From f00597306afcaae3138ecb4a1ebddfe5067046b4 Mon Sep 17 00:00:00 2001
From: "nathan.daelman@physik.hu-berlin.de" <ndaelman@physik.hu-berlin.de>
Date: Tue, 27 Jun 2023 16:39:30 +0200
Subject: [PATCH 1/8] Minor touch ups

---
 electronicparsers/fhiaims/parser.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index 55f3adf3..8fd69715 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -1816,9 +1816,7 @@ def parse_atom_type(species):
                             sec_basis_func.x_fhi_aims_controlInOut_basis_func_primitive_gauss_alpha = alpha
 
         # add inout parameters read from main output
-        # species
-        species = self.out_parser.get('control_inout', {}).get('species')
-        if species is not None:
+        if (species := self.out_parser.get('control_inout', {}).get('species')) is not None:
             for specie in species:
                 parse_atom_type(specie)
 

From 60ce7b5d25bce66e702e7912e3ad508edb370030 Mon Sep 17 00:00:00 2001
From: Nathan <ndaelman.physik.hu-berlin.de>
Date: Fri, 14 Jul 2023 16:03:47 +0200
Subject: [PATCH 2/8] Make control species repeatable

---
 electronicparsers/fhiaims/parser.py | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)

diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index 8fd69715..5f0791e8 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -78,15 +78,12 @@ def str_to_unit(val_in):
     def init_quantities(self):
         def str_to_species(val_in):
             val = val_in.strip().splitlines()
-            data = []
             species = dict()
             for v in val:
                 v = v.strip().split('#')[0]
                 if not v or not v[0].isalpha():
                     continue
                 if v.startswith('species'):
-                    if species:
-                        data.append(species)
                     species = dict(species=v.split()[1:])
                 else:
                     v = v.replace('.d', '.e').split()
@@ -95,8 +92,7 @@ def str_to_species(val_in):
                         species[v[0]].extend([vi])
                     else:
                         species[v[0]] = [vi]
-            data.append(species)
-            return data
+            return species
 
         self._quantities = [
             Quantity(
@@ -154,9 +150,9 @@ def str_to_species(val_in):
                 'xc',
                 rf'{re_n} *xc\s*([\w\. \-\+]+)', repeats=False),
             Quantity(
-                'species', rf'{re_n} *(species\s*[A-Z][a-z]?[\s\S]+?)'
-                r'(?:species\s*[A-Z][a-z]?|Completed|\-{10})',
-                str_operation=str_to_species, repeats=False)]
+                'species', rf'{re_n} *(species\s+[A-Z][a-z]?[\s\S]+?)'
+                r'(FHI-aims code project|\-{10})',
+                str_operation=str_to_species, repeats=True)]
 
 
 class FHIAimsOutParser(TextParser):

From 60dabe48f632639043acb2bf95312e986b2b6160 Mon Sep 17 00:00:00 2001
From: Nathan <ndaelman.physik.hu-berlin.de>
Date: Fri, 14 Jul 2023 17:06:08 +0200
Subject: [PATCH 3/8] Clean `atom_parameters` section and `ControlInOut`
 parsing

---
 .../fhiaims/metainfo/fhi_aims.py              | 103 ++++++------------
 electronicparsers/fhiaims/parser.py           |  69 ++----------
 2 files changed, 42 insertions(+), 130 deletions(-)

diff --git a/electronicparsers/fhiaims/metainfo/fhi_aims.py b/electronicparsers/fhiaims/metainfo/fhi_aims.py
index 72034998..4ec7f6dd 100644
--- a/electronicparsers/fhiaims/metainfo/fhi_aims.py
+++ b/electronicparsers/fhiaims/metainfo/fhi_aims.py
@@ -192,75 +192,6 @@ class x_fhi_aims_section_controlIn_basis_set(MSection):
         repeats=True)
 
 
-class x_fhi_aims_section_controlInOut_atom_species(MSection):
-    '''
-    -
-    '''
-
-    m_def = Section(validate=False)
-
-    x_fhi_aims_controlInOut_pure_gaussian = Quantity(
-        type=str,
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_charge = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='coulomb',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_cut_pot_scale = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_cut_pot_width = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='meter',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_cut_pot = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='meter',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_mass = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='kilogram',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_species_name = Quantity(
-        type=str,
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_section_controlInOut_basis_func = SubSection(
-        sub_section=SectionProxy('x_fhi_aims_section_controlInOut_basis_func'),
-        repeats=True)
-
-    x_fhi_aims_section_vdW_TS = SubSection(
-        sub_section=SectionProxy('x_fhi_aims_section_vdW_TS'),
-        repeats=True)
-
-
 class x_fhi_aims_section_controlInOut_basis_func(MSection):
     '''
     -
@@ -1386,8 +1317,38 @@ class AtomParameters(simulation.method.AtomParameters):
 
     m_def = Section(validate=False, extends_base_section=True)
 
