stats

app.py

@@ -6,10 +6,13 @@ from datetime import date, datetime, timedelta
 from io import BytesIO
 
 import dotenv
+import matplotlib.pyplot as plt
+import seaborn as sns
 from datasets import load_dataset
 from dateutil.parser import parse
 from dateutil.tz import tzutc
 from fasthtml.common import *
+from fh_matplotlib import matplotlib2fasthtml
 from huggingface_hub import login, whoami
 
 dotenv.load_dotenv()
@@ -23,6 +26,7 @@ style = Style("""
                 .card a:hover { text-decoration: underline; }
             """)
 
+
 # delete data folder
 if os.path.exists("data"):
     try:
@@ -80,7 +84,7 @@ def generate_week_content(current_week):
     prev_week = weeks[week_index + 1] if week_index < len(weeks) - 1 else None
     next_week = weeks[week_index - 1] if week_index > 0 else None
 
-    nav_buttons = Group(
+    nav_buttons = Div(
         Button(
             "← Previous Week",
             hx_get=f"/week/{prev_week}" if prev_week else "#",
@@ -95,6 +99,7 @@ def generate_week_content(current_week):
             hx_swap="innerHTML",
             disabled=not next_week,
         ),
+        A("View Stats", href="/stats", cls="button"),
     )
 
     articles = week2articles[current_week]
@@ -115,6 +120,7 @@ def generate_week_content(current_week):
 
         if image:
             pil_image = image["image"]  # image[0]["image"]
+            pil_image.thumbnail((500, 500))
             img_byte_arr = BytesIO()
             pil_image.save(img_byte_arr, format="JPEG")
             img_byte_arr = img_byte_arr.getvalue()
@@ -138,7 +144,9 @@ def generate_week_content(current_week):
     week_end = current_week + timedelta(days=6)
     return Div(
         nav_buttons,
-        H3(f"Week of {current_week.strftime('%B %d')} - {week_end.strftime('%B %d, %Y')} ({len(articles)} articles)"),
+        Br(),
+        H5(f"{current_week.strftime('%B %d')} - {week_end.strftime('%B %d, %Y')} ({len(articles)} articles)"),
+        Br(),
         grid,
         nav_buttons,
         id="content",
@@ -159,6 +167,92 @@ def get(date: str):
         return Div(f"Error displaying articles: {str(e)}")
 
 
+@rt("/stats")
+async def get():
+    @matplotlib2fasthtml
+    def generate_chart():
+        end_date = max(weeks)
+        start_date = end_date - timedelta(weeks=11)
+
+        dates = []
+        counts = []
+        current_date = start_date
+        while current_date <= end_date:
+            count = len(week2articles[current_date])
+            date_str = current_date.strftime("%d-%B-%Y")
+            dates.append(date_str)
+            counts.append(count)
+            current_date += timedelta(weeks=1)
+
+        plt.figure(figsize=(12, 6))
+        sns.set_style("darkgrid")
+        # sns.set_palette("deep")
+
+        ax = sns.barplot(x=dates, y=counts)
+
+        plt.title("Papers per Week (Last 12 Weeks)", fontsize=16, fontweight="bold")
+        plt.xlabel("Week", fontsize=12)
+        plt.ylabel("Number of Papers", fontsize=12)
+
+        # Rotate and align the tick labels so they look better
+        plt.xticks(rotation=45, ha="right")
+
+        # Use a tight layout to prevent the labels from being cut off
+        plt.tight_layout()
+
+        # Add value labels on top of each bar
+        for i, v in enumerate(counts):
+            ax.text(i, v + 0.5, str(v), ha="center", va="bottom")
+
+        # Increase y-axis limit slightly to accommodate the value labels
+        plt.ylim(0, max(counts) * 1.1)
+
+    @matplotlib2fasthtml
+    def generate_contributions_chart():
+        article_df = article_ds.data.to_pandas()
+        added_by_df = article_df.groupby("added_by").size().reset_index(name="count")
+        added_by_df = added_by_df.sort_values("count", ascending=False)  # Ascending for bottom-to-top order
+
+        plt.figure(figsize=(12, 8))
+        sns.set_style("darkgrid")
+        sns.set_palette("deep")
+
+        ax = sns.barplot(x="count", y="added_by", data=added_by_df)
+
+        plt.title("Upload Counts", fontsize=16, fontweight="bold")
+        plt.xlabel("Number of Articles Added", fontsize=12)
+        plt.ylabel("User", fontsize=12)
+
+        # Add value labels to the end of each bar
+        for i, v in enumerate(added_by_df["count"]):
+            ax.text(v + 0.5, i, str(v), va="center")
+
+        # Adjust x-axis to make room for labels
+        plt.xlim(0, max(added_by_df["count"]) * 1.1)
+
+        plt.tight_layout()
+
+    # chart = Div(generate_chart(), id="chart")
+    bar_chart = Div(generate_chart(), id="bar-chart")
+    pie_chart = Div(generate_contributions_chart(), id="pie-chart")
+
+    # add contributions
+    article_df = article_ds.data.to_pandas()
+    added_by_df = article_df.groupby("added_by").size().reset_index(name="count")
+    added_by_df = added_by_df.sort_values("count", ascending=False)
+
+    return Titled(
+        "AnswerAI Zotero Stats",
+        H5("Papers per Week (Last 12 Weeks)"),
+        bar_chart,
+        Br(),
+        H5("Contributions by User"),
+        pie_chart,
+        Br(),
+        A("Back to Weekly View", href="/", cls="button"),
+    )
+
+
 # serve()
 
 if __name__ == "__main__":

requirements.txt

@@ -12,3 +12,5 @@ datasets
 PyMuPDF
 pillow
 tqdm
+fh-matplotlib
+seaborn