Upload app.py

app.py

@@ -0,0 +1,247 @@
+import streamlit as st
+import pandas as pd
+import json
+import time # For simulating delays
+
+# --- Configuration and Initial Setup ---
+st.set_page_config(layout="wide", page_title="Dummy URS Doc Processor UI Test")
+st.title("Dummy URS Doc Processor - UI Test Version")
+
+# Initialize session states
+if "show_step2" not in st.session_state:
+    st.session_state.show_step2 = False
+if "show_step3" not in st.session_state:
+    st.session_state.show_step3 = False
+if "protocol_generated" not in st.session_state:
+    st.session_state.protocol_generated = False
+if "extracted_data" not in st.session_state:
+    st.session_state.extracted_data = None
+if "dataframe" not in st.session_state:
+    st.session_state.dataframe = None
+if "processing" not in st.session_state:
+    st.session_state.processing = False
+if "generated_pdf" not in st.session_state:
+    st.session_state.generated_pdf = None
+if "generated_pdf_name" not in st.session_state:
+    st.session_state.generated_pdf_name = None
+
+# Dummy LLM setup (no actual LLM calls)
+@st.cache_resource
+def get_llms():
+    return None, None # No LLMs needed for UI test
+
+# --------------------- Dummy Backend Functions ---------------------
+
+def dummy_run_rag_pipeline(pdf_file, template_file):
+    """
+    Simulates the RAG pipeline for UI testing.
+    Returns dummy extracted data.
+    """
+    st.toast("Simulating RAG Pipeline...")
+    time.sleep(2) # Simulate processing time
+
+    # Dummy extracted data
+    dummy_data = {
+        "ProjectName": "Dummy Project UI",
+        "DocumentNumber": "DUMMY-001",
+        "SystemName": "Test System Beta",
+        "Version": "1.0.0",
+        "Author": "UI Tester",
+        "Date": "2025-07-21",
+        "FunctionalRequirement_1": "The system shall display data responsively on mobile devices.",
+        "FunctionalRequirement_2": "The system shall allow editing of extracted values.",
+        "Utility#1": "Power Supply (240V)",
+        "Utility#2": "Ethernet Connection",
+        "SoftwareRequirements": "Python 3.9+, Streamlit 1.x",
+        "HardwareRequirements": "Any modern smartphone or tablet"
+    }
+    st.toast("RAG Pipeline simulation complete!")
+    return dummy_data
+
+def dummy_json_to_dataframe(json_data):
+    """
+    Converts dummy JSON data to DataFrame for editing.
+    """
+    if not json_data:
+        return pd.DataFrame(columns=['Placeholder', 'Value'])
+
+    data_list = []
+    for key, value in json_data.items():
+        data_list.append((str(key), str(value) if value is not None else ""))
+    
+    df = pd.DataFrame(data_list, columns=['Placeholder', 'Value'])
+    return df
+
+def create_dummy_pdf():
+    """Generates a dummy PDF byte stream for download."""
+    from reportlab.lib.pagesizes import letter
+    from reportlab.pdfgen import canvas
+    from io import BytesIO
+
+    buffer = BytesIO()
+    p = canvas.Canvas(buffer, pagesize=letter)
+    p.drawString(100, 750, "This is a Dummy PDF for UI Testing!")
+    p.drawString(100, 730, "It contains no real data from your URS or protocol.")
+    p.drawString(100, 710, "Generated on: " + time.ctime())
+    p.showPage()
+    p.save()
+    buffer.seek(0)
+    return buffer.getvalue()
+
+# --- Status & Progress Tracking ---
+# This section remains largely the same to simulate your original progress logic
+progress = 0
+status_text = "🟦 Waiting for input files"
+
+if st.session_state.processing:
+    progress = 20
+    status_text = "Extracting values from uploaded files (simulated)..."
+
+if st.session_state.show_step2:
+    progress = 40
+    status_text = "Input form is being generated (simulated)..."
+
+if st.session_state.show_step3:
+    progress = 60
+    status_text = "Saving changes to input form (simulated)..."
+
+if st.session_state.protocol_generated:
+    progress = 100
+    status_text = "Final protocol document is being generated (simulated)..."
+
+st.markdown(f"### Status: {status_text}")
+st.progress(progress / 100.0)
+
+# --- Layout: Horizontal columns for each step ---
+st.write("---") # Horizontal line for separation
+col1, col2, col3 = st.columns(3)
+
+# --- STEP 1 ---
+with col1:
+    st.markdown("### 🔹 Step 1")
+    st.write("📄 **Upload input URS (PDF)**")
+    urs_pdf = st.file_uploader("Upload PDF", type=["pdf"], key="pdf_upload_step1")
+
+    st.write("📊 **Upload template (XLSX)**")
+    template_xlsx = st.file_uploader("Upload XLSX", type=["xlsx"], key="xlsx_upload_step1")
+
+    if st.button("➡️ Generate Input Form", key="generate_form_button"):
+        if urs_pdf and template_xlsx:
+            st.session_state.processing = True
+            st.rerun()
+        else:
+            st.warning("Please upload both PDF and XLSX files for the simulation.")
