arrow-build.sh

#!/usr/bin/env bash

set -eu

PWD="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
SRC_DIR=$(realpath "${PWD}/..")
CXX_SRC=${SRC_DIR}/cpp

# The following can be set
: "${CMAKE:=cmake}"

error() {
  echo "$@" >&2
  exit 1
}

check_installed() {
  which "$1" > /dev/null
}

parse_argv() {
  check_installed docopts || error "docopts is missing, install with 'pip install docopts'"

  # Parse the argv into an associative array stored in a local variable named
  # __args. We cannot write in the `args` implicit scoped variable due to
  # redefinition and shadowing.
  eval "$(docopts -A __args -V - -h - : "$@" <<EOF
Usage: $0 [options] <build_dir>

Prepare a cmake build directory for arrow.

Options:
  --cc=<compiler>      Set C compiler [default: cc].
  --cxx=<compiler>     Set C++ compiler [default: c++].
  --cxx-flags=<extra>  Set extra compiler flags.
  --generator=<gen>    CMake generator type [default: Ninja].

  --build-type=<build>  CMake build type [default: DEBUG].
  --warn-level=<level>  Warning level, possible options are EVERYTHING,
                        CHECKIN, PRODUCTION [default: CHECKIN].
  --cmake-extra=<args>  CMake extra argument to pass.

  --with-tests=<opt>       Activate tests [default: ON].
  --with-benchmarks=<opt>  Activate benchmarks [default: ON].
  --with-python=<opt>      Activate python bindings [default: ON].
  --with-parquet=<opt>     Activate parquet component [default: ON].
  --with-gandiva=<opt>     Activate gandiva component [default: ON].
  --with-plasma=<opt>      Activate plasma component [default: ON].
  --with-flight=<opt>      Activate flight component [default: OFF].
  --with-asan=<opt>        Activate ASAN sanitization [default: OFF].

  --force    Delete exisiting build dir if found.
  --trace    Trace this script.
  --help     Show help options.
----
EOF
  )"

  # copy __args into args
  # disable shellcheck since __args is defined in the previous eval.
  # shellcheck disable=SC2154
  for k in "${!__args[@]}"; do args+=(["$k"]=${__args["$k"]}); done
  # enable `set -x` if tracing is requested
  if [[ "${args[--trace]}" == "true" ]]; then set -x; fi
}

build_cmake_argv() {
  # Define generator
  cmake_argv+=("-G${args[--generator]}")

  # toolchain
  export "CC=${args[--cc]}"
  export "CXX=${args[--cxx]}"

  if [[ ! -z "${args[--cxx-flags]:-}" ]]; then
    cmake_argv+=("-DARROW_CXXFLAGS=${args[--cxx-flags]}")
  fi

  # Build type
  cmake_argv+=("-DCMAKE_BUILD_TYPE=${args[--build-type]}")
  cmake_argv+=("-DARROW_BUILD_TESTS=${args[--with-tests]}")
  cmake_argv+=("-DARROW_BUILD_BENCHMARKS=${args[--with-benchmarks]}")
  cmake_argv+=("-DBUILD_WARNING_LEVEL=${args[--warn-level]}")
  cmake_argv+=("-DARROW_USE_ASAN=${args[--with-asan]}")

  # Components
  cmake_argv+=("-DARROW_PYTHON=${args[--with-python]}")
  cmake_argv+=("-DARROW_PARQUET=${args[--with-parquet]}")
  cmake_argv+=("-DARROW_GANDIVA=${args[--with-gandiva]}")
  cmake_argv+=("-DARROW_PLASMA=${args[--with-plasma]}")
  cmake_argv+=("-DARROW_FLIGHT=${args[--with-flight]}")

  # conda-ism
  if [[ ! -z "${CONDA_PREFIX:-}" ]]; then
    local arrow_home=${CONDA_PREFIX}
    for v in ARROW_HOME ARROW_BUILD_TOOLCHAIN PARQUET_HOME BOOST_HOME; do
      export "${v}=${arrow_home}"
    done

    cmake_argv+=("-DCMAKE_INSTALL_PREFIX=${arrow_home}")
    cmake_argv+=("-DCMAKE_INSTALL_LIBDIR=lib")
  fi

  if [[ ! -z "${args[--cmake-extra]:-}" ]]; then
    cmake_argv+=(${args[--cmake-extra]})
  fi

  cmake_argv+=("$CXX_SRC")
}

invoke_cmake() {
  local build_dir=${args[<build_dir>]}
  local force=${args[--force]}

  # Deal with existing build_dir
  if [[ -e "$build_dir" ]]; then
    if [[ "$force" != "true" ]]; then
      error "$build_dir already exists, pass --force to override"
    fi

    rm -r "$build_dir"
  fi

  # CMake 3.13 supports the `-B<build_dir>` option, until then we resort to the
  # old pushd/popd mechanism.
  mkdir "$build_dir"
  pushd "$build_dir"
  "${CMAKE}" "${cmake_argv[@]}"
  popd
}

arrow_create_cpp_build() {
  # The following local variables are captured by lexical scope due to bash
  # incapacity to accept/return array variables (without using nameref which
  # are found only in bash 4.3 and later).

  # Parse docopts arguments into `args` associative map.
  declare -A args
  # Define a local array to store cmake arguments.
  declare -a cmake_argv

  parse_argv "$@"

  build_cmake_argv

  invoke_cmake
}

main() {
  arrow_create_cpp_build "$@"
}

# Invoke main only if the script is executed and not sourced.
[[ "${BASH_SOURCE[0]}" != "${0}" ]] || main "$@"