configure

#!/bin/sh

# Step 1: Configure CWB
# 
# The CWB includes specialized configurations (for macOS, Linux, Solaris).
# We check whether RcppCWB is installed on a system for which a specialized
# configuration is available. If not, the standard unix configuration is
# used. 

# Get compiler used by R
CC_R=`$R_HOME/bin/R CMD config CC` 

CARBON=""
SOCKETLIB=""

OS=`uname -s`
echo "* operating system detected for CWB configuration: $OS"

case $OS in
  
 Darwin)
    ARCH=`uname -m`
    case $ARCH in
      arm64)
        echo "* compiling for darwin-arm64"
        CWB_PLATFORM_CONFIG_FILE="darwin-arm64"
        CWB_PLATFORM_CONFIG_FILE_CC="unix";;

      *)
        echo "* compiling for darwin-64"
        CWB_PLATFORM_CONFIG_FILE="darwin-64"
        CWB_PLATFORM_CONFIG_FILE_CC="unix";;
    esac
    ;;
  Linux)
    # the 'uname -s' command returns 'Linux' on Fedora, Debian and Ubuntu
    DISTRO=`awk -F= '$1=="ID" { print $2 ;}' /etc/os-release`
    case "$DISTRO" in
      ubuntu)
        SUPPORT="supported";;
      fedora)
        SUPPORT="supported";;
      debian)
        SUPPORT="supported";;
      centos)
        SUPPORT="supported";;
      *)
        SUPPORT="unsupported, adapting the configuration may be necessary";;
    esac
    echo "* Linux distribution ID: $DISTRO ($SUPPORT)";

    # For Debian, Fedora and Ubuntu, the generic Linux configuration works, 
    # there are no known required adaptions so far. If another distribution is used,
    # the previous message informs users that adaptions may be necessary.
    ARCH=`lscpu | head -n 1 | grep -oP '\w+$'`
    case "$ARCH" in 
      x86_64)
        CWB_PLATFORM_CONFIG_FILE="linux-64";;
      i386) 
        CWB_PLATFORM_CONFIG_FILE="linux";;
      *)
        echo "* unknown architecture ($ARCH), using Linux config as default"
        CWB_PLATFORM_CONFIG_FILE="linux"
        ;;
    esac
    
    # The compiler is set in the unix config file for all architectures
    CWB_PLATFORM_CONFIG_FILE_CC="unix";;
  SunOS)
    CPU_INFO=`psrinfo -pv | sed -n '2p'`
    case $CPU_INFO in
      SPARK)
        CWB_PLATFORM_CONFIG_FILE="solaris_spark"
        CWB_PLATFORM_CONFIG_FILE_CC="solaris_spark"
        ;;
      *)
        CWB_PLATFORM_CONFIG_FILE="solaris_x86"
        CWB_PLATFORM_CONFIG_FILE_CC="solaris_x86"
        ;;
    esac
    SOCKETLIB='-lsocket';;
    
  *)
    CWB_PLATFORM_CONFIG_FILE="unix"
    CWB_PLATFORM_CONFIG_FILE_CC="unix"
    echo "* OS unknown, resorting to Unix as CWB default configuration (modifications may be necessary)"
    ;;
esac

# Inform user about CWB configuration used. To safeguard portability, no inplace 
# sed modification (sed option -i is not available on Solaris)
echo "* using CWB platform configuration file: $CWB_PLATFORM_CONFIG_FILE"
sed -e "s/PLATFORM=.*/PLATFORM=$CWB_PLATFORM_CONFIG_FILE/" ./src/cwb/config.mk > ./src/cwb/config.mk.mod
rm ./src/cwb/config.mk
mv ./src/cwb/config.mk.mod ./src/cwb/config.mk


# Adapt CWB configuration file that defines compiler to use compiler chosen
# by user (Makeconf, Makevars file)
echo "* adapt CWB configuration file to use compiler: $CC_R"
sed -e "s/#[[:space:]]*CC[[:space:]]*=[[:space:]]*gcc/CC = $CC_R/" ./src/cwb/config/platform/${CWB_PLATFORM_CONFIG_FILE_CC} > ./tmpfile
rm ./src/cwb/config/platform/${CWB_PLATFORM_CONFIG_FILE_CC}
mv ./tmpfile ./src/cwb/config/platform/${CWB_PLATFORM_CONFIG_FILE_CC}

# Check availability of glib2.0 (download on macOS)
if [ -x "`command -v pkg-config`" ]; then
  printf "* using pkg-config to check whether glib-2.0 is available ... ";
  pkg-config --exists glib-2.0
  PKG_CONFIG_GLIB_INFO=$?
  if [ $PKG_CONFIG_GLIB_INFO = 0 ]; then
    printf "yes\n";
    if [ $OS = "Darwin" ]; then
      CARBON="-framework Carbon"
    fi
  else
    printf "no\n";
    if [ $OS = "Darwin" ]; then
      echo "* glib2.0 missing, downloading static library from GitHub (https://github.com/PolMine/libglib)";
      ${R_HOME}/bin/Rscript ./tools/maclibs.R `pwd`
      export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:`pwd`/macos/libglib-master/pkgconfig
      # Setting further args required to compile agains glib2.0 in this scenario
      CARBON="-framework Carbon"
    else 
      echo "----------------------- CONFIGURATION ERROR --------------------------"
      echo "Configuration failed because 'glib-2.0' was not found. Try installing:"
      echo " * deb: libglib2.0-dev (Debian, Ubuntu, etc)"
      echo " * yum: glib* (Fedora, CentOS, etc)"
      echo " * csw: glib2 (Solaris)"
      echo "----------------------------------------------------------------------"
      /opt/csw/bin/pkgutil -y -i -W glib2
    fi
  fi
else
  echo "WARNING: Aborting as the 'pkg-config' tool required to find include paths is missing."
  echo "Check whether 'pkg-config' is installed and make sure that 'pkg-config' is on your PATH."
  exit 1
fi 