+
+# Process RAG pipeline if processing is True
+if st.session_state.processing:
+    with st.spinner("🔄 Running RAG Pipeline (Simulated)..."):
+        # Call the dummy RAG pipeline
+        result = dummy_run_rag_pipeline(urs_pdf, template_xlsx)
+        
+        if result:
+            st.session_state.extracted_data = result
+            st.session_state.dataframe = dummy_json_to_dataframe(result)
+            st.session_state.show_step2 = True
+            st.session_state.processing = False
+            st.success("✅ RAG Pipeline simulation completed!")
+            st.rerun()
+        else:
+            st.error("❌ Failed to simulate document processing. Try again.")
+            st.session_state.processing = False
+            st.rerun()
+
+st.write("---") # Horizontal line for separation
+
+# --- STEP 2 ---
+with col2:
+    if st.session_state.show_step2:
+        st.markdown("### 🔹 Step 2")
+        
+        if st.session_state.dataframe is not None:
+            st.markdown("✏️ **Edit Input Form (DataFrame)**")
+            
+            # Display editable dataframe
+            edited_df = st.data_editor(
+                st.session_state.dataframe,
+                num_rows="dynamic",
+                key="data_editor",
+                height=400 # Fixed height for better mobile consistency
+            )
+            
+            # Update session state with edited data (for dummy purposes)
+            st.session_state.dataframe = edited_df
+            
+            if st.button("💾 Save Changes", key="save_changes_button"):
+                # Simulate saving changes
+                st.session_state.extracted_data = {
+                    row['Placeholder']: row['Value'] 
+                    for _, row in edited_df.iterrows() 
+                    if pd.notna(row['Value']) and row['Value'] != ''
+                }
+                st.success("✅ Changes saved (simulated)!")
+                st.session_state.show_step3 = True
+                st.rerun()
+    else:
+        st.markdown("### 🔹 Step 2")
+        st.info("Complete Step 1 to unlock.")
+
+st.write("---") # Horizontal line for separation
+
+# --- STEP 3 ---
+with col3:
+    if st.session_state.show_step3:
+        st.markdown("### 🔹 Step 3")
+        st.write("📤 **Upload Protocol Template (DOCX)**")
+        output_docx = st.file_uploader("Upload DOCX", type=["docx"], key="docx_upload_step3")
+
+        # Dummy progress bar for protocol generation steps
+        protocol_status_text = st.empty()
+        protocol_progress_bar_inner = st.progress(0.0)
+
+        if st.button("📝 Generate Protocol Document", key="generate_protocol_button"):
+            if output_docx is None:
+                st.warning("Please upload a DOCX template file for the simulation.")
+            else:
+                protocol_status_text.text("💾 Saving completed DOCX (simulated)...")
+                protocol_progress_bar_inner.progress(0.5)
+                time.sleep(1) # Simulate delay
+
+                protocol_status_text.text("⏳ Converting DOCX to PDF (simulated)...")
+                protocol_progress_bar_inner.progress(0.7)
+                time.sleep(1) # Simulate delay
+
+                # Generate a dummy PDF
+                st.session_state.generated_pdf = create_dummy_pdf()
+                st.session_state.generated_pdf_name = "dummy_final_protocol.pdf"
+                st.session_state.protocol_generated = True
+                
+                protocol_status_text.text("✅ Final protocol document generated (simulated)!")
+                protocol_progress_bar_inner.progress(1.0)
+                st.success("✅ Protocol document generated! You can now view or download the dummy PDF.")
+                st.rerun()
+
+        if st.session_state.protocol_generated and hasattr(st.session_state, 'generated_pdf'):
+            st.markdown("#### 📥 View / Download")
+            col_a, col_b = st.columns(2) # Removed 'Save' column
+            with col_a:
+                # The 'View PDF' button essentially just triggers a download with preview capabilities
+                st.download_button(
+                    label="👁️ View PDF",
+                    data=st.session_state.generated_pdf,
+                    file_name=st.session_state.generated_pdf_name,
+                    mime="application/pdf",
+                    key="view_pdf_button"
+                )
+            with col_b:
+                st.download_button(
+                    label="⬇️ Download PDF",
+                    data=st.session_state.generated_pdf,
+                    file_name=st.session_state.generated_pdf_name,
+                    mime="application/pdf",
+                    key="download_pdf_button"
+                )
+    else:
+        st.markdown("### 🔹 Step 3")
+        st.info("Complete Step 2 to unlock.")
+
+st.write("---") # Final horizontal line