shithub: 9intro

ref: 87288afa5ac476efb3ef1ed40df658427339f13e
dir: /idx/hierarchy/

#!/bin/rc

# input:
#	key (tab) string (tab) page numbers
#		command	command	123
#		command, data	command, [data]	11
#		command, display	command, [display]	11, 54, 63, 75
#		command, model	command, [model]	11
#		command, quit	command, [quit]	5, 16
# output:
#	key (tab) string (tab) page numbers
#		key	command  123
#		key	   [data]  11
#		key	   [display] ...
#		key	   [model] ...
#		key	   [quit] ...

awk '
BEGIN	{ FS = OFS = "\t" }

{	line[NR] = $0; x[NR] = $2 "\t" $3; y[NR] = $1 }

# find a sequence that have the same prefix
# dump prefix, then each instance with spaces instead of prefix
END {
	for (i = 1; i <= NR; i = j+1) {
		j = findrun(i)		# returns last elem of run
		if (j > i)
			printrun(i, j)
		else
			print y[i], x[i]
	}
}

function findrun(s,	j, p, np) {	# find y[s],y[s+1]... with same prefix
	p = prefix(y[s])
	np = length(p)
	for (j = s+1; j <= NR; j++) {
		if (y[j] == p)			# same, so include
			continue
		if (index(y[j], p) != 1)	# no match
			break
		c = substr(y[j], np+1, 1)
		if (c != " " && c != ",")	# has to be whole word prefix
			break
	}
	return j-1
}

function prefix(s,	n) {	# find 1st word of s: same sort key, minus ,
	gsub(/,/, "", s)
	n = index(s, " ")
	if (n > 0)
		return substr(s, 1, n-1)
	else
		return s
}

function printrun(s, e,		i) {	# move [...] to end, "see" to front
	s1 = 0; e1 = 0; p1 = 0; i1 = 0
	for (i = s; i <= e; i++) {
		if (x[i] ~ /{see/) {		# see, see also
			sx[s1] = x[i]
			sy[s1] = y[i]
			s1++
		} else if (x[i] ~ /^\[/) {	# prefix word is [...]
			px[p1] = x[i]
			py[p1] = y[i]
			p1++
		} else if (x[i] ~ /\[.*\]/) {	# [...] somewhere else
			ex[e1] = x[i]
			ey[e1] = y[i]
			e1++
		} else {			# none of the above
			ix[i1] = x[i]
			iy[i1] = y[i]
			i1++
		}
	}
	if (e-s+1 != s1 + p1 + i1 + e1) print "oh shit" >"/dev/stderr"

	for (i = 0; i < s1; i++)	# "see", one/line
		print sy[i], sx[i]
	if (i1 > 1)
		printgroup(ix,iy,0,i1)	# non [...] items
	else if (i1 == 1)
		print iy[0], ix[0]
	if (e1 > 1)
		printgroup(ex,ey,0,e1)	# prefix [...] items
	else if (e1 == 1)
		print ey[0], ex[0]
	# for (i = 0; i < p1; i++)	# [prefix] ... items
	# 	print py[i], px[i]
	if (p1 > 1)
		printgroup(px,py,0,p1)	# [prefix] ... items
	else if (p1 == 1)
		print py[0], px[0]
}

function printgroup(x, y, s, e,		i, j) {
	split(x[s], f23)
	if (split(f23[1], temp, " ") > 1) {
		pfx = temp[1] " " temp[2]	# 2-word prefix
		for (i = s+1; i < e; i++) {
			if (index(x[i], pfx) != 1)
				break
			c = substr(x[i], length(pfx)+1, 1)
			if (c != " " && c != ",")	# has to be whole word prefix
				break
		}
		if (i == e) {
			# print "got a run with", pfx
			sub(/ /, "@", f23[1])
			for (i = s; i < e; i++)
				sub(/ /, "@", x[i])	# take @ out later
		}
	}
	n = sub(/,?[ ~]+.*/, "", f23[1]) # zap rest of line

	sub(/,$/, "", f23[1])
	if (n > 0) {	# some change, so not a single word
		sub(/@/, " ", f23[1])
		print y[s], f23[1]	# print main entry
	}
	for (j = s; j < e; j++) {
		split(x[j], f23)
		sub(/^[^, ]+[, ]+/, "   ", f23[1])
		sub(/@/, " ", f23[1])
		print y[s], f23[1], f23[2]
	}
}

' $*