-info .usernameBadgesWrapper a',
+		AUTHOR_SUBSCRIBER: '.userInfo span:nth-child(4)',
+		PORNSTARS: '.pornstarsWrapper .pstar-list-btn',
+		CATEGORIES: '.categoriesWrapper a:not(.add-btn-small)',
+		TAGS: '.tagsWrapper a:not(.add-btn-small)',
+		PRODUCTION: '.productionWrapper .item',
+		DURATION: 'meta[property="video:duration"]',
+		NUMBER_OF_COMMENT: '#cmtWrapper h2 span',
+		UPLOAD_DATE: 'script[type="application/ld+json"',
+		DESCRIPTION: 'script[type="application/ld+json"',
+		THUMBNAIL_URL: 'script[type="application/ld+json"'
+	},
+	comment_selectors: {
+		AVATAR: '.avatarTrigger',
+		USERNAME: '.usernameLink',
+		DATE: '.date',
+		MESSAGE: '.commentMessage span',
+		TOTAL_VOTE: '.voteTotal'
+	},
+	related_videos_selectors: {
+		LINK: 'a',
+		TITLE: '.title a',
+		HD: 'a .marker-overlays .hd-thumbnail',
+		DURATION: 'a .marker-overlays .duration',
+		VIEWS: '.videoDetailsBlock var',
+		PREMIUM: 'a .marker-overlays .premiumIcon',
+		AUTHOR: '.videoUploaderBlock .usernameWrap a',
+		RATINGS: '.rating-container .value'
+	},
+	javascript: {
+		UPLOAD_DATE: 'uploadDate',
+		DESCRIPTION: 'description',
+		THUMBNAIL_URL: 'thumbnailUrl'
+	}
+};

package/src/constants/consts_queries.js

@@ -0,0 +1,11 @@
+module.exports = {
+	filter: {
+		MOST_VIEWED: 'mv',
+		HOTTEST: 'ht',
+		TOP_RATED: 'tr'
+	},
+	min_duration: [10, 20, 30],
+	max_duration: [10, 20, 30],
+	production: ['homemade', 'professional'],
+	promo: ['modelhub', 'premium']
+};

package/src/constants/consts_sanitizer.js

@@ -0,0 +1,14 @@
+module.exports = {
+	sanitizer_by_type: {
+		STRING: 'sanitizer_string',
+		URL: 'sanitizer_normal',
+		URL_PORNHUB: 'sanitizer_url_pornhub',
+		NUMBER: 'sanitizer_number',
+		NUMBER_KM: 'sanitizer_KM_to_unit',
+		NUMBER_SECONDS: 'sanitizer_normal',
+		ARRAY: 'sanitizer_array',
+		DATE: 'sanitizer_date',
+		OBJECT: 'sanitizer_normal',
+		BOOLEAN: 'sanitizer_boolean'
+	}
+};

package/src/constants/consts_scrap.js

@@ -0,0 +1,10 @@
+module.exports = {
+	scrap_by_attribut_type: {
+		innerHTML: 'scrap_inner_html',
+		dataContent: 'scrap_data_content',
+		textContent: 'scrap_text_content',
+		multi_textContent: 'scrap_multi_text_content',
+		javascript: 'scrap_javascript',
+		null: 'scrap_null'
+	}
+};

package/src/constants/consts_search.js

@@ -0,0 +1,22 @@
+module.exports = {
+	primary_search_selectors: {
+		RELATED_SEARCH: '.relatedSearches li a',
+		RELATED_PORNSTARS: '#relatedPornstarSidebar li a .pornstars-name'
+	},
+	page_search_element_attributs: {
+		LINK: 'href',
+		TITLE: 'title',
+		HD: null,
+		DURATION: 'innerHTML',
+		VIEWS: 'innerHTML',
+		PREMIUM: null,
+		AUTHOR: 'innerHTML',
+		RATINGS: 'innerHTML',
+		RELATED_SEARCH: 'multi_textContent',
+		RELATED_PORNSTARS: 'multi_textContent',
+		ACTOR: 'innerHTML',
+		VIDEO_NUMBER: 'innerHTML',
+		VIEW_NUMBER: 'innerHTML',
+		RANK: 'innerHTML'
+	}
+};

