Update app.py

app.py

@@ -1,93 +1,170 @@
 import streamlit as st
 import os
 import time
-import json
+import re
+import requests
+from PIL import Image
+from io import BytesIO
+from urllib.parse import quote
 from openai import OpenAI
 
-# App config
-st.set_page_config(page_title="🏗️ Forrestdale Technical Drawing Assistant", layout="wide")
-st.title("🏗️ Forrestdale Technical Drawing Assistant")
-st.caption("Ask about plans, drawings or components")
+# ------------------ App Configuration ------------------
+st.set_page_config(page_title="Schlaeger Forrestdale TechDocAIA", layout="wide", initial_sidebar_state="collapsed")
+st.title("📄 Schlaeger Forrestdale Document Assistant")
+st.caption("Explore City of Armadale construction documents using AI + OCR 🧠")
 
-# Load secrets
+# ------------------ Load API Key and Assistant ID ------------------
 OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
 ASSISTANT_ID = os.environ.get("ASSISTANT_ID")
-client = OpenAI(api_key=OPENAI_API_KEY)
 
 if not OPENAI_API_KEY or not ASSISTANT_ID:
-    st.error("Missing secrets: OPENAI_API_KEY or ASSISTANT_ID")
+    st.error("❌ Missing secrets. Please set both OPENAI_API_KEY and ASSISTANT_ID in Hugging Face Space secrets.")
     st.stop()
 
-# Session setup
+client = OpenAI(api_key=OPENAI_API_KEY)
+
+# ------------------ Session State Initialization ------------------
 if "messages" not in st.session_state:
     st.session_state.messages = []
 if "thread_id" not in st.session_state:
     st.session_state.thread_id = None
+if "image_url" not in st.session_state:
+    st.session_state.image_url = None
+if "image_updated" not in st.session_state:
+    st.session_state.image_updated = False
+if "pending_prompt" not in st.session_state:
+    st.session_state.pending_prompt = None
 
-# Prompt input
-prompt = st.chat_input("Ask something like 'Show me all civil drawings' or 'Where are the switchboards?'")
-if prompt:
-    st.session_state.messages.append({"role": "user", "content": prompt})
-
-# Chat + assistant processing
-if st.session_state.messages and st.session_state.messages[-1]["role"] == "user":
-    if st.session_state.thread_id is None:
-        thread = client.beta.threads.create()
-        st.session_state.thread_id = thread.id
-
-    client.beta.threads.messages.create(
-        thread_id=st.session_state.thread_id,
-        role="user",
-        content=st.session_state.messages[-1]["content"]
-    )
-    run = client.beta.threads.runs.create(
-        thread_id=st.session_state.thread_id,
-        assistant_id=ASSISTANT_ID
-    )
-
-    with st.spinner("🔎 Querying assistant..."):
-        while True:
-            run_status = client.beta.threads.runs.retrieve(thread_id=st.session_state.thread_id, run_id=run.id)
-            if run_status.status in ("completed", "failed", "cancelled"):
-                break
-            time.sleep(1)
-
-        if run_status.status != "completed":
-            st.error(f"Assistant failed: {run_status.status}")
-        else:
-            messages = client.beta.threads.messages.list(thread_id=st.session_state.thread_id)
-            for message in reversed(messages.data):
-                if message.role == "assistant":
-                    try:
-                        raw = message.content[0].text.value.strip()
-                        if raw.startswith("```json"):
-                            raw = raw.removeprefix("```json").removesuffix("```").strip()
-                        data = json.loads(raw)
-                        st.session_state.messages.append({"role": "assistant", "content": data})
-                    except Exception as e:
-                        st.error(f"⚠️ Could not parse assistant response as JSON.\n{e}")
-                    break
+# ------------------ Sidebar ------------------
+st.sidebar.header("ℹ️ Information")
+if st.sidebar.button("🧹 Clear Chat"):
+    st.session_state.messages = []
+    st.session_state.thread_id = None
+    st.session_state.image_url = None
+    st.session_state.image_updated = False
+    st.session_state.pending_prompt = None
     st.rerun()
 