-    x_fhi_aims_section_controlInOut_atom_species = SubSection(
-        sub_section=SectionProxy('x_fhi_aims_section_controlInOut_atom_species'),
+    x_fhi_aims_controlInOut_pure_gaussian = Quantity(
+        type=str,
+        shape=[],
+        description='''
+        -
+        ''')
+
+    x_fhi_aims_controlInOut_species_cut_pot_scale = Quantity(
+        type=np.dtype(np.float64),
+        shape=[],
+        description='''
+        -
+        ''')
+
+    x_fhi_aims_controlInOut_species_cut_pot_width = Quantity(
+        type=np.dtype(np.float64),
+        shape=[],
+        unit='meter',
+        description='''
+        -
+        ''')
+
+    x_fhi_aims_controlInOut_species_cut_pot = Quantity(
+        type=np.dtype(np.float64),
+        shape=[],
+        unit='meter',
+        description='''
+        -
+        ''')
+
+    x_fhi_aims_section_vdW_TS = SubSection(
+        sub_section=SectionProxy('x_fhi_aims_section_vdW_TS'),
         repeats=True)
 
 
diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index 5f0791e8..dbc72160 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -51,7 +51,6 @@
 from .metainfo.fhi_aims import Run as xsection_run, Method as xsection_method,\
     x_fhi_aims_section_parallel_task_assignement, x_fhi_aims_section_parallel_tasks,\
     x_fhi_aims_section_controlIn_basis_set, x_fhi_aims_section_controlIn_basis_func,\
-    x_fhi_aims_section_controlInOut_atom_species, x_fhi_aims_section_controlInOut_basis_func,\
     x_fhi_aims_section_vdW_TS
 
 from ..utils import BeyondDFTWorkflowsParser
@@ -790,7 +789,7 @@ def __init__(self):
             'Hybrid M11 gradient-corrected functionals': [{'name': 'MGGA_C_M11'}, {'name': 'HYB_MGGA_X_M11'}]}
 
         # TODO update metainfo to reflect all energy corrections
-        # why section_vdW_TS under x_fhi_aims_section_controlInOut_atom_species?
+        # why section_vdW_TS under atom_parameter?
         self._energy_map = {
             'Total energy uncorrected': 'energy_total',
             'Total energy corrected': 'energy_total_t0',
@@ -1290,7 +1289,6 @@ def parse_scf(iteration):
         def parse_vdW(section):
             # these are not actually vdW outputs but vdW control parameters but are
             # printed within the calculation section.
-            # TODO why is x_fhi_aims_section_vdW_TS under x_fhi_aims_section_controlInOut_atom_species
             # we would then have to split the vdW parameters by species
             atoms = section.get('vdW_TS', {}).get('atom_hirshfeld', [])
             if not atoms:
@@ -1303,8 +1301,7 @@ def parse_vdW(section):
             for sec in sec_atom_type:
                 for atom in atoms:
                     if sec.label == atom['atom']:
-                        sec_vdW_ts = sec.x_fhi_aims_section_controlInOut_atom_species[-1].m_create(
-                            x_fhi_aims_section_vdW_TS)
+                        sec_vdW_ts = sec.m_create(x_fhi_aims_section_vdW_TS)
                         for key, val in atom.items():
                             metainfo_name = self._property_map.get(key, None)
                             if metainfo_name is None:
@@ -1742,26 +1739,19 @@ def parse_topology(self):
 
