Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -13,111 +13,183 @@ st.set_page_config(page_title="Schlaeger Forrestdale DocAIA", layout="wide", ini
|
|
13 |
st.title("π Schlaeger Forrestdale Document Assistant")
|
14 |
st.caption("Explore City of Armadale construction documents using AI + OCR π§ ")
|
15 |
|
16 |
-
# ------------------ Load API Key and Assistant
|
17 |
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
|
18 |
-
|
|
|
19 |
|
20 |
-
if not OPENAI_API_KEY or not
|
21 |
-
st.error("β Missing secrets. Please set
|
22 |
st.stop()
|
23 |
|
24 |
client = OpenAI(api_key=OPENAI_API_KEY)
|
25 |
|
26 |
-
# ------------------
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
if "
|
32 |
-
|
33 |
-
if "
|
34 |
-
|
35 |
-
if "
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
st.
|
40 |
-
if st.sidebar.button("π§Ή Clear Chat"):
|
41 |
-
st.session_state.messages = []
|
42 |
-
st.session_state.thread_id = None
|
43 |
-
st.session_state.image_url = None
|
44 |
-
st.session_state.image_updated = False
|
45 |
-
st.session_state.pending_prompt = None
|
46 |
-
st.rerun()
|
47 |
-
|
48 |
-
show_image = st.sidebar.toggle("π Show Page Image", value=True)
|
49 |
-
|
50 |
-
st.sidebar.subheader("π Document Tools")
|
51 |
-
st.sidebar.markdown("Use the tools below to locate relevant clauses and actions:")
|
52 |
-
|
53 |
-
keyword = st.sidebar.text_input("Search by Keyword", placeholder="e.g. defects, WHS, delay")
|
54 |
-
if st.sidebar.button("π Search Keyword") and keyword:
|
55 |
-
st.session_state.pending_prompt = f"Find clauses or references related to: {keyword}"
|
56 |
-
|
57 |
-
section_options = [
|
58 |
-
"Select a section...",
|
59 |
-
"1. Formal Instrument of Contract",
|
60 |
-
"2. Offer and Acceptance",
|
61 |
-
"3. Key Personnel",
|
62 |
-
"4. Contract Pricing",
|
63 |
-
"5. Specifications",
|
64 |
-
"6. WHS Policies",
|
65 |
-
"7. Penalties and Delays",
|
66 |
-
"8. Dispute Resolution",
|
67 |
-
"9. Principal Obligations"
|
68 |
-
]
|
69 |
-
section_select = st.sidebar.selectbox("π Jump to Section", section_options)
|
70 |
-
if section_select != section_options[0]:
|
71 |
-
st.session_state.pending_prompt = f"Summarize or list key points from section: {section_select}"
|
72 |
-
|
73 |
-
actions = [
|
74 |
-
"Select an action...",
|
75 |
-
"List all contractual obligations",
|
76 |
-
"Summarize payment terms",
|
77 |
-
"List WHS responsibilities",
|
78 |
-
"Find delay-related penalties",
|
79 |
-
"Extract dispute resolution steps"
|
80 |
-
]
|
81 |
-
action_select = st.sidebar.selectbox("βοΈ Common Contract Queries", actions)
|
82 |
-
if action_select != actions[0]:
|
83 |
-
st.session_state.pending_prompt = action_select
|
84 |
-
|
85 |
-
# ------------------ Layout: Chat + Image ------------------
|
86 |
-
chat_col, image_col = st.columns([2, 1])
|
87 |
-
|
88 |
-
# ------------------ Chat Interface ------------------
|
89 |
-
with chat_col:
|
90 |
-
st.markdown("### π§ Ask a Document-Specific Question")
|
91 |
-
user_prompt = st.chat_input("Example: What is the defects liability period?")
|
92 |
-
|
93 |
-
# Use pending prompt from sidebar if no new chat prompt
|
94 |
-
if user_prompt:
|
95 |
-
st.session_state.messages.append({"role": "user", "content": user_prompt})
|
96 |
-
elif st.session_state.pending_prompt:
|
97 |
-
st.session_state.messages.append({"role": "user", "content": st.session_state.pending_prompt})
|
98 |
st.session_state.pending_prompt = None
|
99 |
|
100 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
101 |
try:
|
102 |
-
if st.session_state.
|
103 |
thread = client.beta.threads.create()
|
104 |
-
st.session_state.
|
105 |
|
106 |
client.beta.threads.messages.create(
|
107 |
-
thread_id=st.session_state.
|
108 |
role="user",
|
109 |
-
content=st.session_state.
