Hacker News: murzynalbinos

New comment by murzynalbinos in "Your ePub Is fine"

murzynalbinos — Mon, 15 Jun 2026 16:13:24 +0000

epubcheck doing its job perfectly while Adobe's ancient RMSDK (frozen somewhere around 2013) silently chokes on valid CSS4 like min() is peak digital publishing pain. The fact that Kobo routes normal .epub to the Adobe engine and only .kepub.epub gets the modern WebKit one feels almost malicious.

Show HN: Csv.repair – Free browser-based tool to fix broken CSV files

murzynalbinos — Fri, 27 Feb 2026 16:00:48 +0000

I built csv.repair a free, open source browser-based tool for analyzing, querying, and repairing broken or oversized CSV files. No file uploads - everything runs locally in your browser. Your data never leaves your machine.

What it does:

- Handles CSV files with millions of rows (virtual scrolling + Web Workers)

- Inline cell editing (double-click to edit, Tab to move between cells)

- Run SQL queries directly on your data (powered by AlaSQL)

- Auto-repair: trim whitespace, remove empty/duplicate rows, standardize dates, fix encoding

- Health diagnostics that flag malformed rows, inconsistent columns, encoding problems

- Column statistics and distribution charts

- Undo/redo, search & replace, keyboard shortcuts

- Export the cleaned file when you're done

- Tech stack: React + TypeScript + Vite + PapaParse + AlaSQL + Recharts + Tailwind CSS. Installable as a PWA.

Live: https://csv.repair

I'd love feedback especially from people who regularly deal with messy CSV files.

Comments URL: https://news.ycombinator.com/item?id=47182061

Points: 3

# Comments: 0

Show HN: Log Voyager 2.0 – Analyze 10GB+ log files in the browser

murzynalbinos — Tue, 24 Feb 2026 18:11:06 +0000

I posted the first version here a few months ago, but today I've got a massive update, that I wanted to share... So basically Log Voyager is a client-side log analyzer that can open multi-gigabyte log files (10GB+) in the browser without crashing, using the File Slicing API to stream chunks rather than loading everything into memory.

Technical approach: Instead of reading the entire file into RAM (which crashes browsers on 2GB+ files), LV use File.prototype.slice() to read 50KB chunks via FileReader. This keeps memory usage constant at ~10MB regardless of file size. For GZIP files I use the native DecompressionStream API.

What's new in v2.0:

- Command Palette (Ctrl+K) for keyboard-driven navigation

- Multi-Filter: apply multiple include/exclude filters simultaneously with regex support

- Error Aggregation: groups similar errors by pattern, shows occurrence counts

- Split View: two independent panels for comparing different file sections

- Heatmap Navigation: visual error density on minimap (red=error, orange=warn)

- Log Statistics: real-time level distribution with bar charts

- Export Selection: export specific lines as TXT/JSON/CSV

- Dark/Light theme toggle

- Full keyboard shortcut system (Vim-style navigation for search, bookmarks, etc.)

Thanks for checking it out (and very appreciate feedback)!

Code: https://github.com/hsr88/log-voyager

Comments URL: https://news.ycombinator.com/item?id=47140465

Points: 1

# Comments: 0

Show HN: ConsentScope – detect cookies loaded before user consent

murzynalbinos — Thu, 12 Feb 2026 17:38:36 +0000

So I recently ran a small experiment.

I tested 20 websites (SaaS, e-commerce, agencies, blogs) to check something simple:

Do they actually wait for user consent before loading analytics?

14 out of 20 loaded Google Analytics or other tracking scripts before the user clicked “Accept”.

In most cases:the cookie banner was visible, no interaction had happened yet but GA/GTM requests were already firing. You don’t really notice this unless you’re watching the Network tab closely.

So I built a small tool to monitor:

cookies being set

localStorage/sessionStorage usage

tracking-related network requests

the exact timestamp of the consent click

Originally it was just a script for internal audits. It grew into a Chrome extension called ConsentScope.

What it does:

-compares before vs after consent

-classifies cookies (necessary / analytics / marketing)

-shows a simple timeline of events

-optionally exports a report

It doesn’t auto-click banners. It doesn’t interpret the law. It just shows what technically happens.

I’m curious if others have checked this on their own sites. Feel free to ask anything technical — happy to share details.

Comments URL: https://news.ycombinator.com/item?id=46992058

Points: 1

# Comments: 0

New comment by murzynalbinos in "Show HN: I built Divvy to split restaurant bills from a photo"

murzynalbinos — Sun, 08 Feb 2026 09:34:39 +0000

Good point. But I like the minimalism of this website, it's so charming ;)

New comment by murzynalbinos in "Ask HN: Share your personal website"

murzynalbinos — Sun, 18 Jan 2026 10:22:31 +0000

https://hsr.in.net/ - My personal website

https://shrt.surf/ - and my latest project, a minimalistic URL shortener that doesn't track you, doesn't show ads, and doesn't ask for your email. Just pure utility.

New comment by murzynalbinos in "Show HN: Log Voyager – View 10GB+ log files in browser without crashing RAM"

murzynalbinos — Sun, 04 Jan 2026 11:55:47 +0000

Currently, the highlighter uses a simple keyword heuristic (line.includes('error')). It produces false positives (e.g., "no error found"). I plan to add regex-based custom highlighting to fix this. I treat the file like a video stream. Since the browser doesn't know the total line count of a 10GB file without reading it all (which defeats the purpose), a traditional scrollbar is hard to implement perfectly. The slider acts as a "seek bar". I'll consider adding Prev/Next Chunk buttons for finer control. The "~20MB" claim refers to the file buffer/chunk size held in memory by the FileReader, not the entire browser process footprint (React + DOM + V8 overhead). You are right that the baseline for a React app is higher (60MB+), but the key value proposition is that opening a 10GB file won't spike usage to 10GB—it stays flat relative to the baseline.

I appreciate you testing it out!

Show HN: Log Voyager – View 10GB+ log files in browser without crashing RAM

murzynalbinos — Sun, 04 Jan 2026 09:58:03 +0000

Key features: - Reads 10GB+ files instantly (constant memory usage ~20MB). - 100% Local Execution (Sandbox): No data is ever uploaded to any server. Works offline. - JSON Prettifier: Detects JSON lines and formats them on click. - "Warp Jump" Bookmarks: Save byte-offset positions to jump between lines gigabytes apart.

Comments URL: https://news.ycombinator.com/item?id=46486542

Points: 4

# Comments: 2