Skip to main content

tilearn.data APIs

The tilearn.data module contains low-level file-system helpers used by multi-list scheduling pipelines.

Header expectation​

CSV files are expected to use:

Name,p,r,d,w

tilearn.data.backup(src, dist)​

Recursively copy all *.csv files from src into dist.

tilearn.data.path(original_path, backup_path)​

Return join(backup_path, basename(original_path)).

tilearn.data.read_file(file_path)​

Load CSV rows and coerce types:

  • Name -> str
  • p, w -> float
  • r, d -> int

tilearn.data.precedence(file_path, opt_row)​

Return rows after dropping indices <= opt_row.

tilearn.data.none(file_path, opt_row)​

Return rows after dropping exactly opt_row.

tilearn.data.updated(file_path, prec, opt_row)​

Rewrite a CSV in-place after selection:

  • prec=1: precedence mode (precedence)
  • prec=0: non-precedence mode (none)

tilearn.data.clear(folder_path)​

Delete top-level *.csv files in folder_path.

Side effects

backup, updated, and clear modify the file system directly. Use a dedicated temporary backup directory when running end-to-end workflows.

Example​

from tilearn import data

data.backup("data/source", "data/backup")
rows = data.read_file("data/backup/list1.csv")
data.updated("data/backup/list1.csv", prec=0, opt_row=0)
data.clear("data/backup")
On this page