         def parse_atom_type(species):
             sec_atom_type = sec_method.m_create(AtomParameters)
-            sec_atom_species = sec_atom_type.m_create(
-                x_fhi_aims_section_controlInOut_atom_species)
             for key, val in species.items():
                 if key == 'nuclear charge':
-                    charge = val[0] * ureg.elementary_charge
-                    sec_atom_type.charge = charge
-                    sec_atom_species.x_fhi_aims_controlInOut_species_charge = charge
+                    sec_atom_type.charge = val[0] * ureg.elementary_charge
                 elif key == 'atomic mass':
-                    mass = val[0][0] * ureg.amu
-                    sec_atom_type.mass = mass
-                    sec_atom_species.x_fhi_aims_controlInOut_species_mass = mass
+                    sec_atom_type.mass = val[0][0] * ureg.amu
                 elif key == 'species':
                     sec_atom_type.label = val
-                    sec_atom_species.x_fhi_aims_controlInOut_species_name = val
                 elif 'request to include pure gaussian fns' in key:
-                    sec_atom_species.x_fhi_aims_controlInOut_pure_gaussian = val[0]
+                    sec_atom_type.x_fhi_aims_controlInOut_pure_gaussian = val[0]
                 elif 'cutoff potl' in key:
-                    sec_atom_species.x_fhi_aims_controlInOut_species_cut_pot = val[0][0] * ureg.angstrom
-                    sec_atom_species.x_fhi_aims_controlInOut_species_cut_pot_width = val[0][1] * ureg.angstrom
-                    sec_atom_species.x_fhi_aims_controlInOut_species_cut_pot_scale = val[0][2]
+                    sec_atom_type.x_fhi_aims_controlInOut_species_cut_pot = val[0][0] * ureg.angstrom
+                    sec_atom_type.x_fhi_aims_controlInOut_species_cut_pot_width = val[0][1] * ureg.angstrom
+                    sec_atom_type.x_fhi_aims_controlInOut_species_cut_pot_scale = val[0][2]
                 elif "request for '+U'" in key:
                     sec_hubbard = sec_atom_type.m_create(HubbardKanamoriModel)
                     sec_hubbard.orbital = f'{val[0][0]}{val[0][1]}'
@@ -1769,47 +1759,8 @@ def parse_atom_type(species):
                     sec_hubbard.double_counting_correction = 'Dudarev'
                     sec_hubbard.x_fhi_aims_projection_type = 'Mulliken (dual)'
                     sec_hubbard.x_fhi_aims_petukhov_mixing_factor = self.out_parser.get('petukhov')
-                elif 'free-atom' in key or 'free-ion' in key:
-                    for i in range(len(val)):
-                        sec_basis_func = sec_atom_species.m_create(
-                            x_fhi_aims_section_controlInOut_basis_func)
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_type = ' '.join(key.split()[:-1])
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_n = val[i][0]
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_l = val[i][1]
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_occ = val[i][2]
-                elif 'hydrogenic' in key:
-                    for i in range(len(val)):
-                        sec_basis_func = sec_atom_species.m_create(
-                            x_fhi_aims_section_controlInOut_basis_func)
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_type = ' '.join(key.split()[:-1])
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_n = val[i][0]
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_l = val[i][1]
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_eff_charge = val[i][2]
-                elif 'ionic' in key:
-                    for i in range(len(val)):
-                        sec_basis_func = sec_atom_species.m_create(
-                            x_fhi_aims_section_controlInOut_basis_func)
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_type = 'ionic basis'
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_n = val[i][0]
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_l = val[i][1]
-                elif 'basis function' in key:
-                    for i in range(len(val)):
-                        sec_basis_func = sec_atom_species.m_create(
-                            x_fhi_aims_section_controlInOut_basis_func)
-                        sec_basis_func.x_fhi_aims_controlInOut_basis_func_type = key.split(
-                            'basis')[0].strip()
-                        if val[i][0] == 'L':
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_gauss_l = val[i][2]
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_gauss_N = val[i][3]
-                            alpha = [val[i][j + 2] for j in range(len(val[i])) if val[i][j] == 'alpha']
-                            weight = [val[i][j + 2] for j in range(len(val[i])) if val[i][j] == 'weight']
-                            alpha = np.array(alpha) * (1 / ureg.angstrom ** 2)
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_gauss_alpha = alpha
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_gauss_weight = weight
-                        elif len(val[i]) == 2:
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_gauss_l = val[i][0]
-                            alpha = np.array(val[i][1]) / ureg.angstrom ** 2
-                            sec_basis_func.x_fhi_aims_controlInOut_basis_func_primitive_gauss_alpha = alpha
+                # From legacy versions we know that 'free-atom' or 'free-ion' are connected to 'occ'
+                # and 'hydrogenic' to 'eff_charge'. Nothing for 'ionic'
 
         # add inout parameters read from main output
         if (species := self.out_parser.get('control_inout', {}).get('species')) is not None:

From f5dace35e7959628f67b420cdde8e07b8170bf73 Mon Sep 17 00:00:00 2001
From: Nathan <ndaelman.physik.hu-berlin.de>
Date: Mon, 17 Jul 2023 18:21:45 +0200
Subject: [PATCH 4/8] Add reference to basis function from atom_parameters +
 TODO: add the inverse

---
 electronicparsers/fhiaims/metainfo/fhi_aims.py | 6 ++++++
 electronicparsers/fhiaims/parser.py            | 2 ++
 2 files changed, 8 insertions(+)

diff --git a/electronicparsers/fhiaims/metainfo/fhi_aims.py b/electronicparsers/fhiaims/metainfo/fhi_aims.py
index 4ec7f6dd..41052856 100644
--- a/electronicparsers/fhiaims/metainfo/fhi_aims.py
+++ b/electronicparsers/fhiaims/metainfo/fhi_aims.py
@@ -1351,6 +1351,12 @@ class AtomParameters(simulation.method.AtomParameters):
         sub_section=SectionProxy('x_fhi_aims_section_vdW_TS'),
         repeats=True)
 
+    x_fhi_aims_section_controlIn_basis_set = Quantity(
+        type=Reference(SectionProxy('x_fhi_aims_section_controlIn_basis_set')),
+        shape=[],
+        description='''-''',
+    )
+
 
 class HubbardKanamoriModel(simulation.method.HubbardKanamoriModel):
 
diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index dbc72160..0b4cbe8a 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -1739,6 +1739,8 @@ def parse_topology(self):
 
         def parse_atom_type(species):
             sec_atom_type = sec_method.m_create(AtomParameters)
+            param_index = len(sec_method.atom_parameters) - 1
+            sec_atom_type.x_fhi_aims_section_controlIn_basis_set = sec_method.x_fhi_aims_section_controlIn_basis_set[param_index]
             for key, val in species.items():
                 if key == 'nuclear charge':
                     sec_atom_type.charge = val[0] * ureg.elementary_charge

