stjames 0.0.42__py3-none-any.whl → 0.0.43__py3-none-any.whl

stjames/atom.py

@@ -60,7 +60,7 @@ class Atom(Base):
         Atom(1, [0.00000, 0.00000, 0.00000])
         """
         name, *xyz = xyz_line.split()
-        symbol = int(name) if name.isdigit() else SYMBOL_ELEMENT[name]
+        symbol = int(name) if name.isdigit() else SYMBOL_ELEMENT[name.title()]
         if not len(xyz) == 3:
             raise ValueError("XYZ file should have 3 coordinates per atom")
         return cls(atomic_number=symbol, position=xyz)

stjames/molecule.py

@@ -1,8 +1,9 @@
+import re
 from pathlib import Path
 from typing import Iterable, Optional, Self
 
 import pydantic
-from pydantic import NonNegativeInt, PositiveInt
+from pydantic import NonNegativeInt, PositiveInt, ValidationError
 
 from .atom import Atom
 from .base import Base
@@ -135,6 +136,8 @@ class Molecule(Base):
             match format:
                 case "xyz":
                     return cls.from_xyz_lines(f.readlines(), charge=charge, multiplicity=multiplicity)
+                case "extxyz":
+                    return cls.from_extxyz_lines(f.readlines(), charge=charge, multiplicity=multiplicity)
                 case _:
                     raise ValueError(f"Unsupported {format=}")
 
@@ -161,7 +164,7 @@ class Molecule(Base):
 
         try:
             return cls(atoms=[Atom.from_xyz(line) for line in lines], charge=charge, multiplicity=multiplicity)
-        except Exception as e:
+        except (ValueError, ValidationError) as e:
             raise MoleculeReadError("Error reading molecule from xyz") from e
 
     def to_xyz(self, comment: str = "", out_file: Path | str | None = None) -> str:
@@ -190,3 +193,85 @@ class Molecule(Base):
                 f.write(out)
 
         return out
+
+    @classmethod
+    def from_extxyz(cls: type[Self], extxyz: str, charge: int = 0, multiplicity: PositiveInt = 1) -> Self:
+        r"""
+        Generate a Molecule from a EXTXYZ string. Currently only supporting Lattice and Properties fields.
+
+        >>> Molecule.from_extxyz('''
+        ... 2
+        ... Lattice="6.0 0.0 0.0 6.0 0.0 0.0 6.0 0.0 0.0"Properties=species:S:1:pos:R:3
+        ... H 0 0 0
+        ... H 0 0 1
+        ... ''').cell.lattice_vectors
+        ((6.0, 0.0, 0.0), (6.0, 0.0, 0.0), (6.0, 0.0, 0.0))
+        """
+
+        return cls.from_extxyz_lines(extxyz.strip().splitlines(), charge=charge, multiplicity=multiplicity)
+
+    @classmethod
+    def from_extxyz_lines(cls: type[Self], lines: Iterable[str], charge: int = 0, multiplicity: PositiveInt = 1) -> Self:
+        # ensure first line is number of atoms
+        lines = list(lines)
+        if len(lines[0].split()) == 1:
+            natoms = lines[0].strip()
+            if not natoms.isdigit() or (int(lines[0]) != len(lines) - 2):
+                raise MoleculeReadError(f"First line of EXTXYZ file should be the number of atoms, got: {lines[0]} != {len(lines) - 2}")
+            lines = lines[1:]
+        else:
+            raise MoleculeReadError(f"First line of EXTXYZ should be only an int denoting number of atoms. Got {lines[0].split()}")
+
+        # ensure second line contains key-value pairs
+        if "=" not in lines[0]:
+            raise MoleculeReadError(f"Invalid property line, got {lines[0]}")
+
+        cell = parse_comment_line(lines[0])
+        lines = lines[1:]
+
+        try:
+            return cls(atoms=[Atom.from_xyz(line) for line in lines], cell=cell, charge=charge, multiplicity=multiplicity)
+        except (ValueError, ValidationError) as e:
+            raise MoleculeReadError("Error reading molecule from extxyz") from e
+
+
+def parse_comment_line(line: str) -> PeriodicCell:
+    """
+    currently only supporting lattice and porperites fields from comment line
+    modify in future to support other fields from comment from_xyz_lines
+    ex: name, mulitplicity, charge, etc.
+    """
+    cell = None
+    # Regular expression to match key="value", key='value', or key=value
+    pattern = r"(\S+?=(?:\".*?\"|\'.*?\'|\S+))"
+    pairs = re.findall(pattern, line)
+
+    prop_dict = {}
+    for pair in pairs:
+        key, value = pair.split("=", 1)
+        if key.lower() == "lattice":
+            value = value.strip("'\"").split()
+            if len(value) != 9:
+                raise MoleculeReadError(f"Lattice should have 9 entries got {len(value)}")
+
+            # Convert the value to a 3x3 tuple of tuples of floats
+            try:
+                cell = tuple(tuple(map(float, value[i : i + 3])) for i in range(0, 9, 3))
+            except ValueError:
+                raise MoleculeReadError(f"Lattice should be floats, got {value}")
+
+            prop_dict[key] = value
+
+        elif key.lower() == "properties":
+            if value.lower() != "species:s:1:pos:r:3":
+                raise MoleculeReadError(f"Only accepting properties of form species:S:1:pos:R:3, got {value}")
+            prop_dict[key] = value
+        else:
+            raise MoleculeReadError(f"Currently only accepting lattice and propery keys. Got {key}")
+
+    if cell is None:
+        raise MoleculeReadError("Lattice field is required but missing.")
+
+    if "properties" not in [key.lower() for key in prop_dict.keys()]:
+        raise MoleculeReadError(f"Property field is required, got keys {prop_dict.keys()}")
+    return PeriodicCell(lattice_vectors=cell)

