added more sources

app.py

@@ -4,16 +4,36 @@ import pandas as pd
 
 st.set_page_config(layout="wide")
 
-st.header("HuggingFace 🤗 Posts leaderboard")
+col1, col2 = st.columns([2, 3])  # Adjust the width ratio as needed
 
-st.write(
-    """Data Source: https://huggingface.co/datasets/maxiw/hf-posts"""
-)
+sources = [
+    "https://huggingface.co/datasets/cfahlgren1/hub-stats",
+    "https://huggingface.co/datasets/maxiw/hf-posts",
+]
+
+with col1:
+    st.header("HuggingFace 🤗 Posts leaderboard")
 
-df = pd.read_json("hf://datasets/maxiw/hf-posts/posts.jsonl", lines=True)
+with col2:
+    selected_source = st.selectbox(
+        "Data Source:",
+        options=sources,
+        index=0,
+    )
 
-df["publishedAt"] = pd.to_datetime(df.publishedAt)
-# print(df.columns)
+if selected_source == sources[0]:
+    df = pd.read_parquet("hf://datasets/cfahlgren1/hub-stats/posts.parquet")
+    df["Name"] = df.fullname
+    df["username"] = df.name
+
+if selected_source == sources[1]:
+    df = pd.read_json("hf://datasets/maxiw/hf-posts/posts.jsonl", lines=True)
+
+    df["publishedAt"] = pd.to_datetime(df.publishedAt)
+    print(">>> ", df.columns)
+
+    df["Name"] = df.author.apply(lambda x: x["fullname"])
+    df["username"] = df.author.apply(lambda x: x["name"])
 
 # Define the metrics
 metrics = ["totalUniqueImpressions", "totalReactions", "numComments", "Num of posts"]
@@ -48,8 +68,6 @@ mask = df["publishedAt"].between(*date_range)
 df = df[mask]
 
 
-df["Name"] = df.author.apply(lambda x: x["fullname"])
-df["username"] = df.author.apply(lambda x: x["name"])
 df["totalReactions"] = df.reactions.apply(lambda x: sum([_["count"] for _ in x]))
 df["Num of posts"] = 1
 
@@ -68,6 +86,7 @@ data.index.name = "Rank"
 # Format metrics columns with commas
 data[metrics] = data[metrics].applymap(lambda x: f"{x:,}")
 
+
 def make_clickable(val):
     return f'<a target="_blank" href="https://huggingface.co/{val}">{val}</a>'
 
@@ -77,6 +96,3 @@ st.write(
     f"""<center>{df_styled.to_html(escape=False, index=False)}""",
     unsafe_allow_html=True,
 )
-
-
-# st.dataframe(data=df_styled, width=100_000)