#!/bin/bash
## chg -- for systematic revisions of the "replace A by B" sort -- by Eugene Reimer 1998-Feb;
##	including character-substitutions and/or destuttering ala Unix tr;  can also apply any Unix filter;  always processes directories recursively;
## PREREQ:  fullnameNOSLASH -- from http://ereimer.net/programs/ertools.zip
##
## Copyright © 1998,2009 Eugene Reimer;  can be used, modified, copied, and distributed or sold under the terms of either the LGPL or the GPL (your choice);
## see http://www.gnu.org/licenses for the details of these terms.

difterse () { diff -bBs -U0 "$@" |egrep -v '^(---|\+\+\+|@@) '; }	##show diffs in terse style
cmdargs="$@"
RC=1
while [ $# -gt 0 ] && [[ $1 == -* ]];do		##parse the options
  case "$1" in
  --sed)	CMD="sed"		;;	##for chgsed;  NOTE: sedscript need not follow immediately==!!==
  --tr)		CMD="tr"		;;	##for chgtr
  -s|--des*)	CMD="tr"; ARG="-s"	;;	##destutter, implies --tr
  -a|--fixhtma)	CMD="tidy"; ARG="-omit -wrap 140 -b -ascii"	;;	##tidy options: -omit:omit optional end-tags;  -b:cvt MS-chars;  -win1252/-latin1/-ascii: charset
  -u|--fixhtmu)	CMD="tidy"; ARG="-omit -wrap 140 -b -utf8"	;;	##tidy options: -omit:omit optional end-tags;  -b:cvt MS-chars;  -utf8: charset
  -c|--cmd)	CMD=$2; shift; if [[ $CMD == *\ * ]];then ARG="${CMD#* }"; CMD="${CMD%% *}"; fi	;;	##07aug: ==!!may need fixups for strophs etc??
  -k|--kee*)	KEEPTS=1		;;	##keep-timestamp option
  -n|--nob*)	NOBKUP=1		;;	##no-backup option
  -T|--tmp*)	TMPBKUP=1		;;	##backups under /tmp option
  -q|--qui*)	QUIET=1; VERBOSE=	;;	##quiet option
  -t|--tes*)	TEST=1; PROMPT=; QUIET= ;;	##test option
  -p|--pro*)	PROMPT=1; TEST=		;;	##prompt option
  -m|--must*)	MUST=1			;;	##must-match option (user wants a message for "file-unchanged" case)
  -v|--verb*)	VERBOSE=1; QUIET=	;;	##verbose output option (2010-04)
  -V|--vers*)	CMD="vers"; VERS=1	;;	##produce version info only
  -\?|--hel*)	CMD="help"; HELP=0	;;	##produce usage info only, exit-code=0
  --)		shift; break		;;	##ender for options
  esac
  shift;					##remove that arg
done
if [ "$CMD" = "" ];then				##expect A B positionals
  if [ $# -lt 3 ];then HELP=2;  else  CMD="sed";  BAR=$'\x01';  ARG="'s$BAR${1//\'/\\x27}$BAR${2//\'/\\x27}${BAR}g'";  shift; shift;  fi
elif [ "$CMD" = "sed" ] && [ "$ARG" = "" ];then
  if [ $# -lt 2 ];then HELP=2;  else ARG="'${1//\'/\\x27}'"; shift; fi
elif [ "$CMD" = "tr" ];then
