Tier 2: input validation, rate limiting, and bug fixes

app.py

@@ -5,6 +5,7 @@ Deployed on Hugging Face Spaces
 """
 
 import os
+import re
 import gradio as gr
 from openai import OpenAI
 from geopy.geocoders import Nominatim
@@ -129,6 +130,7 @@ TRANSLATIONS = {
         "income_label": "Annual Household Income ($)",
         "household_label": "Number of People in Household",
         "snap_label": "Enrolled in SNAP or WIC?",
+        "insurance_label": "Do you have health insurance?",
         "yes": "Yes",
         "no": "No",
         "check_button": "Check Eligibility",
@@ -156,6 +158,7 @@ TRANSLATIONS = {
         "income_label": "Ingreso Anual del Hogar ($)",
         "household_label": "Número de Personas en el Hogar",
         "snap_label": "¿Inscrito en SNAP o WIC?",
+        "insurance_label": "¿Tiene seguro de salud?",
         "yes": "Sí",
         "no": "No",
         "check_button": "Verificar Elegibilidad",
@@ -183,6 +186,10 @@ def calculate_fpl_percentage(income, household_size):
 def find_nearby_hospitals(zip_code, lang):
     t = TRANSLATIONS[lang]
     try:
+        if not re.match(r'^06[0-9]{3}$', str(zip_code).strip()):
+            msg = "Please enter a valid Connecticut ZIP code (060xx–069xx)." if lang == "en" else "Ingrese un código postal válido de Connecticut (060xx–069xx)."
+            return msg, []
+
         location = geolocator.geocode(f"{zip_code}, Connecticut, USA")
         if not location:
             return "Invalid ZIP code" if lang == "en" else "Código postal inválido", []
@@ -213,11 +220,32 @@ def find_nearby_hospitals(zip_code, lang):
     except Exception as e:
         return f"Error: {str(e)}", []
 
-def determine_eligibility(hospital_name, income, household_size, has_snap_wic, lang):
+def determine_eligibility(hospital_name, income, household_size, has_snap_wic, lang, has_insurance=False):
     if "(" in hospital_name:
         hospital_name = hospital_name.split(" (")[0]
 
     t = TRANSLATIONS[lang]
+
+    if hospital_name not in HOSPITALS:
+        error_msg = "Hospital not found. Please go back and select a valid hospital." if lang == "en" else "Hospital no encontrado. Vuelva atrás y seleccione un hospital válido."
+        return {
+            "hospital": hospital_name,
+            "income": income,
+            "household_size": household_size,
+            "fpl_percentage": 0,
+            "has_snap_wic": has_snap_wic,
+            "has_insurance": has_insurance,
+            "pa_24_81_eligible": False,
+            "contact": "N/A",
+            "special_notes": "",
+            "asset_limit": None,
+            "fap_url": "",
+            "lang": lang,
+            "eligibility_status": error_msg,
+            "discount_level": "N/A",
+            "error": True
+        }
+
     hospital = HOSPITALS[hospital_name]
     fpl_percentage = calculate_fpl_percentage(income, household_size)
 
@@ -227,6 +255,7 @@ def determine_eligibility(hospital_name, income, household_size, has_snap_wic, l
         "household_size": household_size,
         "fpl_percentage": fpl_percentage,
         "has_snap_wic": has_snap_wic,
+        "has_insurance": has_insurance,
         "pa_24_81_eligible": False,
         "contact": hospital["contact"],
         "special_notes": hospital["special_notes"],
@@ -266,9 +295,10 @@ def generate_explanation_streaming(data):
     fpl_pct = data["fpl_percentage"]
 
     # Confidence-based language selection
+    hospital_threshold = HOSPITALS[data["hospital"]]["free_care_threshold"] if data["hospital"] in HOSPITALS else 200
     if is_pa_24_81:
         confidence = "statutory"  # Law-based, highest confidence
-    elif is_eligible and not has_asset_limit and fpl_pct < data.get("threshold", 200) * 0.8:
+    elif is_eligible and not has_asset_limit and fpl_pct < hospital_threshold * 0.8:
         confidence = "strong"  # Clear match, no complications
     elif is_eligible and has_asset_limit:
         confidence = "moderate"  # Additional requirements exist
@@ -333,10 +363,16 @@ Explica en 2 párrafos claros y cálidos. Máximo 150 palabras."""
 
 Explain in 2 clear, warm paragraphs. Maximum 150 words."""
 
+    insurance_note = ""
+    if data.get("has_insurance"):
+        insurance_note = "\nInsurance: Yes — Note that the hospital will bill insurance first; financial assistance applies to remaining balances."
+    else:
+        insurance_note = "\nInsurance: No"
+
     user_prompt = f"""Hospital: {data['hospital']}
 Income: ${data['income']:,.0f} ({data['fpl_percentage']}% FPL)
 Household: {data['household_size']} people
-SNAP/WIC: {'Yes' if data['has_snap_wic'] else 'No'}
+SNAP/WIC: {'Yes' if data['has_snap_wic'] else 'No'}{insurance_note}
 Status: {data['eligibility_status']}
 Discount: {data['discount_level']}
 Contact: {data['contact']}
@@ -345,7 +381,7 @@ PA 24-81 Eligible: {'Yes' if data['pa_24_81_eligible'] else 'No'}"""
 
     try:
         stream = client.chat.completions.create(
-            model="gpt-4o",
+            model="gpt-4o-mini",
             messages=[
                 {"role": "system", "content": system_msg},
                 {"role": "user", "content": user_prompt}
@@ -361,8 +397,9 @@ PA 24-81 Eligible: {'Yes' if data['pa_24_81_eligible'] else 'No'}"""
                 collected_text += chunk.choices[0].delta.content
                 yield collected_text
 
