Is there a way to remove a line fully? - linux

Question

I'm using a one-line command to compile and print all of the animal names listed in a log file.
The WILD names are all listed in capital letters under the /wild directory.
The output should appear in the format of one name per line, with no duplicates:
ANT
BAT
CAT
I tried
grep 'wild' animal.txt | awk '{print $7}' | sed 's/[a-z0-9./]//g' | sort -u
It showed what I want but I want to remove the whole string which contains special characters like -, # ? %
Below is a sample of the file animal.txt
191.21.66.100 - - [21/Aug/1995:05:17:57 -0400] "GET /wild/elvpage.htm#ZOO HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:22:35 -0400] "GET /wild/S/s_26s.jpg HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:22:41 -0400] "GET /wild/struct.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:34 -0400] "GET /wild/elvpage.htm HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:36 -0400] "GET /wild/endball.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:37 -0400] "GET /wild/hot.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:38 -0400] "GET /wild/elvhead3.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:38 -0400] "GET /wild/PEGASUS/minpeg1.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:39 -0400] "GET /wild/DOG/DOG.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:39 -0400] "GET /wild/SWAN/SWAN.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:39 -0400] "GET /wild/ATLAS/atlas.gif HTTP/1.0"
191.21.66.100 - - [01/Aug/1995:02:27:40 -0400] "GET /wild/LIZARD/lizard.gif HTTP/1.0"
Below is a sample of my output after running the command:
ATLAS
ATLAS-
CAT_
DOG
%FACT
-KWM
?TIL-
#ZOO

Answer 1

Why not allow only capital A-Z and remove everything else:
grep 'wild' animal.txt | awk '{print $7}' | sed 's/[^A-Z]//g'
from your example input, this will return:
PEGASUS
DOGDOG
SWANSWAN
ATLAS
LIZARD
If you need to: you can further cleanup empty lines by appending |sed "/^$/d" and then sort

Answer 2

You can use a single GNU sed command:
sed -n 's!.*/wild/\([A-Z][A-Z]\+\)/.*!\1!p' animal.txt
Means:
-n: Do not print every line.
s!X!Y! Substitute X with Y.
.*/wild/\([A-Z][A-Z]\+\)/*: find a capital letter followed by at least one capital letter and preceded by wild/. These should be followed by a / and anything. Capture (remember) the capital letters.
!\1!: Replace whatever you found with the capital letter sequence.
p: If it was a match then print the line.
Gives:
PEGASUS
DOG
SWAN
ATLAS
LIZARD

Answer 3

This might work for you (GNU sed):
sed -E '/.*\/wild\/[^A-Z ]*([A-Z]+).*/!d # delete lines with no uppercase letters
s//\1/ # remove everything but uppercases letters
H # append word to the hold space
$!d # delete all lines but the last
x # swap to the hold space
:a # loop name space
s/((\n[^\n]+).*)\2/\1/ # remove duplicates
ta # repeat until failure
s/.//' file # remove introduced newline

Answer 4

GNU awk to get result:
grep 'wild' animal.txt | awk '
($0 = $7)
{gsub(/\//, " ", $0)}; #replace '/' with space so we can separate $0 to ($1, $2, $3);
(NF == 3 && length($2) > 2) #check if there is three word in line ($1, $2, $3) and then check if length($2) is more then 2 character
{print $2}'
| sort -u
Answer:
grep 'wild' animal.txt | awk '
($0 = $7) {gsub(/\//, " ", $0)};
(NF == 3 && length($2) > 2) {print $2}' | sort -u