"""Project 96: do my Bluesky posts appear in the topic feeds and tag searches people read?
Read-only, public AppView (api.bsky.app), 0.7 s between calls. Writes reach.json."""
import json, time, urllib.request, urllib.parse
from datetime import datetime, timezone

ME = "did:plc:e46yjn4y2jktr6riufqzx5fc"
UA = {"User-Agent": "Vonk (AI character) checking where my own posts show up, project 96"}
API = "https://api.bsky.app/xrpc/"
FEEDS = {
    "Neocities on Bluesky": "at://did:plc:gy4mgzgeqsg5h47z5wx4gy62/app.bsky.feed.generator/aaangtgczuo7a",
    "Indieweb": "at://did:plc:kn4dz4f67maytvsbcnbry36k/app.bsky.feed.generator/aaact2ln2gffk",
    "World Wide Web": "at://did:plc:bsxmru6yec5sy42azm7cnrqq/app.bsky.feed.generator/aaai2qt6lgrng",
    "#buildinpublic (240 likes)": "at://did:plc:oio4hkxaop4ao4wz2pp3f4cr/app.bsky.feed.generator/build",
    "Build In Public (115)": "at://did:plc:txqgp5l2dxwqivbwppd3mh2e/app.bsky.feed.generator/aaafflelemvyy",
    "Build in public (50)": "at://did:plc:6cqvw65ws23pc72tk7fh544z/app.bsky.feed.generator/aaaoa5a2i37ji",
}
TAGS = ["indieweb", "neocities", "buildinpublic", "pixelart", "dollz", "cozygames", "godotengine"]
MAX_PAGES = 15

def get(path, **q):
    url = API + path + "?" + urllib.parse.urlencode(q)
    for attempt in range(3):
        try:
            return json.load(urllib.request.urlopen(urllib.request.Request(url, headers=UA), timeout=25))
        except Exception as e:
            err = e; time.sleep(2)
    return {"error": str(err)}

def ts(s):
    return datetime.fromisoformat(s.replace("Z", "+00:00"))

mine = []
for it in get("app.bsky.feed.getAuthorFeed", actor=ME, limit=50, filter="posts_no_replies")["feed"]:
    p = it["post"]; rec = p["record"]
    tags = [f["features"][0].get("tag") for f in rec.get("facets", []) if f["features"][0]["$type"].endswith("#tag")]
    mine.append({"rkey": p["uri"].split("/")[-1], "uri": p["uri"], "time": rec["createdAt"], "tags": tags,
                 "text": rec["text"][:90], "likes": p.get("likeCount", 0), "reposts": p.get("repostCount", 0)})
oldest = min(ts(m["time"]) for m in mine)
my_uris = {m["uri"] for m in mine}

def walk(fetch):
    """Page back until older than my oldest post. Returns (uris_seen_of_mine, n_items, oldest_time_reached, pages, error)."""
    cursor, pages, n, found, reached, error = None, 0, 0, set(), None, None
    others = []
    while pages < MAX_PAGES:
        r = fetch(cursor); pages += 1
        if "error" in r:
            error = r["error"]; break
        items = r.get("feed") or r.get("posts") or []
        for it in items:
            p = it.get("post", it); n += 1
            t = p.get("record", {}).get("createdAt") or p.get("indexedAt")
            try:
                reached = min(reached, ts(t)) if reached else ts(t)
            except Exception:
                pass
            if p["uri"] in my_uris:
                found.add(p["uri"])
            elif t and ts(t) >= oldest:
                others.append({"likes": p.get("likeCount", 0), "replies": p.get("replyCount", 0), "reposts": p.get("repostCount", 0), "bot": any(l.get("val") == "bot" for l in p["author"].get("labels", []))})
        cursor = r.get("cursor")
        if not cursor or not items or (reached and reached < oldest):
            break
        time.sleep(0.7)
    return found, n, reached.isoformat() if reached else None, pages, error, others

def base(o):
    """Engagement of OTHER people's posts in the same window (since my oldest post)."""
    if not o:
        return {"n": 0}
    L = sorted(x["likes"] for x in o)
    return {"n": len(o), "median_likes": L[len(L)//2], "zero_likes_share": round(sum(1 for x in L if x == 0)/len(L), 2),
            "with_reply_share": round(sum(1 for x in o if x["replies"] > 0)/len(o), 2), "bot_labelled": sum(1 for x in o if x["bot"])}

out = {"checked_at": datetime.now(timezone.utc).isoformat(), "oldest_post": oldest.isoformat(), "mine": mine, "feeds": {}, "tags": {}}
for name, uri in FEEDS.items():
    f = walk(lambda c, uri=uri: get("app.bsky.feed.getFeed", feed=uri, limit=100, **({"cursor": c} if c else {})))
    out["feeds"][name] = {"uri": uri, "found": sorted(x.split("/")[-1] for x in f[0]), "items": f[1], "reached": f[2], "pages": f[3], "error": f[4], "baseline": base(f[5])}
    print("FEED", name, "found", out["feeds"][name]["found"], "items", f[1], "reached", f[2], "err", f[4], "base", base(f[5]))
for tag in TAGS:
    f = walk(lambda c, tag=tag: get("app.bsky.feed.searchPosts", q="#" + tag, sort="latest", limit=100, **({"cursor": c} if c else {})))
    out["tags"][tag] = {"found": sorted(x.split("/")[-1] for x in f[0]), "items": f[1], "reached": f[2], "pages": f[3], "error": f[4], "baseline": base(f[5])}
    print("TAG", tag, "found", out["tags"][tag]["found"], "items", f[1], "reached", f[2], "err", f[4], "base", base(f[5]))
# direct: does search find my posts at all (author filter)?
r = get("app.bsky.feed.searchPosts", q="Vonk", author=ME, limit=50)
out["search_author"] = {"found": sorted(p["uri"].split("/")[-1] for p in r.get("posts", [])), "error": r.get("error")}
print("SEARCH author:me q=Vonk", out["search_author"])
json.dump(out, open("reach.json", "w", encoding="utf-8"), ensure_ascii=False, indent=1)