Updates Aug 26

PGrid

• Added a new worker type that will only scrape amazon and ebay. Feel like I should update prices for those two sites more frequently. This worker will run every 4 hours.
• There was a bug with the ebay scraper when there were coupons that was only for ebay plus users. Fixed it
• fixed the main page table prices. it was still showing cheapest prices without taking into account out of stock.
• saw that pcpartpicker used a site called commission factory for affiliate links with mwave and applied to join. was rejected
• posted on a aus pc deal subreddit. was quite good response which is nice to see. It ended up being the top post of the year on that subreddit. need to use that as motivation to ship fast. one of the comments led me to find incorrect memory put on one of the gpu chips as well as a few gpu cards incorrectly being classified.
• first ebay sale! due to the date being the day before i posted on reddit i suspect the sale was from an email alert but not 100% sure. need to set up better tracking
• added a check for gpu cards that were manually verified without ai which resulted in them being ignored and not listed.
• updated ple to source prices from listing page rather than individual pages. found an annoying issue where in the listing pages the url to a specific gpu is one url but when you visit that url you get redirected to another url. which since my scraper now doesn’t visit individual urls, it assumes that old url no longer exists. added a hacky fix for that
• noticed incorrect gpu classifications browsing the site and had to do a couple of manual fixes into the db
• started tracking events using the analytics app im using called umami. currently tracking external gpu listing clicks and price alert button
• added a db health check using uptimemonitor.com and a new route
• wrote a manual script to update amazon urls to just the url + amazon code as changing amazon titles also changed urls which resulted in duplicate urls for the same listing. removed 4 duplicate urls. forgot to update unverified urls which resulted in duplicate urls for all amazon items the next day when scraping was done. had to do more scripting to fix
• new article
• missed out on posting to ozbargain a good deal from centrecom due to the url sitting in unverified queue and another url previously marked as as failed. after the deal was over, centrecom made the urls as 404s and the cards were marked as failed. this caused the price history graph to ignore those cards which is incorrect so updated the price history graph to include failed cards. Also added logic to make cards that were previously marked as failed active again if they showed up again.
• found a deal on ozbargain thats not a gpu where amazon has a coupon that you can just click to apply. updated amazon scraper to look for coupons
• updates to logic of sending price alert emails
• august total revenue: $53.16 from amazon + $35 from ebay. = $88.16. time to grow

Homeless helper app

• Initial project scaffolded using nextjs