|
110 |
)
|
111 |
|
112 |
run = client.beta.threads.runs.create(
|
113 |
-
thread_id=st.session_state.
|
114 |
-
assistant_id=
|
115 |
)
|
116 |
|
117 |
-
with st.spinner("
|
118 |
while True:
|
119 |
run_status = client.beta.threads.runs.retrieve(
|
120 |
-
thread_id=st.session_state.
|
121 |
run_id=run.id
|
122 |
)
|
123 |
if run_status.status in ("completed", "failed", "cancelled"):
|
@@ -127,44 +199,17 @@ with chat_col:
|
|
127 |
if run_status.status != "completed":
|
128 |
st.error(f"β οΈ Assistant failed: {run_status.status}")
|
129 |
else:
|
130 |
-
messages = client.beta.threads.messages.list(thread_id=st.session_state.
|
131 |
for message in reversed(messages.data):
|
132 |
if message.role == "assistant":
|
133 |
-
|
134 |
-
st.session_state.
|
135 |
-
|
136 |
-
# Parse Document Reference and Page, then construct image URL with encoding
|
137 |
-
match = re.search(r'Document Reference:\s*(.*?),\s*Page\s*(\d+)', assistant_reply)
|
138 |
-
if match:
|
139 |
-
doc_name = match.group(1).strip()
|
140 |
-
page = int(match.group(2))
|
141 |
-
page_str = f"{page:04d}"
|
142 |
-
folder = quote(doc_name)
|
143 |
-
image_url = (
|
144 |
-
f"https://raw.githubusercontent.com/AndrewLORTech/c2ozschlaegerforrestdale/main/"
|
145 |
-
f"{folder}/{folder}_page_{page_str}.png"
|
146 |
-
)
|
147 |
-
st.session_state.image_url = image_url
|
148 |
-
st.session_state.image_updated = True
|
149 |
break
|
150 |
|
151 |
st.rerun()
|
152 |
except Exception as e:
|
153 |
st.error(f"β Error: {e}")
|
154 |
|
155 |
-
for msg in st.session_state.
|
156 |
with st.chat_message(msg["role"]):
|
157 |
st.markdown(msg["content"], unsafe_allow_html=True)
|
158 |
-
|
159 |
-
# ------------------ Image Display ------------------
|
160 |
-
with image_col:
|
161 |
-
if show_image and st.session_state.image_url:
|
162 |
-
with st.spinner("Loading document preview..."):
|
163 |
-
try:
|
164 |
-
response = requests.get(st.session_state.image_url)
|
165 |
-
response.raise_for_status()
|
166 |
-
img = Image.open(BytesIO(response.content))
|
167 |
-
st.image(img, caption="π OCR Page Image", use_container_width=True)
|
168 |
-
st.session_state.image_updated = False
|
169 |
-
except Exception as e:
|
170 |
-
st.error(f"β Failed to load image: {e}")
|
|
|
13 |
st.title("π Schlaeger Forrestdale Document Assistant")
|
14 |
st.caption("Explore City of Armadale construction documents using AI + OCR π§ ")
|
15 |
|
16 |
+
# ------------------ Load API Key and Assistant IDs ------------------
|
17 |
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
|
18 |
+
CONTRACT_ASSISTANT_ID = os.environ.get("ASSISTANT_ID") # Contract assistant
|
19 |
+
TECH_ASSISTANT_ID = "asst_DjvuWBc7tCvMbAhY7n1em4BZ" # Technical drawings assistant
|
20 |
|
21 |
+
if not OPENAI_API_KEY or not CONTRACT_ASSISTANT_ID:
|
22 |
+
st.error("β Missing secrets. Please set OPENAI_API_KEY and ASSISTANT_ID in Hugging Face Space secrets.")
