Streamlit MigrAItion - Making cool Webs from old ones
For the Real Estate and few Wordpress projects, my clients asked me to migrate their existing content.
It’s always great to having a look to the initial pages to know what to expect from them.
Is it everything working?
Hows the initial Performance of the Site?
Checking that a Website works - or in this case, what we will have to Migrate 📌
- WebCheck -
- Gte all url/pages of a website - A must to migrate everything!
- Are there any broken links?
#podman run --rm -it ghcr.io/linkchecker/linkchecker:latest --verbose https://fossengineer.com > linkchecker_output.txt
docker run --rm -it -u $(id -u):$(id -g) ghcr.io/linkchecker/linkchecker:latest --verbose https://www.jmodels.net
pip3 install linkcheckerCheck SiteMap/Robots.txt of a Web 📌
#curl -s https://example.com/sitemap.xml -o /dev/null -w "%{http_code}\n"
curl -s https://jmodels.net/sitemap.xml -o /dev/null -w "%{http_code}\n" #hugo paper mod has it
#optional - check robots.txt
curl -s https://jmodels.net/robots.txt | grep -i sitemap #look for sitemap direction
curl -s https://jmodels.net/robots.txt | head -n 10 #see the first 10 linesI was expecting to just use the latest AI Scrapping tools that I learnt here…
…but Python+BS4 can do very interesting tricks.
Lets have a look.
For the Real Estate Project
Getting All the Images from the Web to Migrate [BS4] 📌
You can do the following:
- Explore the website content
- Find the html section where all the photo links are stored
- Use LLMs so that they give you a Python Code to: Find the photo links + Download them in a folder
- Repeat for every link you have (leverage the previous functions)
- [Optional] Include the folder to the
.gitignore
This is what I made with this sample code here, just using BS4
A Wordpress Migration
Inspect SiteMap with Python
I created this Python Script to Inspect (and save) the SiteMaps
Use AI Generated Script to get all URLs
I used chatgpt to create the following…
Another Py scripts that saves as .csv all the different urls Links as per the SiteMap
For the future, you can do these 2 steps with ONE SCRIPT
Always good to check if there are broken links referenced in each original posts - For correction.
Scrap & Get all info (as it is of one Link)
For Scrapping, there are few options as we saw: Crawl4AI, ScrapeGraph, FireCrawlAI…
But we can have a custom tool with BS4.
Leverage Time - Get info…of ALL Links
In this case, we had to go through ~1k different urls.
That was a good reason to go with a HUGO Theme as well. Fuwari Astro was also an alternative.
Big thanks yo CaiJimmy/hugo-theme-stack-starter.
Each post has 2 different languages.
Custom BS4 Tool
Making sure that all info is there by knowing the web categories
Category 1 Posts
Example - De hombres y maquinas https://jmodels.net/2020/08/10/de-hombres-maquinas-554-ilyushin-db-3/
- See how the html is structured
- In this case, they are under
<div class="content clear fleft" id="content">
- Create a BS4 scrpt to get the content
- Here, it redirects to this link - https://jmodels.net/de-hombres-y-maquinas/aire-air/ilyushin-db-3/
- It was not flagged in the sitexml, but checking the robots…there is another sitemap
https://jmodels.net/news-sitemap.xml
curl -s https://jmodels.net/robots.txt | grep -i sitemap #look for sitemap direction- And its also not there…TBC!!!
- Inspecting a URL with content - https://jmodels.net/de-hombres-y-maquinas/aire-air/ilyushin-db-3/
- Inspecting it, we have content at
<div class="post-content clear">
- Plug the scrapped content to an LLM to get the markdown, while preserving the initial structure
- This script takes the info from the given div, and feeds it to OpenAI API to get
.md
- If you already have a theme selected, you ar every close to have one post migrated