Re: [ferret_users] constructing date strings like 19910825 instead of 25-AUG-1991

["Paulo B. Oliveira" <pboliveira@xxxxxxx>]

Hi,

I must be missing something... doesn't the "ZW" keyword in embedded expressions do the trick ?

def ax/t="10-jul-1987 12:00":"30-may-2016 12:00":1/unit=day/t0=1-jan-1900 t28y

let ymd = DAYS1900TOYMDHMS(t[gt=t28y])

def sym url0 http://data.remss.com

repeat/l=1:`ymd,ret=lend` (\
 def sym path ccmp/v02.0/y`ymd[k=1],ZW=4`/M`ymd[k=2],ZW=2`;\
 def sym fn CCMP_Wind_Analysis_`ymd[k=1],ZW=4``ymd[k=2],ZW=2``ymd[k=3],ZW=2`_V02.0_L3.0_RSS.nc;\
 sp wget ($url0)/($path)/($fn);\
)

On Tue, 2017-01-31 at 17:45 +0900, Ryo Furue wrote:
>     Hi Billy and ferret users,
>     
>     On Tue, Jan 31, 2017 at 4:57 PM, William Kessler <william.s.kessler@xxxxxxxx> wrote:
> >         RSS has released a new version (v2) of the CCMP wind product, consistently analyzed from Jul 1987 through May 2016 (http://www.remss.com/measurements/ccmp). As far as I can tell, this is only available from them, and only by ftp-ing the individual daily files.
> >         
> >         The whole global, 28-year fields are more than 285Gb. I want to write a script that will loop through the 10563 days, make daily averages from the four 6-hr values in each file, select my region of interest, and append that to a much smaller file (then delete the ftp-ed file).
> >         
> >         (Yes, I'll probably have to run it in chunks over several nights).
> >         
> >         My problem is to generate the filenames, which look like the example below:
> >         
> >         CCMP_Wind_Analysis_19910825_V02.0_L3.0_RSS.nc
> >         
> >         It's pretty straightforward to define a daily time axis spanning the CCMP dates, REPEAT-loop on the days, and come up with the year/month/day number needed for the filename (e.g. DAYS1900TOYMDHMS). But can anyone sugest a simple way to construct the filename given the need to have leading zeroes for months before October and days of the month less than 10?
> >     
>     
>     Would you be happy with a solution that combines ferret with another scripting language such as below?  The idea is to generate the filename outside and pass it to Ferret as an argument.
>     
>     
>     Ryo
>     
>     ----------------------------
>     
>     # I just took it from the Internet and modified it.
>     
>     # Run it as
>     
>     #     $ sh thisscript
>     
>     # on the command line.
>     
>     #!/bin/sh
>     startdate=1987-01-01
>     #enddate=2016-05-31
>     enddate=1987-03-01
>     
>     # GNU date command
>     #gdate=/usr/local/bin/gnu/date
>     gdate=gdate
>     
>     curr="$startdate"
>     while true; do
>         echo ferret -script yourscript.jnl "$curr" #<-- remove "echo" to run ferret
>         [ "$curr" \< "$enddate" ] || break
>         curr=$( $gdate +%Y-%m-%d --date "$curr +1 day" )
>     done
>     
>