Using sed to index files [duplicate] - bash

Question

file1 contains multiple alphabetic sequences:
AETYUIOOILAKSJ
EAYEURIOPOSIDK
RYXURIAJSKDMAO
URITORIEJAHSJD
YWQIAKSJDHFKCM
HAJSUDIDSJSIAJ
AJDHDPFDIXSIBJ
JAQIAUXCNCVUFO
while file2 contains indexes of the sequences which I want to pull out and transfer to another file. For example, 3T means I want the sequence with a T at position 3 from within file1.
In reality both files are very large with thousands of indexes and sequences.
file2:
3T
10K
14D
1J
Desired output:
AETYUIOOILAKSJ
RYXURIAJSKDMAO
URITORIEJAHSJD
JAQIAUXCNCVUFO
Ideally the output should match the order of indexes in file2. In other words the first index "3T" matches sequence "AETYUIOOILAKSJ" and thus this is the first sequence in the new file.
Things I have tried:
grep -f file2 file1
grep -fov file2 file1 # possibly to filter for those non-matching entries
I have also used the command line tool sift but am still having difficulty.
Thanks

Answer 1

$ cat tst.awk
NR==FNR {
lgth = length($0)
pos2char[substr($0,1,lgth-1)] = substr($0,lgth,1)
next
}
{
for (pos in pos2char) {
if ( substr($0,pos,1) == pos2char[pos] ) {
print
next
}
}
}
$ awk -f tst.awk file2 file1
AETYUIOOILAKSJ
RYXURIAJSKDMAO
URITORIEJAHSJD
JAQIAUXCNCVUFO

Answer 2

With awk + grep pipeline:
awk '{ pat=sprintf("%*s", int($0)-1, ""); gsub(" ", ".", pat);
printf "^%s%s\n", pat, substr($0, length) }' file2 | grep -f- file1
The output:
AETYUIOOILAKSJ
RYXURIAJSKDMAO
URITORIEJAHSJD
JAQIAUXCNCVUFO

Answer 3

Here you go:
awk 'NR==FNR {b[$0]++;next} {for (i in b) {a=match($0,"[A-Z]");n=substr($0,1,(a-1));s=substr($0,a);t=substr(i,n,1);if (t==s) print i}}' file1 file2
AETYUIOOILAKSJ
RYXURIAJSKDMAO
URITORIEJAHSJD
JAQIAUXCNCVUFO
Some more readable:
awk '
NR==FNR {
b[$0]++;
next
}
{
for (i in b) {
a=match($0,"[A-Z]");
n=substr($0,1,(a-1));
s=substr($0,a);
t=substr(i,n,1);
if (t==s)
print i
}
}
' file1 file2
With comments:
awk '
NR==FNR { # For the first file
b[$0]++; # Store file1 in in array b
next
}
{
for (i in b) { # Loop trough elements in array b
a=match($0,"[A-Z]"); # For file2 find where letters starts
n=substr($0,1,(a-1)); # Store the number part of file2 in n
s=substr($0,a); # Store the letters part of file2 in s
t=substr(i,n,1); # from file1 find string at position n
if (t==s) # test if string found is equal to letter to find s
print i # if yes, print the line
}
}
' file1 file2

Answer 4

awk '(NR==FNR){a[$0]=substr($0,length);next}
{ for(key in a) if (a[key] == substr($0,key+0,1)) { print; break }
}' file2 file1
Here, the array a[key] is a associative array with the following key-value pairs:
key: value
3T T
10K K
... ...
When processing file2 with the line: (NR==FNR){a[$0]=substr($0,length);next}: we extract the value beforehand so we don't have to do it later on. The index is easily extracted with a math operation. Eg. "10K"+0=10 in Awk.
Processing file1 is done with the next line. Here we just check if the character matches for any of the entries in the associative array.

Answer 5

With GNU awk and grep:
awk -v FPAT='[0-9]+|[A-Z]+' '{ print "^.{" $1-1 "}" $2 }' file1 | grep -Ef - file2
Output:
AETYUIOOILAKSJ
RYXURIAJSKDMAO
URITORIEJAHSJD
JAQIAUXCNCVUFO