Extracting just a phone number from a file - regex

Question

I'm sure the answer to this is already online however i don't know what i am looking for. I just started a course in Unix/Linux and my dad asked me to make him something for his work. He has a text file and on every fourth line there is a 10 digit number somewhere. How do i make a list of just the numbers? I assume the file looks something like this:
Random junk
Random junk fake number 1234567809
Random junk
My phone number is 1234567890 and it is here random numbers 32131;1231
Random junk
Random junk another fake number 2345432345
Random junk
Just kidding my phone number is here 1234567890 the date is mon:1231:31231
I assume its something like grep [1-9].\{9\} file but how do i get just lines 4,8,12 etc. Because i tested it and i get all phone numbers on every line. Also how do i get just the number not the whole line?
Any help will be greatly appreciated, even if its pointing me in the right direction so i can research it myself. Thanks.

Answer 1

You can do it in two steps:
$ awk '!(NR%4)' file | grep -Eo '[0-9]{10}'
1234567890
1234567890
awk '!(NR%4)' file prints those lines whose number is multiple of 4. It is the same as saying awk '(NR%4==0) {print}' file.
grep -Eo '[0-9]{10}' prints the numbers that appear on blocks of 10. Note that -o is for "just print the matches" and -E to use extended regular expressions.
Or also
$ awk '!(NR%4)' file | grep -Eo '[1-9][0-9]{9}' #check if first number is <>0

Answer 2

Using GNU sed:
sed -nr '0~4{s/.*\b([0-9]{10})\b.*/\1/p}' inputfile
Saying 0~4 produces every 4th line starting from the 0th line, i.e. produces every 4th line in the file. The substitution part is rather obvious.
For your sample input, it'd produce:
1234567890
1234567890

Answer 3

Since you are looking for one number per line, an awk solution would involve
awk '!(NR%4) && match($0, /[[:digit:]]{10}/){print substr($0, RSTART, RLENGTH)}' file

Answer 4

Using perl:
$ perl -nle 'print /([0-9]{10})/ if !($.%4)' file
1234567890
1234567890

Answer 5

To solve this, first, you should know what should be the length of the phone number. You should also consider area codes to be recognized by your code, and possible phone number start numbers. That way you will filter only the most possible true numbers. But if I write "My number is 028 2233 5674... Just kidding, it's 028 2233 9873." Then the code will consider both numbers as correct. So, to solve this completely, if there are fake numbers in the text, is nearly impossible. But an intelligent code, will filter the ones that are most likely to be correct.