lspcloud/prolific-preferences-personalized
0
1"""2Entry point for the user study Streamlit app.3 4Run from repo root:5 streamlit run src/app.py6 streamlit run src/app.py -- --debug (sets DEBUG_MODE via env)7 8HuggingFace Space secrets required:9 HF_TOKEN — read/write HuggingFace token10 GH_TOKEN — GitHub PAT (ehejin account) for the private lsp submodule11 TINKER_API_KEY — Tinker inference API key12 DEBUG_MODE — "true" to skip all validation (optional)13"""14import os15import sys16import json17import subprocess18from pathlib import Path19 20# ---------------------------------------------------------------------------21# 1. Initialise lsp git submodule before any lsp imports22# On a cold HF Space start the submodule directory exists but is empty;23# GH_TOKEN lets us authenticate against the private GitHub repo.24# ---------------------------------------------------------------------------25_BASE = Path(__file__).resolve().parent.parent26_LSP_PATH = _BASE / "lsp"27 28 29def _init_submodule() -> None:30 prompts_exist = (_LSP_PATH / "src" / "prompts").exists()31 if not prompts_exist:32 token = os.getenv("GH_TOKEN", "")33 if not token:34 raise RuntimeError("GH_TOKEN secret is not set.")35 36 import shutil37 import tarfile38 import urllib.request39 import time as _time40 41 # Clean any stale state42 if _LSP_PATH.exists():43 shutil.rmtree(str(_LSP_PATH), ignore_errors=True)44 git_modules = _BASE / ".git" / "modules" / "lsp"45 if git_modules.exists():46 shutil.rmtree(str(git_modules), ignore_errors=True)47 48 # GitHub serves a tarball of any branch/tag/SHA at this URL.49 # Pinned to a specific commit SHA so future lsp changes don't break us.50 branch = "74582acd911f81309ba8b22cef9286c2887dda18"51 tarball_url = f"https://api.github.com/repos/batu-el/lsp/tarball/{branch}"52 tmp_tar = Path("/tmp/lsp.tar.gz")53 tmp_extract = Path("/tmp/lsp_extract")54 55 for attempt in range(1, 4):56 print(f"[SUBMODULE] tarball download attempt {attempt}/3 ...")57 try:58 req = urllib.request.Request(59 tarball_url,60 headers={61 "Authorization": f"Bearer {token}",62 "Accept": "application/vnd.github+json",63 "User-Agent": "prolific-preferences",64 },65 )66 with urllib.request.urlopen(req, timeout=60) as resp:67 tmp_tar.write_bytes(resp.read())68 print(f"[SUBMODULE] downloaded {tmp_tar.stat().st_size} bytes")69 70 # Extract71 if tmp_extract.exists():72 shutil.rmtree(str(tmp_extract), ignore_errors=True)73 tmp_extract.mkdir(parents=True)74 with tarfile.open(str(tmp_tar)) as tar:75 tar.extractall(str(tmp_extract))76 77 # GitHub tarballs have a top-level dir like batu-el-lsp-abc123/78 subdirs = [d for d in tmp_extract.iterdir() if d.is_dir()]79 if not subdirs:80 raise RuntimeError("tarball had no top-level directory")81 top = subdirs[0]82 83 # Verify the prompts dir is present84 if not (top / "src" / "prompts").exists():85 raise RuntimeError(f"src/prompts not found in extracted tarball at {top}")86 87 # Move extracted contents to /app/lsp88 shutil.copytree(str(top), str(_LSP_PATH))89 tmp_tar.unlink(missing_ok=True)90 shutil.rmtree(str(tmp_extract), ignore_errors=True)91 92 print("[SUBMODULE] ready.")93 break94 except Exception as e:95 msg = str(e).replace(token, "***") if token else str(e)96 print(f"[SUBMODULE] attempt {attempt} failed: {msg}")97 _time.sleep(3)98 else:99 raise RuntimeError(f"Failed to download lsp tarball after 3 attempts.")100 101 lsp_src = str(_LSP_PATH / "src")102 if lsp_src not in sys.path:103 sys.path.insert(0, lsp_src)104 if str(_BASE) not in sys.path:105 sys.path.insert(0, str(_BASE))106 107 108_init_submodule()109 110# Wipe stale local state ONLY on the first container load (not on every Streamlit rerun).111# We use a marker file — once created, subsequent imports skip the wipe.112# Completions stay durable in HF; we re-scan HF fresh after wipe.113_data_root = _BASE / "data"114_data_root.mkdir(parents=True, exist_ok=True)115_wipe_marker = _data_root / ".startup_wiped"116if not _wipe_marker.exists():117 for pattern in ("reservations.json", "local_completions_*.json", "completion_cache_*.json"):118 for f in _data_root.glob(pattern):119 try:120 f.unlink()121 print(f"[STARTUP] Wiped stale file: {f.name}")122 except Exception as e:123 print(f"[STARTUP] Could not wipe {f.name}: {e}")124 _wipe_marker.touch()125 print("[STARTUP] Marked container as wiped")126 127# ---------------------------------------------------------------------------128# 2. App imports (only after submodule is