From 45b42315b777908ca6ea8cbb666edf86675963e7 Mon Sep 17 00:00:00 2001
From: "nathan.daelman@physik.hu-berlin.de" <ndaelman@physik.hu-berlin.de>
Date: Tue, 18 Jul 2023 14:02:08 +0200
Subject: [PATCH 5/8] Add general hash mapper for `MSections` + Store hash by
 element

---
 .../fhiaims/metainfo/fhi_aims.py              |  7 +++
 electronicparsers/fhiaims/parser.py           |  5 +-
 electronicparsers/utils/__init__.py           |  2 +-
 electronicparsers/utils/utils.py              | 48 +++++++++++++++++++
 4 files changed, 60 insertions(+), 2 deletions(-)

diff --git a/electronicparsers/fhiaims/metainfo/fhi_aims.py b/electronicparsers/fhiaims/metainfo/fhi_aims.py
index 41052856..f2f6e0e5 100644
--- a/electronicparsers/fhiaims/metainfo/fhi_aims.py
+++ b/electronicparsers/fhiaims/metainfo/fhi_aims.py
@@ -187,6 +187,13 @@ class x_fhi_aims_section_controlIn_basis_set(MSection):
         -
         ''')
 
+    x_fhi_aims_controlIn_hash = Quantity(
+        type=str,
+        shape=[],
+        description='''
+        -
+        ''')
+
     x_fhi_aims_section_controlIn_basis_func = SubSection(
         sub_section=SectionProxy('x_fhi_aims_section_controlIn_basis_func'),
         repeats=True)
diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index 0b4cbe8a..1258330e 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -53,7 +53,7 @@
     x_fhi_aims_section_controlIn_basis_set, x_fhi_aims_section_controlIn_basis_func,\
     x_fhi_aims_section_vdW_TS
 
-from ..utils import BeyondDFTWorkflowsParser
+from ..utils import BeyondDFTWorkflowsParser, get_basis_hash
 
 
 re_float = r'[-+]?\d+\.\d*(?:[Ee][-+]\d+)?'
@@ -1565,6 +1565,9 @@ def parse_basis_set(species):
                 sec_basis_set.x_fhi_aims_controlIn_number_of_basis_func = len(division)
                 sec_basis_set.x_fhi_aims_controlIn_division = division
 
+            # store hash
+            sec_basis_set.x_fhi_aims_controlIn_hash = get_basis_hash([sec_basis_set], [True])
+
         def _get_elemental_tier(
                 basis_settings: x_fhi_aims_section_controlIn_basis_set,
                 reference: dict = self._native_tier_references) -> tuple[Any, Any]:
diff --git a/electronicparsers/utils/__init__.py b/electronicparsers/utils/__init__.py
index ebeeb1e3..e61060f6 100644
--- a/electronicparsers/utils/__init__.py
+++ b/electronicparsers/utils/__init__.py
@@ -17,5 +17,5 @@
 # limitations under the License.
 
 from .utils import (
-    extract_section, get_files, BeyondDFTWorkflowsParser
+    extract_section, get_files, BeyondDFTWorkflowsParser, get_basis_hash
 )
diff --git a/electronicparsers/utils/utils.py b/electronicparsers/utils/utils.py
index 066816e4..64201d7e 100644
--- a/electronicparsers/utils/utils.py
+++ b/electronicparsers/utils/utils.py
@@ -17,9 +17,13 @@
 # limitations under the License.
 #
 
+import json
+import hashlib
 import os
 from glob import glob
 
+from nomad.metainfo import MSection
+from nomad.metainfo.util import MSubSectionList
 from nomad.datamodel import EntryArchive
 from nomad.datamodel.metainfo.simulation.run import Run
 from nomad.datamodel.metainfo.workflow import Link, TaskReference
@@ -84,6 +88,50 @@ def get_files(pattern: str, filepath: str, stripname: str = '', deep: bool = Tru
     return filenames
 
