Reddit scraper PRs get a review summary; the form/Reddit overlap is measured (#23, #24)
CI / test (pull_request) Successful in 40s

- collection/etl/review_reddit.py (`python -m collection review-reddit`) and .gitea/workflows/reddit-review.yml:
  on every scraper pull request, one comment (refreshed on each push) listing the runs added, changed and gone, the
  runs that break a game rule (left out of the model and stats until corrected), other flags, runs that may also
  be in the form, and result pages already counted; plus how to reject / fix / add. forge.upsert_comment keeps it
  to one comment. collection/README.md: "Reviewing Reddit runs".
- analysis/overlap/overlap.py -> data/public/overlap.json: a Reddit run with at least 4 route answers "matches" a
  form row giving the same answers within a day; chance is the same test a few weeks off. 9 of 36 comparable runs
  match, ~7.3 by chance: about 2 of 82 look reported twice, too few to matter. Nothing is removed. Shown on the
  page's Data & methods tab; in the dataset metadata; checked in test:py; part of `rebuild`.
- Neither the form nor the register records a result-page id, so the only proven copy (comment <-> result page)
  stays the one #18 applies.
- data/public/stats.json: the statistics of #42 (impossible runs left out), which did not make it into that merge.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ
This commit is contained in:
ekcSchiano
2026-11-19 16:44:52 +00:00
co-authored by Claude Opus 5.5
parent 0df6b8d473
commit 98be27fc1f
15 changed files with 497 additions and 6 deletions
+112
View File
@@ -0,0 +1,112 @@
"""Is the same attempt reported twice, in the community form and in a Reddit comment? (issue #24)
python analysis/overlap/overlap.py [--check]
There are no player ids, so a copy can only be proven by a shared dragonkind.com result-page id. Neither the form
nor the register records one, and the comment-to-result-page link is already applied when the runs are built
(collection/etl/runs.py). Everything else is measured, never removed:
- a Reddit comment run is *comparable* when it is dated and gives at least MIN_ANSWERS route answers besides how it
ended (fewer answers match dozens of form rows by chance on a busy day);
- it *matches* a form row submitted within WINDOW days when every answer it gives is the form row's answer;
- the same test with the comment's date shifted by a few weeks (SHIFTS, only where the form has rows that week) is
what chance alone produces, so matched - expected is the estimate of runs reported twice.
Writes data/public/overlap.json (the page's Data & methods tab, and the scraper pull-request summary). Nothing is
dropped: identical routes are expected, and the Reddit runs already have their own reporting-bias group in the
route model (planner/policy.yaml).
"""
import argparse
import collections
import datetime as dt
import json
import sys
from pathlib import Path
ROOT = Path(__file__).resolve().parents[2]
sys.path.insert(0, str(ROOT / "core"))
from atlas import runs as runs_io # noqa: E402
OUT = ROOT / "data/public/overlap.json"
FIELDS = ["company", "left_squadmates", "direction", "first_sighting", "sighting_action", "scream", "resolution",
"combat_first", "combat_second", "post_path", "hour", "branch", "outcome", "outcome_color", "tail", "attempt"]
ENDING = ("outcome", "outcome_color")
MIN_ANSWERS = 4
WINDOW = 1
SHIFTS = (-28, -21, -14, 14, 21, 28)
def given(v):
return v not in (None, "") and v != "Unknown" and not (isinstance(v, str) and v.startswith("("))
def answers(r):
return [k for k in FIELDS if given(r.get(k))]
def comparable(r):
return bool(r.get("date")) and len([k for k in answers(r) if k not in ENDING]) >= MIN_ANSWERS
def days(date, shift=0, window=WINDOW):
d = dt.date.fromisoformat(date[:10]) + dt.timedelta(days=shift)
return [(d + dt.timedelta(days=i)).isoformat() for i in range(-window, window + 1)]
def form_index(runs):
by_day = collections.defaultdict(list)
for r in runs:
if r["source"] == "reddit spreadsheet" and r.get("date"):
by_day[r["date"][:10]].append(r)
return by_day
def matches(c, by_day, shift=0):
"""Form rows within the window that give every answer the comment gives."""
keys = answers(c)
return [f for d in days(c["date"], shift) for f in by_day.get(d, []) if all(f.get(k) == c[k] for k in keys)]
def measure(runs):
by_day = form_index(runs)
comments = [r for r in runs if r["source"] == "reddit comment"]
comp = [c for c in comments if comparable(c)]
found = {c["run_id"]: [f["run_id"] for f in matches(c, by_day)] for c in comp}
matched = sum(1 for v in found.values() if v)
# chance: each comment's match rate on shifted dates, over the shifts whose window has form rows at all
expected = 0.0
for c in comp:
valid = [s for s in SHIFTS if any(by_day.get(d) for d in days(c["date"], s))]
if valid:
expected += sum(1 for s in valid if matches(c, by_day, s)) / len(valid)
return {
"schema": 1,
"rule": {"min_answers": MIN_ANSWERS, "window_days": WINDOW, "baseline_shifts_days": list(SHIFTS)},
"comment_runs": len(comments),
"comparable": len(comp),
"matched": matched,
"expected_by_chance": round(expected, 1),
"likely_double": round(max(0.0, matched - expected), 1),
"matches": [{"comment": k, "form": v} for k, v in sorted(found.items()) if v],
}
def main(argv=None):
ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
ap.add_argument("--check", action="store_true", help="fail if data/public/overlap.json is out of date")
a = ap.parse_args(argv)
text = json.dumps(measure(runs_io.read(drop=())), indent=1, ensure_ascii=False) + "\n"
if a.check:
if not OUT.exists() or OUT.read_text(encoding="utf-8") != text:
print(f"out of date (run `python {Path(__file__).relative_to(ROOT)}`): {OUT.relative_to(ROOT)}")
return 1
return 0
OUT.write_text(text, encoding="utf-8")
m = json.loads(text)
print(f"{OUT.relative_to(ROOT)}: {m['matched']} of {m['comparable']} comparable Reddit runs match a form row "
f"(~{m['expected_by_chance']} by chance; ~{m['likely_double']} likely reported twice)")
return 0
if __name__ == "__main__":
sys.exit(main())