Update src/streamlit_app.py

src/streamlit_app.py

@@ -13,13 +13,12 @@ from report import ReportBuilder
 
 from transformers import AutoTokenizer
 
-# Use a writable cache dir to avoid permission issues on Hugging Face Spaces
-HF_CACHE_DIR = "./hf_cache"
-os.environ["HF_HOME"] = HF_CACHE_DIR
-os.environ["TRANSFORMERS_CACHE"] = HF_CACHE_DIR
-
-# Ensure cache directory exists
-os.makedirs(HF_CACHE_DIR, exist_ok=True)
+# Temp-safe paths
+input_path = "/tmp/input.csv"
+output_path = "/tmp/output.csv"
+report_path = "/tmp/final_report.pdf"
+charts_dir = "/tmp/charts"
+os.makedirs(charts_dir, exist_ok=True)
 
 # Authenticate and load tokenizer to check access
 hf_token = os.environ.get("HUGGINGFACE_TOKEN")
@@ -32,17 +31,19 @@ tokenizer = AutoTokenizer.from_pretrained(
 st.set_page_config(page_title="Smart Data Cleaning Agent", layout="wide")
 st.title("🧠 Smart Data Cleaning Agent")
 
-os.makedirs("charts", exist_ok=True)
-
 uploaded_file = st.file_uploader("📂 Upload a CSV file", type=["csv"])
 
 if uploaded_file:
-    df = pd.read_csv(uploaded_file)
+    # Save file to /tmp/ for processing
+    with open(input_path, "wb") as f:
+        f.write(uploaded_file.read())
+
+    df = pd.read_csv(input_path)
     st.subheader("🔍 Original Data Preview")
     st.dataframe(df.head())
 
     with st.spinner("📊 Analyzing CSV..."):
-        analysis = analyze_csv(uploaded_file)
+        analysis = analyze_csv(input_path)
 
     with st.spinner("🧼 Generating Cleaning Plan..."):
         cleaning_plan, cleaning_summary = generate_cleaning_plan(analysis)
@@ -53,6 +54,7 @@ if uploaded_file:
 
     with st.spinner("🧪 Applying cleaning..."):
         cleaned_df = execute_plan(df.copy(), cleaning_plan)
+        cleaned_df.to_csv(output_path, index=False)
         st.subheader("🧼 Cleaned Data Preview")
         st.dataframe(cleaned_df.head())
         st.download_button("⬇️ Download Cleaned CSV", cleaned_df.to_csv(index=False), file_name="cleaned.csv")
@@ -68,23 +70,26 @@ if uploaded_file:
             st.markdown(f"#### {vis['title']}")
             st.markdown(vis['description'])
             try:
-                exec(vis["code"], {"df": cleaned_df, "plt": plt, "sns": sns, "os": os})
+                safe_code = vis["code"].replace("charts/", f"{charts_dir}/")
+                exec(safe_code, {"df": cleaned_df, "plt": plt, "sns": sns, "os": os})
                 st.pyplot(plt.gcf())
                 plt.clf()
             except Exception as e:
                 st.error(f"❌ Failed to render: {e}")
 
     if st.button("📝 Generate PDF Report"):
-        report = ReportBuilder("report.pdf")
+        report = ReportBuilder(output_path=report_path)
         report.add_title("📊 Smart Data Cleaning Report")
-        report.add_text("Cleaning Summary", cleaning_summary)
-        report.add_text("EDA Insights", insights)
+        report.add_section("Cleaning Summary", cleaning_summary)
+        report.add_section("EDA Insights", insights)
 
         for vis in visuals:
-            if "savefig('" in vis['code']:
-                path = vis['code'].split("savefig('")[-1].split("')")[0]
-                report.add_image(path, vis['description'])
+            if "savefig('" in vis["code"]:
+                path = vis["code"].split("savefig('")[-1].split("')")[0]
+                if not path.startswith("/"):
+                    path = os.path.join(charts_dir, os.path.basename(path))
+                report.add_plot(path, vis["description"])
 
         report.save()
-        with open("report.pdf", "rb") as f:
+        with open(report_path, "rb") as f:
             st.download_button("⬇️ Download PDF Report", f, file_name="smart_data_report.pdf")