-    except Exception as e:
-        yield f"Error: {str(e)}"
+    except Exception:
+        fallback = "Detailed explanation temporarily unavailable. Please review the eligibility summary above and contact the hospital directly." if lang == "en" else "La explicación detallada no está disponible temporalmente. Revise el resumen de elegibilidad anterior y comuníquese directamente con el hospital."
+        yield fallback
 
 def format_results_streaming(data, lang):
     """
@@ -370,7 +407,14 @@ def format_results_streaming(data, lang):
     """
     t = TRANSLATIONS[lang]
 
-    # Start with header (instant)
+    # Disclaimer banner at top
+    disclaimer_banner = f"""> {t['disclaimer']}
+
+---
+
+"""
+
+    # Header
     header = f"""## {data['hospital']}
 
 ### {t['status']}: {data['eligibility_status']}
@@ -381,14 +425,7 @@ def format_results_streaming(data, lang):
 
 """
 
-    # Stream the explanation
-    partial_explanation = ""
-    for partial_text in generate_explanation_streaming(data):
-        partial_explanation = partial_text
-        full_output = header + partial_explanation
-        yield full_output
-
-    # Add footer after streaming completes
+    # Footer (always shown regardless of AI explanation success)
     footer = f"""
 
 ---
@@ -412,8 +449,16 @@ def format_results_streaming(data, lang):
 {t['disclaimer']}
 """
 
-    final_output = header + partial_explanation + footer
-    yield final_output
+    base = disclaimer_banner + header
+
+    # Stream the explanation
+    partial_explanation = ""
+    for partial_text in generate_explanation_streaming(data):
+        partial_explanation = partial_text
+        yield base + partial_explanation
+
+    # Always append footer after streaming completes
+    yield base + partial_explanation + footer
 
 # Custom CSS for CT.gov styling
 custom_css = """
@@ -550,9 +595,10 @@ def create_interface():
         with step3:
             selected_hospital_display = gr.Markdown()
 
-            income_input = gr.Number(label="Annual Household Income ($)", minimum=0, value=35000)
+            income_input = gr.Number(label="Annual Household Income ($)", minimum=1, value=35000)
             household_input = gr.Number(label="Household Size", minimum=1, value=3, precision=0)
             snap_radio = gr.Radio(choices=["Yes", "No"], label="Enrolled in SNAP or WIC?", value="No")
+            insurance_radio = gr.Radio(choices=["Yes", "No"], label="Do you have health insurance?", value="No")
 
             check_btn = gr.Button("Check Eligibility", variant="primary", size="lg")
             back_btn_c = gr.Button("← Back", size="sm")
@@ -607,6 +653,7 @@ def create_interface():
                 gr.update(visible=False),
                 gr.update(visible=True),
                 gr.update(visible=False),
+                gr.update(choices=[t['yes'], t['no']], value=t['no']),
                 gr.update(choices=[t['yes'], t['no']], value=t['no'])
             ]
 
@@ -619,13 +666,26 @@ def create_interface():
                 gr.update(visible=False),
                 gr.update(visible=True),
                 gr.update(visible=False),
+                gr.update(choices=[t['yes'], t['no']], value=t['no']),
                 gr.update(choices=[t['yes'], t['no']], value=t['no'])
             ]
 
-        def check_eligibility_wrapper(hospital, income, household, snap, lang):
+        def check_eligibility_wrapper(hospital, income, household, snap, insurance, lang):
             t = TRANSLATIONS[lang]
+
+            # Input validation
+            if income is None or income <= 0:
+                gr.Warning("Please enter a valid income greater than $0." if lang == "en" else "Ingrese un ingreso válido mayor que $0.")
+                yield [gr.update(), gr.update(), gr.update(), gr.update(), gr.update()]
+                return
+            if household is None or int(household) < 1:
+                gr.Warning("Household size must be at least 1." if lang == "en" else "El tamaño del hogar debe ser al menos 1.")
+                yield [gr.update(), gr.update(), gr.update(), gr.update(), gr.update()]
+                return
+
             has_snap = (snap == t['yes'])
-            data = determine_eligibility(hospital, income, int(household), has_snap, lang)
+            has_insurance = (insurance == t['yes'])
+            data = determine_eligibility(hospital, income, int(household), has_snap, lang, has_insurance=has_insurance)
 
             yield [
                 "",
@@ -666,19 +726,21 @@ def create_interface():
         continue_btn_a.click(
             fn=continue_from_search,
             inputs=[hospital_radio, lang_state],
-            outputs=[selected_hospital, selected_hospital_display, step1, step2a, step3, step4, snap_radio]
+            outputs=[selected_hospital, selected_hospital_display, step1, step2a, step3, step4, snap_radio, insurance_radio]
         )
 
         continue_btn_b.click(
             fn=continue_from_dropdown,
             inputs=[hospital_dropdown, lang_state],
-            outputs=[selected_hospital, selected_hospital_display, step1, step2b, step3, step4, snap_radio]
+            outputs=[selected_hospital, selected_hospital_display, step1, step2b, step3, step4, snap_radio, insurance_radio]
         )
 
         check_btn.click(
             fn=check_eligibility_wrapper,
-            inputs=[selected_hospital, income_input, household_input, snap_radio, lang_state],
-            outputs=[results_output, step1, step2a, step3, step4]
+            inputs=[selected_hospital, income_input, household_input, snap_radio, insurance_radio, lang_state],
+            outputs=[results_output, step1, step2a, step3, step4],
+            api_name=False,
+            concurrency_limit=5
         )
 
     return demo