Replace broken submodule with a download scheme

.gitmodules

@@ -1,3 +0,0 @@
-[submodule "faiss"]
-	path = faiss
-	url = [email protected]:colonelwatch/abstracts-index-faiss

app.py

@@ -10,9 +10,10 @@ from pathlib import Path
 from sys import stderr
 from typing import TypedDict, Self, Any, Callable
 
-from datasets import Dataset, disable_caching
+from datasets import Dataset
 from datasets.search import FaissIndex
 import faiss
+from huggingface_hub import snapshot_download
 import gradio as gr
 import requests
 from sentence_transformers import SentenceTransformer
@@ -242,12 +243,18 @@ def main():
     prompt_name = get_env_var("PROMPT_NAME")
     trust_remote_code = get_env_var("TRUST_REMOTE_CODE", bool, default=False)
     fp16 = get_env_var("FP16", bool, default=False)
-    dir = get_env_var("DIR", Path, default=Path("faiss/index"))
+    dir = get_env_var("DIR", Path)
+    repo = get_env_var("REPO", str)
     search_time_s = get_env_var("SEARCH_TIME_S", float, default=1)
     k = get_env_var("K", int, default=20)  # TODO: can't go higher than 20 yet
     mailto = get_env_var("MAILTO", str, None)
 
-    disable_caching()  # disable caching in the datasets library
+    if dir is None:  # acquire the index if it's not local
+        if repo is None:
+            repo = "colonelwatch/abstracts-faiss"
+        dir = Path(snapshot_download(repo, repo_type="dataset")) / "index"
+    elif repo is not None:
+        print('warning: used "REPO" and also "DIR", ignoring "REPO"...', file=stderr)
 
     normalize, model = get_model(model_name, dir, trust_remote_code)
     index = get_index(dir, search_time_s)

faiss

@@ -1 +0,0 @@
-Subproject commit 652454ca78e307c0d9262da278619570f4ad120d