| 1 | # uniq.awk --- do uniq in awk
|
|---|
| 2 | #
|
|---|
| 3 | # Requires getopt and join library functions
|
|---|
| 4 | #
|
|---|
| 5 | # Arnold Robbins, [email protected], Public Domain
|
|---|
| 6 | # May 1993
|
|---|
| 7 |
|
|---|
| 8 | function usage( e)
|
|---|
| 9 | {
|
|---|
| 10 | e = "Usage: uniq [-udc [-n]] [+n] [ in [ out ]]"
|
|---|
| 11 | print e > "/dev/stderr"
|
|---|
| 12 | exit 1
|
|---|
| 13 | }
|
|---|
| 14 |
|
|---|
| 15 | # -c count lines. overrides -d and -u
|
|---|
| 16 | # -d only repeated lines
|
|---|
| 17 | # -u only non-repeated lines
|
|---|
| 18 | # -n skip n fields
|
|---|
| 19 | # +n skip n characters, skip fields first
|
|---|
| 20 |
|
|---|
| 21 | BEGIN \
|
|---|
| 22 | {
|
|---|
| 23 | count = 1
|
|---|
| 24 | outputfile = "/dev/stdout"
|
|---|
| 25 | opts = "udc0:1:2:3:4:5:6:7:8:9:"
|
|---|
| 26 | while ((c = getopt(ARGC, ARGV, opts)) != -1) {
|
|---|
| 27 | if (c == "u")
|
|---|
| 28 | non_repeated_only++
|
|---|
| 29 | else if (c == "d")
|
|---|
| 30 | repeated_only++
|
|---|
| 31 | else if (c == "c")
|
|---|
| 32 | do_count++
|
|---|
| 33 | else if (index("0123456789", c) != 0) {
|
|---|
| 34 | # getopt requires args to options
|
|---|
| 35 | # this messes us up for things like -5
|
|---|
| 36 | if (Optarg ~ /^[0-9]+$/)
|
|---|
| 37 | fcount = (c Optarg) + 0
|
|---|
| 38 | else {
|
|---|
| 39 | fcount = c + 0
|
|---|
| 40 | Optind--
|
|---|
| 41 | }
|
|---|
| 42 | } else
|
|---|
| 43 | usage()
|
|---|
| 44 | }
|
|---|
| 45 |
|
|---|
| 46 | if (ARGV[Optind] ~ /^\+[0-9]+$/) {
|
|---|
| 47 | charcount = substr(ARGV[Optind], 2) + 0
|
|---|
| 48 | Optind++
|
|---|
| 49 | }
|
|---|
| 50 |
|
|---|
| 51 | for (i = 1; i < Optind; i++)
|
|---|
| 52 | ARGV[i] = ""
|
|---|
| 53 |
|
|---|
| 54 | if (repeated_only == 0 && non_repeated_only == 0)
|
|---|
| 55 | repeated_only = non_repeated_only = 1
|
|---|
| 56 |
|
|---|
| 57 | if (ARGC - Optind == 2) {
|
|---|
| 58 | outputfile = ARGV[ARGC - 1]
|
|---|
| 59 | ARGV[ARGC - 1] = ""
|
|---|
| 60 | }
|
|---|
| 61 | }
|
|---|
| 62 | function are_equal( n, m, clast, cline, alast, aline)
|
|---|
| 63 | {
|
|---|
| 64 | if (fcount == 0 && charcount == 0)
|
|---|
| 65 | return (last == $0)
|
|---|
| 66 |
|
|---|
| 67 | if (fcount > 0) {
|
|---|
| 68 | n = split(last, alast)
|
|---|
| 69 | m = split($0, aline)
|
|---|
| 70 | clast = join(alast, fcount+1, n)
|
|---|
| 71 | cline = join(aline, fcount+1, m)
|
|---|
| 72 | } else {
|
|---|
| 73 | clast = last
|
|---|
| 74 | cline = $0
|
|---|
| 75 | }
|
|---|
| 76 | if (charcount) {
|
|---|
| 77 | clast = substr(clast, charcount + 1)
|
|---|
| 78 | cline = substr(cline, charcount + 1)
|
|---|
| 79 | }
|
|---|
| 80 |
|
|---|
| 81 | return (clast == cline)
|
|---|
| 82 | }
|
|---|
| 83 | NR == 1 {
|
|---|
| 84 | last = $0
|
|---|
| 85 | next
|
|---|
| 86 | }
|
|---|
| 87 |
|
|---|
| 88 | {
|
|---|
| 89 | equal = are_equal()
|
|---|
| 90 |
|
|---|
| 91 | if (do_count) { # overrides -d and -u
|
|---|
| 92 | if (equal)
|
|---|
| 93 | count++
|
|---|
| 94 | else {
|
|---|
| 95 | printf("%4d %s\n", count, last) > outputfile
|
|---|
| 96 | last = $0
|
|---|
| 97 | count = 1 # reset
|
|---|
| 98 | }
|
|---|
| 99 | next
|
|---|
| 100 | }
|
|---|
| 101 |
|
|---|
| 102 | if (equal)
|
|---|
| 103 | count++
|
|---|
| 104 | else {
|
|---|
| 105 | if ((repeated_only && count > 1) ||
|
|---|
| 106 | (non_repeated_only && count == 1))
|
|---|
| 107 | print last > outputfile
|
|---|
| 108 | last = $0
|
|---|
| 109 | count = 1
|
|---|
| 110 | }
|
|---|
| 111 | }
|
|---|
| 112 |
|
|---|
| 113 | END {
|
|---|
| 114 | if (do_count)
|
|---|
| 115 | printf("%4d %s\n", count, last) > outputfile
|
|---|
| 116 | else if ((repeated_only && count > 1) ||
|
|---|
| 117 | (non_repeated_only && count == 1))
|
|---|
| 118 | print last > outputfile
|
|---|
| 119 | }
|
|---|