 
+def get_basis_hash(basis_settings: list[MSection], subsections: list[bool], **kwargs):
+    '''
+    General function for converting basis set sections to a hash for comparison.
+    Basis sets may contain element-specific settings, which typically are tackled separately.
+    The option consists of adding also general settings to the hash.
+
+    There are two modes determining whether sections are defined by the `quantities` provided (`inclusion`)
+    or rather `quantities` are explicitely removed (`exclusion`).
+
+    basis_settings: sections to be hashed together
+    subsections: list of bools, indicating whether to include susbections. Must be of same length as basis_settings.
+    mode: str, either `inclusion` or `exclusion`
+    quantities: list of str, quantities to be included or excluded
+    '''
+    mode: str = kwargs.get('mode', 'exclusion')
+    quantities: list[str] = kwargs.get('quantities', [])
+    # sanity checks
+    try:
+        evaluation_settings = zip(basis_settings, subsections)
+    except Exception:  # TODO: specify exception
+        raise ValueError(
+            f'''basis_settings:{basis_settings} and subsections:{subsections}
+            must be of same length.'''
+        )
+    # filter out subsections
+    to_compare: list[dict[str, any]] = []
+    for section, subsection_bool in evaluation_settings:
+        section_dict = section.m_to_dict()
+        to_write = {}
+        for key, val in section_dict.items():
+            if not subsection_bool and\
+                isinstance(getattr(section, key), (MSection, MSubSectionList)):
+                    continue
+            if key == 'm_def' or\
+                (mode == 'exclusion' and key not in quantities) or\
+                (mode == 'inclusion' and key in quantities):
+                    to_write[key] = val
+        to_compare.append(to_write)
+    # hash the filtered sections
+    hash = hashlib.sha1()
+    hash.update(json.dumps(to_compare, sort_keys=True).encode('utf-8'))
+    return hash.hexdigest()
+
+
 class BeyondDFTWorkflowsParser:
     '''
     Generates automatic beyondDFT (GW, BSE, DMFT) workflows. Main classes for parsers will

From 8e7931e5b7c55ebfc5bb4c109b9c6303c3c082e0 Mon Sep 17 00:00:00 2001
From: "nathan.daelman@physik.hu-berlin.de" <ndaelman@physik.hu-berlin.de>
Date: Tue, 18 Jul 2023 14:21:36 +0200
Subject: [PATCH 6/8] Change dependency from `hashlib` to `nomad.utils`

---
 electronicparsers/utils/utils.py | 27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

diff --git a/electronicparsers/utils/utils.py b/electronicparsers/utils/utils.py
index 64201d7e..b50d438f 100644
--- a/electronicparsers/utils/utils.py
+++ b/electronicparsers/utils/utils.py
@@ -18,7 +18,6 @@
 #
 
 import json
-import hashlib
 import os
 from glob import glob
 
@@ -32,6 +31,8 @@
     ParticleHoleExcitationsMethod, ParticleHoleExcitationsResults,
     PhotonPolarization, PhotonPolarizationMethod, PhotonPolarizationResults
 )
+from nomad.utils import hash
+from typing import Union
 
 
 def extract_section(source: EntryArchive, path: str):
@@ -88,7 +89,11 @@ def get_files(pattern: str, filepath: str, stripname: str = '', deep: bool = Tru
     return filenames
 
 
-def get_basis_hash(basis_settings: list[MSection], subsections: list[bool], **kwargs):
+def hash_section(
+        sections: Union(MSection, list[MSection]),
+        subsections: Union(bool, list[bool]),
+        **kwargs,
+    ) -> str:
     '''
     General function for converting basis set sections to a hash for comparison.
     Basis sets may contain element-specific settings, which typically are tackled separately.
@@ -97,19 +102,21 @@ def get_basis_hash(basis_settings: list[MSection], subsections: list[bool], **kw
     There are two modes determining whether sections are defined by the `quantities` provided (`inclusion`)
     or rather `quantities` are explicitely removed (`exclusion`).
 
-    basis_settings: sections to be hashed together
-    subsections: list of bools, indicating whether to include susbections. Must be of same length as basis_settings.
-    mode: str, either `inclusion` or `exclusion`
-    quantities: list of str, quantities to be included or excluded
+    `sections`: sections to be hashed together
+    `subsections`: list of bools, indicating whether to include susbections. Must be of same length as basis_settings.
+    `mode`: str, either `inclusion` or `exclusion`
+    `quantities`: list of str, quantities to be included or excluded
     '''
+    sections = [sections] if isinstance(sections, MSection) else sections
+    subsections = [subsections] if isinstance(subsections, bool) else subsections
     mode: str = kwargs.get('mode', 'exclusion')
     quantities: list[str] = kwargs.get('quantities', [])
     # sanity checks
     try:
-        evaluation_settings = zip(basis_settings, subsections)
+        evaluation_settings = zip(sections, subsections)
     except Exception:  # TODO: specify exception
         raise ValueError(
-            f'''basis_settings:{basis_settings} and subsections:{subsections}
+            f'''basis_settings:{sections} and subsections:{subsections}
             must be of same length.'''
         )
     # filter out subsections
@@ -127,9 +134,7 @@ def get_basis_hash(basis_settings: list[MSection], subsections: list[bool], **kw
                     to_write[key] = val
         to_compare.append(to_write)
     # hash the filtered sections
-    hash = hashlib.sha1()
-    hash.update(json.dumps(to_compare, sort_keys=True).encode('utf-8'))
-    return hash.hexdigest()
+    return hash(*to_compare)
 
 
 class BeyondDFTWorkflowsParser:

