Skip to content

Latest commit

 

History

History
32 lines (27 loc) · 2.37 KB

README.md

File metadata and controls

32 lines (27 loc) · 2.37 KB
Raw

starplag

Instructions

  1. Extract the solutions using the following structure: .../task/group/user/sub.cpp
    • From terry you can use util/extract_solutions.py folder/with/zips/ target/folder/ --all --source-only --ignore0 (from algorithm-ninja/terry)
    • From cms you can use util/from_cms.py folder/with/solutions/ target/folder/ --ignore0 (you can use --same-school to check only between solutions of the same school)
    • Ignoring the solutions that score 0 will make much cleaner results
  2. Prepare a folder with the templates given to the contestants, grouped by task (e.g. .../task/template1.cpp).
    • You can use util/get_templates.sh round_folder for CMS.
  3. Put the ranking of the contest in a text file:
    • Each line must contain a single word: the id of the user, the same as step 1
    • The ranking must be sorted from the top ranked
  4. Compile starplag by issuing make
  5. Run ./build/main path/to/task/ path/to/templates/of/task/ path/to/ranking.txt cutoff path/to/target/folder/
    • The cutoff should be the limit of the ranking where you are interested in finding plagiarism (e.g. top 200): it only considers pairs of subs where at least one is from the top cutoff users
    • It only checks pairs of users within the same group
    • path/to/target/folder/ will contain the results of the execution as well as the snapshots of the computation used in case of crash
  6. After the execution ends a file named total is created inside the target folder
    • The first line contains the number of processed user, the number of matches H found before the cutoff (limited to 500) and the number of matches L found after the cutoff (limited to 500)
    • The next H lines contain the match information for the top part of the ranking
    • The next L lines contain the match information for the rest of the ranking
  7. To ease the manual checking of those matches you can use ./manual_check.py path/to/total path/to/cache
    • Where the first parameter is the path to the total file generated by the previous step
    • The second parameter is a cache file to save partial results
    • The script will create a file output.tsv with the list of copied solutions

Dependencies

For the main tool you need a C++17 compiler with std::filesystem support.

For manual_check.py you need python3, tmux and you have to run make before using it.