Ako skrotiť 400-tisíc CVE súborov v Gite bez preťaženia disku

for year in $(seq 1999 2026); do
  files=$(find $year -type f -name "*.json" | wc -l | cut -d' ' -f 1);
  size=$(du -chs $year | head -n 1 | cut  -f 1 );
  echo "| $year | $size | $files  |";
done

git clone --depth 1 \
  https://github.com/CVEProject/cvelistV5.git \
  cvelistv5-depth1.git

git clone --filter=blob:none --no-checkout \
  https://github.com/CVEProject/cvelistV5.git \
  cvelistv5-sparse.git

git sparse-checkout init --cone
git worktree add --detach ../cves-2026 main

( cd ../cvelistv5-sparse.git/ && git fetch origin main ); git checkout origin/main

> cat cvelistv5-sparse.git/.git/worktrees/cves-2026/config.worktree
[core]
sparseCheckout = true
sparseCheckoutCone = true

> cat .git
gitdir: /tmp/cvelistv5-sparse.git/.git/worktrees/cves-2026

#!/usr/bin/env bash
set -euo pipefail

# === CONFIGURATION ===
UPSTREAM_URL="https://github.com/CVEProject/cvelistV5.git"
CORE_DIR="cvelist-core"
# Base set of historical years; current year will be dynamically appended
YEARS_TO_SETUP=(2026)

CURRENT_YEAR=$(date +%Y)

# Add current year if not already in array
if [[ ! " ${YEARS_TO_SETUP[*]} " =~ " ${CURRENT_YEAR} " ]]; then
    YEARS_TO_SETUP+=("$CURRENT_YEAR")
fi

# Helper function to create a worktree for a specific year
ensure_worktree() {
    local YEAR="$1"
    local TARGET_DIR="../cves-$YEAR"

    if [ -d "$TARGET_DIR" ]; then
        return 0
    fi

    echo "--> Creating worktree for year $YEAR in $TARGET_DIR..."
    git worktree add --detach "$TARGET_DIR" main
    (
        cd "$TARGET_DIR"
        git sparse-checkout set "cves/$YEAR"
    )
}

# === FUNCTION: INITIALIZATION ===
do_setup() {
    echo "===> 1. Cloning repository metadata (Partial Clone)..."
    if [ ! -d "$CORE_DIR" ]; then
        git clone --filter=blob:none --no-checkout "$UPSTREAM_URL" "$CORE_DIR"
    else
        echo "Directory $CORE_DIR already exists, skipping clone."
    fi

    cd "$CORE_DIR"
    git sparse-checkout init --cone

    echo "===> 2. Creating sparse-checkout worktrees (Target years: ${YEARS_TO_SETUP[*]})..."
    for YEAR in "${YEARS_TO_SETUP[@]}"; do
        ensure_worktree "$YEAR"
    done

    echo ""
    echo "===> Initialization complete. Created directories:"
    ls -d ../cves-* 2>/dev/null || true
}

# === FUNCTION: SYNCHRONIZATION ===
do_sync() {
    if [ ! -d "$CORE_DIR" ]; then
        echo "Error: Main directory $CORE_DIR does not exist. Run setup first." >&2
        exit 1
    fi

    echo "===> 1. Fetching latest commits from upstream..."
    (
        cd "$CORE_DIR"
        git fetch origin main
    )

    echo "===> 2. Ensuring worktree exists for current year ($CURRENT_YEAR)..."
    (
        cd "$CORE_DIR"
        ensure_worktree "$CURRENT_YEAR"
    )

    echo "===> 3. Updating all created worktree directories..."
    for WORKTREE_DIR in ../cves-*; do
        if [ -d "$WORKTREE_DIR" ]; then
            YEAR=$(basename "$WORKTREE_DIR" | sed 's/cves-//')
            echo "--> Synchronizing year $YEAR ($WORKTREE_DIR)..."
            (
                cd "$WORKTREE_DIR"
                git checkout --detach origin/main
            )
        fi
    done

    echo ""
    echo "===> Synchronization completed successfully!"
}

# === EXECUTION LOGIC ===
COMMAND="${1:-setup}"

case "$COMMAND" in
    setup)
        do_setup
        ;;
    sync)
        do_sync
        ;;
    *)
        echo "Usage: $0 {setup|sync}"
        exit 1
        ;;
esac