remove-docker-data-from-git.sh

#!/bin/bash
# Remove docker/data from git tracking (permanent solution)
# This script removes docker/data files from git index without requiring file access

set -e

REPO_DIR="/home/skumar/DaatScience/Sentinel"
cd "$REPO_DIR"

echo "Removing docker/data/ from git tracking..."
echo ""

# Get list of files to remove
echo "Collecting docker/data files from git index..."
git ls-files docker/data/ > /tmp/docker_data_files.txt 2>/dev/null || echo "" > /tmp/docker_data_files.txt

FILE_COUNT=$(wc -l < /tmp/docker_data_files.txt | tr -d ' ')
if [ "$FILE_COUNT" -eq "0" ]; then
    echo "✅ No docker/data files tracked by git. Already clean!"
    exit 0
fi

echo "Found $FILE_COUNT files to remove. Processing in batches..."

# Split into batches to avoid command line length issues
cat /tmp/docker_data_files.txt | split -l 1000 - /tmp/batch_ 2>/dev/null || {
    # If split fails, process all at once
    cat /tmp/docker_data_files.txt | git update-index --force-remove --stdin 2>&1 | grep -v "error\|fatal" || true
    exit 0
}

# Process each batch
BATCH_COUNT=0
for batch in /tmp/batch_*; do
    BATCH_COUNT=$((BATCH_COUNT + 1))
    echo "Processing batch $BATCH_COUNT..."
    cat "$batch" | git update-index --force-remove --stdin 2>&1 | grep -v "error\|fatal" || true
done

# Cleanup temp files
rm -f /tmp/docker_data_files.txt /tmp/batch_* 2>/dev/null || true

# Verify removal
REMAINING=$(git ls-files docker/data/ 2>/dev/null | wc -l || echo "0")
if [ "$REMAINING" -eq "0" ]; then
    echo ""
    echo "✅ Successfully removed all docker/data/ files from git tracking!"
    echo "   You can now commit without permission issues."
else
    echo ""
    echo "⚠️  Warning: $REMAINING files still tracked."
    echo "   You may need to run: sudo chown -R skumar:users docker/data/"
    echo "   Then run this script again."
fi