|
23 |
st.stop()
|
24 |
|
25 |
client = OpenAI(api_key=OPENAI_API_KEY)
|
26 |
|
27 |
+
# ------------------ Tabs ------------------
|
28 |
+
tab1, tab2 = st.tabs(["π Contract Assistant", "π Technical Drawings Assistant"])
|
29 |
+
|
30 |
+
# ------------------ Tab 1: Contract Assistant ------------------
|
31 |
+
with tab1:
|
32 |
+
if "messages" not in st.session_state:
|
33 |
+
st.session_state.messages = []
|
34 |
+
if "thread_id" not in st.session_state:
|
35 |
+
st.session_state.thread_id = None
|
36 |
+
if "image_url" not in st.session_state:
|
37 |
+
st.session_state.image_url = None
|
38 |
+
if "image_updated" not in st.session_state:
|
39 |
+
st.session_state.image_updated = False
|
40 |
+
if "pending_prompt" not in st.session_state:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
st.session_state.pending_prompt = None
|
42 |
|
43 |
+
show_image = st.sidebar.toggle("π Show Page Image", value=True)
|
44 |
+
|
45 |
+
st.sidebar.subheader("π Document Tools")
|
46 |
+
keyword = st.sidebar.text_input("Search by Keyword", placeholder="e.g. defects, WHS, delay")
|
47 |
+
if st.sidebar.button("π Search Keyword") and keyword:
|
48 |
+
st.session_state.pending_prompt = f"Find clauses or references related to: {keyword}"
|
49 |
+
|
50 |
+
section_options = [
|
51 |
+
"Select a section...",
|
52 |
+
"1. Formal Instrument of Contract",
|
53 |
+
"2. Offer and Acceptance",
|
54 |
+
"3. Key Personnel",
|
55 |
+
"4. Contract Pricing",
|
56 |
+
"5. Specifications",
|
57 |
+
"6. WHS Policies",
|
58 |
+
"7. Penalties and Delays",
|
59 |
+
"8. Dispute Resolution",
|
60 |
+
"9. Principal Obligations"
|
61 |
+
]
|
62 |
+
section_select = st.sidebar.selectbox("π Jump to Section", section_options)
|
63 |
+
if section_select != section_options[0]:
|
64 |
+
st.session_state.pending_prompt = f"Summarize or list key points from section: {section_select}"
|
65 |
+
|
66 |
+
actions = [
|
67 |
+
"Select an action...",
|
68 |
+
"List all contractual obligations",
|
69 |
+
"Summarize payment terms",
|
70 |
+
"List WHS responsibilities",
|
71 |
+
"Find delay-related penalties",
|
72 |
+
"Extract dispute resolution steps"
|
73 |
+
]
|
74 |
+
action_select = st.sidebar.selectbox("βοΈ Common Contract Queries", actions)
|
75 |
+
if action_select != actions[0]:
|
76 |
+
st.session_state.pending_prompt = action_select
|
77 |
+
|
78 |
+
chat_col, image_col = st.columns([2, 1])
|
79 |
+
|
80 |
+
with chat_col:
|
81 |
+
st.markdown("### π§ Ask a Document-Specific Question")
|
82 |
+
user_prompt = st.chat_input("Example: What is the defects liability period?")
|
83 |
+
|
84 |
+
if user_prompt:
|
85 |
+
st.session_state.messages.append({"role": "user", "content": user_prompt})
|
86 |
+
elif st.session_state.pending_prompt:
|
87 |
+
st.session_state.messages.append({"role": "user", "content": st.session_state.pending_prompt})
|
88 |
+
st.session_state.pending_prompt = None
|
89 |
+
|
90 |
+
if st.session_state.messages and st.session_state.messages[-1]["role"] == "user":
|
91 |
+
try:
|
92 |
+
if st.session_state.thread_id is None:
|
93 |
+
thread = client.beta.threads.create()
|
94 |
+
st.session_state.thread_id = thread.id
|
95 |
+
|
96 |
+
client.beta.threads.messages.create(
|
97 |
+
thread_id=st.session_state.thread_id,
|
98 |
+
role="user",
|
99 |
+
content=st.session_state.messages[-1]["content"]
|
100 |
+
)
|
101 |
+
|
102 |
+
run = client.beta.threads.runs.create(
|
103 |
+
thread_id=st.session_state.thread_id,
|
104 |
+
assistant_id=CONTRACT_ASSISTANT_ID
|
105 |
+
)
|
106 |
+
|
107 |
+
with st.spinner("π€ Parsing and responding with referenced content..."):
|
108 |
+
while True:
|
109 |
+
run_status = client.beta.threads.runs.retrieve(
|
110 |
+
thread_id=st.session_state.thread_id,
|
111 |
+
run_id=run.id
|
112 |
+
)
|
113 |
+
if run_status.status in ("completed", "failed", "cancelled"):
|
114 |
+
break
|
115 |
+
time.sleep(1)
|
116 |
+
|
117 |
+
if run_status.status != "completed":
|
118 |
+
st.error(f"β οΈ Assistant failed: {run_status.status}")
|
119 |
+
else:
