Code and data for arewedecentralizedyet.online and related projects
Add a --now argument that enables us to pretend we're running in the past
+37
-9
+37
-9
data-processing/fedi-nodeinfo/parse-nodeinfo.py
+37
-9
data-processing/fedi-nodeinfo/parse-nodeinfo.py
···
1
1
#!/usr/bin/env python3
2
+
import argparse
2
3
import sys
3
4
import os
4
5
import json
···
210
211
211
212
return hostname, software_name, software_version, users_total, active_month, protocols, protocols_str
212
213
214
+
def _parse_now(value: Optional[str]) -> "datetime.datetime":
215
+
import datetime
216
+
217
+
if value is None:
218
+
return datetime.datetime.now(datetime.timezone.utc)
219
+
iso_value = value.strip()
220
+
if iso_value.endswith("Z"):
221
+
iso_value = iso_value[:-1] + "+00:00"
222
+
try:
223
+
parsed = datetime.datetime.fromisoformat(iso_value)
224
+
except ValueError as exc:
225
+
raise argparse.ArgumentTypeError(f"Invalid --now value: {value}") from exc
226
+
if parsed.tzinfo is None:
227
+
return parsed.replace(tzinfo=datetime.timezone.utc)
228
+
return parsed.astimezone(datetime.timezone.utc)
229
+
230
+
213
231
def main() -> None:
214
232
import datetime
215
233
216
-
if len(sys.argv) < 3 or len(sys.argv) > 4:
217
-
print(f"Usage: {sys.argv[0]} nodeinfo_dir output.csv [max_age_days=30]", file=sys.stderr)
218
-
sys.exit(1)
234
+
parser = argparse.ArgumentParser(
235
+
description="Parse nodeinfo JSON into a CSV snapshot.",
236
+
)
237
+
parser.add_argument("nodeinfo_dir")
238
+
parser.add_argument("output_csv")
239
+
parser.add_argument("max_age_days", nargs="?", type=int, default=30)
240
+
parser.add_argument(
241
+
"--now",
242
+
type=_parse_now,
243
+
help="ISO-8601 datetime to use as the current time (UTC if naive).",
244
+
)
245
+
args = parser.parse_args()
219
246
220
-
nodeinfo_dir = sys.argv[1]
221
-
output_csv = sys.argv[2]
222
-
max_age_days = int(sys.argv[3]) if len(sys.argv) == 4 else 30
247
+
nodeinfo_dir = args.nodeinfo_dir
248
+
output_csv = args.output_csv
249
+
max_age_days = args.max_age_days
223
250
224
251
cutoff = datetime.timedelta(days=max_age_days)
225
-
# Make 'now' timezone-aware (UTC)
226
-
now = datetime.datetime.now(datetime.timezone.utc)
252
+
now = args.now or datetime.datetime.now(datetime.timezone.utc)
227
253
228
254
hostname_dirs = [
229
255
os.path.join(nodeinfo_dir, d)
···
256
282
if not candidates:
257
283
continue
258
284
259
-
candidates_recent = [c for c in candidates if now - c[0] <= cutoff]
285
+
candidates_recent = [
286
+
c for c in candidates if c[0] <= now and now - c[0] <= cutoff
287
+
]
260
288
if not candidates_recent:
261
289
continue
262
290