{stjames-0.0.42.dist-info → stjames-0.0.43.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: stjames
-Version: 0.0.42
+Version: 0.0.43
 Summary: standardized JSON atom/molecule encoding scheme
 Author-email: Corin Wagen <corin@rowansci.com>
 Project-URL: Homepage, https://github.com/rowansci/stjames

{stjames-0.0.42.dist-info → stjames-0.0.43.dist-info}/RECORD

@@ -1,6 +1,6 @@
 stjames/__init__.py,sha256=LkWCylP4VeXyQL0iu5w6yeYK1UHjcUH55hwWXTy4mQQ,576
 stjames/_deprecated_solvent_settings.py,sha256=gj5j9p3zakIwSTK5_ndqBXJx--IzjZNxZ75z-wipLOo,450
-stjames/atom.py,sha256=2NZ-tyWk6uPrn_rsANFFultNHS0CvgYYXwj8rud991M,2056
+stjames/atom.py,sha256=w7q-x9xpBw4sJ1WGrWt65WAaStxhz-m7dugXCYEOpq4,2064
 stjames/base.py,sha256=9PvUjBeVSkmA3TaruaB0uvjFMbWYTGKXECISNGAj_AU,1201
 stjames/basis_set.py,sha256=wI3M2q9uPf9jhKpAi4E2DrsyKzloDGLRjAlk7krdYgc,949
 stjames/calculation.py,sha256=O2LwwQ_cOLmDOGXTHA9J71YbUZXigUSbvbLA-fSVm3w,915
@@ -12,7 +12,7 @@ stjames/int_settings.py,sha256=5HXp8opt5ZyY1UpmfaK7NVloWVLM5jkG0elEEqpVLUo,896
 stjames/message.py,sha256=Rq6QqmHZKecWxYH8fVyXmuoCCPZv8YinvgykSeorXSU,216
 stjames/method.py,sha256=xnfphxyiWZotxQcmgvpFMJDmGEM2B-_5cbPkgYZBIws,1245
 stjames/mode.py,sha256=xw46Cc7f3eTS8i35qECi-8DocAlANhayK3w4akD4HBU,496
-stjames/molecule.py,sha256=XJdfEJndl4DqY_mg5scWH9xWQU8gNcXN74Qb4n2TN1c,6798
+stjames/molecule.py,sha256=ZyqerT0p0CIykdn0vWfvy2UbU4LsVPCUQ7mSUeILkUI,10524
 stjames/opt_settings.py,sha256=gxXGtjy9l-Q5Wen9eO6T6HHRCuS8rfOofdVQIJj0JcI,550
 stjames/periodic_cell.py,sha256=JDCyynpamggTNi_HnTnnotRbeSMBfYc-srhD-IwUnrg,996
 stjames/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -45,8 +45,8 @@ stjames/workflows/scan.py,sha256=hL4Hco3Ns0dntjh2G2HhhWmED1mbt0gA_hsglPQ5Vjg,814
 stjames/workflows/spin_states.py,sha256=hzGbYfG4KGYYQPV4cF0ufCndDMoIVnZt2ZZt7LqvJgc,4854
 stjames/workflows/tautomer.py,sha256=kZSCHo2Q7LzqtQjF_WyyxjECkndG49T9QOM12hsUkx8,421
 stjames/workflows/workflow.py,sha256=t5Yz-d5WgSQEiHbiZfc4NQ653mEnydiGZ-lEkkPYAKw,994
-stjames-0.0.42.dist-info/LICENSE,sha256=i7ehYBS-6gGmbTcgU4mgk28pyOx2kScJ0kcx8n7bWLM,1084
-stjames-0.0.42.dist-info/METADATA,sha256=mIKojGiwySNE8F0-7hulBNJv2WFORfn08rwCUghPbWM,1628
-stjames-0.0.42.dist-info/WHEEL,sha256=cVxcB9AmuTcXqmwrtPhNK88dr7IR_b6qagTj0UvIEbY,91
-stjames-0.0.42.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
-stjames-0.0.42.dist-info/RECORD,,
+stjames-0.0.43.dist-info/LICENSE,sha256=i7ehYBS-6gGmbTcgU4mgk28pyOx2kScJ0kcx8n7bWLM,1084
+stjames-0.0.43.dist-info/METADATA,sha256=9gZNt8YmerGV47ExAhb0Cie2toEDtPt2SkhgeWf7Yjc,1628
+stjames-0.0.43.dist-info/WHEEL,sha256=cVxcB9AmuTcXqmwrtPhNK88dr7IR_b6qagTj0UvIEbY,91
+stjames-0.0.43.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
+stjames-0.0.43.dist-info/RECORD,,