From 4531b477054efdb3713b63ae032a01b84ec5faf2 Mon Sep 17 00:00:00 2001
From: "nathan.daelman@physik.hu-berlin.de" <ndaelman@physik.hu-berlin.de>
Date: Tue, 18 Jul 2023 19:38:02 +0200
Subject: [PATCH 7/8] Clean Gaussian basis sets

---
 .../fhiaims/metainfo/fhi_aims.py              | 107 +++++-------------
 electronicparsers/fhiaims/parser.py           |  66 +++++++----
 electronicparsers/utils/__init__.py           |   2 +-
 electronicparsers/utils/utils.py              |  16 +--
 4 files changed, 76 insertions(+), 115 deletions(-)

diff --git a/electronicparsers/fhiaims/metainfo/fhi_aims.py b/electronicparsers/fhiaims/metainfo/fhi_aims.py
index f2f6e0e5..3749136a 100644
--- a/electronicparsers/fhiaims/metainfo/fhi_aims.py
+++ b/electronicparsers/fhiaims/metainfo/fhi_aims.py
@@ -88,6 +88,30 @@ class x_fhi_aims_section_controlIn_basis_func(MSection):
         -
         ''')
 
+    x_fhi_aims_controlIn_basis_func_gauss_l = Quantity(
+        type=np.dtype(np.int32),
+        shape=[],
+        description='''
+        "L is an integer number, specifying the angular momentum"
+        - Manual FHI-aims v201716_2
+        ''')
+
+    x_fhi_aims_controlIn_basis_func_gauss_alphas = Quantity(
+        type=np.dtype(np.float64),
+        shape=['*'],
+        unit='1 / meter ** 2',
+        description='''
+        "The exponent defining a (primitive) Gaussian function"
+        - Manual FHI-aims v201716_2
+        ''')
+
+    x_fhi_aims_controlIn_basis_func_gauss_coeffs = Quantity(
+        type=np.dtype(np.float64),
+        shape=['*'],
+        description='''
+        Weights in linearly composed Gaussian functions.
+        ''')
+
     x_fhi_aims_controlIn_basis_func_type = Quantity(
         type=str,
         shape=[],
@@ -145,20 +169,6 @@ class x_fhi_aims_section_controlIn_basis_set(MSection):
         angular leven for the hartreee part
         ''')
 
-    x_fhi_aims_controlIn_mass = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        description='''
-        mass of the nucleus in atomic mass units
-        ''')
-
-    x_fhi_aims_controlIn_nucleus = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        description='''
-        charge of the nucleus
-        ''')
-
     x_fhi_aims_controlIn_outer_grid = Quantity(
         type=np.dtype(np.float64),
         shape=[],
@@ -180,14 +190,14 @@ class x_fhi_aims_section_controlIn_basis_set(MSection):
         radial multiplier
         ''')
 
-    x_fhi_aims_controlIn_species_name = Quantity(
+    x_fhi_aims_controlIn_hash = Quantity(
         type=str,
         shape=[],
         description='''
         -
         ''')
 
-    x_fhi_aims_controlIn_hash = Quantity(
+    x_fhi_aims_controlIn_species_name = Quantity(
         type=str,
         shape=[],
         description='''
@@ -213,49 +223,6 @@ class x_fhi_aims_section_controlInOut_basis_func(MSection):
         -
         ''')
 
-    x_fhi_aims_controlInOut_basis_func_gauss_alpha = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='1 / meter ** 2',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_gauss_l = Quantity(
-        type=np.dtype(np.int32),
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_gauss_N = Quantity(
-        type=np.dtype(np.int32),
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_gauss_weight = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_l = Quantity(
-        type=str,
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_n = Quantity(
-        type=np.dtype(np.int32),
-        shape=[],
-        description='''
-        -
-        ''')
-
     x_fhi_aims_controlInOut_basis_func_occ = Quantity(
         type=np.dtype(np.float64),
         shape=[],
@@ -263,28 +230,6 @@ class x_fhi_aims_section_controlInOut_basis_func(MSection):
         -
         ''')
 
-    x_fhi_aims_controlInOut_basis_func_primitive_gauss_alpha = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        unit='1 / meter ** 2',
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_radius = Quantity(
-        type=np.dtype(np.float64),
-        shape=[],
-        description='''
-        -
-        ''')
-
-    x_fhi_aims_controlInOut_basis_func_type = Quantity(
-        type=str,
-        shape=[],
-        description='''
-        -
-        ''')
-
 
 class x_fhi_aims_section_eigenvalues_group_ZORA(MSection):
     '''
diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index 1258330e..ba4947d2 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -53,7 +53,7 @@
     x_fhi_aims_section_controlIn_basis_set, x_fhi_aims_section_controlIn_basis_func,\
     x_fhi_aims_section_vdW_TS
 