package/src/constants/consts_search_gifs.js

@@ -0,0 +1,15 @@
+module.exports = {
+	GIFS_LIST: '.gifsWrapper .gifVideoBlock',
+	gifs_search_selectors: {
+		TITLE: '.title',
+		THUMBNAIL_URL: '.gifVideo',
+		LINK_MP4: '.gifVideo',
+		LINK_WEBM: '.gifVideo'
+	},
+	gifs_element_attributs: {
+		TITLE: 'textContent',
+		THUMBNAIL_URL: 'data-poster',
+		LINK_MP4: 'data-mp4',
+		LINK_WEBM: 'data-webm'
+	}
+};

package/src/constants/consts_search_pornstars.js

@@ -0,0 +1,25 @@
+module.exports = {
+	PORNSTARS_LIST: '#pornstarsSearchResult li .wrap',
+	pornstars_search_selectors: {
+		ACTOR: '.title',
+		VIDEO_NUMBER: '.videosNumber',
+		VIEW_NUMBER: '.pstarViews',
+		RANK: '.rank_number'
+	},
+	pornstars_element_attributs: {
+		LINK: 'href',
+		TITLE: 'title',
+		HD: null,
+		DURATION: 'innerHTML',
+		VIEWS: 'innerHTML',
+		PREMIUM: null,
+		AUTHOR: 'innerHTML',
+		RATINGS: 'innerHTML',
+		RELATED_SEARCH: 'multi_textContent',
+		RELATED_PORNSTARS: 'multi_textContent',
+		ACTOR: 'innerHTML',
+		VIDEO_NUMBER: 'innerHTML',
+		VIEW_NUMBER: 'innerHTML',
+		RANK: 'innerHTML'
+	}
+};

package/src/constants/consts_search_videos.js

@@ -0,0 +1,29 @@
+module.exports = {
+	VIDEOS_LIST: '.nf-videos ul .pcVideoListItem',
+	videos_search_selectors: {
+		LINK: 'a',
+		TITLE: '.title a',
+		HD: 'a .marker-overlays .hd-thumbnail',
+		DURATION: 'a .marker-overlays .duration',
+		VIEWS: '.videoDetailsBlock var',
+		PREMIUM: 'a .marker-overlays .premiumIcon',
+		AUTHOR: '.videoUploaderBlock .usernameWrap a',
+		RATINGS: '.rating-container .value'
+	},
+	videos_element_attributs: {
+		LINK: 'href',
+		TITLE: 'title',
+		HD: null,
+		DURATION: 'innerHTML',
+		VIEWS: 'innerHTML',
+		PREMIUM: null,
+		AUTHOR: 'innerHTML',
+		RATINGS: 'innerHTML',
+		RELATED_SEARCH: 'multi_textContent',
+		RELATED_PORNSTARS: 'multi_textContent',
+		ACTOR: 'innerHTML',
+		VIDEO_NUMBER: 'innerHTML',
+		VIEW_NUMBER: 'innerHTML',
+		RANK: 'innerHTML'
+	}
+};

package/src/helpers/utils_global.js

@@ -0,0 +1,36 @@
+/**
+ * @file Regroup all the general functions that can be usefull anywhere
+ * @author Justal Kevin
+ */
+
+'use strict';
+
+module.exports = {
+	/**
+	* Remove duplicated string inside an array
+	*
+	* @params {array} array The array with possible duplicate of same data
+	* @return {array} The array with unique value
+	**/
+	remove_duplicate: array => {
+		return array.filter((item, index) => array.indexOf(item) === index);
+	},
+	/**
+	* Check if the parameter is defined and exist
+	*
+	* @params {string} parameter The parameter we want to test
+	* @return {boolean} True if the parameter exist or else false
+	**/
+	is_parameter_missing: parameter => {
+		return parameter === null || parameter === '' || parameter === undefined;
+	},
+	selectors_restriction: (keys, selectors) => {
+		return Object.fromEntries(Object.keys(selectors).map(selector => {
+			if (keys.includes(selector)) {
+				return [selector, selectors[selector]];
+			}
+
+			return null;
+		}).filter(x => x));
+	}
+};

