Support files formatted in general non-PDBx CIF format

[Alex6357]

The CIFBlock.deserialize() method fails to parse CIF data correctly, causing inability to access CIFCategory objects.

Reproduction example:

import biotite.structure.io.pdbx as pdbx

cifblock = pdbx.CIFBlock.deserialize(
    """
_symmetry_space_group_name_H-M   'P 1'
_cell_length_a   3.94513000
_cell_length_b   3.94513000
_cell_length_c   3.94513000
_cell_angle_alpha   90.00000000
_cell_angle_beta   90.00000000
_cell_angle_gamma   90.00000000
_symmetry_Int_Tables_number   1
_chemical_formula_structural   SrTiO3
_chemical_formula_sum   'Sr1 Ti1 O3'
_cell_volume   61.40220340
_cell_formula_units_Z   1
loop_
 _symmetry_equiv_pos_site_id
 _symmetry_equiv_pos_as_xyz
  1  'x, y, z'
loop_
 _atom_site_type_symbol
 _atom_site_label
 _atom_site_symmetry_multiplicity
 _atom_site_fract_x
 _atom_site_fract_y
 _atom_site_fract_z
 _atom_site_occupancy
  Sr  Sr0  1  0.00000000  0.00000000  0.00000000  1.0
  Ti  Ti1  1  0.50000000  0.50000000  0.50000000  1.0
  O  O2  1  0.50000000  0.00000000  0.50000000  1.0
  O  O3  1  0.50000000  0.50000000  0.00000000  1.0
  O  O4  1  0.00000000  0.50000000  0.50000000  1.0
"""
)

print(cifblock._categories)
print(cifblock.get("atom_site"))

Output:

{"symmetry_space_group_name_H-M   'P 1": "_symmetry_space_group_name_H-M   'P 1'\n", 'cell_length_a   3': '_cell_length_a   3.94513000\n', 'cell_length_b   3': '_cell_length_b   3.94513000\n', 'cell_length_c   3': '_cell_length_c   3.94513000\n', 'cell_angle_alpha   90': '_cell_angle_alpha   90.00000000\n', 'cell_angle_beta   90': '_cell_angle_beta   90.00000000\n', 'cell_angle_gamma   90': '_cell_angle_gamma   90.00000000\n', 'symmetry_Int_Tables_number   ': '_symmetry_Int_Tables_number   1\n', 'chemical_formula_structural   SrTiO': '_chemical_formula_structural   SrTiO3\n', "chemical_formula_sum   'Sr1 Ti1 O3": "_chemical_formula_sum   'Sr1 Ti1 O3'\n", 'cell_volume   61': '_cell_volume   61.40220340\n', 'cell_formula_units_Z   ': '_cell_formula_units_Z   1\n', None: 'loop_\n _atom_site_type_symbol\n _atom_site_label\n _atom_site_symmetry_multiplicity\n _atom_site_fract_x\n _atom_site_fract_y\n _atom_site_fract_z\n _atom_site_occupancy\n  Sr  Sr0  1  0.00000000  0.00000000  0.00000000  1.0\n  Ti  Ti1  1  0.50000000  0.50000000  0.50000000  1.0\n  O  O2  1  0.50000000  0.00000000  0.50000000  1.0\n  O  O3  1  0.50000000  0.50000000  0.00000000  1.0\n  O  O4  1  0.00000000  0.50000000  0.50000000  1.0\n'}
None

From my code analysis (based on personal interpretation of the source code), _categories should be a key-value mapping of CIFCategory names to the raw string lines where these names are declared. Currently, the category name parsing is incorrect.

[padix-key]

The problem is that this is not a PDBx-formatted CIF file with keys in the form <category>.<column>, but one formatted according to https://www.iucr.org/resources/cif/dictionaries. pdbx.CIFFile makes the following assumptions which are fulfilled in PDBx files, but not in those CIF files:

• Separation of category and column by a .. This could be circumvented by reading a CIF dictionary (which is also not implemented yet.)
• The categories are sequentially listed (e.g. there is no column from category B within a range of values in category A). This is a central assumption of pdbx.CIFCategory.

So in summary we unfortunately would need a new parser for this format.

[Alex6357]

Thanks for the explanation. I now see the difference.

I believe it would be valuable to add an error message before fully implementing support for other CIF formats, if we can easily distinguish between the formats.