post-commit

#!/bin/bash

# This is a sample post commit file for a git repository that wants to automate model training and data pipeline updates.

# Get the list of changed files in the current commit
FILES=$(git show --name-only HEAD | tail -n +2)
COMMIT_HASH=$(git log -1 --format="%H")
COMMIT_MSG=$(git log -1 --format="%B")

# Check if any files are in src/data/
if echo "$FILES" | grep -qE "src/data/(load_raw|process_data)\.py"; then
  echo "Creating new data experiment..."
  $(.venv/bin/python -m scripts.update_data_version --commit_id $COMMIT_HASH --commit_msg "$COMMIT_MSG")
  git add "data-experiments.yaml"
  git commit -m "Updated data version"
  echo "Rerunning data pipeline..."
  $(.venv/bin/python scripts/run_data_pipeline.py --commit_id $COMMIT_HASH)
fi

# Check if any files are in src/models/
if echo "$FILES" | grep -q "^src/models/"; then
  echo "Creating new model experiment..."
  $(.venv/bin/python -m scripts.update_experiment_version --commit_id $COMMIT_HASH --commit_msg "$COMMIT_MSG")
  git add "experiments.yaml" "src/config.py"
  git commit -m "Updated model version"
  echo "Running new model..."
  export XLA_PYTHON_CLIENT_MEM_FRACTION=0.1
  $(.venv/bin/python -m src.training.train)
fi