Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -16,12 +16,7 @@ logging.basicConfig(level=logging.INFO)
|
|
| 16 |
GROQ_API_KEY = st.secrets.get("GROQ_API_KEY", os.environ.get("GROQ_API_KEY"))
|
| 17 |
GROQ_MODEL = "llama-3.1-8b-instant"
|
| 18 |
|
| 19 |
-
client = None
|
| 20 |
-
if GROQ_API_KEY:
|
| 21 |
-
try:
|
| 22 |
-
client = Groq(api_key=GROQ_API_KEY)
|
| 23 |
-
except Exception as e:
|
| 24 |
-
st.error(e)
|
| 25 |
|
| 26 |
# ---------------- PAGE CONFIG ----------------
|
| 27 |
st.set_page_config(
|
|
@@ -30,124 +25,120 @@ st.set_page_config(
|
|
| 30 |
layout="wide"
|
| 31 |
)
|
| 32 |
|
| 33 |
-
# ---------------- FORCE DARK UI
|
| 34 |
st.markdown("""
|
| 35 |
<style>
|
|
|
|
| 36 |
:root {
|
| 37 |
-
--
|
| 38 |
-
--
|
| 39 |
-
--
|
| 40 |
-
--
|
|
|
|
|
|
|
| 41 |
}
|
| 42 |
|
| 43 |
html, body, [data-testid="stAppViewContainer"], .main {
|
| 44 |
-
background-color:
|
| 45 |
-
color:
|
| 46 |
}
|
| 47 |
|
| 48 |
-
/* Hide Streamlit
|
| 49 |
-
header[data-testid="stHeader"], footer {
|
| 50 |
-
display: none;
|
| 51 |
-
}
|
| 52 |
|
| 53 |
/* Sidebar */
|
| 54 |
[data-testid="stSidebar"] {
|
| 55 |
-
background-color:
|
| 56 |
width: 20rem;
|
| 57 |
-
position: fixed;
|
| 58 |
-
height: 100vh;
|
| 59 |
-
overflow-y: auto;
|
| 60 |
}
|
| 61 |
|
| 62 |
-
/*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
.fixed-header {
|
| 64 |
position: fixed;
|
| 65 |
top: 0;
|
| 66 |
left: 0;
|
| 67 |
-
width: 100%;
|
| 68 |
height: 6rem;
|
| 69 |
-
|
| 70 |
-
|
|
|
|
| 71 |
display: flex;
|
| 72 |
flex-direction: column;
|
| 73 |
align-items: center;
|
| 74 |
justify-content: center;
|
| 75 |
-
|
| 76 |
-
}
|
| 77 |
-
|
| 78 |
-
.title-text {
|
| 79 |
-
font-size: 2.5rem;
|
| 80 |
-
font-weight: 800;
|
| 81 |
-
}
|
| 82 |
-
|
| 83 |
-
.creator-text {
|
| 84 |
-
font-size: 1rem;
|
| 85 |
-
color: #9ca3af;
|
| 86 |
}
|
| 87 |
|
| 88 |
-
.
|
| 89 |
-
|
| 90 |
-
text-decoration: none;
|
| 91 |
-
}
|
| 92 |
|
| 93 |
-
/* Main
|
| 94 |
.main .block-container {
|
| 95 |
margin-top: 6rem;
|
| 96 |
-
padding-bottom:
|
| 97 |
}
|
| 98 |
|
| 99 |
-
/*
|
| 100 |
input, textarea {
|
| 101 |
-
background-color:
|
| 102 |
-
color:
|
| 103 |
-
border: 1px solid
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
}
|
| 105 |
|
| 106 |
-
/*
|
| 107 |
button {
|
| 108 |
-
|
|
|
|
| 109 |
color: white !important;
|
| 110 |
-
border-radius:
|
| 111 |
-
font-weight:
|
| 112 |
}
|
| 113 |
|
| 114 |
-
button:hover {
|
| 115 |
-
background-color: #2563eb !important;
|
| 116 |
-
}
|
| 117 |
|
| 118 |
/* Chat bubbles */
|
| 119 |
.chat-user {
|
| 120 |
background: #2d3748;
|
| 121 |
-
padding:
|
| 122 |
-
border-radius:
|
| 123 |
margin: 6px 0 6px auto;
|
| 124 |
max-width: 85%;
|
| 125 |
text-align: right;
|
| 126 |
}
|
| 127 |
|
| 128 |
