| Re: IDLVM and execute [message #79020] |
Tue, 24 January 2012 16:25 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
This problem has been a problem for me before so I wrote a simple
arithmetic parser:
docs: http://docs.idldev.com/idllib/analysis/mg_evalexpr.html
source: http://docs.idldev.com/idllib/analysis/mg_evalexpr.pro
Evaluates a mathematical expression using the basic arithmetic operators
+, -, *, /, and ^ along with parentheses for grouping and simple
function calls of a single variable.
This routine does not use `EXECUTE`, so it is safe to use in the Virtual
Machine.
For example, simple arithmetic expressions can be evaluated::
IDL> print, mg_evalexpr('1 + 2 + 3', error=error), error
6 0
Note that the `ERROR` keyword returns whether there was an error in
evaluating the expression. Expressions can also take variables, if their
values are provided via a structure or hash-like object::
IDL> print, mg_evalexpr('exp(i * pi)', { pi: !dpi, i: complex(0, 1) })
( -1.0000000, 1.2246468e-16)
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL, A Guide to Learning IDL: http://modernidl.idldev.com
Research Mathematician
Tech-X Corporation
|
|
|
|
| Re: IDLVM and execute [message #79035 is a reply to message #79020] |
Mon, 23 January 2012 08:17  |
Reimar Bauer
Messages: 14
Registered: August 2002
|
Junior Member |
|
|
On Jan 23, 1:57 pm, greg.a...@googlemail.com wrote:
> Yes, I've wanted to do this before. I suppose the reason execute() is not allowed is to prevent you from writing a wrapper of IDL interpreter for the VM. If this is so, perhaps it could be possible for Exelis to provide a reduced version of execute which would simply evaluate an arithmetic expression. For my purposes - and yours, too, it appears - it would be enough if only +-*/^ were allowed (although probably sqrt, alog10, exp, sin... and a bunch of other built-in fns could sensibly be included, too): I don't think you could build anything interpreter-like from those. A user-defined function is a useful thing to have in a scientific data analysis language...
>
> Greg
If you read this article you find what you need
http://cow.physics.wisc.edu/~craigm/idl/down/routine_names.t xt
Reimar
|
|
|
|
| Re: IDLVM and execute [message #79036 is a reply to message #79035] |
Mon, 23 January 2012 04:57 |
greg.addr
Messages: 160
Registered: May 2007
|
Senior Member |
|
|
Yes, I've wanted to do this before. I suppose the reason execute() is not allowed is to prevent you from writing a wrapper of IDL interpreter for the VM. If this is so, perhaps it could be possible for Exelis to provide a reduced version of execute which would simply evaluate an arithmetic expression. For my purposes - and yours, too, it appears - it would be enough if only +-*/^ were allowed (although probably sqrt, alog10, exp, sin... and a bunch of other built-in fns could sensibly be included, too): I don't think you could build anything interpreter-like from those. A user-defined function is a useful thing to have in a scientific data analysis language...
Greg
|
|
|
|
| Re: IDLVM and execute [message #79040 is a reply to message #79036] |
Sun, 22 January 2012 11:59 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On 1/22/12 8:30 am, Russell wrote:
> On Jan 20, 3:36 pm, TonyLanz<tolan...@gmail.com> wrote:
>> Hi all,
>> I realize this has been asked on the group before. I'm hoping someone
>> might have a "simple" workaround in IDLVM to what would be easily
>> accomplished using the execute function in IDL proper. I have a widget
>> program that contains a compound widget field that allows the user to
>> define an equation for manipulating a set of four variables. For the
>> sake of argument let's say var1, var2, var3, and var4. In this
>> compound widget field they can define what math they want to do,
>> they're restricted to addition, subtraction, multiplication and
>> division.
>>
>> so they could for example enter
>>
>> var1/var2
>> or
>> var1-var2*0.2
>> or
>> var1+var2+var3
>>
>> etc., you get the idea. Now in IDL I can just get the value from the
>> widget and simply do something like
>>
>> s='result='+cwidget_value(0)
>> status=execute(s)
>> print,result
>>
>> of course in IDLVM I'm not allowed to use EXECUTE()
>>
>> Anyone have any suggestions for implementing this simply? I know in
>> the past there was discussion of writing code to break down the string
>> and pass the pieces to the right operators or functions (using
>> call_function() ) to assemble the result.
