Calculation aid script!

assasukasse · 10-25-2006, 05:21 AM

Hello
i have a very tedious task to be automated:
i am running a computational analysis, and i need to edit MANY time the coordinates on a file..
the file is a txt, and the coordinate to be edited are those:

lines before (which i don't need to edit)
SDEF pos= x y z blabla blablabla
lines after (which i don't need to edit)

i need to edit them taking from a list of coordinates in another file txt i can provide; and the script should output N new files (where N is the number of coordinates i have to examine) each one with a tag number (example sample1, sample2..)
So in short it should edit a file N times, saving each time it edits with a different name.
please help me i have to edit 3580 times the file, the coordinates needed are just 70 but i've to consider many different cases..
Thanks!!

bigearsbilly · 10-25-2006, 05:40 AM

so the text file coords is just 70 lines?
and you need 3580 files output.

need more example data and example output

assasukasse · 10-25-2006, 05:54 AM

yes:
the file with the coordinates is a 70 lines simple TXT file made like that:
1 x y z
2 x y z
3 x y z
where the line number should be used in the output file name.
i need only 70 output files because this is the most tedious work:
i am calculating for energy ranges so what i do is basically this, i choose an energy range and write to the file
then i make 70 files for all the coordinates and launch the analysis
once is done i collect the data in a excel sheet and i pass to the next energy by editing the main file
then again i make 70 files out of the template changing only the coordinates.
and so on..
what i need the script for is avoiding the work of editing 70 times the main file, by getting 70 duplicate files where only the coordinate value has been changed..
i need 3580 analisys in total

but this i need to do myself, the important is automating the coordinate change or i will get crazy or very likely make an error..

assasukasse · 10-25-2006, 06:00 AM

i post you an example data:
is part of the file since is pretty long

/cut
13 RCC 0 0 1.634 0 0 9.5 7
14 RCC 6.45 0 10.134 0 0 1 0.55 $cella sorgente
c fine surface cards

c definizione della sorgente
SDEF pos=6.45 0 10.634 axs=0 0 1 par=2 ERG=0.662 CEL=11
RAD=D1 EXT=D2
SI1 0 0.55
SI2 0.5
c fine definizione della sorgente
c mat cards
/cut

what i need to edit is the value pos=x y z

the output file should be a copy of the imput file with the coordinate changes according to the coordinate file that should have this form:
1 x y z
2 x y z
3 x y z
...

for example if i have only 3 coordinate files i should get 4 files:
one is the original, and 3 are the modified coordinate ones

so if original is sample
i will have the following 4 files: sample, sample1 (which has been modified to have pos=x y z as in line 1 of coordinate file), sample2, sample3

Thanks

bigearsbilly · 10-25-2006, 06:27 AM

So this is the ONLY line to change,

Code:

SDEF pos=6.45 0 10.634 axs=0 0 1 par=2 ERG=0.662 CEL=11

and only pos=n n n

and will it always start with SDEF

assasukasse · 10-25-2006, 06:56 AM

yes this is the ONLY line to change
will always start with SDEF 5spaces then pos=x y z
actually what is needed to change is the coordinates after pos=

bigearsbilly · 10-25-2006, 07:55 AM

Try this.

It's probably not totally robust, think of it as a prototype.

It's a quick dirty solution,
but I've added some rudimentary integrity checks
on the data.

will create files named after the template with a suffix

use it like:

do_it.pl template_file coord_file

Code:

#!/usr/bin/perl -w

sub usage {

    die "Usage: $0 template-file coord-file\n";
}
usage() unless (scalar(@ARGV) == 2); # count params



($template, $data_file) = @ARGV; # get filenames


@ARGV = ($data_file);
@coords = <>;  # slurp coordinates
warn "Is '$data_file' a proper coordinate file? It contains non-numerics\n"
     if grep /[a-z]/i, @coords;

@ARGV = ($template);
@data = <>; # slurp template
die "No SDEF found! in $template" unless grep /SDEF/, @data;


foreach (@coords) {

    next unless /./; # jump blank lines

    ($nm, $x, $y, $z) = split; 

    @copy = @data; 

# do the sub for SDEF
    map {s/(SDEF pos=)[\d .]+(.*)/$1$x $y $z $2/} @copy;
    $filename = "$template.$nm";

    open OUT, ">$filename";
    print OUT @copy;
    close OUT;

