website-scripts/Git/Migrate/github-gitea-migration.sh

#!/bin/bash

#clear the screen
clear

echo "This script will mirror a GitHub user's public repositories into gitea and assign them to a specific organiztion within Gitea."
echo "Before starting, you will need to create:"
echo "#1) A token with the appropriate rights (at least write:repository)"
echo "#2) An organization within Gitea to place all mirrored repositories into"
echo "If you enter a GitHub API token, you will be able to mirror non-public projects. This is optional!"
echo ""
echo "This application supports using a config file for user input variables"

#Config file:
CONFIG_FILE="migration_config.conf"

# Migration state file:
STATE_FILE="migration_state.txt"

#Check for config file. If none, ask user for input.
if [[ -f $CONFIG_FILE ]]; then
    source $CONFIG_FILE
	echo "Config file found! Continuing without asking for user input."
else
    echo "Configuration file not found! Using manual input..."

    ### BEGIN User input ###
    echo -n "Enter the GitHub username to migrate: "
    read GITHUB_USERNAME

    echo -n "Enter your GitHub Access Token (optional&hidden): "
    read -s GITHUB_TOKEN
    echo

    echo -n "Enter your Gitea username: "
    read GITEA_USER

    echo -n "Enter your Gitea token (input will be hidden): "
    read -s GITEA_TOKEN
    echo

    echo -n "Enter your Gitea URL (default: https://code.flatironnetworks.com): "
    read GITEA_URL
    GITEA_URL=${GITEA_URL:-https://code.flatironnetworks.com}

    echo -n "Enter the Gitea organization to migrate into: "
    read GITEA_ORG
    ### END User input ###
fi

# Check GitHub rate limit
check_rate_limit() {
    local headers
    local remaining

    if [ -z "$GITHUB_TOKEN" ]; then
        headers=$(curl -sI "https://api.github.com/users/$USERNAME/repos")
    else
        headers=$(curl -sI -H "Authorization: token $GITHUB_TOKEN" "https://api.github.com/users/$USERNAME/repos")
    fi
    
    remaining=$(echo "$headers" | grep X-RateLimit-Remaining: | awk '{print $2}')

    # If the rate limit headers are not present, assume we can continue without delay.
    if [ -z "$remaining" ]; then
        return
    fi

    while [ "$remaining" -le 5 ]; do
        echo "Rate limited by GitHub. Waiting for 10 minutes..."
        sleep 600

        if [ -z "$GITHUB_TOKEN" ]; then
            remaining=$(curl -sI "https://api.github.com/users/$USERNAME/repos" | grep X-RateLimit-Remaining: | awk '{print $2}')
        else
            remaining=$(curl -sI -H "Authorization: token $GITHUB_TOKEN" "https://api.github.com/users/$USERNAME/repos" | grep X-RateLimit-Remaining: | awk '{print $2}')
        fi
    done
}

# Check if Gitea UID is valid
USER_ID=$(curl -s "$GITEA_URL/api/v1/users/$GITEA_USER" -H "authorization: token $GITEA_TOKEN" | jq '.id')
if [ -z "$USER_ID" ] || [ "$USER_ID" == "null" ]; then
    echo "Failed to retrieve UID for $GITEA_USER. Please check your Gitea token and username."
    exit 1
fi

# If the state file exists, read the last page processed
if [ -f "$STATE_FILE" ]; then
    page=$(cat $STATE_FILE)
else
    page=1
fi

# Fetch repositories from GitHub
fetch_github_repos() {
    local page=$1
    if [ -z "$GITHUB_TOKEN" ]; then
        echo $(curl -s "https://api.github.com/users/$USERNAME/repos?per_page=100&page=$page")
    else
        echo $(curl -s "https://api.github.com/user/repos?visibility=all&per_page=100&page=$page" -H "Authorization: token $GITHUB_TOKEN")
    fi
}

#Main script:
while : ; do
    check_rate_limit

    repos_json=$(fetch_github_repos $page)
    repos=$(echo "$repos_json" | jq -r '.[].clone_url')

    if [ -z "$GITHUB_TOKEN" ]; then
        repos=$(curl -s "https://api.github.com/users/$USERNAME/repos?per_page=100&page=$page" | jq -r '.[].clone_url')
    else
        repos=$(curl -s "https://api.github.com/users/$USERNAME/repos?per_page=100&page=$page" -H "Authorization: token $GITHUB_TOKEN" | jq -r '.[].clone_url')
    fi

    if [ -z "$repos" ]; then
        echo "Migration complete!"
        rm $STATE_FILE
        break
    fi

    for repo_url in $repos; do
    repo_name=$(echo $repo_url | awk -F'/' '{print $NF}' | sed 's/.git$//')

    # Check if repository is private and modify the clone URL accordingly
    github_response=$(curl -s "https://api.github.com/repos/$USERNAME/$repo_name" -H "Authorization: token $GITHUB_TOKEN")
    is_private=$(echo "$github_response" | jq '.private')
    if [ "$is_private" == "true" ]; then
        repo_url="https://$GITHUB_TOKEN@github.com/$USERNAME/$repo_name.git"
    fi

    echo "Transferring repository: $repo_name..."

    response=$(curl -X POST "$GITEA_URL/api/v1/repos/migrate" \
             -H "accept: application/json" \
             -H "authorization: token $GITEA_TOKEN" \
             -H "Content-Type: application/json" \
             -d '{
                   "clone_addr": "'$repo_url'",
                   "mirror": true,
                   "private": false,
                   "repo_name": "'$repo_name'",
                   "uid": '$USER_ID',
                   "repo_owner": "'$REPO_OWNER'"
                 }' -w "%{http_code}" -s -o /tmp/gitea_response.txt)

    if [ "$response" == "409" ]; then
        echo "Repository $repo_name already exists in $REPO_OWNER. Skipping."
    elif [ "$response" != "201" ]; then
        echo "Error migrating $repo_name. HTTP status: $response"
        cat /tmp/gitea_response.txt
    fi
done

    echo $page > $STATE_FILE
    page=$((page + 1))
done