UNIX for Dummies Questions & Answers

I have a long list of numbers; let's call them integers. Here is an excerpt:

01072523
0371
037408
0382028
038224
01

I want to find integers that contain a certain pattern that I'll describe as a "relationship between adjacent digits".

The pattern is a triplet:

digit digit digit

where:

#1 the relationship between the first and second digit is an increment of at least 3 and

#2 the relationship between the second and third digit is also an increment of at least 3 and

#3 the first digit in the triplet is not at "position 1" in the integer i.e. exclude the first digit of the integer i.e. exclude the first triplet.

Examples:

If integer is 01072523
then two matching triplets

01072523
...^^^

01072523
....^^^

If integer is 037408
then three matching triplets

037408
.^^^

037408
..^^^

037408
...^^^

(The digit at position 1 in the integer 037408 is "0" and the first triplet "037" is excluded from the pattern matching according to rule #3 above.)

If I can use Bourne sh and awk that would be the best.

Cheers!

what have you tried??
sounds like homework to me

I guess I write too much like a professor. I triggered the homework alert.

This can't be homework because I'm not studying Comp Sci or taking any UNIX courses. I already have several degrees; alas none in Maths, Physics, Info Sci or any degree that required it. I am self-taught.

I learn via simple examples, oftentimes from sites like this one.

I must be doing OK if I can write questions that could qualify as UNIX homeworks. Knowing how to ask the right questions is the biggest challenge in learning.

Cheers

If it's not homeworks, indeed, it very looks like...any problem can be solved with bash, using any other external program...when you know exactly what it's about...
This is mathematics puzzle to me, and as such I won't be helpful.
It's not undoable: parsing each line of the file, getting for each line what section(s) of it you want to keep, using arithmetics evaluation to compare with what section of the line...

When I see it 'bash coded', I might understand...
...by now, I suggest maintainers/moderators to transfer this post to a more apropriate place such as 'experts...', erasing my message.

wouldn't you?

Thanks for some perspective dapeach. I'm flattered that you guys think I sound like a UNIX professor.

I was thinking this could be done simply and was just a matter of me not knowing the shell arithmetic functions and use of conditionals.

Perhaps it's more complex than I thought. Maybe it could it be done with string comparison (since there are only so many matching triplet possibilities)?

I'll leave this question up for a few more hours. If no one answers I'll remove it.

I apologise if this question is not really a good one for unix.com (i.e. it wouldn't really be a good learning exercise). I didn't think about it enough. But I did make it sound like a homework assigment... so maybe I get a +1 for writing style.

OK OK!!!! I agree its not home work question
but i didn't understand how 072 is triplet in 01072523??
you mean 0-7>3 and 7-2>3 {first-sec or sec-first>3}???
if so below code will work.. I know it can be done much smaller way..but for better understanding i wrote like this

Code:

awk '{len=length($0)}{if(len>3){for(i=2;i<=len-2;i++)
{first=substr($0,i,1);sec=substr($0,i+1,1);third=substr($0,i+2,1);
if(((sec-first)>=3 || (first-sec)>=3) && ((third-sec)>=3 || (sec-third)>=3)) {print substr($0,i,3)}}}}' filename

Alternatively ksh or bash with a file called 'numbers'

Code:

#!/bin/ksh
while read number; do
  for (( i=4; i<=${#number}; i++ )) ; do
    (( incr1=${number:i-3:1}-${number:i-2:1}))
    (( incr2=${number:i-2:1}-${number:i-1:1}))
    if (( ${incr1#-} >= 3 && ${incr2#-} >= 3 )); then
      echo $number
      echo ${number:i-3:3}
    fi
  done
done<numbers