Update app.py

app.py

@@ -1,31 +1,10 @@
 import gradio as gr
 from huggingface_hub import hf_hub_download
-import json_stream as json
 import gzip
 import urllib
-from collections import defaultdict
-import gc
-import sys
-
-usernames = defaultdict(dict)
-
-versions = ["v1.0", "v1.1", "v1.2", "v2.0", "v2.0.1", "v2.1.0"]
-versions = [sys.intern(version) for version in versions]
-
-for version in versions:
-    print(f"Loading {version}")
-    branch = version if version != "v1.0" else "main"
-    filepath = hf_hub_download(repo_id="bigcode/the-stack-username-to-repo", filename="username_to_repo.json.gz", repo_type="dataset", revision=branch)
-    with gzip.open(filepath, 'r') as f:
-        data = json.load(f)
-        for username, repos in data.items():
-            username = username.lower()
-            for repo in repos:
-                if repo not in usernames[username]:
-                    usernames[username][repo] = []
-                usernames[username][repo].append(version)
-        del data
-    gc.collect()
+import duckdb
+
+db = duckdb.connect("repos.duckdb")
 
 text = """\
 ![](https://huggingface.co/spaces/lvwerra/in-the-stack-gr/resolve/main/banner.png)
@@ -80,9 +59,8 @@ def issue_url(username, repos):
 def check_username(username, version):
     username = username.lower()
     output_md = ""
-    repos = []
-    if username in usernames:
-        repos = [repo for repo, versions in usernames[username].items() if version in versions]
+    repos = db.sql(f"SELECT repo FROM repos WHERE user='{username}' AND version='{version}' ORDER BY repo").collectall()
+    repos = [repo[0] for repo in repos]
 
     if repos:
         repo_word = "repository" if len(repos)==1 else "repositories"