fix-python-headers - bash tool to fix python file header references

fix-python-header.sh

#!/bin/bash
#
# Fix the first line of python a file to point to the
# new python interpreter.
#
# This mess required to make sure that script works on
# the Mac. Normally we would just use readlink.
set -e
RootDir=$(cd $(dirname $0) && pwd)
source $RootDir/libutils.sh
set +e

# ========================================================================
# Functions
# ========================================================================
function _help() {
    cat <<EOF
USAGE
    $BASENAME [OPTIONS] [FILES]

DESCRIPTION
    Fix the shebang headers in python files.

    This tool is useful for modifying python scripts
    to use a different interpreter by default.

OPTIONS
    -d, --dryrun    Do a fry run.
                    Do not change any files.

    -h, --help      This help message.

    -s HDR, --shebang-header HDR
                    The shebang header for the python file.
                    It must start with '#!'.
                    Here is an examle:
                        -s '#!/opt/cr/bin/python'

    -v, --verbose   Increase the verbosity.
                    Normally the program runs silently.
                    This option allows the user to see
                    what is happening.
                    -v       Shows the phases of the operation.
                    -v -v    Shows the files being modified.
                    -v -v -v Shows skipped non-python files.

    -V, --version   Print the program version and exit.

EXAMPLE
    # Help
    \$ $BASENAME -h

    # Change all my programs to use python3.5.
    \$ $BASENAME -s '#!/usr/local/bin/python3.5' myprogs/*

    # Change all my programs to use python3.5 and
    # show me what is happening.
    \$ $BASENAME -v -s '#!/usr/local/bin/python3.5' myprogs/*

    # Change all my programs to use python3.5 and
    # really show me what is happening.
    \$ $BASENAME -v -v -s '#!/usr/local/bin/python3.5' myprogs/*

    # Change all my programs to the currently available python.
    \$ $BASENAME -s '#!/usr/bin/env python'
EOF
    exit 0
}

# ========================================================================
# Main
# ========================================================================
BASENAME=$(basename $0)
VERSION="1.0.0"
HEADER=""
FILES=()
DRYRUN=0
VERBOSE=0

while (( $# )) ; do
    OPT="$1"
    shift
    case "$OPT" in
        -d|--dryrun)
            DRYRUN=1
            ;;
        -h|--help)
            _help
            ;;
        -s|--shebang-header)
            HEADER="$1"
            shift
            ;;
        -v|--verbose)
            VERBOSE=$(( VERBOSE + 1 ))
            ;;
        -V|--version)
            echo "$BASENAME $VERSION"
            exit 0
            ;;
        -*)
            _err "Unrecognized option '$OPT'."
            ;;
        *)
            FILES+=($OPT)
            ;;
    esac
done

# Tell'em who we are.
(( VERBOSE )) && _banner "Fixing python shebang headers."

# Make sure that the user specified a header with a
# shebang. The reason that we enforce the existence
# of the shebang is purely for consistency. The user
# is replacing the entire header, not just a single
# part of it.
[[ -z "$HEADER" ]] && _err "Shebang header not specified."
[[ ! "$HEADER" =~ "#!" ]] && _err "Shebang header missing shebang (#!)."

# Walk through all of the files and directories specified
# and fix them.
# Ignore files that are not python executables.
# Provide a verbose mode that allows the developer
# to see what is going on (-v -v).
for FILE in ${FILES[@]} ; do
    PYFILES=()
    if [[ -e $FILE ]] ; then
        if [[ -d $FILE ]] ; then
            # This is a directory, process all of the files in it.
            #PYFILES=($(file $FILE/* | grep -i 'python script text' | awk -F: '{print $1}'))
            # Cannot reliably use the file command, it fails under certain
            # conditions. Awk works much better. It is also very fast.
            PYFILES=($(awk 'FNR==1 && /#!.*python/ {print FILENAME ": PYTHON : " $0; }; FNR>1 {nextfile}' $FILE/* | awk -F: '{print $1}'))
        elif [[ -r $FILE ]] ; then
            PYFILES=($(awk 'FNR==1 && /#!.*python/ {print FILENAME ": PYTHON : " $0; }; FNR>1 {nextfile}' $FILE | awk -F: '{print $1}'))
            if (( ${#PYFILES} == 0 )) ; then
                (( VERBOSE > 2 )) && _warn "File is not a python file: $FILE."
            fi
        else
            (( VERBOSE )) && _warn "File is not readable, skipping: $FILE."
        fi
    else
        (( VERBOSE )) && _warn "File does not exist: $FILE."
    fi

    # Now process the python files and update if necessary.
    for PYFILE in ${PYFILES[@]} ; do
        if [[ ! -w $PYFILE ]] ; then
            (( VERBOSE )) && _warn "File is not writable: $FILE"
            continue
        fi
        CURR_HEADER=$(head -1 $PYFILE)
        if echo "$CURR_HEADER" | grep -q '#[ ]*!' ; then
            if [[ "$CURR_HEADER" == "$HEADER" ]] ; then
                (( VERBOSE > 1 )) && _info "Skipping $PYFILE, already updated."
            else
                if (( DRYRUN == 0 )) ; then
                    # This is not a dryrun, change the first line of
                    # the file.
                    (( VERBOSE > 1 )) && _info "Updating $PYFILE."
                    