|
120 |
+
messages = client.beta.threads.messages.list(thread_id=st.session_state.thread_id)
|
121 |
+
for message in reversed(messages.data):
|
122 |
+
if message.role == "assistant":
|
123 |
+
assistant_reply = message.content[0].text.value
|
124 |
+
st.session_state.messages.append({"role": "assistant", "content": assistant_reply})
|
125 |
+
|
126 |
+
match = re.search(r'Document Reference:\s*(.*?),\s*Page\s*(\d+)', assistant_reply)
|
127 |
+
if match:
|
128 |
+
doc_name = match.group(1).strip()
|
129 |
+
page = int(match.group(2))
|
130 |
+
page_str = f"{page:04d}"
|
131 |
+
folder = quote(doc_name)
|
132 |
+
image_url = (
|
133 |
+
f"https://raw.githubusercontent.com/AndrewLORTech/c2ozschlaegerforrestdale/main/"
|
134 |
+
f"{folder}/{folder}_page_{page_str}.png"
|
135 |
+
)
|
136 |
+
st.session_state.image_url = image_url
|
137 |
+
st.session_state.image_updated = True
|
138 |
+
break
|
139 |
+
|
140 |
+
st.rerun()
|
141 |
+
except Exception as e:
|
142 |
+
st.error(f"β Error: {e}")
|
143 |
+
|
144 |
+
for msg in st.session_state.messages:
|
145 |
+
with st.chat_message(msg["role"]):
|
146 |
+
st.markdown(msg["content"], unsafe_allow_html=True)
|
147 |
+
|
148 |
+
with image_col:
|
149 |
+
if show_image and st.session_state.image_url:
|
150 |
+
with st.spinner("Loading document preview..."):
|
151 |
+
try:
|
152 |
+
response = requests.get(st.session_state.image_url)
|
153 |
+
response.raise_for_status()
|
154 |
+
img = Image.open(BytesIO(response.content))
|
155 |
+
st.image(img, caption="π OCR Page Image", use_container_width=True)
|
156 |
+
st.session_state.image_updated = False
|
157 |
+
except Exception as e:
|
158 |
+
st.error(f"β Failed to load image: {e}")
|
159 |
+
|
160 |
+
# ------------------ Tab 2: Technical Drawing Assistant ------------------
|
161 |
+
with tab2:
|
162 |
+
if "tech_messages" not in st.session_state:
|
163 |
+
st.session_state.tech_messages = []
|
164 |
+
st.session_state.tech_thread_id = None
|
165 |
+
|
166 |
+
st.markdown("### ποΈ Ask a Question About Drawings or Diagrams")
|
167 |
+
tech_prompt = st.chat_input("Example: Show me all architectural drawings")
|
168 |
+
|
169 |
+
if tech_prompt:
|
170 |
+
st.session_state.tech_messages.append({"role": "user", "content": tech_prompt})
|
171 |
+
|
172 |
+
if st.session_state.tech_messages and st.session_state.tech_messages[-1]["role"] == "user":
|
173 |
try:
|
174 |
+
if st.session_state.tech_thread_id is None:
|
175 |
thread = client.beta.threads.create()
|
176 |
+
st.session_state.tech_thread_id = thread.id
|
177 |
|
178 |
client.beta.threads.messages.create(
|
179 |
+
thread_id=st.session_state.tech_thread_id,
|
180 |
role="user",
|
181 |
+
content=st.session_state.tech_messages[-1]["content"]
|
182 |
)
|
183 |
|
184 |
run = client.beta.threads.runs.create(
|
185 |
+
thread_id=st.session_state.tech_thread_id,
|
186 |
+
assistant_id=TECH_ASSISTANT_ID
|
187 |
)
|
188 |
|
189 |
+
with st.spinner("π Querying Technical Drawing Assistant..."):
|
190 |
while True:
|
191 |
run_status = client.beta.threads.runs.retrieve(
|
192 |
+
thread_id=st.session_state.tech_thread_id,
|
193 |
run_id=run.id
|
194 |
)
|
195 |
if run_status.status in ("completed", "failed", "cancelled"):
|
|
|
199 |
if run_status.status != "completed":
|
200 |
st.error(f"β οΈ Assistant failed: {run_status.status}")
|
201 |
else:
|
202 |
+
messages = client.beta.threads.messages.list(thread_id=st.session_state.tech_thread_id)
|
203 |
for message in reversed(messages.data):
|
204 |
if message.role == "assistant":
|
205 |
+
reply = message.content[0].text.value
|
206 |
+
st.session_state.tech_messages.append({"role": "assistant", "content": reply})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
207 |
break
|
208 |
|
209 |
st.rerun()
|
210 |
except Exception as e:
|
211 |
st.error(f"β Error: {e}")
|
212 |
|
213 |
+
for msg in st.session_state.tech_messages:
|
214 |
with st.chat_message(msg["role"]):
|
215 |
st.markdown(msg["content"], unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|