Skip to main content
  • EN
Solved

Python & global Variable

  • 6 March 2013
  • 4 replies
  • 117 views
Hi all

 

 

I use a PythonCaller to compare the output from some transformers with a textfile. Depending on the results I would like to set a global variable or modify a published parameter.

 

In the further workflow I would like to use the parameter to determine if the features should be written out or not.

 

As a workaround I used a VaraibleSetter and VariableRetriever to set a global variable which I can use in a Tester. It yould be more elegant to set and expose the global variable or modify a published parameter within the python Caller.

 

  1. Is it possible to change the value of a published parameter within a PythonCaller? If I try to fo modify the published parameter with FME_MacroValuesa'NameOfPublishedParameter'] = "Yes", it doesn't affect the value. The default vlaue will be used.
  2. Is it possible to set and expose a glaobal variable (function or factory: @GlobalVariable) analog to the VariableSetter/VariableRetriever combo?

Thanks for any advice

 

 

felix

4 replies

david_r
Rank
Userlevel 5
Hi Felix,

 

 

unfortunately, FME_MacroValues is considered read-only by FME, so any changes you make to the dictionary values will not be propagated back to the running session. In fact, I believe that it isn't possible to change the value of the user parameters during run-time at all, regardless (I'd be glad to be proved wrong on this one!)...

 

 

Two alternative strategies:
  1. Initialize the parameter value using a Python scripted parameter, or
  2. Have the PythonCaller create a new attribute, which you then reference in your workspace rather than the user parameter.
Hope this helps.

 

 

David
F
Hi David,

 

 

Thanks for your answer. That's too bad.

 

The whole workbench is for data processing and quality assurance. After processing some data I do quality checks. The output is then compared with a text file, where exceptions can be defined, in the PyhtonCaller.

 

So, I can't use a scripted Parameter. I think I have to stick to the second strategie and use the VariableSetter & VariableRetriever combo.

 

 

Is it possible to use a transformer within the PyhtonCaller (VariableSetter) or set a GlobalVariable?

 

 

thanks

 

felix
david_r
Rank
Userlevel 5
Hi Felix,

 

 

unfortunately, I do not know of any way to set/get variables like the VariableSetter does.

 

 

You can, however, set a session variable in one PythonCaller that you can retrieve in another PythonCaller later on, if that is of any help. The easiest is probably to define a variable (or dictionary, etc) in the startup scrip, such as:

 

 

my_global_value = 0

 

global my_global_value

 

You can then reference and modify my_global_value from the various PythonCallers in your workspace.

 

 

If this does not solve your problem, I would consider taking a step back (figuratively :-) and see if it is possible to separate the loading and the validation of the data. This is a common way to do it when importin into a database schema with strict constraints:
  • load input data (e.g. csv, shape, etc) into a validation schema (copy of production schema, but without any hard constraints, relationships, etc)
  • try to apply constraints one by one on validation schema
  • if everything ok, transfer data to production schema, if not, report error
This is of course very simplified, but you get the idea.

 

 

Hope this helps.

 

 

David
F
Hi David

 

 

Thanks, that solved most of my problems. First I tried to access the global variable within a scripted parameter. As you mentioned earlier, that doesn't work.

 

Now I set the global variable in the PythonCaller (btw. you flipped the order, first global parameter, then parameter ="text"). Instead of assigning the variable to each feature and unsing a Tester I use a second PythonCaller and test programatically for the global variable.

 

 

Thanks for your help

 

felix

 

 

 

P.S.

 

With the VariableSetter you can set a variable. Later you can asign this variable to an attribute (VariableRetriever).

 

In my case I createed one feature with an attribute in the PythonCaller and used the VariableSetter. All processed features went into an FeatureHolder. I asigned to each feature the attribute from the PythonCaller with the VariableRetriever. In the end I was able to test for his attribute.

 

Reply


Community Stats

29,635
Posts
112,628
Replies
37,285
Members
Terms & ConditionsCookie settings