#!/bin/bash
## Unix-diff with enhanced options;  by Eugene Reimer  1998-02;

AT="$@"									##remember cmdline args for msgs
SAME="-s"								##default for report-identical option
WHITE="-b -B"								##default for whitespace-matters option
##=====new improved arg-parsing:
USAGE="USAGE:  dif  [OPTION]...  OLDFILENAME  NEWFILENAME
where OPTION can be one of:
	--sed=SEDCMD (-S=SEDCMD)	--special-case of --cmd=FILTER;
	--cmd=FILTER (-c=FILTER)	--filter applied to each file prior to comparison;
	--ignore-case (-i)		--GNU-diff option to be case-blind;
	--whitespace-matters (-m)	--by default dif uses -b and -B GNU-diff options;  this option suppresses those;
	--ignore-all-space (-w)		--GNU-diff option to ignore all whitespace
	--terse (-T)			--for -U0 output without the "@@" lines;  two of these also suppresses the heading;
	++report-identical (+s)		--inverse of the GNU-diff option;  default is to report when the 2 files are identical
	any option recognized by GNU-diff, except -c -T;"		##NOTE: ending quote, ends the USAGE=... statement
getopnd () { if [[ $ARG == *=* ]];then OPX="${ARG#*=}";else ((++J)); OPX="$NXT"; fi; }	##Note: implicit param J is updated in next-word case;  result $OPX
N=$#									##set N = nbr-args;  could use $# in for-expr but need to evaluate just once??
for((J=1;J<=N;++J));do  ARG=${!J}; ((X=J+1));NXT=${!X}			##for J from 1..N do, setting ARG to the Jth arg...
  if [[ $ARG != [-+]* ]];then break; fi					##leave on encountering non-option
  if [[ $ARG == -- ]];then shift; break; fi				##leave on encountering "--" allowing subsequent non-option to start with dash|plus
  case "$ARG" in
  -h*|--help)      echo "$USAGE"; exit 0			;;
  -S*|--sed*)      getopnd;SEDCMD=$OPX				;;	##sed-filter;  getopnd procedure-call to handle same- or next-word opnd
  -c*|--cmd*)      getopnd;FILTER=$OPX				;;	##other filter
  -m*|--white*)    WHITE=					;;	##suppress the use of -b and -B as GNU-diff options
  +s*|++report-i*) SAME=					;;	##turn off the report-identical option
  -T*|--terse)     ((++TERSE))					;;	##increment terseness-level TERSE
  -I*|--ignore-m*) getopnd; OPTIONS="$OPTIONS-I'$OPX' "		;;	##handle GNU-diff -I|--ignore-matching-lines=regexp option  (broken, needs eval on diff-cmdline??)
  *)               OPTIONS="$OPTIONS$1 "			;;	##assume anything else is a GNU-diff option  (may add complete parsing?)
  *)  echo "unrecognized option: $ARG"; echo "$USAGE"; exit 1	;;	##errmsg for unknown option  (presently useless)
  esac
done
for((J--;J--;));do shift; done						##remove args 1..J-1  (an extra shift has been in the "--" case)
if [ $# -ne 2 ];then echo -e "dif $AT - 2 non-options required;\n$USAGE";exit 1;fi	##errmsg if OLDFILENAME NEWFILENAME missing, or too many non-options
##=====endof arg-parsing
FOLD=$1;  FNEW=$2
if [[ $SEDCMD && $FILTER ]];then					##apply sed-filter plus FILTER to each file before comparison...
  cat $FOLD |sed "$SEDCMD" |$FILTER  >tmp-Old$$;  FOLD=tmp-Old$$
  cat $FNEW |sed "$SEDCMD" |$FILTER  >tmp-New$$;  FNEW=tmp-New$$
elif [[ $SEDCMD ]];then							##apply sed-filter to each file before comparison...
  cat $FOLD |sed "$SEDCMD"  >tmp-Old$$;		  FOLD=tmp-Old$$
  cat $FNEW |sed "$SEDCMD"  >tmp-New$$;		  FNEW=tmp-New$$
elif [[ $FILTER ]];then							##apply FILTER to each file before comparison...