-from ..utils import BeyondDFTWorkflowsParser, get_basis_hash
+from ..utils import BeyondDFTWorkflowsParser, hash_section
 
 
 re_float = r'[-+]?\d+\.\d*(?:[Ee][-+]\d+)?'
@@ -76,21 +76,26 @@ def str_to_unit(val_in):
 
     def init_quantities(self):
         def str_to_species(val_in):
-            val = val_in.strip().splitlines()
-            species = dict()
-            for v in val:
-                v = v.strip().split('#')[0]
-                if not v or not v[0].isalpha():
+            lines = []
+            line = ''
+            val_in = val_in.strip().splitlines()
+            val_in.reverse()
+            for v in val_in:
+                line = v.strip().split('#')[0].replace('.d', '.e') + ' '+ line
+                if not line:
                     continue
-                if v.startswith('species'):
-                    species = dict(species=v.split()[1:])
+                if line[0].isalpha():
+                    lines = [line.split()] + lines
+                    if line.startswith('species'):
+                        break
+                    line = ''
+            species = {}
+            for line in lines:
+                content = [line[1]] if len(line) == 2 else [line[1:]]
+                if line[0] in species:
+                    species[line[0]].extend(content)
                 else:
-                    v = v.replace('.d', '.e').split()
-                    vi = v[1] if len(v[1:]) == 1 else v[1:]
-                    if v[0] in species:
-                        species[v[0]].extend([vi])
-                    else:
-                        species[v[0]] = [vi]
+                    species[line[0]] = content
             return species
 
         self._quantities = [
@@ -146,12 +151,12 @@ def str_to_species(val_in):
                 xsection_method.x_fhi_aims_controlIn_verbatim_writeout,
                 rf'{re_n} *verbatim_writeout\s*([\w]+)', repeats=False),
             Quantity(
-                'xc',
-                rf'{re_n} *xc\s*([\w\. \-\+]+)', repeats=False),
+                'xc', rf'{re_n} *xc\s*([\w\. \-\+]+)', repeats=False),
             Quantity(
                 'species', rf'{re_n} *(species\s+[A-Z][a-z]?[\s\S]+?)'
-                r'(FHI-aims code project|\-{10})',
-                str_operation=str_to_species, repeats=True)]
+                r'FHI-aims code project|\-{10}',
+                str_operation=str_to_species, repeats=True,),
+        ]
 
 
 class FHIAimsOutParser(TextParser):
@@ -1542,20 +1547,31 @@ def parse_basis_set(species):
                 elif key == 'division':
                     pass
                 elif key in basis_funcs:
-                    for i in range(len(val)):
+                    for v in val:
                         sec_basis_func = sec_basis_set.m_create(
                             x_fhi_aims_section_controlIn_basis_func)
                         sec_basis_func.x_fhi_aims_controlIn_basis_func_type = key
-                        sec_basis_func.x_fhi_aims_controlIn_basis_func_n = int(val[i][0])
-                        sec_basis_func.x_fhi_aims_controlIn_basis_func_l = str(val[i][1])
-                        if len(val[i]) == 3 and hasattr(val[i][2], 'real'):
-                            sec_basis_func.x_fhi_aims_controlIn_basis_func_radius = val[i][2]
+                        if key == 'gaussian':
+                            sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_l = int(v[0])
+                            gauss_alphas, gauss_coeffs = [], []
+                            for gaussian_index, gaussian_extra in enumerate(v[2:]):
+                                if gaussian_index % 2:
+                                    gauss_alphas.append(float(gaussian_extra))
+                                else:
+                                    gauss_coeffs.append(float(gaussian_extra))
+                            sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_alphas = np.array(gauss_alphas) / ureg.bohr ** 2
+                            sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_coeffs = gauss_coeffs
+                        else:
+                            sec_basis_func.x_fhi_aims_controlIn_basis_func_n = int(v[0])
+                            sec_basis_func.x_fhi_aims_controlIn_basis_func_l = str(v[1])
+                            if len(v) == 3 and hasattr(v[2], 'real'):
+                                sec_basis_func.x_fhi_aims_controlIn_basis_func_radius = v[2]
                 elif key in ['cut_pot', 'radial_base']:
                     setattr(sec_basis_set, 'x_fhi_aims_controlIn_%s' % key, np.array(
                         val[0], dtype=float))
                 else:
                     try:
-                        setattr(sec_basis_set, 'x_fhi_aims_controlIn_%s' % key, val[0])
+                        setattr(sec_basis_set, 'x_fhi_aims_controlIn_%s' % key, v[0])
                     except Exception:
                         self.logger.warning('Error setting controlIn metainfo.', details={key: key})
 
@@ -1566,7 +1582,7 @@ def parse_basis_set(species):
                 sec_basis_set.x_fhi_aims_controlIn_division = division
 
             # store hash
-            sec_basis_set.x_fhi_aims_controlIn_hash = get_basis_hash([sec_basis_set], [True])
+            sec_basis_set.x_fhi_aims_controlIn_hash = hash_section([sec_basis_set], [True])
 
         def _get_elemental_tier(
                 basis_settings: x_fhi_aims_section_controlIn_basis_set,
diff --git a/electronicparsers/utils/__init__.py b/electronicparsers/utils/__init__.py
index e61060f6..ee10a158 100644
--- a/electronicparsers/utils/__init__.py
+++ b/electronicparsers/utils/__init__.py
@@ -17,5 +17,5 @@
 # limitations under the License.
 