package/src/helpers/utils_sanitizer.js

@@ -0,0 +1,144 @@
+/**
+ * @file Regroup all the functions use for sanitize the informations coming and going in the api
+ * @author Justal Kevin
+ */
+
+'use strict';
+
+const consts_global = require('./../constants/consts_global');
+const consts_sanitizer = require('./../constants/consts_sanitizer');
+const utils_global = require('./utils_global');
+const Entities = require('html-entities').AllHtmlEntities;
+const entities = new Entities();
+
+module.exports = {
+	/**
+	* Sanitize the number of pornhub to a javascript type Number
+	* Most of the number in pornhub contains letters or commas, so it need to be filter
+	*
+	* @params {string} value The string that represents a number
+	* @return {number} The string converted into a number javascript
+	**/
+	sanitizer_number: value => {
+		value = value.replace(/[()&A-Za-z,%]/g, '');
+		value = Number(value);
+		return value;
+	},
+	/**
+	* Sanitize the string from pornhub to a single ligne.
+	* This function remove the tabs, break line and trim the string
+	*
+	* @params {string} value The string that we want to sanitize
+	* @return {string} The string sanitized
+	**/
+	sanitizer_string: value => {
+		value = value.replace(/[\t\n]/g, '');
+		value = value.trim();
+		value = entities.decode(value);
+		return value;
+	},
+	/**
+	* Convert a string or number into a boolean
+	*
+	* @params {(string|number)} value The string or number representing a boolean
+	* @return {boolean} True or False depending of the value passed
+	**/
+	sanitizer_boolean: value => {
+		return Boolean(value);
+	},
+	/**
+	* Sanitize the key passed to the API
+	* Replace space and : and put the key in uppercase
+	*
+	* @params {string} value The API key that we want to sanitize
+	* @return {string} The key sanitized
+	**/
+	sanitizer_key: value => {
+		value = module.exports.sanitizer_string(value);
+		value = value.replace(/\s/g, '_');
+		value = value.replace(/:/g, '');
+		value = value.toUpperCase();
+		return value;
+	},
+	/**
+	* Convert the value with an unit to a number javascript
+	*
+	* @params {string} value The string with an unit representing a number
+	* @return {number} The javascript number
+	**/
+	sanitizer_KM_to_unit: value => {
+		if (value.includes('K')) {
+			return Number(value.replace('K', '')) * 1000;
+		}
+
+		if (value.includes('M')) {
+			return Number(value.replace('M', '')) * 1000000;
+		}
+
+		return value;
+	},
+	/**
+	* Convert a string date to a javascript date
+	*
+	* @params {string} value A string representing a date
+	* @return {Date} The javascript date representing the value passed in argument
+	**/
+	sanitizer_date: value => {
+		return new Date(value);
+	},
+	/**
+	* Create a complete url pornhub from a route passed in argument
+	*
+	* @params {string} value A pornhub route
+	* @return {string} A complete url pornhub
+	**/
+	sanitizer_url_pornhub: value => {
+		return consts_global.links.BASE_URL + value;
+	},
+	/**
+	* Simply return the variable passed in argument
+	* This function is used for getting the function sanitizer possible and givig every type a function
+	*
+	* @params {Object} A value
+	* @return {Object} The same value without any change
+	**/
+	sanitizer_normal: value => {
+		return value;
+	},
+	/**
+	* Sanitize every string element of an array of string
+	* and remove duplicate value
+	*
+	* @params {array} array An array containing string
+	* @return {array} Return the array with unique value and sanitized string
+	**/
+	sanitizer_array: array => {
+		if (Array.isArray(array)) {
+			array = array.map(x => module.exports.sanitizer_string(x));
+			return utils_global.remove_duplicate(array);
+		}
+
+		return module.exports.sanitizer_string(array);
+	},
+	/**
+	* Sanitize all the data receive from pornhub to a more flexible format
+	*
+	* @params {Object} datas The datas that we want to sanitize
+	* @return {Object} The datas sanitized
+	**/
+	sanitizer: datas => {
+		const rsl = Object.keys(consts_global.type).map(x => {
+			if (datas[x] === null || datas[x] === undefined) {
+				return;
+			}
+
+			// Choose the sanitizer to apply to the current value depending of the type
+			const sanitizer = consts_sanitizer.sanitizer_by_type[consts_global.type[x].toUpperCase()];
+			const sanitize_data = module.exports[sanitizer](datas[x]);
+
+			return [x.toLowerCase(), sanitize_data];
+		}).filter(x => x);
+
+		return Object.fromEntries(rsl);
+	}
+};