History: Crawl
Preview of version: 11
Tiki Crawl
A crawler tool for checking links or gathering content from websites.
https://gitlab.com/tikiwiki/tiki-crawl
This is the alpha version of a crawler tool for checking links or gathering content from websites. It relies on the Crawler library from Spatie. Kudo to those guys.
Installation
This piece of code has been tested with php 8.2.6 and node v18.16.0, make sure you have those version installed (but it may very well be that it works on older versions, code is not that complicated).
composer update npm install
Usage
First you need to configure your options. You can override any options from config/config.default.php by creating a config/config.php file where you will be adding new values that will override the one in config/config.default.php, for example:
<?php $options['timeout'] = 600; $options['js'] = true;
> Note: You don't have to change values from config/config.default.php, any configuration needed you can change the value from the created file config/config.php
Then you can launch the crawling
./bin/crawl https://books.toscrape.com
> Note: toscrape.com is useful for testing crawlers
!! Config options
| Name | Type | Default | Description |
|---|---|---|---|
| timeout | integer | 60 | the maximum time that should be waited before getting a response for any page crawled |
| cli | boolean | true | show a progress dotted line while launching it in console |
| log | boolean | true | keeps logs in `logs/` directory (one for access one for errors) |
| store | boolean | false | store crawled pages in `collected-data/` directory |
| max_size | integer | 2 | when `store` is enabled, the maximum size of stored documents, if you want to get pdfs you should up |
| limit | ?integer | 5 | number of pages to crawl, set to `false` for unlimited crawls |
| js | boolean | false | use headless chrome with puppeteer (takes much longer, and requires to have puppeteer installed) |
| internal | boolean | true | only crawl urls that are on the same host |
| concurrency | integer | 10 | number of concurrent requests to make |
| max_depth | ?integer | false | if you want to only get the immediate links on the page you are crawl, set `1`, you can decide how deep you want to crawl from initial page |
| delay | ?integer | false | add a delay in milliseconds between requests |
| skip_urls | ?array | false | if you stumble upon problematic urls that make the crawler crash, you can skip them by listing them in this array |
| allow_redirect | boolean | false | wether or not follow 30x redirects |
| ignore_robots | boolean | false | bypass instructions contained in `robots.txt` file |
| allow_nofollow | boolean | false | bypass the `rel="nofollow"` directive in links |
| user_agent | string | TikiCrawl | the user-agent header sent with http requests |
Those options will be passed to spatie, you can learn more about those at https://github.com/spatie/crawler/
Roadmap
- handle the case where headless browser fails miserably and crawler lib don't catch it
- improve crawling on sites by using the canonical url that may be declared inside the html
- add an option to try gathering or checking http status on images (for finding missing ones)
- publish as a composer package
- add a feature in Tiki to make use of it
Notes
- It relies on the Crawler library from Spatie (Over 5 million downloads).
- It starts off as a tool just for developers, and will eventually be integrated in Tiki, so accessible to power users.
This was used to gather data for an upcoming AI Chatbot we are working on (more news on this later, along with some code).