Gracefully handle empty CAS xmi files

Describe the bug We encountered the scenario that our pipeline has created empty .xmi files (because the import data was not clean and contained empty documents). This is a cleaned example of the .xmi file:

<?xml version="1.0" encoding="UTF-8"?>
<xmi:XMI xmlns:xmi="http://www.omg.org/XMI" xmlns:tcas="http:///uima/tcas.ecore" xmlns:cas="http:///uima/cas.ecore" xmi:version="2.0">
    <cas:NULL xmi:id="0" />
    <tcas:DocumentAnnotation xmi:id="2" sofa="1" begin="0" end="0" language="en" />
    <cas:Sofa xmi:id="1" sofaNum="1" sofaID="_InitialView" mimeType="text" sofaString="" />
    <cas:View sofa="1" members="2" />
</xmi:XMI>

Trying to open this file causes an error.

To Reproduce Steps to reproduce the behavior:

from cassis import load_cas_from_xmi
xmi="""<?xml version="1.0" encoding="UTF-8"?>
<xmi:XMI xmlns:xmi="http://www.omg.org/XMI" xmlns:tcas="http:///uima/tcas.ecore" xmlns:cas="http:///uima/cas.ecore" xmi:version="2.0">
    <cas:NULL xmi:id="0" />
    <tcas:DocumentAnnotation xmi:id="2" sofa="1" begin="0" end="0" language="en" />
    <cas:Sofa xmi:id="1" sofaNum="1" sofaID="_InitialView" mimeType="text" sofaString="" />
    <cas:View sofa="1" members="2" />
</xmi:XMI>"""
load_cas_from_xmi(xmi)

Expected behavior I would like to see that cassis succesfully creates a CAS (that is empty). This is important to us because it avoids downstream problems.

Error message

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
/tmp/ipykernel_32727/2857839666.py in <module>
----> 1 cas = load_cas_from_xmi(xmi)

~/.local/lib/python3.8/site-packages/cassis/xmi.py in load_cas_from_xmi(source, typesystem, lenient, trusted)
     41     deserializer = CasXmiDeserializer()
     42     if isinstance(source, str):
---> 43         return deserializer.deserialize(
     44             BytesIO(source.encode("utf-8")), typesystem=typesystem, lenient=lenient, trusted=trusted
     45         )

~/.local/lib/python3.8/site-packages/cassis/xmi.py in deserialize(self, source, typesystem, lenient, trusted)
    225                 # Map from offsets in UIMA UTF-16 based offsets to Unicode codepoints
    226                 if typesystem.is_instance_of(fs.type, "uima.tcas.Annotation"):
--> 227                     fs.begin = sofa._offset_converter.uima_to_cassis(fs.begin)
    228                     fs.end = sofa._offset_converter.uima_to_cassis(fs.end)
    229 

~/.local/lib/python3.8/site-packages/cassis/cas.py in uima_to_cassis(self, idx)
     66         if idx is None:
     67             return None
---> 68         return self._uima_to_cassis[idx]
     69 
     70     def cassis_to_uima(self, idx: Optional[int]) -> Optional[int]:

KeyError: 0

Please complete the following information:

Version: 0.5.3
OS: Ubuntu 20.04

dkpro / dkpro-cassis

Gracefully handle empty CAS xmi files #193