  if [ $# -lt 3 ];then HELP=2;  else ARG="$ARG '${1//\'/\\x27}' '${2//\'/\\x27}'"; shift;shift; fi
fi
##[ $PROMPT$TEST ] || PROMPT=1;			##had PROMPT as default til recent changes trusted - soon became tiresome...
[ $TEST ] && echo "---chg $cmdargs---"		##show the cmdline;  2007feb: $PROMPT now done per-file below;  2010-04: $VERBOSE also done per-file
[ $VERS ] && echo "chg version:$(date -r$0 +%Y-%m-%d). Copyright 1998,2008 Eugene Reimer. Licensed under the GPL" && exit 0
##echo "chg:  CMD=$CMD;  ARG=$ARG;  HELP=$HELP; KEEPTS=$KEEPTS; NOBKUP=$NOBKUP; PROMPT=$PROMPT; QUIET=$QUIET; TEST=$TEST; TMPBKUP=$TMPBKUP;"	##DEBUG
##(note: ARG displays badly wrt the x01 chars)
if [ $HELP ];then
  if [ $HELP -gt 0 ];then echo "CMD=$CMD; ARG=$ARG;"; fi
  echo "usage:  chg options   'A'  'B'        file-or-directory..."
  echo "   or:  chg options --sed sedpattern  file-or-directory..."
  echo "   or:  chg options --tr  set1 set2   file-or-directory..."
  echo "   or:  chg options --cmd filtercmd   file-or-directory..."
  echo "   or:  chg options --fixhtma         file-or-directory..."
  echo "   or:  chg options --fixhtmu         file-or-directory..."
  echo "Revises files with s|A|B|g or sedpattern or tr-set1-set2-substitutions or filtercmd."
  echo "  -s | --destutter      implies --tr;  removes stuttering repetition"
  echo "  -k | --keeptimestamp  revised file keeps its previous modification-timestamp"
  echo "  -n | --nobackup       no copy kept of previous file contents"
  echo "  -T | --tmpbackup      backup made under /tmp, instead of as tilde-file in current directory"
  echo "  -q | --quiet          suppress messages about file modification and backup"
  echo "  -t | --test           just show which files would be affected, but do not revise them"
  echo "  -p | --prompt         for each affected file, show diff's then prompt before revising it"
  echo "  -? | --help           just show this info on usage"
  echo "  -v | --version        just show version info"
  echo "For a directory, all subfiles are revised recursively."
  echo "Normally makes a backup copy when revising, but not for a file less than 5min old that has"
  echo "  a backup copy (using the --keeptimestamp option can result in more frequent backups)."
  echo "See unchg if needing to undo."
  exit $HELP;					##exit
fi
##[ $# -eq 0 ] && set ..?* .[^.]* *;		##used to match all files by default  (now require at least one file on cmdline)

