1 #!/bin/sh
2 
3 # input:
4 #	key (tab) string (tab) page numbers
5 #		command	command	123
6 #		command, data	command, [data]	11
7 #		command, display	command, [display]	11, 54, 63, 75
8 #		command, model	command, [model]	11
9 #		command, quit	command, [quit]	5, 16
10 # output:
11 #	key (tab) string (tab) page numbers
12 #		key	command  123
13 #		key	   [data]  11
14 #		key	   [display] ...
15 #		key	   [model] ...
16 #		key	   [quit] ...
17 
18 awk '
19 BEGIN	{ FS = OFS = "\t" }
20 
21 {	line[NR] = $0; x[NR] = $2 "\t" $3; y[NR] = $1 }
22 
23 # find a sequence that have the same prefix
24 # dump prefix, then each instance with spaces instead of prefix
25 END {
26 	for (i = 1; i <= NR; i = j+1) {
27 		j = findrun(i)		# returns last elem of run
28 		if (j > i)
29 			printrun(i, j)
30 		else
31 			print y[i], x[i]
32 	}
33 }
34 
35 function findrun(s,	j, p, np) {	# find y[s],y[s+1]... with same prefix
36 	p = prefix(y[s])
37 	np = length(p)
38 	for (j = s+1; j <= NR; j++) {
39 		if (y[j] == p)			# same, so include
40 			continue
41 		if (index(y[j], p) != 1)	# no match
42 			break
43 		c = substr(y[j], np+1, 1)
44 		if (c != " " && c != ",")	# has to be whole word prefix
45 			break
46 	}
47 	return j-1
48 }
49 
50 function prefix(s,	n) {	# find 1st word of s: same sort key, minus ,
51 	gsub(/,/, "", s)
52 	n = index(s, " ")
53 	if (n > 0)
54 		return substr(s, 1, n-1)
55 	else
56 		return s
57 }
58 
59 function printrun(s, e,		i) {	# move [...] to end, "see" to front
60 	s1 = 0; e1 = 0; p1 = 0; i1 = 0
61 	for (i = s; i <= e; i++) {
62 		if (x[i] ~ /{see/) {		# see, see also
63 			sx[s1] = x[i]
64 			sy[s1] = y[i]
65 			s1++
66 		} else if (x[i] ~ /^\[/) {	# prefix word is [...]
67 			px[p1] = x[i]
68 			py[p1] = y[i]
69 			p1++
70 		} else if (x[i] ~ /\[.*\]/) {	# [...] somewhere else
71 			ex[e1] = x[i]
72 			ey[e1] = y[i]
73 			e1++
74 		} else {			# none of the above
75 			ix[i1] = x[i]
76 			iy[i1] = y[i]
77 			i1++
78 		}
79 	}
80 	if (e-s+1 != s1 + p1 + i1 + e1) print "oh shit" >"/dev/stderr"
81 
82 	for (i = 0; i < s1; i++)	# "see", one/line
83 		print sy[i], sx[i]
84 	if (i1 > 1)
85 		printgroup(ix,iy,0,i1)	# non [...] items
86 	else if (i1 == 1)
87 		print iy[0], ix[0]
88 	if (e1 > 1)
89 		printgroup(ex,ey,0,e1)	# prefix [...] items
90 	else if (e1 == 1)
91 		print ey[0], ex[0]
92 	# for (i = 0; i < p1; i++)	# [prefix] ... items
93 	# 	print py[i], px[i]
94 	if (p1 > 1)
95 		printgroup(px,py,0,p1)	# [prefix] ... items
96 	else if (p1 == 1)
97 		print py[0], px[0]
98 }
99 
100 function printgroup(x, y, s, e,		i, j) {
101 	split(x[s], f23)
102 	if (split(f23[1], temp, " ") > 1) {
103 		pfx = temp[1] " " temp[2]	# 2-word prefix
104 		for (i = s+1; i < e; i++) {
105 			if (index(x[i], pfx) != 1)
106 				break
107 			c = substr(x[i], length(pfx)+1, 1)
108 			if (c != " " && c != ",")	# has to be whole word prefix
109 				break
110 		}
111 		if (i == e) {
112 			# print "got a run with", pfx
113 			sub(/ /, "@", f23[1])
114 			for (i = s; i < e; i++)
115 				sub(/ /, "@", x[i])	# take @ out later
116 		}
117 	}
118 	n = sub(/,?[ ~]+.*/, "", f23[1]) # zap rest of line
119 
120 	sub(/,$/, "", f23[1])
121 	if (n > 0) {	# some change, so not a single word
122 		sub(/@/, " ", f23[1])
123 		print y[s], f23[1]	# print main entry
124 	}
125 	for (j = s; j < e; j++) {
126 		split(x[j], f23)
127 		sub(/^[^, ]+[, ]+/, "   ", f23[1])
128 		sub(/@/, " ", f23[1])
129 		print y[s], f23[1], f23[2]
130 	}
131 }
132 
133 ' $*