                    # There some odd cases where surrounding quotes are
                    # added. This causes execution to fail. The quote
                    # stripping removes them.
                    sed -i -e "1c$HEADER" $PYFILE
                    sed -i -e '1s/"//g' -e "1s/'//g" $PYFILE
                else
                    (( VERBOSE > 1 )) && _info "Updating $PYFILE, dryrun, no change."
                fi
            fi
        else
            (( VERBOSE > 1 )) && _warn "Skipping $PYFILE, no header."
        fi
    done
done

# Tell them we done if they are interested.
(( VERBOSE )) && _info "Done."

libutils.sh

#!/bin/bash
#
# Utilities for setup scripts.
#

# ================================================================
# Functions
# ================================================================
function __msg() {
    local LineNo=$1
    local Type=$2
    local Code="$3"
    shift
    shift
    shift
    printf "$Code"
    printf "%-28s %-7s %5s: " "$(date +'%Y-%m-%d %H:%M:%S %z %Z')" "$Type" $LineNo
    echo "$*"
    printf "\033[0m"
}

# Print an info message to stdout.
function _info() {
    __msg ${BASH_LINENO[0]} "INFO" "\033[0m" $*
}

# Print an info message to stdout in green.
function _info_green() {
    __msg ${BASH_LINENO[0]} "INFO" "\033[32m" $*
}

# Print an info message to stdout in red.
function _info_red() {
    __msg ${BASH_LINENO[0]} "INFO" "\033[31m" $*
}

# Print an info message to stdout in bold.
function _info_bold() {
    __msg ${BASH_LINENO[0]} "INFO" "\033[1m" $*
}

# Print a warning message to stderr and exit.
function _warn() {
    __msg ${BASH_LINENO[0]} "WARNING" "\033[34m" $*
}

# Print an error message to stderr and exit.
function _err() {
    __msg ${BASH_LINENO[0]} "ERROR" "\033[31m" $* >&2
    exit 1
}

# Print an error message to stderr.
# Do not exit.
function _err_nox() {
    __msg ${BASH_LINENO[0]} "ERROR" "\033[31m" $* >&2
}

# Decorate a command and exit if the return code is not zero.
function _exec() {
    local Cmd="$*"
    __msg ${BASH_LINENO[0]} "INFO" "\033[1m" "cmd.cmd=$Cmd"
    eval "$Cmd"
    local Status=$?
    if (( $Status )) ; then
        __msg ${BASH_LINENO[0]} "INFO" "\033[31m" "cmd.code=$Status"
        __msg ${BASH_LINENO[0]} "ERROR" "\033[31m" "cmd.status=FAILED"
        exit 1
    else
         __msg ${BASH_LINENO[0]} "INFO" "\033[32m" "cmd.code=$Status"
         __msg ${BASH_LINENO[0]} "INFO" "\033[32m" "cmd.status=PASSED"
    fi
}

# Execute a command quietly.
# Only decorate if an error occurs.
function _exeq() {
    local Cmd="$*"
    eval "$Cmd"
    local Status=$?
    if (( $Status )) ; then
        __msg ${BASH_LINENO[0]} "INFO" "\033[1m" "cmd.cmd=$Cmd"
        __msg ${BASH_LINENO[0]} "INFO" "\033[31m" "cmd.code=$Status"
        __msg ${BASH_LINENO[0]} "ERROR" "\033[31m" "cmd.status=FAILED"
        exit 1
    fi
}

# Decorate a command. Do not exit if the return code is not zero.
function _exec_nox() {
    local Cmd="$*"
    __msg ${BASH_LINENO[0]} "INFO" "\033[1m" "cmd.cmd=$Cmd"
    eval "$Cmd"
    local Status=$?
    if (( $Status )) ; then
        __msg ${BASH_LINENO[0]} "INFO" "\033[31m" "cmd.code=$Status"
        __msg ${BASH_LINENO[0]} "ERROR" "\033[31m" "cmd.status=FAILED (IGNORED)"
    else
        __msg ${BASH_LINENO[0]} "INFO" "\033[32m" "cmd.code=$Status"
        __msg ${BASH_LINENO[0]} "INFO" "\033[32m" "cmd.status=PASSED"
    fi
    return $Status
}

# Banner.
function _banner() {
    echo
    echo "# ================================================================"
    echo "# $*"
    echo "# ================================================================"
}

# Get the version number.
function _get_version() {
    local VERSION='1.0.0'
    echo $VERSION
}