Skip to content
Extraits de code Groupes Projets
install.sh 12,5 ko
Newer Older
  • Learn to ignore specific revisions
  • : '
    see https://github.com/jamiew/git-friendly for more ideas
    
    maybe this should be two files
    one which includes cloning diaspora/diaspora, and one that assumes you already cloned it yourself
    maybe one script just calls another?
    
    other ideas what we could do
    
     1. check that you have ruby installed, if not, point to wiki page and exit
     2. check to see if we need sudo (generally, if it is a system ruby you need sudo, which you can check
        if which ruby is /usr/bin/ruby, or does not have rvm in the path)
     3. check if you have bundle installed and install it, and install with/without sudo if you need it
    
     check if you have mysql and/or postgres installed, point to wiki page if neither is found.
     (maybe even switch database.yml if this is the case?)
    
     make it work if you have just cloned diapsora and want a quick setup, or
     support magic install, like this http://docs.meteor.com/#quickstart
    '
    
    ####                                                             ####
    #                                                                   #
    #                           DEFAULT VARS                            #
    #                                                                   #
    ####                                                             ####
    
    
    # required programs
    declare -A BINARIES
    BINARIES["git"]="git"
    BINARIES["ruby"]="ruby"
    BINARIES["rubygems"]="gem"
    BINARIES["bundler"]="bundle"
    BINARIES["sed"]="sed"
    BINARIES["mktemp"]="mktemp"
    
    
    D_GIT_CLONE_PATH="/srv/diaspora"     # path for diaspora
    
    
    N.S.Devaraj's avatar
    N.S.Devaraj a validé
    D_REMOTE_REPO_URL="https://github.com/diaspora/diaspora.git"
    
    D_INSTALL_SCRIPT_URL="https://raw.github.com/diaspora/diaspora/master/script/install.sh"
    
    
    D_WIKI_URL="https://github.com/diaspora/diaspora/wiki"
    
    D_IRC_URL="irc://freenode.net/diaspora"
    
    
    D_DB="mysql"
    
    D_DB_CONFIG_FILE="config/database.yml"
    
    D_DB_HOST="localhost"
    
    D_DB_USER="diaspora"
    
    D_DB_PASS="diaspora"
    
    # TODO: read this from ./script/env/ruby_env
    D_RUBY_VERSION="1.9.3-p194"
    
    
    ####                        INTERNAL VARS                        ####
    
    RVM_DETECTED=false
    JS_RUNTIME_DETECTED=false
    
    ####                                                             ####
    #                                                                   #
    #                          FUNCTIONS, etc.                          #
    #                                                                   #
    ####                                                             ####
    
    #... could be put in a separate file and sourced here
    
    # heredoc for variables - very readable, http://stackoverflow.com/a/8088167
    
    # define VAR <<'EOF'
    # somecontent
    # EOF
    define(){ IFS='\n' read -r -d '' ${1}; }
    
    
    # add padding to the left of a given string to
    # fill to a given amount of characters with a 
    # given char or space
    # example:
    #   lpad 7 "test" "-"
    lpad() {
      LEN=$1
      TXT=$2
      CHR=$3
      PAD=""
    
      L_PAD=$(($LEN - ${#TXT}))
      if [ $L_PAD -ne 0 ] ; then
        PAD=$(printf "%*s" ${L_PAD} " ")
      fi
      if [ ${#CHR} -ne 0 ] ; then
        PAD=$(printf "$PAD" | tr " " "$CHR")
      fi
      PAD="${PAD}${TXT}"
    
      printf "%s" "$PAD"
    }
    
    # log function
    # prints a given message with the given log level to STDOUT
    logf() {
      MSG=$1
      LVL=$2
      L_LEN=7
    
      if [ ${#LVL} -ne 0 ] ; then
        LVL="[$(lpad $(($L_LEN-2)) $LVL " ")]"
      else
        LVL=$(lpad $L_LEN "" "-")
      fi
    
      printf "%s -- %s\\n" "$LVL" "$MSG"
    }
    
    # short functions for various log levels
    log_err() {
      logf "$1" "error"
    }
    
    
    log_wrn() {
      logf "$1" "warn"
    }
    
    
    log_dbg() {
      logf "$1" "debug"
    }
    
    log_inf() {
      logf "$1" "info"
    }
    
    
    # run a command or print the error
    run_or_error() {
      eval "$1"
      if [ $? -ne 0 ]; then
        error "executing '$1' failed."
      fi
    }
    
    
    # nicely output error messages and quit
    error() {
    
      log_err "$1"
      logf "have a look at our wiki: $D_WIKI_URL"
      logf "or join us on IRC: $D_IRC_URL"
    
    # check for functions
    fn_exists() {
      type -t $1 | grep -q 'function'
    }
    
    
    # shell interactive or not
    interactive_check() {
    
      fd=0 #stdin
      if [[ -t "$fd" || -p /dev/stdin ]]; then
        # all is well
    
        printf "\n"
    
      else
        # non-interactive
    
        curl -s -o "$TMPFILE" "$D_INSTALL_SCRIPT_URL"
        chmod +x "$TMPFILE"
        exec 0< /dev/tty
        bash -i "$TMPFILE"
        rm "$TMPFILE"
        exit 0
      fi
    
