Fixes export failures on nested parses

[imaduro]

While using Python2, if a grammar can have multiple nested expressions, the exporter will fail to export these if Strings are present within them.

Example.tx

Expression:
    ArithmeticExpression | SubExpression
;

ArithmeticExpression:
    operand=Operand (operator=Operator operand=Operand)+
;

SubExpression:
    leftParen="(" sub=Expression rightParen=")"
;

Operand:
    ID | INT | SubExpression
;

Operator:
    "+" | "-"
;

then using the string (value + (exp + bla)) - blabla

Python 2.7.15 (default, May  1 2018, 16:44:08)
[GCC 4.2.1 Compatible Apple LLVM 9.1.0 (clang-902.0.39.1)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from textx import metamodel_from_file
>>> example_meta = metamodel_from_file('example.tx')
>>> my_string = "(value + (exp + bla)) - blabla"
>>> parsed = example_meta.model_from_str(my_string)
>>> from textx.export import model_export
>>> model_export("./", "my_file.dot")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 163, in model_export
    model_export_to_file(f, model, repo)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 266, in model_export_to_file
    _export(model)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 194, in _export
    for attr_name, attr in obj_cls._tx_attrs.items():
AttributeError: type object 'str' has no attribute '_tx_attrs'

This bug only affects Python 2, Python 3 does not run into this error.

[igordejanovic]

Your example works fine with both Python 2.7 and 3.x.
The problem is incorrect use of model_export.
Instead of model_export("./", "my_file.dot") you should write model_export(parsed, "my_file.dot") where parsed is your model. With the example above you've tried to export string "./".

[imaduro]

My bad, though I tried it again with the same model and code:

Python 2.7.15 (default, May  1 2018, 16:44:08)
[GCC 4.2.1 Compatible Apple LLVM 9.1.0 (clang-902.0.39.1)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from textx import metamodel_from_file
>>> example_meta = metamodel_from_file('example.tx')
>>> my_string = "(value + (exp + bla)) - blabla"
>>> parsed = example_meta.model_from_str(my_string)
>>> from textx.export import model_export

>>> model_export(parsed, "my_file.dot")

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 163, in model_export
    model_export_to_file(f, model, repo)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 267, in model_export_to_file
    _export(model)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 222, in _export
    _export(list_obj)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 242, in _export
    _export(attr_value)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 222, in _export
    _export(list_obj)
  File "/usr/local/lib/python2.7/site-packages/textx/export.py", line 195, in _export
    for attr_name, attr in obj_cls._tx_attrs.items():
AttributeError: type object 'str' has no attribute '_tx_attrs'

I wonder if it's specific to my machine's python2 then.

[igordejanovic]

@imaduro I follow your exact steps and managed to reproduce. I'll investigate. Thanks for reporting.

[igordejanovic]

Ah, I got it. textX is using unicode strings. Your input string must be a unicode. I guess we should make some checks in *_from_str functions/methods to give better error responses.
Please leave this issue open until we do. In the mean time just use u in front of your input strings or from __future__ import unicode_literals at the top of your files.

[igordejanovic]

Please leave this issue open until we do.

I meant please open an issue for this since this is a PR :)

[imaduro]

Will do! Thanks!