############################################################################
#
#	File:     stripunb.icn
#
#	Subject:  Procedures to strip unbalanced material
#
#	Author:   Richard L. Goerwitz
#
#	Date:	  May 2, 2001
#
############################################################################
#
#   This file is in the public domain.
#
############################################################################
#
#	Version:  1.7
#
############################################################################
#  
#  This routine strips material from a line which is unbalanced with
#  respect to the characters defined in arguments 1 and 2 (unbalanced
#  being defined as bal() defines it, except that characters preceded
#  by a backslash are counted as regular characters, and are not taken
#  into account by the balancing algorithm).
#
#  One little bit of weirdness I added in is a table argument. Put
#  simply, if you call stripunb() as follows,
#
#      stripunb('<','>',s,&null,&null,t)
#
#  and if t is a table having the form,
#
#      key:  "bold"        value: outstr("\e[2m", "\e1m")
#      key:  "underline"   value: outstr("\e[4m", "\e1m")
#      etc.
#
#  then every instance of "<bold>" in string s will be mapped to
#  "\e2m," and every instance of "</bold>" will be mapped to "\e[1m."
#  Values in table t must be records of type output(on, off).  When
#  "</>" is encountered, stripunb will output the .off value for the
#  preceding .on string encountered.
#
############################################################################
#
#  Links: scan
#
############################################################################

link scan

global last_k
record outstr(on, off)


procedure stripunb(c1,c2,s,i,j,t)

    # NB:  Stripunb() returns a string - not an integer (like find,
    # upto).

    local lookinfor, bothcs, s2, k, new_s, c, compl
    #global last_k
    initial last_k := list()

    /c1 := '<'
    /c2 := '>'
    bothcs := c1 ++ c2
    lookinfor := c1 ++ '\\'
    c := &cset -- c1 -- c2

    /s := &subject
    if \i then {
	if i < 1 then
	    i := *s + (i+1)
    }
    else i := \&pos | 1
    if \j then {
	if j < 1 then
	    j := *s + (j+1)
    }
    else j := *s + 1

    s2 := ""
    s ? {
	while s2 ||:= tab(upto(lookinfor)) do {
	    if ="\\" then {
		if not any(bothcs) then
		    s2 ||:= "\\"
		&pos+1 > j & (return s2)
		s2 ||:= move(1)
		next
	    }
	    else {
		&pos > j & (return s2)
		any(c1) |
		    stop("stripunb:  Unbalanced string, pos(",&pos,").\n",s)
		if not (k := tab(&pos <= slashbal(c,c1,c2,&subject)))
		then {
		    # If the last char on the line is the right-delim...
		    if (.&subject[&pos:0]||" ") ? slashbal(c,c1,c2)
		    # ...then, naturally, the rest of the line is the tag.
		    then k := tab(0)
		    else {
			# BUT, if it's not the right-delim, then we have a
			# tag split by a line break.  Blasted things.
			return stripunb(c1,c2,&subject||read(&input),
					*.&subject,,t) |
			# Can't find the right delimiter.  Parsing error.
			stop("stripunb:  Incomplete tag\n",s[1:80] | s)
		    }
		}
		# T is the maptable.
		if \t then {
		    k ?:= 2(tab(any(c1)), tab(upto(c2)), move(1), pos(0))
		    if k ?:= (="/", tab(0)) then {
			compl:= pop(last_k) | stop("Incomplete tag, ",&subject) 
			if k == ""
			then k := compl
			else k == compl | stop("Incorrectly paired tag,/tag.")
			s2 ||:= \(\t[k]).off
		    }
		    else {
			s2 ||:= \(\t[k]).on
			push(last_k, k)
		    }
		}
	    }
	}
	s2 ||:= tab(0)
    }

    return s2

end