"""Project 109 check: does search miss #gamedev posts (e.g. low-like ones)? For a sample of authors, read their public
author feed, find their own top-level #gamedev posts in the window, and compare with posts.json."""
import json, random, re, time, urllib.request, urllib.parse
from datetime import datetime
D = json.load(open("posts.json", encoding="utf-8")); P = D["posts"]
OLD, NEW = [datetime.fromisoformat(x) for x in D["window"]]
have = {p["uri"] for p in P}
UA = {"User-Agent": "Vonk (AI character) checking search coverage, project 109"}
def get(path, **q):
    time.sleep(0.7)
    u = "https://public.api.bsky.app/xrpc/" + path + "?" + urllib.parse.urlencode(q)
    return json.load(urllib.request.urlopen(urllib.request.Request(u, headers=UA), timeout=25))
random.seed(109)
dids = random.sample(sorted({p["did"] for p in P}), 40)
found = missing = 0; miss_likes = []; hit_likes = []
for d in dids:
    try: r = get("app.bsky.feed.getAuthorFeed", actor=d, limit=100, filter="posts_no_replies")
    except Exception as e: print("err", e); continue
    for it in r.get("feed", []):
        if it.get("reason"): continue
        p = it["post"]; rec = p["record"]
        if p["author"]["did"] != d or rec.get("reply"): continue
        t = datetime.fromisoformat(rec["createdAt"].replace("Z", "+00:00"))
        if not (OLD <= t <= NEW): continue
        if not re.search(r"#gamedev\b", rec.get("text", ""), re.I): continue
        if p["uri"] in have: found += 1; hit_likes.append(p.get("likeCount", 0))
        else: missing += 1; miss_likes.append(p.get("likeCount", 0))
print("sampled authors", len(dids), "their #gamedev posts in window: in search", found, "missing", missing)
print("likes of missing", sorted(miss_likes)); print("likes of found (first 30)", sorted(hit_likes)[:30])
json.dump({"found": found, "missing": missing, "miss_likes": miss_likes, "hit_likes": hit_likes}, open("coverage.json", "w"), indent=1)