>>
>> Tony
>
> Yes, like all things, there are several ways to skin this cat. The
> simplest way is to write the string to a file, compile that file, then
> call it as a function (and presumably delete it after you're
> finished). So if you have the command as a string, such as "result =
> var1 + var2*0.2" then just write it to a file with the appropriate
> declarations and closings. So, you'd execute:
>
> ;save the data:
> vars=[var1,var2]
>
> ;write the file:
> openw,lun,'file.pro',/get_lun
> printf,lun,'function file,vars'
> printf,lun,'return,vars(0)+vars(1)*0.2'
> printf,lun,'end'
> close,lun& free_lun,lun
>
> ;compile the file
> resolve_routine,'file.pro',/is_function
>
> ;call the function:
> result=call_function('file',var1,var2)
>
> ;clean up
> file_delete,'file.pro',/allow_non
>
>
> now the rub here is that you need to know how many variables there are
> to store in that temporary variable. Hopefully, there's some way for
> you to loop over that.
You can't compile a file from the VM, so I think you are stuck actually
parsing the expression.
> The alternative way, would be to parse the command using the usual
> order of operations rules. Then process that parsed things using pre-
> defined functions for the four arithmetic operations. Such as:
>
> t=multiply(v1,v2)
> t=divide(v1,v2)
>
> etc. WHere they have the obvious "innards". Of course, this can get
> very tedious, but is very useful. If you go down this path, I
> recommend using the byte typecasting to find the major operators:
> t=byte(cmd).
>
>
> Good Luck
> Russell
>
> PS, the reason this is so complicated, is that they don't want you
> doing this.
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL, A Guide to Learning IDL: http://modernidl.idldev.com
Research Mathematician
Tech-X Corporation
|
|
|
|
| Re: IDLVM and execute [message #79041 is a reply to message #79040] |
Sun, 22 January 2012 07:30 |
Russell[1]
Messages: 101
Registered: August 2011
|
Senior Member |
|
|
On Jan 20, 3:36 pm, TonyLanz <tolan...@gmail.com> wrote:
> Hi all,
> I realize this has been asked on the group before. I'm hoping someone
> might have a "simple" workaround in IDLVM to what would be easily
> accomplished using the execute function in IDL proper. I have a widget
> program that contains a compound widget field that allows the user to
> define an equation for manipulating a set of four variables. For the
> sake of argument let's say var1, var2, var3, and var4. In this
> compound widget field they can define what math they want to do,
> they're restricted to addition, subtraction, multiplication and
> division.
>
> so they could for example enter
>
> var1/var2
> or
> var1-var2*0.2
> or
> var1+var2+var3
>
> etc., you get the idea. Now in IDL I can just get the value from the
> widget and simply do something like
>
> s='result='+cwidget_value(0)
> status=execute(s)
> print,result
>
> of course in IDLVM I'm not allowed to use EXECUTE()
>
> Anyone have any suggestions for implementing this simply? I know in
> the past there was discussion of writing code to break down the string
> and pass the pieces to the right operators or functions (using
> call_function() ) to assemble the result.
>
> Tony
Yes, like all things, there are several ways to skin this cat. The
simplest way is to write the string to a file, compile that file, then
call it as a function (and presumably delete it after you're
finished). So if you have the command as a string, such as "result =
var1 + var2*0.2" then just write it to a file with the appropriate
declarations and closings. So, you'd execute:
;save the data:
vars=[var1,var2]
;write the file:
openw,lun,'file.pro',/get_lun
printf,lun,'function file,vars'
printf,lun,'return,vars(0)+vars(1)*0.2'
printf,lun,'end'
close,lun & free_lun,lun
;compile the file
resolve_routine,'file.pro',/is_function
;call the function:
result=call_function('file',var1,var2)
;clean up
file_delete,'file.pro',/allow_non
now the rub here is that you need to know how many variables there are
to store in that temporary variable. Hopefully, there's some way for
you to loop over that.
The alternative way, would be to parse the command using the usual
order of operations rules. Then process that parsed things using pre-
defined functions for the four arithmetic operations. Such as:
t=multiply(v1,v2)
t=divide(v1,v2)
etc. WHere they have the obvious "innards". Of course, this can get
very tedious, but is very useful. If you go down this path, I
recommend using the byte typecasting to find the major operators:
t=byte(cmd).
Good Luck
Russell
PS, the reason this is so complicated, is that they don't want you
doing this.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
