Extend parsers (#19)

* Add system quantities to qe

* Add output quantities to qe

* Extend octopus parser

* Extend ams parser

Author: ladinesa

src/nomad_simulation_parsers/parsers/ams/file_parser.py

@@ -24,7 +24,10 @@ def str_to_labels_positions(val_in):
             unit = val.pop(0)[0].lower()
             unit = ureg.bohr if unit.startswith('bohr') else ureg.angstrom
             val = np.transpose(val)
-            return val[1], np.array(val[2:5].T, dtype=np.dtype(np.float64)) * unit
+            return [
+                list(val[1]),
+                np.array(val[2:5].T, dtype=np.dtype(np.float64)) * unit,
+            ]
 
         def str_to_lattice_vectors(val_in):
             val = [v.split()[-3:] for v in val_in.strip().splitlines()]
@@ -697,8 +700,8 @@ def to_band_energy_ranges(val_in):
                 Quantity(
                     'forces_total',
                     rf'Gradients \(hartree/bohr\)\s+'
-                    rf'Index.+\s+((?:\d+ +[A-Z][a-z]*'
-                    rf' +{RE_FLOAT} +{RE_FLOAT} +{RE_FLOAT}\s+)+)',
+                    rf'Index.+\s+'
+                    rf'((?:\d+ +[A-Z][a-z]* +{RE_FLOAT} +{RE_FLOAT} +{RE_FLOAT}\s+)+)',
                     str_operation=str_to_forces,
                 ),
                 Quantity(
@@ -1253,8 +1256,7 @@ def parse_results(self) -> None:
                 )
 
     def parse_properties(self) -> None:
-        if properties := self.data.get('Properties'):
-            return
+        properties = self.data.get('Properties', {})
 
         nspin = self.data.get('General', {}).get('nspin', 1)
 
@@ -1342,7 +1344,9 @@ def parse_properties(self) -> None:
         )
         bottom_conduction_band = band_structure.get('BottomConductionBand')
         band_gap['energy_lowest_unoccupied'] = (
-            (bottom_conduction_band * ureg.hartree) if not None else None
+            (bottom_conduction_band * ureg.hartree)
+            if bottom_conduction_band is not None
+            else None
         )
         self._results['band_gap'] = band_gap
 

src/nomad_simulation_parsers/parsers/ams/parser.py

@@ -1,5 +1,8 @@
+import os
 from importlib import reload
+from typing import Any
 
+import numpy as np
 from nomad.datamodel import EntryArchive
 from nomad.parsing.file_parser import ArchiveWriter
 from nomad.parsing.file_parser.mapping_parser import MetainfoParser, TextParser
@@ -8,9 +11,11 @@
 from nomad_simulations.schema_packages.general import Program, Simulation
 from structlog.stdlib import BoundLogger
 
+from nomad_simulation_parsers.parsers.utils.general import search_files
 from nomad_simulation_parsers.schema_packages import ams
 
 from .file_parser import OutParser
+from .file_parser import RKFParser as RKFTextParser
 
 LOGGER = get_logger(__name__)
 
@@ -21,6 +26,44 @@ class MainfileParser(TextParser):
     def logger(self):
         return LOGGER
 