  cat $FOLD |$FILTER  >tmp-Old$$;		  FOLD=tmp-Old$$
  cat $FNEW |$FILTER  >tmp-New$$;		  FNEW=tmp-New$$
fi
cmp -s $FOLD $FNEW;  same=$?						##set same;  same==zero iff files identical
OUTFIL="cat"; if((TERSE>0));then OUTFIL='egrep -v ^(---|\+\+\+|@@)';fi	##output-filter uses egrep for TERSE>=1
if((TERSE<2)) && [[ $SAME == "-s" || $same -ne 0 ]];then		##supply one-line heading, if TERSE<2 and (SAME or not same)
  echo "==dif $AT"							##(had -e and leading "\n")
fi
diff -U0 $SAME $WHITE $OPTIONS  $FOLD  $FNEW  |$OUTFIL			##run GNU-diff with options and output-filtering
rm -f tmp-{Old,New}$$							##cleanup any tmpfiles that were created


exit
2010-05-06:
	my diff-variants are getting out of hand, especially since many more combinations are needed;
	==NEEDED: make <dif> be a cmd with options;  name it <dif> since need name to be distinct from Unix-diff;
2010-05-07:
	renamed diff-->dif:  ==looked at output from:  egrep $'(^|[^-\\./a-zA-Z])diff[ \t]'  *
	cd /etc/sbin
	mv diff dif
	egrep    $'(^|[^-\\./a-zA-Z])diff[ \t]'  *								&>/tmp/dif-renaming-01
	egrep -l $'(^|[^-\\./a-zA-Z])diff[ \t]'  *								&>/tmp/dif-renaming-02
	chgsed --tmp --verbose 's!\(^\|[^-\./a-zA-Z]\)diff\([ \t]\)!\1dif\2!g'  $(cat /tmp/dif-renaming-02)	&>/tmp/dif-renaming-03
	chgsed --tmp --verbose 's!/usr/bin/diff!diff!g'                         $(grep -l '/usr/bin/diff' *)	&>/tmp/dif-renaming-04
	--er programs affected:  tesseract-training-from-source  diff-by-charset  chg  weblinkcheck.cron  webcommentcheck
		chg needed manual revision, went with dif-->difterse for PROMPT-output (difterse is inline-function);
		diff-by-charset  <==NEEDS RENAME==??==
	==renaming the other diff-XX scripts (to dif-XX) will be much easier, but isn't essential;  2011-01: have also renamed & published difsed;
2011-01-11:
    NEW OPTIONS:
	--sed=SED-CMDLINE	<--special-case of --cmd=FILTER
	--cmd=FILTER		<--filter applied to each file prior to comparison
	--ignore-case (-i)	<--NOT NEEDED;  GNU-diff has exact same option
	--ignore-whitespace	<--2/3 levels of ignore-whitespace;  GNU-diff:  --ignore-space-change (-b)  --ignore-all-space (-w)  --ignore-blank-lines (-B);
				   using -b -B by default, need whitespace-matters option to avoid those;  ignore-whitespace == GNU ignore-all-space
	--terse			<--2 levels of terseness:  with or without a one-line heading;
				   terse-output is -U0 without its 2-line heading & without the "@@" lines;  one-line heading is an echoing of cmdline options + filenames
	--sort			<--NOT NEEDED;  --cmd=sort does that
	also support any option of GNU-diff;  ergo --ignore-case and --ignore-whitespace not needed since diff has such options;  mine ignores space-changes by default;
	OBSOLETE CMDLINE:  diff -bBs -U0 "$@"	##Unix-diff with my usual options
	(SEE ALSO:  NOTES-dif-REWRITE-201011)
    OBSOLETE CMDS:
	diffsed				-- rename to difsed==!!==  becomes:  dif --sed
	diffterse			-- becomes:  dif -t
	diff-strict			-- becomes:  dif -m
	diff-ignoring-whitespace	-- becomes:  dif -w
	diffsed-ignoring-case		-- becomes:  difsed -i
	diffsed-ignoring-whitespace	-- becomes:  difsed -w
	diffsedsort			-- becomes:  difsed --cmd=sort
	diffsort			-- becomes:  dif --cmd=sort
	finishing:  (1) rename existing dif-variant cmds to XX-OLD;  (2) replace each with one-liner shown above;  (3) provide both diffsed & difsed;  ==DONE;
	NOTE: the above are not "obsolete" but rather have become trivial one-liners;  any published ones will be given a "dif" (one eff) name;
    DIF-VARIANTS THAT REMAIN:
	diff-by-charset			-- this one is fundamentally different...  (the two-eff name seems appropriate)
	diff-redline			-- may add --redline option??
	diff-ls				-- may add --ls option??  may also offer a SIZE-ONLY variant??
	diff+  diff-			-- variants of terse that may be needed again??
    ALTERNATE CODE FOR FOLD FNEW FILTERING (simpler, yet more chance of quoting grief?):
	if [[ $SEDCMD && $FILTER ]];then  FIL="sed '$SEDCMD' |$FILTER"		##sed-filter plus FILTER
	elif [[ $SEDCMD ]];then		  FIL="sed '$SEDCMD'"			##sed-filter
	elif [[ $FILTER ]];then		  FIL="$FILTER"				##FILTER
	fi
	if [[ $FIL ]];then							##apply filter FIL to each file before comparison...
	  cat $FOLD |$FILTER  >tmp-Old$$;		  FOLD=tmp-Old$$
	  cat $FNEW |$FILTER  >tmp-New$$;		  FNEW=tmp-New$$
	fi
    OUTFIL is troublesome WRT quoting??  a space in pattern, after right-paren, leads to:  egrep: ': No such file or directory==??==
    the OUTFIL filtering NOT working as intended due to bash doing incomplete parsing after such substitution;  removing the nested quotes solves the problem!!
2011-01-13:
	webput: several uses of dif are getting the USAGE msg:
	in genByCaption-ER:		dif -I 'This page was generated mechanically on .*'  /tmp/$OUT  $OUT		<--note the quoted and separate-word operand;
	in genOrchidsBySpecies-ER:	dif -I 'This page was generated mechanically on .*'  /tmp/$OUT  $OUT		<--note the quoted and separate-word operand;
	==added parsing-code for -I*|--ignore-m* -- to handle separate-word operand and quoting on -I operand;  ==NOT WORKING;  get:  diff: extra operand 'generated'
	==genByCaption-ER, genOrchidsBySpecies-ER:  dif -I 'This page was.*'  -->  difsed 's|This page was.*||'		<--so webput works, tho -I still broken...
	==consider:  use eval on the diff-cmdline in order to get complete bash-parsing??
20110-01-19:
	for the dif-cmds in the genXX scripts (from webput), want my dif to say nothing for identical files in those cases;
	also want default to be that it reports when 2 files are the same;  ergo need the inverse of -s option;
	using:  +s as the inverse of -s;  not possible with getopt...
	genByCaption-ER, genOrchidsBySpecies-ER:  add +s option on dif-cmdlines to suppress output if files identical;  revise difsed +s-->dif +s --sed  so +s works;
	revised arg-parsing:  -s --> -S;  -t --> -T;  added to help-output: any GNU-option except ...