#!/bin/bash
## Split an HTML file into subsets -- for things formerly handled by hash-links;  Eugene Reimer  2009;
##
## EXAMPLE:
## if about.htm contains the following A-NAME tags:
##	<a name=Memberships>
##	<a name=Elist>
##	<a name=CopyrightPolicy>
## then the commandline  webmksubsets about.htm  will produce the following subset-files:
##	about-Memberships.htm
##	about-Elist.htm
##	about-CopyrightPolicy.htm
## with each containing everything from its A-NAME tag until the next such tag, and having a timestamp identical to the source-file's;

BACKUP=1
while [ $# -gt 0 ] && [[ $1 == -* ]];do			##parse options
  case "$1" in
  -n|--nob*)	BACKUP=			;;		##no-backup option
  --)		shift; break		;;		##ender for options
  esac
  shift;						##remove arg
done
if [ $# -ne 1 ];then  echo "Usage:  webmksubsets  [--nobackup]  SOURCEFILE";  exit;  fi

export NMB=${1%.*}
##port HD=$(cat $1 |sed -n '0,/<body/p'        |tr '\n' '\v')						##save HEAD+BODYtag;  see also webnavreplicate
export HD=$(cat $1 |sed -n '0,/<!-- end nav/p' |tr '\n' '\v')						##save HEAD+NavCode;  see also webnavreplicate

export CART=
if grep -q nopercart.js $1;then CART=$(cat $1 |sed -n '/<NOSCRIPT>/,/=nopercart.js/p' |tr '\n' '\v');fi	##save Shopping-Cart code  (needed for noci shopping.htm)

BKDIR=/tmp/bkmksubsets-$NMB
if [ $BACKUP ];then  mkdir -p $BKDIR;  mv -fv $NMB-* $BKDIR;  fi					##keep old versions as backups;  and start with clean slate

cat $1 |tr '\n' '\v' |sed 's|<a name=|\n&|g' |grep '<a name='  |while read;do				##convert to one A-NAME-section per line;  read line-by-line...
  NMA=${REPLY%%>*};  NMA=${NMA#*=}
  TO=$NMB-$NMA.htm											##form filename for output-file
  echo "webmksubsets: writing $TO"
  echo "$HD$CART$REPLY<p><br>"  |sed "s|</title| $NMA&|; s|<hr>||g; s|h3>|h2>|g"  |tr '\v' '\n'  >$TO	##note: the TITLE, HR, and H3-->H2 conventions
  touch --reference="$1"  $TO										##make subset-file's timestamp be same as parent's
done
##for S in $NMB-*;do [ -e $S ] && echo -e "\n\n==" && diffsed '0,/<!-- end nav/d' $BKDIR/$S $S; done	##DEBUG -- compare omitting nav-code
##d -l $NMB-* $BKDIR/$NMB-*										##DEBUG


exit
webput (noci-part) uses this for:  projects  pictures  shopping;
shopping-subsets:  each getting the META KEYWORDS stuff seems inappropriate, especially for shopping-Donations.htm??
	shopping-PinsBooksDVDs.htm:  losing the HR-separators seems wrong -- may never need that subset, using entire shopping.htm instead??
	WANTED: Membership+Donation shopping-subset??
	==RETHINK: instead of shopping-Subsets, better to make re-ordered but complete pages==??==
CONSIDER: an OOM-subset for each newletter==??==