I archived OI tasks, editorials, and test data (+ grader & result) because official OI websites shut down often and valuable content disappears. E.g. APIO, BalkanOI, BalticOI & CEOI official websites.

Even when problems are mirrored on online judges, the full package is often missing. Many OJs only host the statement and a way to submit, but not the editorial, grader or results. That creates a real risk of permanent loss if those platforms ever go down as well (e.g., dunjudge.me).

The Wayback Machine helps but it’s not reliable as its often skip testdata ZIP files & in rare cases entire websites. Also, finding & downloading is very time consuming.

That’s why I built OI-TED-ARCHIVE — a project dedicated to preserving complete OI content in one place.

Thanks to DMOJ.ca, LOJ.ac, CSES.fi, and many more (some anonymous) for helping me build this archive.

Even I didn’t expect to recover more lost OI material than most online judges combined — especially editorials and official results, which are usually the first things to vanish.

And just to be clear: no disrespect to OJs at all. They gave me tons of valuable resources. Honestly, they’re a big reason why OI-TED-ARCHIVE could become this complete.

The difference is: I didn’t stop at common sources. I executed recovery for each and every problem individually.

General Thumb Of Rule: "If it ever touched the internet, it’s probably still somewhere — you just haven’t searched deep enough."

My Methods

Additional Sources which i may have forgotten to mention in sheet

• Google Search + Google Hacking
• Codeforces Comments: Yes!, The community helped me.
  • Below the Announcement / Discussion Blog: many author/participants share drive/dropbox links
• The Chinese Website
  • I couldn't gotten access to csdn.net because they were locked behind Login Screen (Phone Number Verification). I am looking for a friend from CHINA to share me account access or download stuff for me.
  • Thanks to Chinese platform for preserving massive OI history.
• National/private hosting sites: Countries like Russia, Bulgaria, etc sometimes keep resources on there platforms.
• Emailing and DMing authors/organisers
• Common Ones
  • Github Repos
  • Github Projects like OI Result.
  • Internet Archive: Wayback Machine
  • Official OI Pages: currently active on web or preserved in wayback.

Google Hacking/Dorking is just a fancy name for google advance search 🤣

Fully recovered but low confidence due to scattered sources

Fully Recovered With Confidence

Minor Patch: removed dot (.) as prefix for one of file name as they were skipped by qBitorrent (Linux).

[CONFIRMED] Moving to .tar.zst level 19 (previously .tar.xz). Why? Faster + Similar to TAR.XZ

• BOI 2026 ✓ (With boi-result.tar.zst updated)

Upcoming Version v1.1.0

RMI (Romanian Master of Informatics) from 2017 till 2025 will be added in next update.

• I made a little bit progress for BalkanOI but still not enough to publish :<

Call for HELP

• I am missing following resources:
• RMI 2022: (Testdata, Editorial) aka BalkanOI 2022 testdata and editorial.

• Option 1: Internet Archive Direct Download
• Internet Archive: CHECK ARCHIVE SHEET (Links mention in above sections)

It’s done mainly to reduce size. In most cases, this compression allows you to download an entire year’s worth of material in the space that a single uncompressed testdata package would otherwise take.

Why not zip? Using TAR.XZ/TAR.ZST as ZIP is very space inefficient. (Atleast 30%, Generally 50%-70%)

• Uncompressed Data: 3.1GB
• ZST level 3: 1.2GB (1 min to compress)
• ZST level 19: 0.89GB (10 min to compress) (2-3% larger than XZ)
• XZ (-cJF): 0.86GB ( 30 min to compress)

Newer archives will be in ZST level 19 format.

• On Windows, You can use 7-zip / PeaZip
• On Linux, KDE Ark / Gnome File Roller :)

Integrity Check

• ZIP integrity checks are performed on recovered files from various sources before converting to .tar.xz and uploading, to ensure nothing is corrupted or incomplete.

Checksum Checked

• Checksum (MD5 & SHA1) are verified for uploaded files to be sure that files are successfully uploaded.

OI's Results

Results are preserved primarily in PDF format, or as a single plain HTML file when available.

Complete HTML webpages are avoided unless absolutely necessary, since they can easily break in the future due to missing assets or changing browser support.

When they are used, I keep a simplified backup as well — usually a custom Excel summary exported to PDF — to ensure the results and scores remain accessible even if the webpage format fails.

• IZhO: available on codeforces.com, qoj.ac, oj.uz. I made a significant progress on testdata.
• BalkanOI (till 2018): available on acmicpc.net (https://www.acmicpc.net/category/94). (unable to make any significant progress)
• PolishOI: available on szkopul.edu.pl (https://szkopul.edu.pl/task_archive/oi/). (many failed attempts to scrape english version.)