printf "* using pkg-config to get linker flags for glib-2.0: "
GLIB_LINKER_FLAGS="`pkg-config --libs glib-2.0`"
printf "%s\n" "$GLIB_LINKER_FLAGS"


# Check availability of pcre2
printf "* check whether pcre2 is available ... ";
if [ -x "`command -v pcre2-config`" ]; then
  PCRE2_VERSION=`pcre2-config --version`
  printf "yes (version: $PCRE2_VERSION)\n"
else
  printf "no\n"
  echo "----------------------- CONFIGURATION ERROR --------------------------"
  echo "Configuration failed because 'pcre2' is not available. Try installing:"
  echo " * deb: libpcre2 libpcre2-dev (Debian, Ubuntu, etc)"
  echo " * yum: pcre2 pcre2-devel (Fedora, CentOS, etc)"
  echo " * csw: libpcre2_dev (Solaris)"
  echo "----------------------------------------------------------------------"
fi

printf "* checking whether pcre2test is available ... ";
if [ -x "`command -v pcre2test`" ]; then
  printf "yes\n";
  printf "* check whether pcre2 has been configured with unicode properties support ... ";
  case "`pcre2test -C unicode`" in
    0)
      printf "no\n";
      echo "-------------------------------  WARNING -------------------------------"
      echo "No Unicode properties support found for pcre2. RcppCWB may not work as"
      echo "intended and throw errors. Recommended solution:"
      echo "Configure pcre2 without '--disable-unicode' option and re-compile."
      echo "------------------------------------------------------------------------";;
    1)
      printf "yes\n";;
    *)
      printf "unknown\n";;
  esac
else 
  printf "no\n";
fi

printf "* using pcre2-config to get libdirs for pcre2 (8bit): "
PCRE2_LIBDIRS="`pcre2-config --libs8`"
printf "%s\n" "$PCRE2_LIBDIRS"

printf "* using pcre2-config to get cflags for pcre2: "
PCRE2_CFLAGS="`pcre2-config --cflags`"
printf "%s\n" "$PCRE2_CFLAGS"

# Step 2: Generate Makevars file
#
# The Makevars file is generated here, to pass the directory where RcppCWB
# is build (variable PKG_DIR) into the Makevars file.

BUILD_DIR="`pwd`"
printf "* using build directory: %s\n" ${BUILD_DIR}

CWB_DIR="`pwd`/src/cwb"

if [ -f ./src/Makevars ]; then rm ./src/Makevars; fi
printf "PKG_CPPFLAGS=-I%s/src/cwb/cqp -I%s/src/cwb/cl -I%s/src/cwb/CQi %s\n" ${BUILD_DIR} ${BUILD_DIR} ${BUILD_DIR} "$PCRE2_CFLAGS" > ./src/Makevars
printf "PKG_LIBS=-L%s/cl -L%s/cqp -L%s/utils -lcwb -lcqp -lcl %s %s %s %s\n" ${CWB_DIR} ${CWB_DIR} ${CWB_DIR} "$GLIB_LINKER_FLAGS" "$PCRE2_LIBDIRS" "$SOCKETLIB" "$CARBON" >> ./src/Makevars
printf "\${SHLIB}: libcl.a libcqp.a libcwb.a\n" >>./src/Makevars
printf "libcl.a: depend\n" >> ./src/Makevars
printf "\tcd cwb; R_PACKAGE_SOURCE=%s PKG_CONFIG_PATH=%s \${MAKE} cl\n" ${CWB_DIR} ${PKG_CONFIG_PATH} >> ./src/Makevars
printf "libcqp.a: depend\n" >> ./src/Makevars
printf "\tcd cwb; R_PACKAGE_SOURCE=%s PKG_CONFIG_PATH=%s \${MAKE} cqp\n" ${CWB_DIR} ${PKG_CONFIG_PATH} >> ./src/Makevars
printf "libcwb.a: depend\n" >> ./src/Makevars
printf "\tcd cwb; R_PACKAGE_SOURCE=%s PKG_CONFIG_PATH=%s \${MAKE} utils\n" ${CWB_DIR} ${PKG_CONFIG_PATH} >> ./src/Makevars
printf "depend: clean\n" >> ./src/Makevars
printf "\tcd cwb; R_PACKAGE_SOURCE=%s PKG_CONFIG_PATH=%s \${MAKE} depend\n" ${CWB_DIR} ${PKG_CONFIG_PATH} >> ./src/Makevars
printf "clean:\n" >> ./src/Makevars
printf "\tcd cwb; R_PACKAGE_SOURCE=%s PKG_CONFIG_PATH=%s \${MAKE} clean\n" ${CWB_DIR} ${PKG_CONFIG_PATH} >> ./src/Makevars