billmills / awk-lesson Goto Github PK

Short version: FS should be [/ :]+ not [/ :].

Long version:

On lines 4, 5, and 7 of earthquakes.dat $10 will be an empty string because their is a double space in the input file. Try changing print $3 to print $3, $10 to illustrate this.

You don't notice this because it only happens when the depth is an integer less than 10 and in awk "" < 10 (the empty string is mostly treated like 0).

In lesson.md after -F is introduced, we have:

{FS="[/ :]"}

This is problematic:

• This code is executed for every line of input, but the way it's introduced it looks like it only runs once (of course, in this case it isn't a problem, because repeating the assignment has the same effect).
• The assignment doesn't take place until after the first line is split. In this case you're not processing the first line, so you can't see the effect that this has. But if you do something like this: awk '{FS="[/ :]"}{print $1}' /etc/passwd | sed 9q you can see that the first line isn't split.

You don't get either of these problems when you use the -F option, so I would just do:

awk '-F[/ :]+' '{if (NR > 1) print $3}' earthquakes.dat

instead (the extra + fixes a different bug that I'm about to write up).

The awk code isn't idiomatic and looks unusual to awk programmers (like me).

For example: '{if (NR > 1) print $3}' would really be better as:

NR > 1 { print $3 }

(there is a rough equivalence between these two forms). This changes the pedagogy a bit, but I love awk, so I'm prepared to submit a Pull Request if you're okay with that.