chgfunc() {
  for f in "$@";do
    [ -L "$f" ] && continue
    [ -f "$f" ] || continue
    [[ $f == *~ ]] && continue				##skip a tilde-bkup file
    
    ##eval args "$CMD" "$ARG"; echo -n "continue?"; read	##DEBUG
    ##echo -n "CMD=$CMD;  ARG=$ARG; f=$f; continue?"; read	##DEBUG
							##note: need apostrophes around an s|x|y| arg, to prevent piping...
    eval "$CMD" "$ARG" <"$f" >/tmp/chg$$		##execute sed or tr to make replacements, output into a tmp-file, for now;  WHY USE eval==??==
    if [ $? -ne 0 ];then echo "chg: BAD RC=$? FROM $CMD"; exit; fi	##shouldmot happen;  06jun:DID w --fixhtma on USDAplantlist;  removing exit was BAD idea w chgsed
    if ! cmp -s -- "$f" /tmp/chg$$;then	##--file was revised--!!--
      TSOLD=$(date --reference="$f"       +%s)		##timestamp of old file, in seconds since 1969
      TSNEW=$(date --reference=/tmp/chg$$ +%s)		##timestamp of new file, in seconds since 1969
      ((AGE=TSNEW-TSOLD))				##nbr-seconds since previous revision to file
      chmod --reference="$f" /tmp/chg$$			##preserve file-attributes, always
      ##chown --reference="$f" /tmp/chg$$		##preserve file-ownership, always == YANKED because of spurious msgs in FC4
      [ $KEEPTS ] &&
        touch --reference="$f" /tmp/chg$$		##preserve timestamp, only if --keep-timestamp specified
      fbk="$f~"							##for tilde-bkup in same dir
      [ $TMPBKUP ] && fbk=/tmp/bk-chg-$(fullnameNOSLASH "$f")	##for bkup under /tmp dir
      Changed="changed"; [ $TEST ] && Changed="would change";	##need different messages for --test
      Quiet=$QUIET; Test=$TEST				##Test gets set if --test OR (--prompt AND user says NO);  and serves to suppress file-replacement
      if [ $PROMPT ];then
        echo "---chg $cmdargs---"			##show the cmdline;  07feb: do here per-file if prompting  ==NOTE: want $cmdopts...
        difterse -- "$f" /tmp/chg$$			##show differences;  when using expand / unexpand, better without -b on diff
	echo -n "replace $f -- Y/N?";  read;		##prompt, asking user whether or not to revise
	[[ $REPLY == [Nn]* ]] && Test=1 && Quiet=1	##note: just CR means YES
      fi
      if [ $TEST ] || ! [ $Test ];then RC=0; fi		##supply return-code=zero if at least one file Changed - 07feb02
      if [ $VERBOSE ] && ! [ $Test ];then
        echo; echo "---chg $cmdargs---"			##show the cmdline; ==NOTE: want $cmdopts rather than $cmdargs with potentially long list of filenames (2010-04)
        difterse -- "$f" /tmp/chg$$			##show changes if $VERBOSE (2010-04)
      fi
      if [ $NOBKUP ];then
        [ $Test ]  || /bin/mv -f -- /tmp/chg$$ "$f"	##replace file $f, without making backup-copy
        [ $Quiet ] || echo "$Changed $f"
      elif [ -e "$fbk" ] && ((AGE<=300));then		##note: decision based on age of file, not on age of backup-copy
        [ $Test ]  || /bin/mv -f -- /tmp/chg$$ "$f"	##replace file $f, without making backup-copy
        [ $Quiet ] || echo "$Changed $f, without backup, as previous update $AGE seconds ago"
      else
        [ $Test ]  || /bin/mv -f -- "$f" "$fbk"		##replace the backup-copy, since more than 5 minutes ago (or nonexistent)
        [ $Test ]  || /bin/mv -f -- /tmp/chg$$ "$f"	##replace file $f
        [ $Quiet ] || echo "$Changed $f, with previous contents saved as $fbk"
      fi
    elif [ $MUST ];then
      echo "File $f was NOT changed by $cmdargs"	##sometimes want message for non-matching case (06dec)
    fi
    [ -e "/tmp/chg$$" ] && rm -f /tmp/chg$$		##cleanup unneeded tmp-file
  done
  for f in "$@";do					##do directories recursively
    [ -L "$f" ] && continue
    [ -d "$f" ] || continue
    chgfunc "$f"/..?* "$f"/.[^.]* "$f"/*		##could be simplified using bash's GLOBIGNORE='.'
  done
}

chgfunc "$@"
exit $RC


2005May: major revision, combining my chg, chgsed, chgsed-keeping-timestamp, chgtr scripts.
	wrote trivial chgsed, chgtr scripts to avoid having to kick old habits;
2006apr: new --fixhtm options, a replacement for my fixhtml-hi script, to convert non-ASCII chars to ampersand-entities.
2006jul: fixed the quirks with apostrophes in sed-patterns, I hope??
	 ==!!need similar change for sed-patterns that alter text containing backslash--??--
2007Aug: added --cmd option, to get a general-purpose way to make any filter into a file-changing-with-backup-and/or-prompting command;
   	 eg: chg --cmd "expand"   -- for a tabs-to-spaces reviser;
	 eg: chg --cmd "sort -u"  -- for a uniquifier
	 ==WILL NOW SCRAP the --sed and --tr params;  eg: chgsed will use --cmd instead;  but need to resolve some param-passing issues...
2008-12-28: very weird bug: chgsed refusing to modify a file named webalizer_lang.h;  ok on other files;  ok on same file named tmp.h;  WENT AWAY after reboot;
2010-04: added VERBOSE option;  ==both its and PROMPT's per-file showing of cmdline-args would be better without ALL filenames, just current one==!!==
	 added difterse function;  using it for VERBOSE-output and for PROMPT-output;