 from .utils import (
-    extract_section, get_files, BeyondDFTWorkflowsParser, get_basis_hash
+    extract_section, get_files, BeyondDFTWorkflowsParser, hash_section
 )
diff --git a/electronicparsers/utils/utils.py b/electronicparsers/utils/utils.py
index b50d438f..88a73b0f 100644
--- a/electronicparsers/utils/utils.py
+++ b/electronicparsers/utils/utils.py
@@ -90,8 +90,8 @@ def get_files(pattern: str, filepath: str, stripname: str = '', deep: bool = Tru
 
 
 def hash_section(
-        sections: Union(MSection, list[MSection]),
-        subsections: Union(bool, list[bool]),
+        sections: Union[MSection, list[MSection]],
+        subsections: Union[bool, list[bool]],
         **kwargs,
     ) -> str:
     '''
@@ -100,16 +100,16 @@ def hash_section(
     The option consists of adding also general settings to the hash.
 
     There are two modes determining whether sections are defined by the `quantities` provided (`inclusion`)
-    or rather `quantities` are explicitely removed (`exclusion`).
+    or rather `quantities` are explicitly removed (`exclusion`).
 
     `sections`: sections to be hashed together
-    `subsections`: list of bools, indicating whether to include susbections. Must be of same length as basis_settings.
-    `mode`: str, either `inclusion` or `exclusion`
+    `subsections`: list of bools, indicating whether to include subsections. Must be of same length as basis_settings.
+    `mode`: str, either `include` or `exclude` (default)
     `quantities`: list of str, quantities to be included or excluded
     '''
     sections = [sections] if isinstance(sections, MSection) else sections
     subsections = [subsections] if isinstance(subsections, bool) else subsections
-    mode: str = kwargs.get('mode', 'exclusion')
+    mode: str = kwargs.get('mode', 'exclude')
     quantities: list[str] = kwargs.get('quantities', [])
     # sanity checks
     try:
@@ -129,8 +129,8 @@ def hash_section(
                 isinstance(getattr(section, key), (MSection, MSubSectionList)):
                     continue
             if key == 'm_def' or\
-                (mode == 'exclusion' and key not in quantities) or\
-                (mode == 'inclusion' and key in quantities):
+                (mode == 'exclude' and key not in quantities) or\
+                (mode == 'include' and key in quantities):
                     to_write[key] = val
         to_compare.append(to_write)
     # hash the filtered sections

From 615dd2dc9733e7f130a8aedbd966e2175ecfbd20 Mon Sep 17 00:00:00 2001
From: "nathan.daelman@physik.hu-berlin.de" <ndaelman@physik.hu-berlin.de>
Date: Tue, 18 Jul 2023 19:59:33 +0200
Subject: [PATCH 8/8] Fix Gaussian parsing

---
 electronicparsers/fhiaims/parser.py | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/electronicparsers/fhiaims/parser.py b/electronicparsers/fhiaims/parser.py
index ba4947d2..9642a276 100644
--- a/electronicparsers/fhiaims/parser.py
+++ b/electronicparsers/fhiaims/parser.py
@@ -78,7 +78,7 @@ def init_quantities(self):
         def str_to_species(val_in):
             lines = []
             line = ''
-            val_in = val_in.strip().splitlines()
+            val_in = val_in.strip().splitlines()[:-1]
             val_in.reverse()
             for v in val_in:
                 line = v.strip().split('#')[0].replace('.d', '.e') + ' '+ line
@@ -154,7 +154,7 @@ def str_to_species(val_in):
                 'xc', rf'{re_n} *xc\s*([\w\. \-\+]+)', repeats=False),
             Quantity(
                 'species', rf'{re_n} *(species\s+[A-Z][a-z]?[\s\S]+?)'
-                r'FHI-aims code project|\-{10}',
+                r'(FHI-aims code project|\-{10})',
                 str_operation=str_to_species, repeats=True,),
         ]
 
@@ -1556,11 +1556,12 @@ def parse_basis_set(species):
                             gauss_alphas, gauss_coeffs = [], []
                             for gaussian_index, gaussian_extra in enumerate(v[2:]):
                                 if gaussian_index % 2:
-                                    gauss_alphas.append(float(gaussian_extra))
-                                else:
                                     gauss_coeffs.append(float(gaussian_extra))
+                                else:
+                                    gauss_alphas.append(float(gaussian_extra))
                             sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_alphas = np.array(gauss_alphas) / ureg.bohr ** 2
-                            sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_coeffs = gauss_coeffs
+                            if gauss_coeffs:
+                                sec_basis_func.x_fhi_aims_controlIn_basis_func_gauss_coeffs = gauss_coeffs
                         else:
                             sec_basis_func.x_fhi_aims_controlIn_basis_func_n = int(v[0])
                             sec_basis_func.x_fhi_aims_controlIn_basis_func_l = str(v[1])