Spaces:
Sleeping
Sleeping
Create ingest/timeline.py
Browse files- ingest/timeline.py +18 -23
ingest/timeline.py
CHANGED
|
@@ -1,31 +1,26 @@
|
|
| 1 |
-
|
| 2 |
from datetime import datetime
|
| 3 |
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
|
|
|
|
|
|
| 8 |
|
| 9 |
for r in results:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
try:
|
| 11 |
-
|
|
|
|
|
|
|
| 12 |
except Exception:
|
| 13 |
continue
|
| 14 |
-
labels.append(r["title"])
|
| 15 |
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
y=list(range(len(dates))),
|
| 21 |
-
mode="markers",
|
| 22 |
-
text=labels
|
| 23 |
-
)
|
| 24 |
-
],
|
| 25 |
-
layout=go.Layout(
|
| 26 |
-
title="Document Release Timeline",
|
| 27 |
-
xaxis_title="Release Date",
|
| 28 |
-
yaxis_title="Document Index"
|
| 29 |
-
)
|
| 30 |
-
)
|
| 31 |
-
return fig
|
|
|
|
| 1 |
+
from collections import defaultdict
|
| 2 |
from datetime import datetime
|
| 3 |
|
| 4 |
+
def release_timeline(results):
|
| 5 |
+
"""
|
| 6 |
+
Aggregate histogram by year-month.
|
| 7 |
+
Safe for HF: no user tracking, no persistence.
|
| 8 |
+
"""
|
| 9 |
+
buckets = defaultdict(int)
|
| 10 |
|
| 11 |
for r in results:
|
| 12 |
+
date_str = r.get("date")
|
| 13 |
+
if not date_str:
|
| 14 |
+
continue
|
| 15 |
+
|
| 16 |
try:
|
| 17 |
+
dt = datetime.fromisoformat(date_str[:10])
|
| 18 |
+
key = dt.strftime("%Y-%m")
|
| 19 |
+
buckets[key] += 1
|
| 20 |
except Exception:
|
| 21 |
continue
|
|
|
|
| 22 |
|
| 23 |
+
return [
|
| 24 |
+
{"Period": k, "Documents": v}
|
| 25 |
+
for k, v in sorted(buckets.items())
|
| 26 |
+
]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|