History: Crawl

• «
• »

            ! 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 [https://github.com/spatie/crawler|Crawler library from Spatie]. Kudo to those guys.


! Installation
This piece of code has been tested with -+~np~php 8.2.6~/np~+- and -+~np~node v18.16.0~/np~+-, make sure you have those version installed (but it may very well be that it works on older versions, code is not that complicated).


{CODE()}composer update
npm install
{CODE}


! Usage
First you need to configure your options. You can override any options from -+~np~config/config.default.php~/np~+- by creating a -+~np~config/config.php~/np~+- file where you will be adding new values that will override the one in -+~np~config/config.default.php~/np~+-, for example:

{CODE()}<?php
$options['timeout'] = 600;
$options['js'] = true;
{CODE}

> Note: You don't have to change values from -+~np~config/config.default.php~/np~+-.


Then you can launch the crawling with the command -+~np~./bin/crawl~/np~+- following with the link of the website that you need to crawl. like the bellow command:

{CODE()}./bin/crawl https://books.toscrape.com{CODE}

> Note: -+~np~toscrape.com~/np~+- is useful for testing crawler, you can replace it with the link of any website that you need to crawl



!!! Config options
This are configuration options that you can override in you new -+~np~config/config.php~/np~+- file

{FANCYTABLE(head="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
{FANCYTABLE}

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 [https://packagist.org/packages/spatie/crawler|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).