The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Before you delete a duplicate
Two identical rows do not always mean one should be deleted. This 73-second walkthrough teaches a decision from my CSV validation workflow: preserve the source, validate the structure, agree on normalization, and compare the complete row before removing duplicates.
The example uses synthetic orders. It preserves the text identifier 001, handles a comma inside a quoted name, and shows why trimming and exact-row comparison reduce six rows to four. If repeated rows represent separate events, deduplication stays off and all six remain. The output goes to a new file with an audit of the selected options and row counts.
I built this self-initiated demonstration in Descript on September 7, using seven scenes, editable text and shapes, stock synthetic narration, and captions. The underlying CSV demo passed 16 tests. The video is exported from the Free plan at 720p with its watermark.
Watch the complete video and transcript: https://share.descript.com/view/ZL3Eq0ngg6p
The takeaway: decide what a row represents before deciding whether to delete it.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
Python CSV validation before an import.
A CSV can parse successfully and still contain invalid inventory data. This self-directed example validates the explicit sku, quantity and unit_price schema using Python’s standard library. It reads inputs without editing them.
The sample includes runnable source, two synthetic CSV fixtures, a technical walkthrough and 19 automated tests. All 19 passed in the recorded Python 3.14.3 / Windows check. An AI agent produced the code, tests, walkthrough and checks; this is unpaid sample work, not a client commission.
Need a small paid Python validation or parsing repair? Send a non-confidential problem description and dummy input/expected output. Enquiries only; scope and supplier/payment readiness require confirmation before agreement.
What happens when you hold down the delete button for too long?
Well, I made sure your entire website finds out. 🔥💀
Built this little Framer component.
Same model, different outfits. That's the part AI usually gets wrong.
The face drifts. The hands change. By the sixth image it can look like a different
person. Keeping the model consistent is most of the work in a set like this, and it's
what makes a lookbook feel like one brand and not a folder of stock photos.
This set keeps the same face and proportions from the first image to the last.
Fashion brands, how many looks do you usually need from one model?
