parse

Parser classes for different file formats. Input- and output files.

We need the following basic Unix tools installed:

grep/egrep

sed

awk (better mawk)

tail

wc

…

The tested egrep versions don’t know the \s character class for whitespace as sed, Perl, Python or any other sane regex implementation does. Use [ ] instead.

Using Parsing classes

All parsing classes:

Pw*OutputFile
Cpmd*OutputFile
Cp2k*OutputFile
Lammps*OutputFile

are derived from FlexibleGetters -> UnitsHandler -> {Structure,Trajectory}FileParser

As a general rule: If a getter (self.get_<attr>() or self._get_<attr>_raw() cannot find anything in the file, it returns None. All getters which depend on it will also return None.

• After initialization

  pp = SomeParsingClass(<filename>)

  all attrs whoose name is in pp.attr_lst will be set to None.

• parse() will invoke self.try_set_attr(<attr>), which does

  self.<attr> = self.get_<attr>()

  for each <attr> in self.attr_lst, thus setting self.<attr> to a defined value: None if nothing was found in the file or not None else

• All getters get_<attr>() will do their parsing, possibly looking for a file self.filename, regardless of the fact that the attribute self.<attr> may already be defined (e.g. if parse() has been called before).

• For interactive use (you need <attr> only once), prefer get_<attr>() over parse().

• Use dump(‘foo.pk’) only for temporary storage and fast re-reading. Use pwtool.io.read_pickle(‘foo.pk’). See also the *FileParser.load() docstring.

• Use relative paths in <filename>.

• If loading a dump()’ed pickle file from disk,

  pp=io.read_pickle(…)

  then use direct attr access

  pp.<attr>

  instead of

  pp.get_<attr>()

  b/c latter would simply parse self.filname again.

For debugging, we still have many getters which produce redundant information, e.g.

cell + cryst_const

_<attr>_raw + <attr> (where <attr> = cell, forces, …)

…

especially in MD parsers, not so much in StructureFileParser drived classes. If parse() is used, all this information retrieved and stored.

• All parsers try to return the default units of the program output, e.g. Ry, Bohr, tryd for PWscf; Ha, Bohr, thart for Abinit and CPMD.

• Use get_struct() / get_traj() to get a Structure / Trajectory object with pwtools standard units (eV, Ang, fs).

Using parse():

Pro:

• Simplicity. All getters are called when parse() is invoked. You get it all.

• In theory, you can delete the file pointed to by self.filename, assuming all getters have extracted all information that you need.

Con:

• The object is full of (potentially big) arrays holding redundant information. Thus, the dump()’ed file may be large. Use the compress() method.

• Parsing may be slow if each getter (of possibly many) is called.

Using get_<attr>():

Pro:

• You only parse what you really need.

Con:

• self.<attr> will NOT be set, since get_<attr>() only returns <attr> but doesn’t set self.<attr> = self.get_<attr>(), so dump() would save an “empty” file.

AWK	str(object='') -> str str(bytes_or_buffer[, encoding[, errors]]) -> str
Ang	Convert a string or number to a floating point number, if possible.
Angstrom	Convert a string or number to a floating point number, if possible.
Bohr	Convert a string or number to a floating point number, if possible.
CifFile([filename, block])	Parse Cif file.
Cp2kDcdMDOutputFile(*args, **kwds)	Same as Cp2kMDOutputFile (all PROJECT* files are text), only that the coordinates file is a dcd format binary file PROJECT-pos-1.dcd.
Cp2kMDOutputFile(*args, **kwds)	CP2K MD output parser.
Cp2kRelaxOutputFile(*args, **kwds)	Parse cp2k global/run_type cell_opt.
Cp2kSCFOutputFile(*args, **kwds)	CP2K SCF output parser ("global/run_type energy_force,print_level low").
CpmdMDOutputFile(*args, **kwds)	Parse CPMD MD output.
CpmdSCFOutputFile(*args, **kwds)	Parse output from a CPMD "single point calculation" (wave function optimization).
DcdOutputFile()	Base class which implements dcd file reading.
Ha	Convert a string or number to a floating point number, if possible.
LammpsDcdMDOutputFile(*args, **kwds)	Parse Lammps DCD binary output + log.lammps text output.
LammpsTextMDOutputFile([filename, order])	Parse LAMMPS text output.
PDBFile([filename])	Very very simple pdb file parser.
PwMDOutputFile([filename, use_alat])	Parse pw.x MD-like output.
PwSCFOutputFile([filename, use_alat])	Parse a pw.x SCF output file (calculation='scf').
PwVCMDOutputFile(*args, **kwds)	Parse only calculation='vc-md'.
Ry	Convert a string or number to a floating point number, if possible.
StructureFileParser([filename, units])	Base class for single-structure parsers.
TrajectoryFileParser([filename, units])	Base class for MD-like parsers.
arr1d_from_txt(txt[, dtype])
arr2d_from_txt(txt[, dtype])
axis_lens(seq[, axis])	Return length of axis of all arrays in seq.
eV	Convert a string or number to a floating point number, if possible.
float_from_txt(txt)
fs	Convert a string or number to a floating point number, if possible.
int_from_txt(txt)
nstep_from_txt(txt)
pi	Convert a string or number to a floating point number, if possible.
ps	Convert a string or number to a floating point number, if possible.
thart	Convert a string or number to a floating point number, if possible.
traj_from_txt(txt, shape[, axis, dtype, sep])	Used for 3d trajectories where the exact shape of the array as written by the MD code must be known, e.g. (nstep,N,3) where N=3 (cell, stress) or N=natoms (coords, forces, ...).