cdl-parser

pip install gemmology-cdl-parser

from cdl_parser import parse_cdl

desc = parse_cdl("cubic[m3m]:{111}@1.0 + {100}@1.3")

print(desc.system)       # 'cubic'
print(desc.point_group)  # 'm3m'
print(len(desc.forms))   # 2
print(desc.forms[0])     # {111}

# With features
desc = parse_cdl("cubic[m3m]:{111}@1.0[phantom:3] + {100}@1.3")
print(desc.forms[0].features)  # [Feature(name='phantom', values=[3])]

# With definitions and references
desc = parse_cdl("""
@base = {111}@1.0 + {100}@1.3
cubic[m3m]:$base
""")
print(desc.definitions)  # [Definition(name='base', ...)]

# Amorphous materials (v2.0)
desc = parse_cdl("amorphous[opalescent]:{nodular, botryoidal} | phenomenon[play_of_color:spectral]")
print(type(desc))        # <class 'AmorphousDescription'>
print(desc.subtype)      # 'opalescent'
print(desc.shapes)       # ['nodular', 'botryoidal']

# Nested growth (v2.0)
desc = parse_cdl("cubic[m3m]:{111}@1.0 > {111}@0.6")
print(desc.forms[0])     # {111} > {111}@0.6

# Aggregates (v2.0)
# The aggregate wraps the form as a top-level FormNode (forms[0] IS the AggregateSpec)
desc = parse_cdl("cubic[m3m]:{100}@1.0 ~ cluster[6]")
agg = desc.forms[0]
print(agg)               # {100} ~ cluster[6]
print(agg.arrangement)   # 'cluster'
print(agg.count)         # 6

from cdl_parser import validate_cdl

valid, error = validate_cdl("cubic[m3m]:{111}")
print(valid)   # True
print(error)   # None

valid, error = validate_cdl("invalid[xyz]:{abc}")
print(valid)   # False
print(error)   # "Expected SYSTEM, got IDENTIFIER ..."

from cdl_parser import strip_comments

text = """
#! Mineral: Diamond
#! System: Cubic
# This is a comment
cubic[m3m]:{111} /* inline block */
"""
cleaned, doc_comments = strip_comments(text)
print(doc_comments)  # ['Mineral: Diamond', 'System: Cubic']

# Get a flat list of all CrystalForm objects (flattens FormGroups)
# Features from parent groups are merged into child forms
flat = desc.flat_forms()  # list[CrystalForm]

# Convert to dictionary representation
d = desc.to_dict()

# String representation (reconstructs CDL; scale is omitted when it equals 1.0)
print(str(desc))  # "cubic[m3m]:{111} + {100}@1.3"

# Features are parsed from [name:value] syntax on forms
desc = parse_cdl("cubic[m3m]:{111}@1.0[phantom:3, colour:blue]")
form = desc.forms[0]  # CrystalForm
print(form.features)
# [Feature(name='phantom', values=[3]), Feature(name='colour', values=['blue'])]

desc = parse_cdl("cubic[m3m]:{111} | twin(spinel)")
print(desc.twin)       # twin(spinel)
print(desc.twin.law)   # 'spinel'
print(desc.twin.count) # 2

desc = parse_cdl("cubic[m3m]:{111} | flatten(c:0.5)")
print(desc.modifications)
# [Modification(type='flatten', params={'axis': 'c', 'ratio': 0.5})]

desc = parse_cdl("trigonal[-3m]:{10-10}@1.0 | phenomenon[asterism:6]")
print(desc.phenomenon)       # phenomenon[asterism, value:6]
print(desc.phenomenon.type)  # 'asterism'

desc = parse_cdl("amorphous[opalescent]:{nodular, botryoidal} | phenomenon[play_of_color:spectral]")
print(desc.subtype)             # 'opalescent'
print(desc.shapes)              # ['nodular', 'botryoidal']
print(desc.phenomenon.type)     # 'play_of_color'

desc = parse_cdl("cubic[m3m]:{111}@1.0 > {111}@0.6")
ng = desc.forms[0]              # NestedGrowth
print(ng.base)                  # {111} (scale 1.0 is omitted in str())
print(ng.overgrowth)            # {111}@0.6

desc = parse_cdl("cubic[m3m]:{100}@1.0 ~ cluster[6]")
agg = desc.forms[0]             # AggregateSpec
print(agg.form)                 # {100} — the wrapped CrystalForm
print(agg.arrangement)          # 'cluster'
print(agg.count)                # 6
print(agg.spacing)              # None

from cdl_parser import MillerIndex

# 3-index notation
mi = MillerIndex(1, 1, 1)
print(mi)            # {111}
print(mi.as_tuple()) # (1, 1, 1)

# 4-index notation (Miller-Bravais)
mi = MillerIndex(1, 0, 1, i=-1)
print(mi)              # {10-11}
print(mi.as_tuple())   # (1, 0, -1, 1)
print(mi.as_3index())  # (1, 0, 1)

from cdl_parser import parse_cdl, ParseError

try:
    parse_cdl("invalid")
except ParseError as e:
    print(e.message)    # Error description
    print(e.position)   # Character position in input

from cdl_parser import parse_cdl, ValidationError

try:
    parse_cdl("cubic[-3m]:{111}")  # -3m is trigonal, not cubic
except ValidationError as e:
    print(e.message)  # "Point group '-3m' not valid for cubic system"
    print(e.field)    # "point_group"
    print(e.value)    # "-3m"

from cdl_parser import CRYSTAL_SYSTEMS

print(CRYSTAL_SYSTEMS)
# {'cubic', 'tetragonal', 'orthorhombic', 'hexagonal', 'trigonal', 'monoclinic', 'triclinic'}

from cdl_parser import POINT_GROUPS

print(POINT_GROUPS['cubic'])
# {'m3m', '432', '-43m', 'm-3', '23'}

from cdl_parser import DEFAULT_POINT_GROUPS

print(DEFAULT_POINT_GROUPS['cubic'])     # 'm3m'
print(DEFAULT_POINT_GROUPS['trigonal'])  # '-3m'

from cdl_parser import NAMED_FORMS

print(NAMED_FORMS['octahedron'])     # (1, 1, 1)
print(NAMED_FORMS['cube'])           # (1, 0, 0)
print(NAMED_FORMS['dodecahedron'])   # (1, 1, 0)
print(NAMED_FORMS['rhombohedron'])   # (1, 0, 1)

from cdl_parser import TWIN_LAWS

print(sorted(TWIN_LAWS))
# ['albite', 'baveno', 'brazil', 'carlsbad', 'dauphine', 'fluorite',
#  'gypsum_swallow', 'iron_cross', 'japan', 'manebach', 'pericline',
#  'spinel', 'spinel_law', 'staurolite_60', 'staurolite_90', 'trilling']

from cdl_parser import Lexer, Token, TokenType

lexer = Lexer("cubic[m3m]:{111}")
tokens = lexer.tokenize()
for token in tokens:
    print(f"{token.type.value}: {token.value}")
# SYSTEM: cubic
# LBRACKET: [
# POINT_GROUP: m3m
# RBRACKET: ]
# COLON: :
# LBRACE: {
# INTEGER: 111
# RBRACE: }
# EOF: None