-# Card view rendering
-for msg in st.session_state.messages:
-    if msg["role"] == "assistant":
-        data = msg["content"]
-        if isinstance(data, list) and len(data) > 0:
-            st.markdown("### 📋 Results")
-            cols = st.columns(4)
-            for i, item in enumerate(data):
-                with cols[i % 4]:
-                    st.image(item.get("images", [item.get("image")])[0], caption=item["drawing_number"], use_column_width=True)
-                    with st.expander("Details"):
-                        st.write(f"**Discipline:** {item['discipline']}")
-                        st.write(f"**Summary:** {item['summary']}")
-                        if "question" in item:
-                            st.write(f"**Question:** {item['question']}")
-                        for idx, img in enumerate(item.get("images", [])):
-                            st.image(img, caption=f"Page {idx+1}", use_column_width=True)
-        else:
-            st.warning("Assistant returned no matching drawings.")
-    elif msg["role"] == "user":
-        with st.chat_message("user"):
-            st.markdown(msg["content"])
+show_image = st.sidebar.toggle("📑 Show Page Image", value=True)
+
+st.sidebar.subheader("📘 Document Tools")
+st.sidebar.markdown("Use the tools below to locate relevant clauses and actions:")
+
+keyword = st.sidebar.text_input("Search by Keyword", placeholder="e.g. defects, WHS, delay")
+if st.sidebar.button("🔎 Search Keyword") and keyword:
+    st.session_state.pending_prompt = f"Find clauses or references related to: {keyword}"
+
+section_options = [
+    "Select a section...",
+    "1. Formal Instrument of Contract",
+    "2. Offer and Acceptance",
+    "3. Key Personnel",
+    "4. Contract Pricing",
+    "5. Specifications",
+    "6. WHS Policies",
+    "7. Penalties and Delays",
+    "8. Dispute Resolution",
+    "9. Principal Obligations"
+]
+section_select = st.sidebar.selectbox("📄 Jump to Section", section_options)
+if section_select != section_options[0]:
+    st.session_state.pending_prompt = f"Summarize or list key points from section: {section_select}"
+
+actions = [
+    "Select an action...",
+    "List all contractual obligations",
+    "Summarize payment terms",
+    "List WHS responsibilities",
+    "Find delay-related penalties",
+    "Extract dispute resolution steps"
+]
+action_select = st.sidebar.selectbox("⚙️ Common Contract Queries", actions)
+if action_select != actions[0]:
+    st.session_state.pending_prompt = action_select
+
+# ------------------ Layout: Chat + Image ------------------
+chat_col, image_col = st.columns([2, 1])
+
+# ------------------ Chat Interface ------------------
+with chat_col:
+    st.markdown("### 🧠 Ask a Document-Specific Question")
+    user_prompt = st.chat_input("Example: What is the defects liability period?")
+
+    # Use pending prompt from sidebar if no new chat prompt
+    if user_prompt:
+        st.session_state.messages.append({"role": "user", "content": user_prompt})
+    elif st.session_state.pending_prompt:
+        st.session_state.messages.append({"role": "user", "content": st.session_state.pending_prompt})
+        st.session_state.pending_prompt = None
+
+    if st.session_state.messages and st.session_state.messages[-1]["role"] == "user":
+        try:
+            if st.session_state.thread_id is None:
+                thread = client.beta.threads.create()
+                st.session_state.thread_id = thread.id
+
+            client.beta.threads.messages.create(
+                thread_id=st.session_state.thread_id,
+                role="user",
+                content=st.session_state.messages[-1]["content"]
+            )
+
+            run = client.beta.threads.runs.create(
+                thread_id=st.session_state.thread_id,
+                assistant_id=ASSISTANT_ID
+            )
+
+            with st.spinner("🤖 Parsing and responding with referenced content..."):
+                while True:
+                    run_status = client.beta.threads.runs.retrieve(
+                        thread_id=st.session_state.thread_id,
+                        run_id=run.id
+                    )
+                    if run_status.status in ("completed", "failed", "cancelled"):
+                        break
+                    time.sleep(1)
+
+            if run_status.status != "completed":
+                st.error(f"⚠️ Assistant failed: {run_status.status}")
+            else:
+                messages = client.beta.threads.messages.list(thread_id=st.session_state.thread_id)
+                for message in reversed(messages.data):
+                    if message.role == "assistant":
+                        assistant_reply = message.content[0].text.value
+                        st.session_state.messages.append({"role": "assistant", "content": assistant_reply})
+
+                        # Parse Document Reference and Page, then construct image URL with encoding
+                        match = re.search(r'Document Reference:\s*(.*?),\s*Page\s*(\d+)', assistant_reply)
+                        if match:
+                            doc_name = match.group(1).strip()
+                            page = int(match.group(2))
+                            page_str = f"{page:04d}"
+                            folder = quote(doc_name)
+                            image_url = (
+                                f"https://raw.githubusercontent.com/AndrewLORTech/c2ozschlaegerforrestdale/main/"
+                                f"{folder}/{folder}_page_{page_str}.png"
+                            )
+                            st.session_state.image_url = image_url
+                            st.session_state.image_updated = True
+                        break
+
+            st.rerun()
+        except Exception as e:
+            st.error(f"❌ Error: {e}")
+
+    for msg in st.session_state.messages:
+        with st.chat_message(msg["role"]):
+            st.markdown(msg["content"], unsafe_allow_html=True)
+
+# ------------------ Image Display ------------------
+with image_col:
+    if show_image and st.session_state.image_url:
+        with st.spinner("Loading document preview..."):
+            try:
+                response = requests.get(st.session_state.image_url)
+                response.raise_for_status()
+                img = Image.open(BytesIO(response.content))
+                st.image(img, caption="📄 OCR Page Image", use_container_width=True)
+                st.session_state.image_updated = False
+            except Exception as e:
+                st.error(f"❗ Failed to load image: {e}")