| Re: variable types [message #10993] |
Mon, 23 February 1998 00:00 |
David Foster
Messages: 341
Registered: January 1996
|
Senior Member |
|
|
Martin Schultz wrote:
>
> David R. Klassen wrote:
>>
>> I'm in the process of writing a quick and dirty program to
>> go through a text file and grab the lines that are ordered
>> pairs of numbers and plot them. The problem comes with the
>> fact that some of these data files have a line or two of
>> text at the top.
>>
>> When I read each line of the file I parse it along white-spaces
>> or tabs uisng the PARTS function (no problem so far). Then I
>> assign the first two parts to my x and y arrays. The problem
>> is that if the line read was a line of text, the parts can not
>> be converted from string type to float type (it turns out that
>> a string ' 13.456' *can* be turned into a float=13.456).
>>
>> My question: is there a way to test the variable type before
>> I make the assignments?
>>
>
> yes, you will probably have to read in your line as string (it is really
> a pity that the '$' format code does not work during input !), then you
> can either go David's way to catch the error or test for a number
> yourself:
> readf,s [,format='(A)']
> test = strpos('0123456789.+-',strmid(strtrim(s,1),0,1)) ge 0
>
> Fortran data files sometimes use a line format like
> N x1 x2 x3 ... xN, so you need to extract the first number
> before you can read the others. In these cases, you should also read
> the line into a string, [test the first character for a number] and
> then extract the numbers using READS.
>
Martin -
If you do a lot of text parsing you may want to try out my
GET_TOKEN.PRO routine that reads data of BYTE, INT, LONG, FLOAT or
DOUBLE type from a string, maintaining a pointer into the string
to allow for sequential parsing of data. The code is fairly ugly
but it works really well. For your data it would look like:
for i = 0, n_elements(lines) - 1 do begin
readf, unit, string, format='(a)'
p = 0
val1 = get_token(string, p, /flt, error_value='ERROR')
if (strtrim(val1,2) ne 'ERROR') then begin
; p = p + 1 ; SEE NOTE BELOW (**)
array(i,0) = val1
val2 = get_token(string, p, /flt, error_value='ERROR')
array(i,1) = val2 ; Maybe check this value too
endif
endfor
(**) If the numbers are separated by non-whitespace character(s), you
will need to increment the pointer appropriately.
You can download GET_TOKEN.PRO by anonymous FTP:
bial8.ucsd.edu : pub/software/idl/share/idl_share.tar.gz
This includes many other routines as well, so don't extract into !PATH !
Hope this helps.
Dave
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~
David S. Foster Univ. of California, San Diego
Programmer/Analyst Brain Image Analysis Laboratory
foster@bial1.ucsd.edu Department of Psychiatry
(619) 622-5892 8950 Via La Jolla Drive, Suite 2240
La Jolla, CA 92037
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~
|
|
|
|
| Re: variable types [message #11011 is a reply to message #10993] |
Fri, 20 February 1998 00:00  |
Martin Schultz
Messages: 515
Registered: August 1997
|
Senior Member |
|
|
David R. Klassen wrote:
>
> I'm in the process of writing a quick and dirty program to
> go through a text file and grab the lines that are ordered
> pairs of numbers and plot them. The problem comes with the
> fact that some of these data files have a line or two of
> text at the top.
>
> When I read each line of the file I parse it along white-spaces
> or tabs uisng the PARTS function (no problem so far). Then I
> assign the first two parts to my x and y arrays. The problem
> is that if the line read was a line of text, the parts can not
> be converted from string type to float type (it turns out that
> a string ' 13.456' *can* be turned into a float=13.456).
>
> My question: is there a way to test the variable type before
> I make the assignments?
>
yes, you will probably have to read in your line as string (it is really
a pity that the '$' format code does not work during input !), then you
can either go David's way to catch the error or test for a number
yourself:
readf,s [,format='(A)']
test = strpos('0123456789.+-',strmid(strtrim(s,1),0,1)) ge 0
Fortran data files sometimes use a line format like
N x1 x2 x3 ... xN, so you need to extract the first number
before you can read the others. In these cases, you should also read
the line into a string, [test the first character for a number] and
then extract the numbers using READS.
Regards,
Martin.
------------------------------------------------------------ -------
Dr. Martin Schultz
Department for Earth&Planetary Sciences, Harvard University
186 Pierce Hall, 29 Oxford St., Cambridge, MA-02138, USA
phone: (617)-496-8318
fax : (617)-495-4551
e-mail: mgs@io.harvard.edu
IDL-homepage: http://www-as.harvard.edu/people/staff/mgs/idl/
------------------------------------------------------------ -------
|
|
|
|
| Re: variable types [message #11018 is a reply to message #11011] |
Thu, 19 February 1998 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
David R. Klassen (klassen@marswatch.tn.cornell.edu) writes:
> I'm in the process of writing a quick and dirty program to
> go through a text file and grab the lines that are ordered
> pairs of numbers and plot them. The problem comes with the
> fact that some of these data files have a line or two of
> text at the top.
>
> When I read each line of the file I parse it along white-spaces
> or tabs uisng the PARTS function (no problem so far). Then I
> assign the first two parts to my x and y arrays. The problem
> is that if the line read was a line of text, the parts can not
> be converted from string type to float type (it turns out that
> a string ' 13.456' *can* be turned into a float=13.456).
>
> My question: is there a way to test the variable type before
> I make the assignments?
If I understand the question correctly the variable type
before the assignments is always STRING, whether the
assignment succeeds or not.
What I would try to do is CATCH the assignment error, thinking
that if the assignment to a FLOAT succeeds, the string must
have been a "number". If it doesn't, I'll just read the next
line. My code might look like this:
line = ''
FOR j=0,n-1 DO BEGIN
Catch, error
IF error NE 0 THEN line = ''
ReadF, lun, line
partA = Part(line)
thisNum = Float(partA) ; This is where error occurs.
ENDFOR
The assignment error causes IDL to set the error variable
to the error number and execution jumps to the next line
of code *after* the Catch error handler. In this case, you
just reinitialize the line variable to a string and away you
go.
Later on you might want to cancel the Catch error handler:
Catch, /Cancel
or set another one, etc. Remember that ON_IOERROR will take
precedent over the Catch, so be sure you have it turned off.
Cheers,
David
-----------------------------------------------------------
David Fanning, Ph.D.
Fanning Software Consulting
E-Mail: davidf@dfanning.com
Phone: 970-221-0438
Coyote's Guide to IDL Programming: http://www.dfanning.com/
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
