#!/bin/sh -
#
#   This is a filter that prints parts of lines (or whole
#   lines) that match a regular expression in bold.
#   Non-matching text is printed unmodified.
#   Examples:
#
#      mount | bold soft-updates
#         Show "mount" ouput with all occurences of the word
#         "soft-updates" highlighted.
#
#      tail -f /var/log/messages | bold -l kernel
#         Print syslog output (and wait for more to come,
#         see the tail(1) manpage) and highlight all lines
#         that conmtain the word "kernel".
#
#   Tested on BSD and Solaris.
#   (Solaris requires using ksh or /usr/xpg4/bin/sh.)
#
#   Copyright (C) 1998-2008 Oliver Fromme  <olli@fromme.com>
#   Use and distribute under BSD license.
#

IGN_CASE=0
WHOLE_LINES=0
RUN_TESTS=0
VIDEO_CMODE=""

Usage()
{
	ME="${0##*/}"
	cat <<-tac >&2
		Usage:  $ME [-burBil] <regex> [<file> ...]
		   or:  $ME -t
		Options for video attributes (default depends on executable name):
		   -b   bold
		   -u   underline
		   -r   reverse
		   -B   blink
		Other options:
		   -i   Ignore case (i.e. comparison is case-insensitive).
		   -l   Apply attributes to whole lines, not just the matching part.
		   -t   Run a built-in attribute test on your terminal.
		Notes:  Attributes can be combined (e.g. -bu = bold + underline).
		        Not all attributes are supported on all terminals.
		        <regex> is an extended regular expression.
	tac
}

CM_NORM='\033[m'
CM_BOLD='\033[1m'
CM_UNDL='\033[4m'
CM_BLNK='\033[5m'
CM_REVS='\033[7m'

while :; do
	case "$1" in
		""|[!-]*)
			break
			;;
		--)
			shift
			break
			;;
		-[!ltbuBir]*)
			Usage
			exit 1
			;;
	esac
	case "$1" in -*i*)	IGN_CASE=1	;; esac
	case "$1" in -*l*)	WHOLE_LINES=1	;; esac
	case "$1" in -*t*)	RUN_TESTS=1	;; esac
	case "$1" in -*b*)	VIDEO_CMODE="${VIDEO_CMODE}$CM_BOLD" ;; esac
	case "$1" in -*u*)	VIDEO_CMODE="${VIDEO_CMODE}$CM_UNDL" ;; esac
	case "$1" in -*B*)	VIDEO_CMODE="${VIDEO_CMODE}$CM_BLNK" ;; esac
	case "$1" in -*r*)	VIDEO_CMODE="${VIDEO_CMODE}$CM_REVS" ;; esac
	shift
done

case "$RUN_TESTS" in
	1)
		INDENT=""
		SPACES="`printf '%30s' ''`"

		for BO in "" b+bold; do
		 for UL in "" u+underline; do
		  for IV in "" r+reverse; do
		   for BL in "" B+blink; do

			OPTS="-${BO%+*}${UL%+*}${IV%+*}${BL%+*}"
			TEXT="${BO#*+} ${UL#*+} ${IV#*+} ${BL#*+} "
			REXP="[^ ].*"

			case "$OPTS" in
				"-")
					OPTS=""
					TEXT="none"
					REXP="xxx"
					;;
			esac

			echo "$INDENT" $TEXT | $0 $OPTS "$REXP"

			case "$INDENT" in
				"")	INDENT="$SPACES" ;;
				*)	INDENT="" ;;
			esac

		   done
		  done
		 done
		done

		exit
		;;
esac

if [ -z "$VIDEO_CMODE" ]; then
	case "$0" in
		*link)	VIDEO_CMODE="$CM_BLNK"	;;
		*line)	VIDEO_CMODE="$CM_UNDL"	;;
		*erse)	VIDEO_CMODE="$CM_REVS"	;;
		*)	VIDEO_CMODE="$CM_BOLD"	;;
	esac
fi

if [ $# -lt 1 ]; then
	Usage
	exit 1
fi

RE="$1"
if [ -z "$1" ]; then
	RE=".*"
fi
shift

#
#   The following requires some explanations.
#
#   Why awk instead of sed?  Because sed always buffers the
#   last line, so this wont work with "tail -f" (the last
#   line would always be missing until more output occurs;
#   see the second example above).  On the other hand, awk
#   always processes each line as it comes, so this works
#   fine with "tail -f".
#
#   Why "case/esac" instead of "if [ ]; then"?  Because
#   "[" (which is a link to "test") is sometimes an external
#   program, but "case" is _always_ a shell-builtin.  So
#   the latter is generally more efficient.
#
#   The quoting around "$RE" might look confusing and
#   strange at first, but it's correct and necessary to
#   protect the variable, which could contain anything
#   (including whitespace).
#

case "${WHOLE_LINES}${IGN_CASE}" in
10)
	nawk '{
		if ($0 ~ "'"$RE"'")
			print "'"$VIDEO_CMODE"'" $0 "'"$CM_NORM"'"
		else
			print
	}' "$@"
	;;
11)
	nawk 'BEGIN{
		re = tolower("'"$RE"'")
	}
	{
		if (tolower($0) ~ re)
			print "'"$VIDEO_CMODE"'" $0 "'"$CM_NORM"'"
		else
			print
	}' "$@"
	;;
00)
	nawk '{
		gsub("'"$RE"'", "'"$VIDEO_CMODE"'&'"$CM_NORM"'")
		print
	}' "$@"
	;;
01)
	nawk 'BEGIN{
		re = tolower("'"$RE"'")
		cm_offset = length("'"${VIDEO_CMODE}${CM_NORM}"'") - 1
	}
	{
		lin = $0
		low = tolower(lin)
		offset = 0
		while (low && match(low, re) && RLENGTH) {
			lin = substr(lin, 1, offset + RSTART - 1)   \
			      "'"$VIDEO_CMODE"'"   \
			      substr(lin, offset + RSTART, RLENGTH)   \
			      "'"$CM_NORM"'"   \
			      substr(lin, offset + RSTART + RLENGTH)
			low = substr(low, RSTART + RLENGTH)
			offset += RSTART + RLENGTH + cm_offset
		}
		print lin
	}' "$@"
	;;
*)
	echo "${0}: Internal error." >&2
	exit 1
	;;
esac

#--