support hf hosting

README.md

@@ -1,3 +1,15 @@
+---
+title: Amazon Spending Visualizer
+emoji: 📦
+colorFrom: blue
+colorTo: indigo
+sdk: streamlit
+sdk_version: 1.56.0
+app_file: app.py
+pinned: false
+python_version: "3.12"
+---
+
 # Amazon Spending Visualizer
 
 Have you ever wondered how much you spend on Amazon? Wonder no more!
@@ -18,7 +30,9 @@ sanctioned way to get the data is the GDPR-style "Request My Data" export.
    - Amazon emails a download link within a few hours to a few days
 2. Drop the resulting `Your Orders.zip` into `data/`. The app reads
    `Your Amazon Orders/Order History.csv` straight out of the zip, so the zip is
-   always the source of truth — replace it to refresh the data.
+   always the source of truth — replace it to refresh the data. (You can also
+   skip this step and upload the zip through the in-app uploader at runtime,
+   which is how the hosted Hugging Face Space works.)
 3. Install deps:
    ```
    python3 -m venv .venv

requirements.txt

@@ -1,3 +1,3 @@
-streamlit>=1.32
-pandas>=2.0
-plotly>=5.18
+streamlit==1.56.0
+pandas==3.0.2
+plotly==6.7.0

src/data.py

@@ -1,14 +1,10 @@
+import io
 import zipfile
 
 import pandas as pd
 import streamlit as st
 
-from src.constants import (
-    EXCLUDED_WEBSITES,
-    ORDERS_CSV_ENTRY,
-    ORDERS_ZIP,
-    REFUNDS_CSV_ENTRY,
-)
+from src.constants import EXCLUDED_WEBSITES, ORDERS_CSV_ENTRY, REFUNDS_CSV_ENTRY
 
 # Tolerance (USD) for matching a refund row to a specific line item by amount.
 # Refund Details.csv has Order ID but no ASIN, so to flag the actual returned
@@ -40,17 +36,9 @@ def _match_refunds_to_lines(orders: pd.DataFrame, refunds: pd.DataFrame) -> set:
     return refunded_idx
 
 
-# `zip_mtime` is unused inside the function — it's an `@st.cache_data` key so
-# the cache invalidates when the zip is replaced (e.g. user uploads a fresh
-# Amazon export). Without it, the first load would be served forever.
 @st.cache_data
-def load_data(zip_mtime: float) -> tuple[pd.DataFrame, pd.DataFrame]:
-    del zip_mtime  # cache-key only; see comment above
-    if not ORDERS_ZIP.exists():
-        raise FileNotFoundError(
-            f"Missing {ORDERS_ZIP}. Place the Amazon 'Your Orders.zip' export in data/."
-        )
-    with zipfile.ZipFile(ORDERS_ZIP) as z:
+def load_data(zip_bytes: bytes) -> tuple[pd.DataFrame, pd.DataFrame]:
+    with zipfile.ZipFile(io.BytesIO(zip_bytes)) as z:
         with z.open(ORDERS_CSV_ENTRY) as f:
             orders = pd.read_csv(f)
         with z.open(REFUNDS_CSV_ENTRY) as f:

src/main.py

@@ -7,15 +7,25 @@ from src.data import load_data
 from src.plots import monthly_spend, top_products
 
 
+def _resolve_zip_bytes() -> bytes | None:
+    # Disk wins so local users keep the "drop once into data/" UX. Session-state
+    # is the upload path used on Hugging Face Spaces (and any other hosted
+    # deployment) where there's no persistent filesystem.
+    if ORDERS_ZIP.exists():
+        return ORDERS_ZIP.read_bytes()
+    return st.session_state.get("uploaded_zip")
+
+
 def run() -> None:
     st.set_page_config(page_title=APP_NAME, layout="wide")
     st.title(APP_NAME)
 
-    if not ORDERS_ZIP.exists():
+    zip_bytes = _resolve_zip_bytes()
+    if zip_bytes is None:
         onboarding.render()
         return
 
-    orders, refunds = load_data(ORDERS_ZIP.stat().st_mtime)
+    orders, refunds = load_data(zip_bytes)
 
     full_net = monthly_spend.compute_full_net(orders, refunds)
     sma = monthly_spend.compute_sma(full_net)

src/onboarding.py

@@ -1,18 +1,27 @@
 import streamlit as st
 
-from src.constants import DATA_DIR, ORDERS_ZIP
+from src.constants import ORDERS_ZIP
 
 AMAZON_REQUEST_URL = "https://amazon.com/gp/privacycentral/dsar/preview.html"
 
 
 def render() -> None:
-    st.info(
-        f"No Amazon export found yet. Drop `Your Orders.zip` into `{DATA_DIR}/` " "to get started."
-    )
+    st.info("No Amazon export loaded yet. Upload `Your Orders.zip` below to get started.")
 
     st.subheader("Already have the export?")
-    st.markdown("Place `Your Orders.zip` at this exact path, then click **Refresh** below:")
-    st.code(str(ORDERS_ZIP), language=None)
+    uploaded = st.file_uploader(
+        "Upload `Your Orders.zip`",
+        type=["zip"],
+        accept_multiple_files=False,
+    )
+    if uploaded is not None:
+        st.session_state["uploaded_zip"] = uploaded.getvalue()
+        st.rerun()
+
+    st.caption(
+        f"Running locally? You can also drop the zip at `{ORDERS_ZIP}` and reload "
+        "to skip the upload step on every session."
+    )
 
     st.subheader("Don't have it yet?")
     st.markdown(
@@ -23,9 +32,6 @@ request a data export:
 1. Open Amazon's data request page: [{AMAZON_REQUEST_URL}]({AMAZON_REQUEST_URL})
 2. Select **Your Orders** and submit the request
 3. Wait for Amazon to email a download link (typically a few hours to a few days)
-4. Drop the resulting `Your Orders.zip` into `data/` and click **Refresh**
+4. Upload the resulting `Your Orders.zip` above
 """
     )
-
-    if st.button("Refresh", type="primary"):
-        st.rerun()