Pulls backups from all repositories in BitBucket only if changed since last run

bitbucket-backup.sh

#!/bin/bash
# 
#  backup.sh
#  
#  Created by David Alger on 2015-03-13
#  Copyright 2015 David Alger. All rights reserved.
# 

##################################################
## Initialize our environment

set -o pipefail;
set -o nounset;
set -o errexit;

# source credential file if present
if [[ -f ~/.bitbucket_backup ]]; then
    source ~/.bitbucket_backup
fi

BASE_DIR="$(dirname $0)";
LOG_FILE=""; # this is to prevent an unset error... uncomment the following line to send output to a log file.
# LOG_FILE="$BASE_DIR/activity.log";

PHP_BIN=/usr/bin/php
API_ENDPOINT="https://bitbucket.org/api/1.0"

BAK_DEST=/server/backups/vcsdata/bitbucket-archive

##################################################
## Initialize our functions...

function log () {
    if [[ "$1" = "CLEAR" ]]; then
        if [[ ! "$LOG_FILE" = "" ]]; then
            echo > "$LOG_FILE";
        fi
        return;
    fi
    if [[ "$(echo "$1" | tr -d [:blank:])" = "" && "$1" != "" ]]; then
        return;
    fi
    if [[ "$LOG_FILE" = "" ]]; then
        echo -e "$1";
    else
        echo -e "$1" >> "$LOG_FILE";
    fi
    return;
}

function assert_failure () {
    echo "command failed with status of: $?";
    exit $?;
}

function pre_run {
    # log start time and header
    log "############################################################
## START (backup.sh): $(date)
##
";

    # creating directories
    log "## creating $BAK_DEST directory";
    log "$(mkdir -p "$BAK_DEST")"
}

function post_run {
    #log finish time and footer
    log "
##
## FINISH (backup.sh): `date`
############################################################
";
}

function hunt_repositories {
    log "## hunting for repositories...";
    REPO_LIST="$(curl -s --user "$BB_USER:$BB_PASS" "$API_ENDPOINT/user/repositories/")"
    REPO_LIST="$(echo "$REPO_LIST" | $PHP_BIN -r '
        $data = json_decode(file_get_contents("php://stdin"), true);
        foreach ($data as $repo) {
            echo "{$repo["owner"]}/{$repo["slug"]}/".str_replace(["+00:00", " ", ":"], ["", "_", ""], $repo["utc_last_updated"])."\n";
        }
    ')"

    log "   found $(echo "$REPO_LIST" | wc -l | tr -d [:blank:]) repositories"
    log ""
}

function loop_repositories {
    log "## iterating repositories for backup"
    log ""

    for REPO_INFO in $REPO_LIST; do
        REPO_ACCT=$(echo "$REPO_INFO" | cut -d / -f1)
        REPO_NAME=$(echo "$REPO_INFO" | cut -d / -f2)
        REPO_MODT=$(echo "$REPO_INFO" | cut -d / -f3)
        REPO_DEST="$REPO_ACCT/$REPO_NAME.git"
        REPO_TAR="$REPO_ACCT/${REPO_MODT}_${REPO_ACCT}_${REPO_NAME}.git.tar"

        if [[ -f "$BAK_DEST/$REPO_TAR" ]]; then
            log " # skipping $REPO_ACCT/$REPO_NAME\n"
            log "   utc_last_mod: $REPO_MODT"
            log "   backup_date: $(ls -l "$BAK_DEST/$REPO_TAR" | cut -d' ' -f9-11)"
            log ""
        else
            backup_repository
            # TODO: cleanup archives with previous modification date in filename
        fi
    done
}

function backup_repository {
    log " # processing $REPO_ACCT/$REPO_NAME\n"
    log "   utc_last_mod: $REPO_MODT"
    log "   bare cloning to $REPO_DEST"
    log "   $(git clone --quiet --bare "[email protected]:$REPO_ACCT/$REPO_NAME.git" "$BAK_DEST/$REPO_DEST" 2>&1)"
    
    log "   creating tar archive at $REPO_TAR"
    log "   $(tar -cf "$BAK_DEST/$REPO_TAR" -C "$BAK_DEST" "$REPO_DEST")"
    log "    - archive size: $(ls -lh "$BAK_DEST/$REPO_TAR" | awk '{print $5}')"
    
    log "   cleaning up bare clone"
    log "   $(rm -rf "$BAK_DEST/$REPO_DEST")"
    
    log ""
}

##################################################
## Begin the backup sequence...

function main {
    pre_run
    hunt_repositories
    loop_repositories
    post_run
}

main