    # check if this script is run as root
    root_check() {
      if [ `id -u` -eq 0 ] ; then
        error "don't run this script as root!"
      fi
    }
    
    
    # check if all necessary binaries are available
    
      for exe in "${!BINARIES[@]}"; do
    
        LOG_MSG="checking for $exe... "
        log_inf "$LOG_MSG"
    
        EXE_PATH=$(which "${BINARIES[$exe]}")
    
        if [ $? -ne 0 ]; then
    
          error "you are missing the '${BINARIES[$exe]}' command, please install '$exe'";
    
        else
          printf "$ONE_UP"
          log_inf "$LOG_MSG  found"
    
      printf "\n"
    
    # check for rvm
    define RVM_MSG <<'EOT'
    RVM was not found on your system (or it isn't working properly).
    It is higly recommended to use it, since it's making it extremely easy
    to install, manage and work with multiple ruby environments.
    
    For more details check out https://rvm.io//
    EOT
    rvm_check() {
    
      LOG_MSG="checking for rvm... "
      log_inf "$LOG_MSG"
    
    
      fn_exists rvm
      if [ $? -eq 0 ] ; then
        RVM_DETECTED=true
    
      # seems we don't have it loaded, try to do so
    
      elif [ -s "$HOME/.rvm/scripts/rvm" ] ; then
    
        source "$HOME/.rvm/scripts/rvm" >/dev/null 2>&1
        RVM_DETECTED=true
    
      elif [ -s "/usr/local/rvm/scripts/rvm" ] ; then
    
        source "/usr/local/rvm/scripts/rvm" >/dev/null 2>&1
        RVM_DETECTED=true
      fi
    
      if $RVM_DETECTED ; then
    
        printf "$ONE_UP"
        log_inf "$LOG_MSG  found"
    
        log_wrn "not found"
        logf "$RVM_MSG"
    
        read -p "Press [Enter] to continue without RVM or abort this script and install RVM..."
    
      printf "\n"
    
    # prepare ruby with rvm
    install_or_use_ruby() {
      if ! $RVM_DETECTED ; then
        return
      fi
    
      # make sure we have the correct ruby version available
    
      LOG_MSG="checking your ruby version... "
      log_inf "$LOG_MSG"
    
    
      rvm use $D_RUBY_VERSION >/dev/null 2>&1
      if [ $? -ne 0 ] ; then
    
        log_wrn "not ok"
    
        printf "$ONE_UP"
        log_inf "$LOG_MSG  ok"
    
      printf "\n"
    
    }
    
    # trust and load rvmrc
    # do this in a directory that has a .rvmrc, only :)
    load_rvmrc() {
    
      if ! $RVM_DETECTED || [[ ! -s ".rvmrc" ]] ; then
    
        return
      fi
    
      # trust rvmrc
      rvm rvmrc is_trusted
      if [ $? -ne 0 ] ; then
        rvm rvmrc trust
      fi
    
      # load .rvmrc
    
      LOG_MSG="loading .rvmrc ... "
      log_inf "$LOG_MSG"
    
    
        printf "$ONE_UP"
        log_inf "$LOG_MSG  ok"
    
        log_wrn "not ok"
    
      printf "\n"
    
    # rvm doesn't need sudo, otherwise we do have to use it :(
    rvm_or_sudo() {
      if $RVM_DETECTED ; then
    
        run_or_error "$1"
    
      else
        eval "$1"
        if [ $? -ne 0 ] ; then
    
          log_wrn "running '$1' didn't succeed, trying again with sudo..."
    
    # we need a valid js runtime...
    define JS_RT_MSG <<'EOT'
    This script was unable to find a JavaScript runtime compatible to ExecJS on
    your system. We recommend you install either Node.js or TheRubyRacer, since
    those have been proven to work.
    
    Node.js      -- http://nodejs.org/
    TheRubyRacer -- https://github.com/cowboyd/therubyracer
    
    For more information on ExecJS, visit
    -- https://github.com/sstephenson/execjs
    EOT
    js_runtime_check() {
    
      LOG_MSG="checking for a JavaScript runtime... "
      log_inf "$LOG_MSG"
    
    
      # Node.js
      which node >/dev/null 2>&1
      if [ $? -eq 0 ] ; then
        JS_RUNTIME_DETECTED=true
      fi
    
      # TheRubyRacer
    
      (printf "require 'v8'" | ruby) >/dev/null 2>&1
    
      # add a check for your favourite js runtime here...
    
        printf "$ONE_UP"
        log_inf "$LOG_MSG  found"
    
        log_err "not ok"
    
        error "can't continue without a JS runtime"
      fi
    
      printf "\n"
    
    }
    
    # make ourselves comfy
    prepare_install_env() {
      install_or_use_ruby
      load_rvmrc
      js_runtime_check
    
      log_inf "making sure the 'bundler' gem is installed"
    
      rvm_or_sudo "gem install bundler"
    
