| Re: point_lun question [message #17582] |
Tue, 02 November 1999 00:00 |
Pavel Romashkin
Messages: 166
Registered: April 1999
|
Senior Member |
|
|
Frank,
I have not found POINT_LUN in the code you sent and therefore see no
problem with it. I wrote several file reading procedures for very
convoluted file formats and had to liberally use cursor positioning
information. I never did it the way you are doing it and don't know why is
FSTAT field not working for you. I also missed the point of attempting to
store positions in the code you sent, to start with. I guess you have a
reason for that, which is not clear to me from the code clip. I also
anticipate that the code you sent is undue slow due to re-creation of
arrays and reading variables throug a string.
I know that using POINT_LUN worked perfectly fine for me for both setting
and retrieving cursor position. I checked the code on Macintosh, Windows
and UNIX and never had failures. You may want to try this approach if you
really want to get positions, although I see you make no use of them in the
code.
I tried to see what is your code doing here:
function rd_data_header
current_line=""
count=0
filen=pickfile()
openr,unit,filen,/get_lun
while not (EOF(unit)) do begin
; Read one line from the file
readf,unit,current_line
; If it is a comment line, remember the position at its end.
if (strpos(current_line,'#S ') eq 0) then begin
current_pos=fstat(unit) ; Try replacing with point_lun, -unit, current_pos
; For the very first time, do the following:
if (count=0) then begin
result_header=current_line ; This is a string!
result_pos=current_pos.cur_ptr
count=1
endif
; All other times, do this:
if (count=1) then begin
; Produce array of size +1
tmp=strarr(size(result_header,/n_elements)+1)
; Produce array of size +1
tmp_pos=lon64arr(size(result_pos,/n_elements)+1)
; Place result_header in up to -1 elements of tmp.
tmp[0:size(result_header,/n_elements)-1]=result_header
; Place the position in up to -1 elements of tmp_pos
tmp_pos[0:size(result_pos,/n_elements)-1]=result_pos
; record the same line in [1] position of tmp
tmp[size(result_header,/n_elements)]=current_line
; Do the same with position
tmp_pos[size(result_pos,/n_elements)]=current_pos.cur_ptr
result_header=tmp
result_pos=tmp_pos ; set tmp vars to current result
; Positions are stored and never used.
; Data not placed anywhere for future use. The result contains byte
;locations in file.
endif
endif
endwhile
free_lun,unit
result={header:result_header, position:result_pos}
return,result
end
************
It looks to me that the code needs to be heavily optimized.
Good luck,
Pavel
|
|
|
|
| Re: point_lun question [message #17585 is a reply to message #17582] |
Tue, 02 November 1999 00:00  |
Wayne Landsman
Messages: 117
Registered: January 1997
|
Senior Member |
|
|
Frank wrote:
> Hi,
> I have a data file with several measurements separated by some header
> lines followed by data points. The whole file containes many
> measurements. So I want to write a procedure, that reads the header and
> the position of the file pointer at the end of the header and returns
> this in a structure for further data processing.
> Actually the routine finds the header, however, point_lun returns the
> 'correct' value only for the first header. If I check the second
> position of the pointer using an hexditor the positions are not correct.
> Whats wrong ? Do I use the wrong data type
>
Are you running under Windows? If so, then you may need to add a /BINARY
(or /NOAUTOMODE) switch to your openr statement.-- otherwise your use of
the EOF() function can corrupt the file pointer. It has something to
do with whether the file is being read in text mode or binary mode, as
discussed in the Windows-specific documentation for openr. (However, the
way that the EOF() function can corrupt the file pointer without these
keywords present still feels like a bug to me.)
Peter Mason wrote a note on this problem to comp.lang.idl-pvwave last
March.
--Wayne
Landsman
landsman@mpb.gsfc.nasa.gov
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