initialised)129# ---------------------------------------------------------------------------130import streamlit as st131 132from src.config import load_config133from src.data import ensure_datasets, init_state134from src.ui.components import inject_css135from src.ui.screens_shared import (136 screen_background,137 screen_chat,138 screen_demographics,139 screen_done,140 screen_post_rating,141 screen_reflection,142 screen_welcome,143)144from src.ui.screens_likelihood import screen_item_intro145from src.ui.screens_preference import screen_pair_intro146 147 148# ---------------------------------------------------------------------------149# 3. Admin dashboard — visit ?admin=1150# ---------------------------------------------------------------------------151def _screen_admin(cfg: dict) -> None:152 """Coverage dashboard — visit ?admin=1 to see this."""153 from src.data import (154 _get_accepted_counts, _load_pool, _pool_path,155 _load_reservations, _save_reservations,156 _expire_reservations, _release_returned_reservations,157 _reservation_lock_path,158 )159 from filelock import FileLock160 161 st.markdown("## 📊 Study Coverage Dashboard")162 st.caption(163 f"Study type: `{cfg['study_type']}` · "164 f"Seed: `{cfg['pair_selection_seed']}` · "165 f"Output repo: `{cfg['output_dataset_repo']}`"166 )167 168 if st.button("🔄 Refresh", type="primary"):169 # Invalidate caches so we re-scan HF and re-poll Prolific170 from src.data import _data_dir171 for f in _data_dir(cfg).glob("completion_cache*"):172 f.unlink()173 prolific_cache = _data_dir(cfg) / "prolific_returned_cache.json"174 if prolific_cache.exists():175 prolific_cache.unlink()176 st.rerun()177 178 # Release expired + returned/timed-out reservations before displaying179 lock = FileLock(str(_reservation_lock_path(cfg)), timeout=10)180 with lock:181 reservations = _load_reservations(cfg)182 _expire_reservations(reservations)183 _release_returned_reservations(reservations, cfg)184 _save_reservations(reservations, cfg)185 186 for cat_cfg in cfg["categories"]:187 cat = cat_cfg["name"]188 pool = _load_pool(str(_pool_path(cat, cfg)))189 total = len(pool)190 191 counts = _get_accepted_counts(cat, cfg)192 193 covered = sum(1 for v in counts.values() if v >= 1)194 reserved_uncovered = sum(195 1 for k in reservations196 if counts.get(k, 0) == 0197 )198 truly_uncovered = total - covered - reserved_uncovered199 200 st.markdown(f"### {cat.capitalize()}")201 col1, col2, col3, col4 = st.columns(4)202 col1.metric("Total items", total)203 col2.metric("Covered ✅", covered)204 col3.metric("In progress 🔄", reserved_uncovered,205 help="Reserved by active Prolific participants")206 col4.metric("Still needed ⚠️", truly_uncovered,207 delta=f"-{truly_uncovered}" if truly_uncovered > 0 else None,208 delta_color="inverse")209 210 if truly_uncovered == 0 and reserved_uncovered == 0:211 st.success(f"✅ All {total} items covered!")212 elif truly_uncovered == 0:213 st.info(f"🔄 {reserved_uncovered} item(s) in progress.")214 else:215 st.warning(216 f"⚠️ {truly_uncovered} item(s) still need a participant. "217 f"Send more Prolific slots."218 )219 220 st.markdown("---")221 222 223# ---------------------------------------------------------------------------224# 4. Main225# ---------------------------------------------------------------------------226def main() -> None:227 cfg = load_config()228 229 st.set_page_config(230 page_title="Product Study",231 page_icon="🛒",232 layout="centered",233 )234 inject_css()235 236 # Admin dashboard — visit ?admin=1237 try:238 params = st.query_params239 except Exception:240 params = {}241 if params.get("admin") == "1":242 ensure_datasets(cfg)243 _screen_admin(cfg)244 return245 246 if "study_state" not in st.session_state:247 ensure_datasets(cfg)248 st.session_state.study_state = init_state(cfg)249 250 s = st.session_state.study_state251 screen = s.get("screen", "welcome")252 253 dispatch = {254 "welcome": lambda: screen_welcome(s, cfg),255 "demographics": lambda: screen_demographics(s, cfg),256 "background": lambda: screen_background(s, cfg),257 "item_intro": lambda: (258 screen_pair_intro(s, cfg)259 if cfg["study_type"] == "preference"260 else screen_item_intro(s, cfg)261 ),262 "chat": lambda: screen_chat(s, cfg),263 "post_rating": lambda: screen_post_rating(s, cfg),264 "reflection": lambda: screen_reflection(s, cfg),265 "done": lambda: screen_done(s, cfg),266 }267 268 handler = dispatch.get(screen)269 if handler:270 handler()271 else:272 st.error(f"Unknown screen: {screen!r}")273 274 275if __name__ == "__main__":276 main()