    # do some sanity checking
    sane_environment_check() {
      binaries_check
      rvm_check
    }
    
    
    # find or set up a working git environment
    git_stuff_check() {
    
      printf "Where would you like to put the git clone, or, where is your existing git clone?\n"
      printf "(please use a full path, not '~' or '.')\n"
    
      read -e -p "-> " D_GIT_CLONE_PATH
    
      printf "\n"
    
    
      test -d "$D_GIT_CLONE_PATH" \
        && cd "$D_GIT_CLONE_PATH" \
    
        && git status # folder exists? go there. is a good git clone?
    
      if [ $? -ne 0 ]; then
    
        # not a git repo, create it?
    
        printf "the folder you specified does not exist or doesn't contain a git repo\n"
    
        read -p "Press [Enter] to create it and contine... "
    
        run_or_error "mkdir -p -v \"$D_GIT_CLONE_PATH\""  # only if it doesn't exist
        run_or_error "git clone \"$D_REMOTE_REPO_URL\" \"$D_GIT_CLONE_PATH\""
    
        run_or_error "git checkout master"
        run_or_error "git pull"
    
      printf "\n"
    
    # handle database decision
    database_question() {
    
      printf "Which database type are you using? [1|2]\n"
    
      select choice in "MySQL" "PgSQL"; do
        case $choice in
          MySQL )
            D_DB="mysql"
            # we're done, mysql is default
            break
            ;;
          PgSQL )
            D_DB="postgres"
            # replace default with postgres
    
            run_or_error "sed -i'' -e 's/\(<<: \*mysql\)/#\1/g' \"$D_DB_CONFIG_FILE\""
            run_or_error "sed -i'' -e 's/\(#\(<<: \*postgres\)\)/\2/g' \"$D_DB_CONFIG_FILE\""
    
            break
            ;;
        esac
      done
    }
    
    # ask for database credentials
    database_credentials() {
    
      read -e -p "DB hostname: " D_DB_HOST
      read -e -p "DB username: " D_DB_USER
      read -e -p "DB password: " D_DB_PASS
    
      run_or_error "sed -i'' -e \"s/\(host:\)[^\n]*/\1 $D_DB_HOST/g\" \"$D_DB_CONFIG_FILE\""
      run_or_error "sed -i'' -e \"s/\(username:\)[^\n]*/\1 $D_DB_USER/g\" \"$D_DB_CONFIG_FILE\""
      run_or_error "sed -i'' -e \"s/\(password:\)[^\n]*/\1 $D_DB_PASS/g\" \"$D_DB_CONFIG_FILE\""
    
    }
    
    # setup database
    # (assume we are in the Diaspora directory)
    
    define DATABASE_CHK_MSG << 'EOT'
    you can now check the generated database config file in './config/database.yml'
    and see if the specified values are correct.
    
    EOT
    
      log_inf "Database setup"
    
      run_or_error "cp config/database.yml.example \"$D_DB_CONFIG_FILE\""
    
      database_question
      database_credentials
    
    
      printf "$DATABASE_CHK_MSG"
      read -p "Press [Enter] to continue... "
    
    
      printf "\n"
    
    # install all the gems with bundler
    # (assume we are in the Diaspora directory)
    prepare_gem_bundle() {
    
      log_inf "installing all required gems..."
    
      rvm_or_sudo "bundle install"
    
      printf "\n"
    
    # display a nice welcome message
    define WELCOME_MSG <<'EOT'
    #####################################################################
    
    This script will guide you through the basic steps
    
    to get a DEVELOPMENT setup of Diaspora* up and running
    
    For a PRODUCTION installation, please do *not* use this script!
    Follow the guide in our wiki, instead:
    -- https://github.com/diaspora/diaspora/wiki/Installation-Guides
    
    
    #####################################################################
    
    EOT
    
    printf "$WELCOME_MSG"
    
    read -p "Press [Enter] to continue... "
    
    
    
    # check if we have everything we need
    sane_environment_check
    
    
    # check git stuff and pull if necessary
    git_stuff_check
    
    
    # goto working directory
    
    run_or_error "cd \"$D_GIT_CLONE_PATH\""
    
    # configure database setup
    database_setup
    
    
    # diaspora config
    
    log_inf "copying diaspora.yml.example to diaspora.yml"
    
    run_or_error "cp config/diaspora.yml.example config/diaspora.yml"
    
    printf "\n"
    
    log_inf "creating the default database specified in config/database.yml. please wait..."
    
    run_or_error "bundle exec rake db:schema:load_if_ruby --trace"
    
    printf "\n"
    
    define GOODBYE_MSG <<EOT
    
    #####################################################################
    
    It worked! :)
    
    Now, you should have a look at
    
      - config/database.yml      and
    
    
    and change them to your liking. Then you should be able to
    start Diaspora* in development mode with:
    
        `rails s`
    
    
    
    For further information read the wiki at $D_WIKI_URL
    or join us on IRC $D_IRC_URL
    
    
    printf "$GOODBYE_MSG"