"""Project 109: what gets a #gamedev post seen on Bluesky?
Read-only public AppView (searchPosts sort=latest, getProfiles), 0.7 s between calls.
Keeps top-level posts created 24 h to 48 h before the fetch (likes have mostly settled). Writes posts.json.
No reply bodies, no personal data beyond what the public post shows; raw posts stay in this folder."""
import json, time, urllib.request, urllib.parse, re
from datetime import datetime, timezone, timedelta

UA = {"User-Agent": "Vonk (AI character) counting public #gamedev posts to publish open reach statistics, project 109"}
API = "https://api.bsky.app/xrpc/"
ME = ""  # your own did, to skip your posts
NOW = datetime.now(timezone.utc)
NEW, OLD = NOW - timedelta(hours=24), NOW - timedelta(hours=48)

def get(path, **q):
    url = API + path + "?" + urllib.parse.urlencode(q, doseq=True)
    err = None
    for _ in range(3):
        try:
            time.sleep(0.7)
            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"))

# The public AppView refuses search cursors (403), so search pages need a logged-in account.
# Use your OWN account with an app password: set BSKY_HANDLE and BSKY_APP_PASSWORD. Still read-only, same pacing.
import os
def _post(url, body):
    req = urllib.request.Request(url, json.dumps(body).encode(), {"Content-Type": "application/json", **UA})
    return json.load(urllib.request.urlopen(req, timeout=25))
class Session:
    def __init__(self):
        self.pds = os.environ.get("BSKY_PDS", "https://bsky.social")
        self.token = _post(self.pds + "/xrpc/com.atproto.server.createSession",
                           {"identifier": os.environ["BSKY_HANDLE"], "password": os.environ["BSKY_APP_PASSWORD"]})["accessJwt"]
    def get(self, nsid, params, proxy=True):
        req = urllib.request.Request(f"{self.pds}/xrpc/{nsid}?{params}", headers={**UA, "Authorization": "Bearer " + self.token,
                                     "atproto-proxy": "did:web:api.bsky.app#bsky_appview"})
        return json.load(urllib.request.urlopen(req, timeout=25))
S = Session()
def search(**q):
    err = None
    for _ in range(3):
        try:
            time.sleep(0.7)
            return S.get("app.bsky.feed.searchPosts", urllib.parse.urlencode(q), proxy=True)
        except Exception as e:
            err = e; time.sleep(2)
    return {"error": str(err)}

def media(e):
    t = e.get("$type", "")
    if "recordWithMedia" in t:
        e = e.get("media", {}); t = e.get("$type", "")
    if "video" in t: return "video"
    if "images" in t: return "images"
    if "external" in t:
        u = (e.get("external") or {}).get("uri", "").lower()
        if ".gif" in u or "tenor.com" in u or "giphy" in u: return "gif"
        return "link card"
    if "record" in t: return "quote only"
    return "text only"

posts, seen_pages, oldest = {}, 0, None
cursor = None
for _ in range(80):
    q = {"q": "#gamedev", "sort": "latest", "limit": 100}
    if cursor: q["cursor"] = cursor
    r = search(**q)
    if "error" in r: print("search error", r["error"]); break
    items = r.get("posts", []); seen_pages += 1
    for p in items:
        rec = p.get("record", {})
        t = ts(rec.get("createdAt", p.get("indexedAt")))
        oldest = t if oldest is None or t < oldest else oldest
        if rec.get("reply") or not (OLD <= t <= NEW) or p["author"]["did"] == ME: continue
        e = p.get("embed") or {}
        text = rec.get("text", "")
        tags = {f["features"][0].get("tag", "").lower() for f in rec.get("facets", []) if f["features"][0]["$type"].endswith("#tag")}
        tags |= {m.lower() for m in re.findall(r"#(\w+)", text)}
        n_img = len(e.get("images", []) or (e.get("media", {}) or {}).get("images", []) or [])
        links = [f["features"][0].get("uri", "") for f in rec.get("facets", []) if f["features"][0]["$type"].endswith("#link")]
        ext = ((e.get("external") or (e.get("media", {}) or {}).get("external") or {}).get("uri", ""))
        posts[p["uri"]] = {
            "uri": p["uri"], "url": f"https://bsky.app/profile/{p['author']['handle']}/post/{p['uri'].split('/')[-1]}",
            "did": p["author"]["did"], "handle": p["author"]["handle"], "time": rec.get("createdAt"),
            "text": text, "langs": rec.get("langs", []), "media": media(e), "n_img": n_img,
            "tags": sorted(tags), "links": links + ([ext] if ext else []),
            "bot": any(l.get("val") == "bot" for l in p["author"].get("labels", [])),
            "likes": p.get("likeCount", 0), "reposts": p.get("repostCount", 0),
            "replies": p.get("replyCount", 0), "quotes": p.get("quoteCount", 0)}
    cursor = r.get("cursor")
    if not items or not cursor or oldest < OLD: break
print("pages", seen_pages, "oldest seen", oldest, "posts in window", len(posts))

dids = sorted({p["did"] for p in posts.values()})
prof = {}
for i in range(0, len(dids), 25):
    r = get("app.bsky.actor.getProfiles", actors=dids[i:i+25])
    for a in r.get("profiles", []): prof[a["did"]] = {"followers": a.get("followersCount", 0), "posts": a.get("postsCount", 0)}
for p in posts.values():
    p["followers"] = prof.get(p["did"], {}).get("followers", 0)
    p["author_posts"] = prof.get(p["did"], {}).get("posts", 0)
json.dump({"fetched": NOW.isoformat(), "window": [OLD.isoformat(), NEW.isoformat()], "pages": seen_pages,
           "posts": list(posts.values())}, open("posts.json", "w", encoding="utf-8"), ensure_ascii=False, indent=1)
print("authors", len(dids), "profiles", len(prof))