111 lines
3.7 KiB
Python
111 lines
3.7 KiB
Python
"""
|
|
Rule taxonomy for BZO extraction.
|
|
Defines fixed rule types and their patterns for deterministic rule detection.
|
|
"""
|
|
|
|
RULE_TAXONOMY = {
|
|
"max_building_height": {
|
|
"patterns": ["fassadenhöhe", "fassadenhöhen", "gebäudehöhe", "firsthöhe", "traufhöhe", "höchsthöhe", "gesamt höhe", "gesamt höhen"],
|
|
"units": ["m", "meter", "metern"],
|
|
"value_type": "numeric",
|
|
"keywords": ["max", "maximal", "höchstens"]
|
|
},
|
|
"max_floors": {
|
|
"patterns": ["vollgeschosse", "vollgeschoss", "geschosse", "geschosszahl"],
|
|
"units": [],
|
|
"value_type": "integer",
|
|
"keywords": ["max", "maximal", "höchstens"]
|
|
},
|
|
"max_attachable_attics": {
|
|
"patterns": ["anrechenbare dachgeschosse", "anrechenbares attikageschoss", "dachgeschosse", "attikageschoss"],
|
|
"units": [],
|
|
"value_type": "integer",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"max_attachable_basement": {
|
|
"patterns": ["anrechenbares untergeschoss", "untergeschoss"],
|
|
"units": [],
|
|
"value_type": "integer",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"density": {
|
|
"patterns": ["ausnützungsziffer", "az", "ausnützung"],
|
|
"units": ["%", "prozent"],
|
|
"value_type": "numeric",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"building_coverage": {
|
|
"patterns": ["überbauungsziffer", "überbauungsziffer max", "uz"],
|
|
"units": ["%", "prozent"],
|
|
"value_type": "numeric",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"building_mass_index": {
|
|
"patterns": ["baumassenziffer", "bmz"],
|
|
"units": [],
|
|
"value_type": "numeric",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"green_space_index": {
|
|
"patterns": ["grünflächenziffer", "gfz"],
|
|
"units": ["%", "prozent"],
|
|
"value_type": "numeric",
|
|
"keywords": ["min", "mindestens"]
|
|
},
|
|
"boundary_distance": {
|
|
"patterns": ["grenzabstand", "grundabstand", "abstand"],
|
|
"units": ["m", "meter", "metern"],
|
|
"value_type": "numeric",
|
|
"keywords": ["min", "mindestens", "max", "maximal"]
|
|
},
|
|
"boundary_distance_length_surcharge": {
|
|
"patterns": ["mehrlängenzuschlag", "längenzuschlag"],
|
|
"units": [],
|
|
"value_type": "fraction", # e.g., "1/3", "1/5"
|
|
"keywords": []
|
|
},
|
|
"boundary_distance_max": {
|
|
"patterns": ["höchstmass", "höchstmass grenzabstand", "höchstmass abstand"],
|
|
"units": ["m", "meter"],
|
|
"value_type": "numeric",
|
|
"keywords": ["max", "maximal"]
|
|
},
|
|
"building_length": {
|
|
"patterns": ["gebäudelänge"],
|
|
"units": ["m", "meter"],
|
|
"value_type": "numeric",
|
|
"keywords": ["min", "mindestens", "max", "maximal"]
|
|
},
|
|
"building_width": {
|
|
"patterns": ["gebäudebreite"],
|
|
"units": ["m", "meter"],
|
|
"value_type": "numeric",
|
|
"keywords": ["min", "mindestens", "max", "maximal"]
|
|
},
|
|
"residential_area_share": {
|
|
"patterns": ["wohnflächenanteil", "wohnanteil"],
|
|
"units": ["%", "prozent"],
|
|
"value_type": "numeric",
|
|
"keywords": ["min", "mindestens", "max", "maximal"]
|
|
}
|
|
}
|
|
|
|
|
|
def get_rule_taxonomy() -> dict:
|
|
"""Get the rule taxonomy dictionary."""
|
|
return RULE_TAXONOMY
|
|
|
|
|
|
def get_rule_types() -> list:
|
|
"""Get list of all rule types."""
|
|
return list(RULE_TAXONOMY.keys())
|
|
|
|
|
|
def get_rule_patterns(rule_type: str) -> list:
|
|
"""Get patterns for a specific rule type."""
|
|
return RULE_TAXONOMY.get(rule_type, {}).get("patterns", [])
|
|
|
|
|
|
def get_rule_units(rule_type: str) -> list:
|
|
"""Get units for a specific rule type."""
|
|
return RULE_TAXONOMY.get(rule_type, {}).get("units", [])
|