Can you switch to Python 3.x in FME? That would probably save you a lot of headaches.

Although I'm curious to hear if it helps for this particular problem. Could well be a bug or some use-case that hasn't been anticipated by the mostly anglophile developers at Safe.

---

Can you switch to Python 3.x in FME? That would probably save you a lot of headaches.

Although I'm curious to hear if it helps for this particular problem. Could well be a bug or some use-case that hasn't been anticipated by the mostly anglophile developers at Safe.

Ah, I wish ^^ unfortunately Python 3.x can't work with FME 2016

 

c.f. This question

---

Hi @vchalmel

Thanks for reporting your encoding issue with Python.  The first thing I wanted to do is reproduce your problem.

So, with the following:

• FME 2016.1 build 16717 64-bit on Windows 10
• Python 2.7

I tried the following script in the PythonCaller:

import fme
import fmeobjects

class FeatureProcessor(object):
    def __init__(self):
        self._log = fmeobjects.FMELogFile()
        pass
    def input(self,feature):
        e = u'é'
        self._log.logMessageString('e type: ' + str(type(e)))
        self._log.logMessageString('e: ' + e);
        print(e.encode('utf-8'))
        
        feature.setAttribute('e', e)
        
        feat_e = feature.getAttribute('e')
        self._log.logMessageString('feat_e type: ' + str(type(feat_e)))
        self._log.logMessageString('feat_e: ' + feat_e)
        
        self.pyoutput(feature)
    def close(self):
        pass

Here is what I get as a result in the Translation Log:

Notice the following:
1) I did not add a `# -*- coding : windows-1252 -*-` or similar line to the top of the script.  Through some experimentation, it appears those lines have no affect on Windows when we pass the script to the Python interpreter for running.  On Mac, it appears the encoding line does do something and in those cases we always recommend passing `utf-8` as the encoding because that is the 'system' encoding for Mac.
In general, I don't believe you need to worry about the encoding of the script you are writing.  FME will store it as a unicode type and will pass it into the interpreter with the appropriate encoding ('system' encoding for Python 2.7 and utf-8 encoding for Python 3.x).
2) The result from `feat_e = feature.getAttribute('e')` is a `unicode` type object that has a value of é.  Since it's a unicode object, there isn't an encoding associated with it.  A byte-string of any appropriately desired encoding can be retrieved through the unicode.encode(<encoding name>) method.
3) True, when `print()` is called for output to the Translation Log, unicode strings need a bit of help and need to be encoded to a utf-8 byte-string.
There is still some confusion on what problems you are experiencing.  Can you provide a sample PythonCaller script showing the issue, so we can try it on our end?  Thanks.

---