Show ungrouped repos better

Author: zarybnicky

fetch_known_repos.py

@@ -10,10 +10,12 @@
 def fetch_known_repos():
     load_dotenv(override=False)
 
-    conn = duckdb.connect("git.duckdb")
+    conn = duckdb.connect(":memory:")
     with open('schema.sql') as f:
         conn.sql(f.read())
 
+    conn.sql("COPY providers FROM 'data/providers.csv' WITH (HEADER 1, DELIMITER E'\\t')")
+    conn.sql("COPY repos FROM 'data/repos*.csv' WITH (HEADER 1, DELIMITER E'\\t')")
     repos = conn.sql("SELECT repos.repo, root || repo as root, origin || repo as origin FROM repos JOIN providers ON provider=providers.name").df()
 
     for _, repo in track(list(repos.iterrows())):

pages/05_Projects-Groups.py

@@ -52,10 +52,10 @@ def main():
     with colored as (
       select repos.*, unnest(list_distinct({colors})) as color from repos where {filters.repo_filter}
     )
-    SELECT repo FROM repos
-    WHERE {filters.repo_filter} AND id not in (SELECT id FROM colored)
-    group by repo ORDER BY repo DESC LIMIT 500
-    """, params=filters.repo_params).df()
+    SELECT repo, count(*) as count FROM repos join git_commits on repo_id=repos.id
+    WHERE {filters.commit_filter} AND id not in (SELECT id FROM colored)
+    group by repo ORDER BY count(*) DESC LIMIT 500
+    """, params=filters.commit_params).df()
     st.markdown(f"### Uncategorized")
     st.dataframe(last_commits, hide_index=True, use_container_width=True)