use of outdated CIF keys, breaking the parser of ASE 3.20

[ltalirz]

• [x] _symmetry_space_group_name_Hall should be replaced by _space_group_name_Hall
• [x] _symmetry_space_group_name_H-M should be replaced by _space_group_name_H-M_alt

P.S. For some reason, from version 3.20, ASE starts complaining about the symmetry specification in current CIF files (now limiting ase version to <3.20) We should check whether updating the cif keys fixes this.

[ltalirz]

@danieleongari This is a very simple change that is necessary to comply with the latest CIF standard. Can we make this change or is there a parser "upstream" that does not understand these keys?

[danieleongari]

I'll change it in https://github.com/danieleongari/CURATED-COFs/pull/22, and we'll see if problems emerge

[danieleongari]

It loads fine with ASE-gui and Zeopp

[ltalirz]

sorry, which version of ase did you test?

I'm testing with ase 3.21.1 and on your updated branch I'm getting

In [5]: atoms = read('19456N2.cif')
----------------------------------------------------
AssertionError     Traceback (most recent call last)
<ipython-input-5-9db16add7238> in <module>
----> 1 atoms = read('19456N2.cif')

~/Applications/miniconda3/lib/python3.6/site-packages/ase/io/formats.py in read(filename, index, format, parallel, do_not_split_by_at_sign, **kwargs)
    681     else:
    682         return next(_iread(filename, slice(index, None), format, io,
--> 683                            parallel=parallel, **kwargs))
    684
    685

~/Applications/miniconda3/lib/python3.6/site-packages/ase/parallel.py in new_generator(*args, **kwargs)
    271             not kwargs.pop('parallel', True)):
    272             # Disable:
--> 273             for result in generator(*args, **kwargs):
    274                 yield result
    275             return

~/Applications/miniconda3/lib/python3.6/site-packages/ase/io/formats.py in _iread(filename, index, format, io, parallel, full_output, **kwargs)
    747     # Make sure fd is closed in case loop doesn't finish:
    748     try:
--> 749         for dct in io.read(fd, *args, **kwargs):
    750             if not isinstance(dct, dict):
    751                 dct = {'atoms': dct}

~/Applications/miniconda3/lib/python3.6/site-packages/ase/io/cif.py in read_cif(fileobj, index, store_tags, primitive_cell, subtrans_included, fractional_occupancies, reader)
    600     # Find all CIF blocks with valid crystal data
    601     images = []
--> 602     for block in parse_cif(fileobj, reader):
    603         if not block.has_structure():
    604             continue

~/Applications/miniconda3/lib/python3.6/site-packages/ase/io/cif.py in parse_cif_ase(fileobj)
    536             continue
    537
--> 538         yield parse_block(lines, line)
    539
    540

~/Applications/miniconda3/lib/python3.6/site-packages/ase/io/cif.py in parse_block(lines, line)
    497
    498 def parse_block(lines: List[str], line: str) -> CIFBlock:
--> 499     assert line.lower().startswith('data_')
    500     blockname = line.split('_', 1)[1].rstrip()
    501     tags = parse_items(lines, line)

AssertionError:

I've looked into it and it turns out the issue is a line with a space after the symmetry specification

E.g. In the case of FAPTOH.cif it fails with this

loop_
_symmetry_equiv_pos_as_xyz
 'x,y,z'

(last line has one space)

and it works with this

loop_
_symmetry_equiv_pos_as_xyz
 'x,y,z'

(last line is empty)

[danieleongari]

Right, I was testing most recent CIFs. Fixed (on Mac) with:

for f in *; do sed -i '' "s/^ *//" $f; done

See commit: