fuzzy sequence match in a text file

Shell Programming and Scripting
1

Hi Forum:

I have struggle with it and decide to use my eye ball to accomplish this.

Basically I am looking for sequence of date inside a file.
If one of the sequence repeat 2-3 time or skip once; it's still consider a match.

input text file:

Sep 6 A
Sep 6 A
Sep 10 A
Sep 7 B
Sep 8 B
Sep 9 B
Sep 10 B
Sep 11 B
Sep 7 C
Sep 7 C
Sep 7 C
Sep 11 C
Sep 8 D
Sep 9 E
Sep 7 F
Sep 8 F
Sep 9 F
Sep 10 F
Sep 11 F
Sep 7 G
Sep 8 G
Sep 9 G
Sep 10 G
Sep 7 H
Sep 8 H
Sep 9 H
Sep 7 I
Sep 8 I
Sep 8 I
Sep 9 I
Sep 10 I
Sep 7 J
Sep 7 J
Sep 7 J
Sep 9 J
Sep 10 J

Desired filtered output:

Sep 7 B
Sep 8 B
Sep 9 B
Sep 10 B
Sep 11 B
Sep 7 F
Sep 8 F
Sep 9 F
Sep 10 F
Sep 11 F
Sep 7 G
Sep 8 G
Sep 9 G
Sep 10 G
Sep 7 H
Sep 8 H
Sep 9 H
Sep 7 I
Sep 8 I
Sep 8 I
Sep 9 I
Sep 10 I
Sep 7 J
Sep 7 J
Sep 7 J
Sep 9 J
Sep 10 J

Cheers!!
Chirish.

2

Try this:

awk '
{ min=day
  max=skip?day+1:day+2
  if($1==mth && $2+0>=min && $2 <=max) {
    if($2+0>min)diff++
    skip=skip||$2+0==day+2
    day=$2+0
    out=start out"\n"$0
    start=""
    next
  }
  if(diff>2) printf "%s\n",out
  mth=$1
  start=$0
  day=$2+0
  diff=1; skip=0; out="" }
END {if(diff>2) printf "%s\n",out}' infile

Edit: rename variables for more clarity

3

Wow, Chubler_XL, I stand in awe... After thirty years or so using Unix, Linux, and awk (among others, see, this is my work AND my hobby too), I am completely stupefied at:

skip=skip||$2+0==day+2

and

max=skip?day+1:day+2

Please, don't get me wrong: It is amazing to cut your code, paste it in my terminal and see the expected output... Is like listening "Bazinga!" in the background!
Would you please give us mere mortals a bit of feedback?

4

Apologies, this site seem focused on smallest number of chars, not clarity.

if(skip != 0 || $2+0 == day + 2) skip=1
 
# or
if (skip == 0) {
    if ($2+0 == day + 2) skip =1
}

if(skip ==0) max=day+2 else max=day+1
1 Like
5

I don't follow the logic behind

if(skip != 0 || $2+0 == day + 2) skip=1

# or
if (skip == 0) {
    if ($2+0 == day + 2) skip =1
}

being equivalent... The first statement is "if ((skip is not 0) OR (($2+0) equals (day + 2)))", but your second expression is "if (skip equals 0) THEN if (($2+0) is equal to (day + 2))", and as such it is NOT equivalent to the first one; in the first statement it suffices that "(skip is not 0)" to make "skip=1", but in the second expression this is not true. As I see it, the second expression is equivalent to "if ((skip equals 0) AND (($2+0) equals (day + 2)))", which is clearly different from the first statement.

7

hexram is correct. Those two expressions are not logically equivalent (you cannot manipulate one into the form of the other using boolean identities/properties/theorems). They behave differently when skip is not zero. However, the difference in behavior does not affect the outcome if the non-zero value is 1.

Regards,
Alister

1 Like
8

Thanks for the clarification Alister,

In this implementation skip is supposed to be a boolean variable (only has value of zero or 1).

I use this "locked on" expression with booleans a lot in programming, and using the var=var?var:expr one-liner method is almost unconscious. So while explaining what is really happening, it's quite easy to forget about any illegal (non zero/1) values the flag could have.