.chat-bot {
|
| 129 |
-
background:
|
| 130 |
-
padding:
|
| 131 |
-
border-radius:
|
| 132 |
margin: 6px auto 6px 0;
|
| 133 |
max-width: 85%;
|
| 134 |
}
|
| 135 |
|
| 136 |
-
/*
|
| 137 |
-
|
| 138 |
-
section[data-testid="stFileUploader"] ul,
|
| 139 |
-
section[data-testid="stFileUploader"] small {
|
| 140 |
-
display: none;
|
| 141 |
-
}
|
| 142 |
-
|
| 143 |
-
/* Scrollbar */
|
| 144 |
-
*::-webkit-scrollbar {
|
| 145 |
-
width: 8px;
|
| 146 |
-
}
|
| 147 |
*::-webkit-scrollbar-thumb {
|
| 148 |
background: #2d3748;
|
| 149 |
border-radius: 4px;
|
| 150 |
}
|
|
|
|
|
|
|
|
|
|
| 151 |
</style>
|
| 152 |
""", unsafe_allow_html=True)
|
| 153 |
|
|
@@ -156,7 +147,7 @@ st.markdown("""
|
|
| 156 |
<div class="fixed-header">
|
| 157 |
<div class="title-text">π PDF Assistant</div>
|
| 158 |
<div class="creator-text">
|
| 159 |
-
by <a href="https://www.linkedin.com/in/abhishek-iitr/" target="_blank">
|
| 160 |
Abhishek Saxena</a>
|
| 161 |
</div>
|
| 162 |
</div>
|
|
@@ -192,53 +183,38 @@ def process_pdf(uploaded):
|
|
| 192 |
tmp.write(uploaded.getvalue())
|
| 193 |
path = tmp.name
|
| 194 |
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
chunks = splitter.split_documents(docs)
|
| 200 |
|
| 201 |
embeddings = HuggingFaceEmbeddings(
|
| 202 |
model_name="sentence-transformers/all-MiniLM-L6-v2",
|
| 203 |
-
model_kwargs={"device": "cpu"},
|
| 204 |
encode_kwargs={"normalize_embeddings": True}
|
| 205 |
)
|
| 206 |
|
| 207 |
-
|
| 208 |
-
st.session_state.retriever =
|
| 209 |
os.unlink(path)
|
| 210 |
|
| 211 |
-
def ask(
|
| 212 |
-
docs = st.session_state.retriever.invoke(
|
| 213 |
context = "\n\n".join(d.page_content for d in docs)
|
| 214 |
|
| 215 |
-
prompt = f"""
|
| 216 |
-
You are a strict RAG assistant.
|
| 217 |
-
Use ONLY the PDF content.
|
| 218 |
-
If the answer is not found, say:
|
| 219 |
-
"I cannot find this in the PDF."
|
| 220 |
-
|
| 221 |
-
CONTEXT:
|
| 222 |
-
{context}
|
| 223 |
-
|
| 224 |
-
QUESTION:
|
| 225 |
-
{question}
|
| 226 |
-
"""
|
| 227 |
-
|
| 228 |
res = client.chat.completions.create(
|
| 229 |
model=GROQ_MODEL,
|
| 230 |
-
messages=[{
|
|
|
|
|
|
|
|
|
|
| 231 |
temperature=0.1
|
| 232 |
)
|
| 233 |
return res.choices[0].message.content.strip()
|
| 234 |
|
| 235 |
# ---------------- SIDEBAR ----------------
|
| 236 |
with st.sidebar:
|
| 237 |
-
|
| 238 |
-
|
| 239 |
-
if st.button("π₯ Clear PDF Memory", use_container_width=True):
|
| 240 |
-
clear_memory()
|
| 241 |
-
st.success("Memory cleared")
|
| 242 |
|
| 243 |
st.markdown("---")
|
| 244 |
|
|
@@ -262,8 +238,8 @@ with st.sidebar:
|
|
| 262 |
# ---------------- INPUT ----------------
|
| 263 |
disabled = not st.session_state.uploaded_file_name or not client
|
| 264 |
|
| 265 |
-
with st.form(
|
| 266 |
-
col1, col2 = st.columns([0.
