INNER CODE UNIT · Python
prepare_tag_lists
Zverik/every_door · tools/add_taginfo.py:66
def prepare_tag_lists(cur, tcur, wcur):
def build_rows(tags: set[str], usage: dict[str, int]) -> Generator[tuple]:
# We duplicate rows for multi-word values.
for kv in tags:
count = usage[kv]
k, v = kv.split('=', 1)
words = v.split('_')
for i, word in enumerate(words):
yield (word.lower(), k, v, count if i == 0 else count // 2)
# Query all tags for POI keys with enough usage.
ph = ','.join('?' * len(POI_KEYS))
tcur.execute(
"select key, value, count_all from tags "
f"where key in ({ph}) and count_nodes >= 100 "
"and value not in ('yes', 'no', 'fixme')",
POI_KEYS)
# Under 1000 results, no point in memory management.