bash - awk not capturing first line / separator -

i don't understand following behaviour:

this text file:

example.txt 12345   4321    hello hello test  blobb 14324   2131    test , incoming !   blubb 52341   1231    last test shutting down bla ... 

it consists of x rows of text, 4 tab-separated columns each. need first three, used awk (for first time):

awk '{fs="\t"; ofs="\t"; print $1,$2,$3}' < example.txt > excerpt.txt 

the result this:

excerpt.txt 12345   4321    hello 14324   2131    test , incoming ! 52341   1231    last test shutting down ... 

the first entry not contain full third column, , printing $1,$2,$3,$4 gives 12345 4321 hello hello first row. so, apparently separates @ whitespace (both after first , second hello), , not @ tab. checked if tab snuck in there, not case:

i find confusing, since works correctly other rows.

you setting field separators on every line. then, awk reads record (line) in way when cursor reached it, first time setting is too late apply first record.

since default field separator space, on first line uses it. then, second record, takes account set.

you need set either in begin block or before (the effect same):

awk 'begin{fs=ofs="\t"} {print $1,$2,$3}' example.txt > excerpt.txt  awk -f"\t" -v ofs="\t" '{print $1,$2,$3}' example.txt > excerpt.txt 

alternatively, can "recompile" record $1=$1. reinterprets record based on current field separators. should work well:

awk '{fs=ofs="\t"; $0=$0; print $1,$2,$3}' example.txt > excerpt.txt 

test

testing last thing, on recompiling field.

$ cat hello      me , here     doing awk $ awk '{fs="\t"; print $2}' doing awk $ awk '{fs="\t"; $0=$0; print $2}' me , doing awk