Doogle
Doogle is a search engine and web crawler which can search indexed websites and images, and then use keywords to be searched later.
Written primarily in OOP style PHP with the intent of better understanding OOP and how web crawlers work.
Features
- Search sites
- Displays title, URL and description
- Search images
- Hover over images to preview description (alt tag)
- Masonry layout for searched images
- Image preview using Fancybox
- Image search page responds dynamically
- Clean homepage
- Filters broken image results
- Organises search results by clicks/visits
- Pagination system at the bottom of the search page
- Shows 'results found' for search term
Table of Contents
Setup and Usage
Server Setup
v1.0.0-beta.1 is supported and tested in PHP 7.4, 8.0 and 8.1.
Please refer to XAMPP for the web server, PHP server and MySQL server configuration. XAMPP is the simplest method as several servers are required to use Doogle.
MySQL Setup on XAMPP will use PHPMyAdmin as a GUI method of setting up the database.
Once logged into the database via PHPMyAdmin under the PHPMyAdmin > SQL tab, the content of 'doogle-tables-no-data.sql' can be pasted into the field
PHP Dependencies
mysql
pdo_mysql
SQL User Creation
Amend the password PASSWORD_HERE using a strong random password.
mysql> CREATE USER IF NOT EXISTS 'doogle'@'localhost' IDENTIFIED BY 'PASSWORD_HERE';
SQL User Permissions
The SQL user 'doogle' must have SELECT, INSERT and UPDATE privileges:
mysql> GRANT SELECT, INSERT, UPDATE ON `doogle`.* TO 'doogle'@'localhost';
- INSERT is used for crawling
- SELECT is required for the search engine to return queries
- UPDATE is required to amend the clicks and broken results (see ./ajax/)
Connecting PHP to MySQL Server
In the file config.php the following must be entered correctly for your database configuration:
$dbname = "doogle";
$dbhost = "localhost";
$dbuser = "doogle";
$dbpass = "";
In the file 'doogle-tables-no-data.sql' the database will be created as 'doogle'.
Crawling Websites to Populate Images and Sites tables
Form-based crawl
In your browser go to where the file is hosted http://localhost/crawl-formSubmit.php
Paste the URL into the input field and press the Crawl button.
Manual crawl
At the bottom of crawl-manual.php the variable $startUrl is where to paste the URL of the website to be crawled:
$startUrl = "https://thehackernews.com/";
Then in your browser go to where the file is hosted http://localhost/crawl-manual.php
Explanation
The crawling process will take some time, it will completely depend on the size of the website being crawled. The page will continue to load (without output) until the crawl.php
script finishes.
Check the tables images
and sites
in the database to ensure they are being populated.
Once the tables are populated visit the Doogle homepage and search! See preview images.
Programming Logic
Pagination
Logic of pagination system
Inside search.php, pagination is implemented
In the example above, currentPage=11. The number of pages to show is always 10.
Results Per Page
Site search will return 20 results per page and image search will return 30 results per page.
The results per page can be changed inside search.php on lines {83, 88} respectively. As indicated by the $pageSize variables:
Handling an edge case
An edge case can occur when no more pages are available.
So, for 331 results, 17 pages will be available. However, without an edge case scenario consider, the UI for the pagination system will allow scrolling through pages which don't exist; which would return an empty result.
To handle an edge case the following logic is implemented in the while-loop:
if($currentPage + $pagesLeft > $numPages + 1)
$currentPage = $numPages + 1 - $pagesLeft;
while($pagesLeft != 0 && $currentPage <= $numPages)
{ ... }
Image Search
Image Captions
To make image searches more informative, the 'alt' tag is part of the search term. As shown in ./classes/ImageResultsProvider.php line 34
Loading Images with JavaScript
In the 'images' table, there is a row 'broken' which tracks images which return an error.
Because images are already loaded with a pure server-side solution, AJAX must be leveraged, loading images dynamically. Which is shown in ./assets/js/script.js
Masonry
Image searches are using Masonry - Cascading grid layout library.
Masonry allows images a grid layout which is responsive due to jQuery. The image below shows an example layout:
Site Search - Trimming Results
As shown in the preview images, Doogle when performing a site search will return (title, URL and description) for each result.
However, to make some results easier to read, a trimming process is performed. Inside ./classes/SiteResultsProvider.php the function trimField() is called:
Title's are trimmed at 55 characters and description's are trimmed at 230 characters.
Telemetry
Both the 'images' and 'sites' tables in the database have a row containing 'clicks' for each column.
The 'clicks' field is increased each time a site is visited or image is previewed.
When performing a search, results returned are organised in descending order of clicks. This behaviour is shown by the $query inside ./classes/SiteResultsProvider.php function getResultsHtml(). See line 43.
User-Agent
Inside ./classes/DomDocumentParser.php the user-agent data used during crawling is located. As indicated on line 9:
Preview Images
Doogle Homepage
Doogle Search - Sites
Doogle Search - Images
Image Preview
Image preview is done using Fancybox.
The title, image URL and site URL are available on the bottom left corner.
Pagination System
Naturally, certain search terms may return many results like 'bbc'.
To which Doogle only displays 20 sites per page. At the bottom of the page, we can view the next 10 pages.
Results Shown
Bottom of Page
Bottom of Page 13
doogleBot Crawl Form
An HTML form to submit a URL for crawling