}

muha · 10-25-2006, 08:55 AM

I'd just like to mention that sed or awk might work for you as well.
Although bigearsbilly's perl solution looks good

assasukasse · 10-25-2006, 08:56 AM

thanks!
i have only one question
the line

map {s/(SDEF pos=)[\d .]+(.*)/$1$x $y $z $2/} @copy;

make me think that the correct sintax for the coordinate file is this:
1x y z

and not this
1 x y z

is that true?

i don't know perl but looks a nice a powerful language!
Thanks

assasukasse · 10-25-2006, 08:59 AM

Quote:

Originally Posted by muha

I'd just like to mention that sed or awk might work for you as well.
Although bigearsbilly's perl solution looks good

thanks muha
do u have a link to a guide how to use sed or awk?
i want to learn how to use them too!

bigearsbilly · 10-25-2006, 09:11 AM

Quote:

make me think that the correct sintax for the coordinate file is this:
1x y z

no, that's incorrect

the $1$x you refer to
is for the SDEF pos=n part grabbed by the parenthesised
expression (SDEF pos=)
hence there is no space for the output.

in this expression $1 represents "SDEF pos=" and $x = n
so $1$x interpolates to SDEF pos=n

the data is a list of lines,
map applies the substitution expression to each line in turn,
(easy but slightly redundant for most lines

)

perl is a very powerful language for text processing.

sed and awk are a subset of perl pretty much
I never bother with sed and awk now.

except for sed one liners.

assasukasse · 10-29-2006, 09:12 AM

i found a small problem with the script:
if i choose a clearly wrong file as a template for the coordinates
it tells me that maybe i am wrong since there are letters too
but proceed anyway making a bunch of wrong files...
is there any way to stop the generation of files in case a letter is encountered?
thanks

assasukasse · 01-04-2007, 11:32 AM

hi bigersbilli i incurred in a small problem
i have the need to process files that have tabs instead of spaces between the coordinates.
or even better, to ignore if there is 1-4-5-6 or whatever spaces in between the numbers, or even tabs..
do i need to change anything?
Thanks

muha · 01-05-2007, 07:06 PM

Quote:

Originally Posted by assasukasse

thanks muha
do u have a link to a guide how to use sed or awk?
i want to learn how to use them too!

I missed that before, here you go:
[sed]
http://sed.sourceforge.net/
http://www.linuxquestions.org/bookmarks/tags/sed
http://wiki.linuxquestions.org/wiki/Sed

[awk]
http://www.vectorsite.net/tsawk_1.html#m1
http://sparky.rice.edu/~hartigan/awk.html
http://www.gnu.org/software/gawk/manual/gawk.html

With sed you could do something like this (if the files are called 1, 2 and 3)

Code:

sed -n '/SDEF/s/\(.*\)axs.*/\1/p' 1 2 3
SDEF pos=6.45 0 10.634
SDEF pos=7.45 1 11.634
SDEF pos=8.45 2 12.634

Or, to get all files in this directory and feed them into sed one at a time:

Code:

for i in $(ls); do echo -n $i; sed -n '/SDEF/s/SDEF\(.*\)axs.*/\1/p' $i; done
1 pos=6.45 0 10.634
2 pos=7.45 1 11.634
3 pos=8.45 2 12.634

Now just grab the standard output and feed that into a textfile, one directory higher

Code:

for i in $(ls); do echo -n $i; sed -n '/SDEF/s/SDEF\(.*\)axs.*/\1/p' $i; done > ../output.txt

If you need more info post back or pm.
/ah, you need to include spaces in the filenames, that can be done also.

Code:

for i in *; do echo -n "$i"; sed -n '/SDEF/s/SDEF\(.*\)axs.*/\1/p' "$i"; done

assasukasse · 01-06-2007, 12:09 PM

Thanks muha
i need more time to study sed, i made some tests but i didn't go much farther..
however, i was thinking i missed something when in the file bigearsbilly gave me:
there is another line that needs to be edited
the line is BEFORE the SDEF line
and starts with 14 RCC then 5 spaces, then the same coordinates of SDEF then other numbers..
now i added the following line to the pl file:

Code:

map {s/(14 RCC     )[\d .]+(.*)/$1$x $y $z $2/} @copy;

however it doesnt work
DOES work if i remove the 5 spaces, but deletes the other numbers that should remain there..
what should i do?