regex needed (?)

czezz · 06-22-2009, 08:06 AM

I have a text file /export/home/tester/file

Code:

Sbalbalbav
erypkerioghjy
tyilerjgyioerjgy
erterilyer
r;yke5ophy
rghryrxxxxxx 
WHERE
UINETDATA_DAILY.begT >= 1245103200000 AND UINETDATA_DAILY.begT < 1245189600000 AND
ufhwerukgher
rklgherio;

What I need to do here is to change/replace values at UINETDATA_DAILY.begT >=
and UINETDATA_DAILY.begT <

1st value should be replace with 1245189600000 (1245103200000+86400000)
2nd value should be replace with 1245276000000 (1245189600000+86400000)

Can anybody help me with that please ?

colucix · 06-22-2009, 08:39 AM

Code:

awk '/^UINETDATA/{
  # Read the value of nit
  for (i=1; i<=NF; i++){
  if ($i == ">=")
     gsub($(i+1),$(i+1)+86400000,$(i+1))
  if ($i == "<")
     gsub($(i+1),$(i+1)+86400000,$(i+1))
  }
}1' /export/home/tester/file

Use nawk if running on Solaris.

czezz · 06-25-2009, 03:06 AM

MAGIC!

Can you help me to better understand code ?
My comments/questions followed by ##

Code:

## looking for line with string UINETDATA in /export/home/tester/file
nawk '/^UINETDATA/{

  # Read the value of nit

## some loop (?)
  for (i=1; i<=NF; i++){

## if the sign next to UINETDATA is ">=" then to $i... which is the value after ">=" sign ?
  if ($i == ">=")
     gsub($(i+1),$(i+1)+86400000,$(i+1))

  if ($i == "<")
     gsub($(i+1),$(i+1)+86400000,$(i+1))
  }
}1' /export/home/tester/file

colucix · 06-25-2009, 04:11 AM

The loop parses field after field

Code:

for (i=1; i<=NF; i++)

it loops from the first field to the last (e.g. the NF-th field). The line containing UINETDATA in your example has 8 fields. Then it checks if the i-th field is ">=" and substitute the next field (which is the value you want) with itself plus 86400000. The same when it encounters the "<" sign. Take in mind that the gsub function actually change the value of the field, so that when the line is printed out the new value is shown.

The print out of every line is performed by the 1 at the end of the awk script. It is simply an expression always true. In that case you don't need to use the print statement, since awk performs a print as "default action". In other words, doing

Code:

or

Code:

{print}

is the same.