+    def get_xc_functionals(self, source: dict[str, Any]) -> list[str]:
+        xc_functionals = []
+        for xc_type in ['LDA', 'GGA', 'MGGA']:
+            functionals = source.get(xc_type, '').split()
+            kind = ['XC'] if len(functionals) == 1 else ['X', 'C']
+            for n, functional in enumerate(functionals):
+                xc_functionals.append(
+                    f'{xc_type}_{kind[n]}_{functional.rstrip("x").rstrip("c").upper()}'
+                )
+        return xc_functionals
+
+    def get_contributions(self, source: dict[str, Any]) -> list[dict[str, Any]]:
+        return [
+            dict(name=key, value=val) for key, val in source.items() if val is not None
+        ]
+
+    def get_eigenvalues(
+        self, source: dict[str, Any] | list[np.ndarray]
+    ) -> list[dict[str, np.ndarray]]:
+        energies = source.get('energies', []) if isinstance(source, dict) else []
+        occupations = (
+            source.get('occupations', []) if isinstance(source, dict) else source[2]
+        )
+        nspin = max(len(energies), len(occupations))
+        eigenvalues = [dict() for _ in range(nspin)]
+        for n, energy in enumerate(energies):
+            eigenvalues[n]['eigenvalues'] = energy
+        for n, occupations in enumerate(occupations):
+            eigenvalues[n]['occupations'] = occupations
+        return [eig for eig in eigenvalues if eig]
+
+
+class RKFParser(MainfileParser):
+    # TODO temporary fix for structlog unable to propagate logger
+    @property
+    def logger(self):
+        return LOGGER
+
 
 # TODO temporary fix for structlog unable to propagate logger
 class AMSMetainfoParser(MetainfoParser):
@@ -32,6 +75,7 @@ def logger(self):
 class AMSArchiveWriter(ArchiveWriter):
     mainfile_parser = MainfileParser(text_parser=OutParser())
     metainfo_parser = AMSMetainfoParser()
+    rkf_parser = RKFParser(text_parser=RKFTextParser())
 
     def write_to_archive(self):
         # reload schema package to use correct annotations
@@ -41,8 +85,17 @@ def write_to_archive(self):
         self.archive.data = Simulation(program=Program(name='AMS'))
         self.metainfo_parser.data_object = self.archive.data
 
-        self.mainfile_parser.filepath = self.mainfile
-        self.mainfile_parser.convert(self.metainfo_parser)
+        rkf_files = search_files('ams.rkf', os.path.dirname(self.mainfile))
+        self.parser = self.mainfile_parser
+        self.parser.filepath = self.mainfile
+        if rkf_files:
+            if len(rkf_files) > 1:
+                self.logger.warning('Multiple ams.rkf files found.')
+            self.parser = self.rkf_parser
+            self.parser.filepath = rkf_files[0]
+            self.parser.data_object.parse()
+
+        self.parser.convert(self.metainfo_parser)
 
 
 class AMSParser(MatchingParser):

…_parsers/parsers/octopus/file_parsers.py …n_parsers/parsers/octopus/file_parser.pysrc/nomad_simulation_parsers/parsers/octopus/file_parsers.py renamed to src/nomad_simulation_parsers/parsers/octopus/file_parser.py src/nomad_simulation_parsers/parsers/octopus/file_parsers.py renamed to src/nomad_simulation_parsers/parsers/octopus/file_parser.py

@@ -184,18 +184,10 @@ def str_to_line(val_in: str) -> list[str]:
             )
         ]
 
-    @property
-    def mainfile(self):
-        return super().mainfile
-
-    @mainfile.setter
-    def mainfile(self, val):
-        # TODO pylint complains when inheriting property
-        # self._results = None
-        # self._file_handler = None
-        # self._mainfile = os.path.abspath(val) if val is not None else val
-        super().mainfile = val
-        super().__init__()
+    def reset(self):
+        super().reset()
+        self._info = None
+        self._keys_mapping = dict()
 
     def evaluate_value(self, value: Any) -> Any:
         """
@@ -357,10 +349,10 @@ def get_coordinates(self) -> tuple[str, np.ndarray]:
         val = self.info.get('Coordinates', self.info.get('ReducedCoordinates', [[]]))[0]
 
         for v in val:
-            if isinstance(v, str):
-                symbols.append(v.strip())
-            else:
+            if v[0].isdecimal() or not isinstance(v, str):
                 coordinates.append(v)
+            else:
+                symbols.append(v.strip())
 
         coordinates = np.array(coordinates, dtype=float)
         coordinates.shape = (len(symbols), 3)