6774254cb0
Task 4/6: Add 5 macros to compress repeated country code patterns: - @country_name / @country_slug: 20-country CASE in dim_cities, dim_locations - @normalize_eurostat_country / @normalize_eurostat_nuts: EL→GR, UK→GB - @infer_country_from_coords: bounding box for 8 markets Net: +91 lines in macros, -135 lines in models = -44 lines total. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
108 lines
3.9 KiB
Python
108 lines
3.9 KiB
Python
import os
|
|
|
|
from sqlmesh import macro
|
|
|
|
|
|
@macro()
|
|
def padelnomics_glob(evaluator) -> str:
|
|
"""Return a quoted glob path for all padelnomics CSV gz files under LANDING_DIR.
|
|
|
|
Used in raw models: SELECT * FROM read_csv(@padelnomics_glob(), ...)
|
|
|
|
The LANDING_DIR variable is read from the SQLMesh config variables block first,
|
|
then falls back to the LANDING_DIR environment variable, then to 'data/landing'.
|
|
"""
|
|
landing_dir = evaluator.var("LANDING_DIR") or os.environ.get("LANDING_DIR", "data/landing")
|
|
return f"'{landing_dir}/padelnomics/**/*.csv.gz'"
|
|
|
|
|
|
# ── Country code helpers ─────────────────────────────────────────────────────
|
|
# Shared lookup used by dim_cities and dim_locations.
|
|
|
|
_COUNTRY_NAMES = {
|
|
"DE": "Germany", "ES": "Spain", "GB": "United Kingdom",
|
|
"FR": "France", "IT": "Italy", "PT": "Portugal",
|
|
"AT": "Austria", "CH": "Switzerland", "NL": "Netherlands",
|
|
"BE": "Belgium", "SE": "Sweden", "NO": "Norway",
|
|
"DK": "Denmark", "FI": "Finland", "US": "United States",
|
|
"AR": "Argentina", "MX": "Mexico", "AE": "UAE",
|
|
"AU": "Australia", "IE": "Ireland",
|
|
}
|
|
|
|
|
|
def _country_case(col: str) -> str:
|
|
"""Build a CASE expression mapping ISO 3166-1 alpha-2 → English name."""
|
|
whens = "\n ".join(
|
|
f"WHEN '{code}' THEN '{name}'" for code, name in _COUNTRY_NAMES.items()
|
|
)
|
|
return f"CASE {col}\n {whens}\n ELSE {col}\n END"
|
|
|
|
|
|
@macro()
|
|
def country_name(evaluator, code_col) -> str:
|
|
"""CASE expression: country code → English name.
|
|
|
|
Usage in SQL: @country_name(vc.country_code) AS country_name_en
|
|
"""
|
|
return _country_case(str(code_col))
|
|
|
|
|
|
@macro()
|
|
def country_slug(evaluator, code_col) -> str:
|
|
"""CASE expression: country code → URL-safe slug (lowercased, spaces → dashes).
|
|
|
|
Usage in SQL: @country_slug(vc.country_code) AS country_slug
|
|
"""
|
|
return f"LOWER(REGEXP_REPLACE({_country_case(str(code_col))}, '[^a-zA-Z0-9]+', '-'))"
|
|
|
|
|
|
@macro()
|
|
def normalize_eurostat_country(evaluator, code_col) -> str:
|
|
"""Normalize Eurostat country codes to ISO 3166-1 alpha-2: EL→GR, UK→GB.
|
|
|
|
Usage in SQL: @normalize_eurostat_country(geo_code) AS country_code
|
|
"""
|
|
col = str(code_col)
|
|
return f"CASE {col} WHEN 'EL' THEN 'GR' WHEN 'UK' THEN 'GB' ELSE {col} END"
|
|
|
|
|
|
@macro()
|
|
def normalize_eurostat_nuts(evaluator, code_col) -> str:
|
|
"""Normalize NUTS code prefix: EL→GR, UK→GB, preserving the suffix.
|
|
|
|
Usage in SQL: @normalize_eurostat_nuts(geo_code) AS nuts_code
|
|
"""
|
|
col = str(code_col)
|
|
return (
|
|
f"CASE"
|
|
f" WHEN {col} LIKE 'EL%' THEN 'GR' || SUBSTR({col}, 3)"
|
|
f" WHEN {col} LIKE 'UK%' THEN 'GB' || SUBSTR({col}, 3)"
|
|
f" ELSE {col}"
|
|
f" END"
|
|
)
|
|
|
|
|
|
@macro()
|
|
def infer_country_from_coords(evaluator, lat_col, lon_col) -> str:
|
|
"""Infer ISO country code from lat/lon using bounding boxes for 8 European markets.
|
|
|
|
Usage in SQL:
|
|
COALESCE(NULLIF(TRIM(UPPER(country_code)), ''),
|
|
@infer_country_from_coords(lat, lon)) AS country_code
|
|
"""
|
|
lat = str(lat_col)
|
|
lon = str(lon_col)
|
|
return (
|
|
f"CASE"
|
|
f" WHEN {lat} BETWEEN 47.27 AND 55.06 AND {lon} BETWEEN 5.87 AND 15.04 THEN 'DE'"
|
|
f" WHEN {lat} BETWEEN 35.95 AND 43.79 AND {lon} BETWEEN -9.39 AND 4.33 THEN 'ES'"
|
|
f" WHEN {lat} BETWEEN 49.90 AND 60.85 AND {lon} BETWEEN -8.62 AND 1.77 THEN 'GB'"
|
|
f" WHEN {lat} BETWEEN 41.36 AND 51.09 AND {lon} BETWEEN -5.14 AND 9.56 THEN 'FR'"
|
|
f" WHEN {lat} BETWEEN 45.46 AND 47.80 AND {lon} BETWEEN 5.96 AND 10.49 THEN 'CH'"
|
|
f" WHEN {lat} BETWEEN 46.37 AND 49.02 AND {lon} BETWEEN 9.53 AND 17.16 THEN 'AT'"
|
|
f" WHEN {lat} BETWEEN 36.35 AND 47.09 AND {lon} BETWEEN 6.62 AND 18.51 THEN 'IT'"
|
|
f" WHEN {lat} BETWEEN 37.00 AND 42.15 AND {lon} BETWEEN -9.50 AND -6.19 THEN 'PT'"
|
|
f" ELSE NULL"
|
|
f" END"
|
|
)
|