#!/bin/bash
## uniquify a list of email-addresses -- by Eugene Reimer  2007-Oct;
## combining items such as:
##	Gord Thomson <GordThomson@sasktel.net>
##	Marge Thomson <gordthomson@sasktel.net>		<--combine to: Gord Thomson or Marge Thomson <gordthomson@sasktel.net>
##
##	Lynwood Cundall <lcundall@saskpower.com>
##	Lynwood Cundall <LCundall@saskpower.com>	<--combine to: Lynwood Cundall <lcundall@saskpower.com>
##
## Note that it uniquifies by lowercased email-address, and leaves the list in that order;  to aid in the elimination of duplicates that have arisen due to 
##	someone having changed email-addresses (where the problem is that you may end up having both the old and the new in your list), you'll probably want 
##	to do a case-blind sort by Name (by the entire line) as an aid in spotting such.  I've decided to leave that sort of uniquifying as a semi-manual 
##	process because for one thing a mechanical method would be so easily foiled by minor variations in the spelling of "Name", and also because there 
##	may well be cases where one really wants the same person to receive two copies, at say both home and work email-addresses.
##
## testcase:  /books/ELIST-Regina-PCAP-PCESC-MailingList

cat $1 |sed '
  /</!s|.*|& <&>|;				##convert just email-address to NM <ID> format  (stdization needed for sort)
  /^</s|<\(.*\)>|\1 <\1>|;			##convert just email-address to NM <ID> format  (stdization needed for sort)
  /<.*</s|^|--LIST |;				##KLUDGE for mailing-list items from exported-addressbook...
  s|\(<[^>]*>\)|\L\1|;				##convert pointy-bracketed part to lowercase
  s|mb.sympatico.ca|mts.net|g;			##KLUDGE for items from exported-addressbook...
  s|^ *||; s|[ ,]*$||;				##remove leading spaces and trailing space|comma
  s| |~|g;s|<|	<|;				##space->tilde;  lt->TAB+lt  (kludges for field-oriented sort)
' |sort -k2,2 |sed '				##SORT on email-address only
  s|~| |g; s|	<|<|;				##undo the sort-kludges
' >tmpuniqify$$A				##to tmpfile - due to subshell rules...

while read;do					##read line-by-line to combine (partial) duplicates
  NM="${REPLY%%<*}"; ID="<${REPLY#*<}"				##split into 2 parts
  NM=${NM% }							##revise NM removing trailing space
  if [[ $NM == *@* ]];then NM=${NM%%@*}; fi			##revise NM containing at-sign  (decided not to undo first stdization)
  NM=${NM//[_.]/ };  NM=${NM//  / }				##revise NM revising some troublesome characters to spaces
  NM=${NM//[,+\'()]/}						##revise NM removing other troublesome characters
  if [[ $ID == $PRVID ]];then
    if [[ $NM != $PRVNM ]];then NM="$PRVNM or $NM"; fi
    echo "combining 2 lines for $ID to get:$NM"  >&2		##msg about combining 2 lines
  elif [[ $PRVID != "" ]]; then
    echo "$PRVNM $PRVID"					##emit one line
  fi
  PRVNM=$NM;  PRVID=$ID
done  <tmpuniqify$$A  >tmpuniqify$$B				##from tmpuniqify$$A, to tmpuniqify$$B
[[ $PRVID != "" ]] && echo "$PRVNM $PRVID"  >>tmpuniqify$$B	##emit last line

if cmp -s $1 tmpuniqify$$B;then
  echo "$1 unchanged"						##unchanged msg
else
  mv -f $1 $1~;  mv tmpuniqify$$B $1				##rename with tilde-backup
  echo "previous $1 saved as $1~"				##msg
  wc $1~ $1							##show nbr lines before & after
  ##diffsort $1~ $1|m						##DEBUG -- view diffs in sorted order by entire line (by NM)
fi
rm -f tmpuniqify$$*						##cleanup