cwhere

#!/bin/bash
#
# cwhere
#
# find which C header file defines the requested macro/identifier
#
# Would normally do this in Perl, but doing it in pure bash for
# something different.
#
# Mikel Ward <mikel@mikelward.com>
#
# TODO move cleanup trap inside print_definitions()

name=cwhere

trap cleanup EXIT

# remove any temporary files, etc.
# Usage: cleanup
cleanup()
{
    test -f "$sourcefile" && rm "$sourcefile"
}

get_cpp_output()
{
    local headers="$@"

    sourcefile=$(mktemp)
    if test $? -ne 0; then
        echo "Error: Cannot create source file" 1>&2
        exit 1
    fi
    for header in $headers; do
        echo "#include <$header>" >> "$sourcefile"
    done
    # TODO: Do we need -D_GNU_SOURCE?
    cpp -dD "$sourcefile"
    if test $? -ne 0; then
        cat "$sourcefile"
        echo "Error: Cannot run cpp" 1>&2
        exit 1
    fi
}

# Print the definition of <identifier> and which header file it appears in
# Usage: print_definitions <identifier> <header>...
# Example: print_definitions O_WRONLY sys/types.h sys/stat.h fcntl.h
print_definitions()
{
    local identifier="$1"
    shift
    local headers="$@"

    local filenamepattern definepattern typedefpattern declarepattern
    filenamepattern='^# [0-9]+ "([^"]*)"'
    definepattern='^[[:space:]]*#define[[:space:]][[:space:]]*'"$identifier"
    typedefpattern='^[[:space:]]*typedef[[:space:]].*'"$identifier"
    declarepattern='^[[:space:]]*(extern[[:space:]]*)?(_|int|void|char|short).*[[:space:]]'"$identifier"

    get_cpp_output "$@" |
    while read -r line; do
        # this is evil, just doing it to learn something new
        # put the pattern in a variable to avoid bash treating pattern as a string :-(
        if [[ $line =~ $filenamepattern ]]; then
            file=${BASH_REMATCH[1]}
            if [[ $identifier == "INCLUDES" ]]; then
                echo "$file"
            else
                # just save it in case there's a symbol match
                :
            fi
        elif [[ $line =~ $typedefpattern ]] ||
             [[ $line =~ $definepattern ]] ||
             [[ $line =~ $declarepattern ]]; then
            if [[ $identifier != "INCLUDES" ]]; then
                echo "$file: $line"
            fi
        fi
    done | grep -v '^/tmp/' | sort | uniq
}

# Usage: get_headers <function|syscall>
# Example: get_headers select
# Example: get_headers 'open(2)'
get_headers()
{
    desc "$@" | sed -ne '/^[[:space:]]*#include/{s/^[[:space:]]*#include <\(.*\)>.*/\1/;p}'
}

# print a message saying how to run this program
usage()
{
    cat <<EOF 1>&2
Usage: $name IDENTIFIER HEADER|FUNCTION [HEADER|FUNCTION]...
Examples:
    $name O_RDONLY sys/stat.h sys/types.h fcntl.h
    $name O_RDONLY 'open(2)'
    $name 'struct sockaddr(|_in)?' sys/types.h sys/socket.h netinet/in.h
    $name 'struct sockaddr(|_in)?' 'socket(2)' 'ip(7)'
EOF
#Arguments:
#    IDENTIFIER is regex matching a C identifier
#    HEADER is a C include file, without the < >
#    FUNCTION is a function with an optional man section in parens
}

main()
{
    if test $# -lt 2; then
        usage
        exit 2
    fi

    # identifier is always the first argument...
    local args=("$1")

    # now look at the remaining arguments
    # they are either headers, which should be passed on verbatim
    # or are function/syscall names, in which case we have to find
    # which headers should be included according to the man pages
    shift
    i=1
    for arg in "$@"; do
        # accept open() or open(2), convert to open or open(2)
        if [[ $arg =~ (.*)\((.*)\) ]]; then
            page=${BASH_REMATCH[1]}
            section=${BASH_REMATCH[2]}
            section="${section:+($section)}"

            headers=$(get_headers "$page$section")
            for header in $headers; do
                args[$i]=$header
                i=$((i+1))
            done
        else
            args[$i]=$arg
            i=$((i+1))
        fi
    done
    print_definitions "${args[@]}"
}

# only run the main script if it was executed
# allows sourcing of the script to get function definitions
# works out of the box in bash and ksh
# zsh requires unsetopt functionargzero in ~/.zshenv
case "$0" in *$name)
    main "$@"
    ;;
esac