updated parser to handle different basis sets

parser/parser-gpaw/parser.py

 import os
 from contextlib import contextmanager
 import numpy as np
-#from nomadcore.unit_conversion.unit_conversion import convert_unit as cu
+from ase.data import chemical_symbols
+from nomadcore.unit_conversion.unit_conversion import convert_unit as cu
 from nomadcore.local_meta_info import loadJsonFile, InfoKindEl
 from nomadcore.parser_backend import JsonParseEventsWriterBackend
 from tar import Reader
@@ -14,13 +15,16 @@ def open_section(p, name):
     p.closeSection(name, gid)
 
 
-def cu(value, unit=None):
+def c(value, unit=None):
     """ Dummy function for unit conversion"""
-    return value
+    return cu(value, unit)
 
 
 parser_info = {"name": "parser_gpaw", "version": "1.0"}
-metaInfoPath = os.path.normpath(os.path.join(os.path.dirname(os.path.abspath(__file__)), "../../../../nomad-meta-info/meta_info/nomad_meta_info/gpaw.nomadmetainfo.json"))
+path = '../../../../nomad-meta-info/meta_info/nomad_meta_info/' +\
+        'gpaw.nomadmetainfo.json'
+metaInfoPath = os.path.normpath(
+    os.path.join(os.path.dirname(os.path.abspath(__file__)), path))
 
 metaInfoEnv, warns = loadJsonFile(filePath=metaInfoPath,
                                   dependencyLoader=None,
@@ -28,33 +32,53 @@ metaInfoEnv, warns = loadJsonFile(filePath=metaInfoPath,
                                   uri=None)
 
 
-# TODO: convert to SI units
 def parse(filename):
     p = JsonParseEventsWriterBackend(metaInfoEnv)
     o = open_section
     r = Reader(filename)
     p.startedParsingSession(filename, parser_info)
+
     with o(p, 'section_run'):
         p.addValue('program_name', 'GPAW')
+        if r.Mode == 'pw':
+            with o(p, 'section_basis_set_cell_associated'):
+                p.addValue('basis_set_cell_associated_name',
+                           'PW_%.1f_Ry' % (r.PlaneWaveCutoff * 2.0))  # in Ry
+                p.addRealValue('basis_set_plane_wave_cutoff',
+                               c(r.PlaneWaveCutoff, 'hartree'))
+        elif r.Mode == 'fd':
+            with o(p, 'section_basis_set_cell_associated'):
+                h1 = np.linalg.norm(r.UnitCell[0]) / r.dims['ngptsx']
+                h2 = np.linalg.norm(r.UnitCell[1]) / r.dims['ngptsy']
+                h3 = np.linalg.norm(r.UnitCell[2]) / r.dims['ngptsz']
+                h = (h1 + h2 + h3) / 3.0
+                p.addValue('basis_set_cell_associated_name',
+                           'GR_%.1f' % (c(h, 'bohr') * 1.0E15))  # in fm
+        elif r.Mode == 'lcao':
+            pass
         with o(p, 'section_system_describtion'):
-            p.addArrayValues('simulation_cell', r.UnitCell)
-            p.addArrayValues('atom_label', r.CartesianPositions)
+            p.addArrayValues('simulation_cell', c(r.UnitCell, 'bohr'))
+            symbols = np.array([chemical_symbols[z] for z in r.AtomicNumbers])
+            p.addArrayValues('atom_label', symbols)
+            p.addArrayValues('atom_position', c(r.CartesianPositions, 'bohr'))
             p.addArrayValues('configuration_periodic_dimensions',
                              np.array(r.BoundaryConditions, bool))
         with o(p, 'section_single_configuration_calculation'):
-            p.addRealValue('energy_total', r.Epot)
-            p.addRealValue('energy_XC', r.Exc)
-            p.addRealValue('electronic_kinetic_energy', r.Ekin)
+            p.addRealValue('energy_total', c(r.Epot, 'hartree'))
+            p.addRealValue('energy_XC', c(r.Exc, 'hartree'))
+            p.addRealValue('electronic_kinetic_energy', c(r.Ekin, 'hartree'))
             if 'CartesianForces' in r:
-                p.addArrayValues('atom_forces_free', r.CartesianForces)
+                p.addArrayValues('atom_forces_free',
+                                 c(r.CartesianForces, 'bohr/hartree'))
             with o(p, 'section_method'):
                 p.addValue('XC_functional', r.XCFunctional)
             with o(p, 'section_eigenvalues_group'):
                 for eps_kn, occ_kn in zip(r.Eigenvalues, r.OccupationNumbers):
                     with o(p, 'section_eigenvalues'):
-                        p.addArrayValues('eigenvalues_eigenvalues', eps_kn)
+                        p.addArrayValues('eigenvalues_eigenvalues',
+                                         c(eps_kn, 'hartree'))
                         p.addArrayValues('eigenvalues_occupation', occ_kn)
-
+                        p.addArrayValues('eigenvalues_kpoints', r.IBZKPoints)
     p.finishedParsingSession("ParseSuccess", None)
 
 if __name__ == '__main__':

parser/parser-gpaw/tar.py

@@ -34,7 +34,7 @@ class Reader:
 
     def __contains__(self, name):
         return name in self.parameters or name in self.shapes
-        
+
     def __getattr__(self, name):
         if name in self.parameters:
             return self.parameters[name]
@@ -54,11 +54,11 @@ class Reader:
             array = np.asarray(array, int)
         array.shape = shape
         return array
-    
+
     def close(self):
         self.tar.close()
 
-        
+
 if __name__ == '__main__':
     r = Reader('H2.gpw')
     print(r.Eigenvalues, r.FermiLevel, r.UnitCell)