|
| 267 |
with col1:
|
| 268 |
question = st.text_input(
|
| 269 |
"",
|
|
@@ -276,13 +252,10 @@ with st.form(key="chat_form", clear_on_submit=True):
|
|
| 276 |
if send and question:
|
| 277 |
st.session_state.messages.append(("user", question))
|
| 278 |
with st.spinner("Thinking..."):
|
| 279 |
-
|
| 280 |
-
st.session_state.messages.append(("bot", answer))
|
| 281 |
st.rerun()
|
| 282 |
|
| 283 |
-
# ---------------- CHAT
|
| 284 |
for role, msg in reversed(st.session_state.messages):
|
| 285 |
-
if role == "user"
|
| 286 |
-
|
| 287 |
-
else:
|
| 288 |
-
st.markdown(f"<div class='chat-bot'>{msg}</div>", unsafe_allow_html=True)
|
|
|
|
| 16 |
GROQ_API_KEY = st.secrets.get("GROQ_API_KEY", os.environ.get("GROQ_API_KEY"))
|
| 17 |
GROQ_MODEL = "llama-3.1-8b-instant"
|
| 18 |
|
| 19 |
+
client = Groq(api_key=GROQ_API_KEY) if GROQ_API_KEY else None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
# ---------------- PAGE CONFIG ----------------
|
| 22 |
st.set_page_config(
|
|
|
|
| 25 |
layout="wide"
|
| 26 |
)
|
| 27 |
|
| 28 |
+
# ---------------- FORCE DARK UI ----------------
|
| 29 |
st.markdown("""
|
| 30 |
<style>
|
| 31 |
+
/* ================= FORCE DARK THEME ================= */
|
| 32 |
:root {
|
| 33 |
+
--bg: #0e1117;
|
| 34 |
+
--sidebar: #161b22;
|
| 35 |
+
--input: #1f2933;
|
| 36 |
+
--border: #374151;
|
| 37 |
+
--text: #f0f2f6;
|
| 38 |
+
--primary: #1e3a8a;
|
| 39 |
}
|
| 40 |
|
| 41 |
html, body, [data-testid="stAppViewContainer"], .main {
|
| 42 |
+
background-color: var(--bg) !important;
|
| 43 |
+
color: var(--text) !important;
|
| 44 |
}
|
| 45 |
|
| 46 |
+
/* Hide Streamlit chrome */
|
| 47 |
+
header[data-testid="stHeader"], footer { display: none; }
|
|
|
|
|
|
|
| 48 |
|
| 49 |
/* Sidebar */
|
| 50 |
[data-testid="stSidebar"] {
|
| 51 |
+
background-color: var(--sidebar) !important;
|
| 52 |
width: 20rem;
|
|
|
|
|
|
|
|
|
|
| 53 |
}
|
| 54 |
|
| 55 |
+
/* FIX: File uploader white background */
|
| 56 |
+
section[data-testid="stFileUploader"],
|
| 57 |
+
section[data-testid="stFileUploader"] > div,
|
| 58 |
+
section[data-testid="stFileUploader"] label {
|
| 59 |
+
background-color: var(--sidebar) !important;
|
| 60 |
+
color: var(--text) !important;
|
| 61 |
+
border-radius: 12px;
|
| 62 |
+
}
|
| 63 |
+
|
| 64 |
+
/* Fixed Header */
|
| 65 |
.fixed-header {
|
| 66 |
position: fixed;
|
| 67 |
top: 0;
|
| 68 |
left: 0;
|
|
|
|
| 69 |
height: 6rem;
|
| 70 |
+
width: 100%;
|
| 71 |
+
background: var(--bg);
|
| 72 |
+
border-bottom: 1px solid rgba(255,255,255,0.08);
|
| 73 |
display: flex;
|
| 74 |
flex-direction: column;
|
| 75 |
align-items: center;
|
| 76 |
justify-content: center;
|
| 77 |
+
z-index: 9999;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
}
|
| 79 |
|
| 80 |
+
.title-text { font-size: 2.5rem; font-weight: 800; }
|
| 81 |
+
.creator-text { color: #9ca3af; }
|
|
|
|
|
|
|
| 82 |
|
| 83 |
+
/* Main spacing */
|
| 84 |
.main .block-container {
|
| 85 |
margin-top: 6rem;
|
| 86 |
+
padding-bottom: 6rem;
|
| 87 |
}
|
| 88 |
|
| 89 |
+
/* ================= INPUT + BUTTON ALIGNMENT ================= */
|
| 90 |
input, textarea {
|
| 91 |
+
background-color: var(--input) !important;
|
| 92 |
+
color: var(--text) !important;
|
| 93 |
+
border: 1px solid var(--border) !important;
|
| 94 |
+
height: 3.2rem !important;
|
| 95 |
+
border-radius: 10px !important;
|
| 96 |
+
}
|
| 97 |
+
|
| 98 |
+
/* Placeholder FIX (same in light & dark) */
|
| 99 |
+
input::placeholder {
|
| 100 |
+
color: #9ca3af !important;
|
| 101 |
+
opacity: 1 !important;
|
| 102 |
}
|
| 103 |
|
| 104 |
+
/* Button (Arrow) */
|
| 105 |
button {
|
| 106 |
+
height: 3.2rem !important;
|
| 107 |
+
background-color: var(--primary) !important;
|
| 108 |
color: white !important;
|
| 109 |
+
border-radius: 10px !important;
|
| 110 |
+
font-weight: 700 !important;
|
| 111 |
}
|
| 112 |
|
| 113 |
+
button:hover { background-color: #2563eb !important; }
|
|
|
|
|
|
|
| 114 |
|
| 115 |
/* Chat bubbles */
|
| 116 |
.chat-user {
|
| 117 |
background: #2d3748;
|
| 118 |
+
padding: 14px;
|
| 119 |
+
border-radius: 12px 12px 2px 12px;
|
| 120 |
margin: 6px 0 6px auto;
|
| 121 |
max-width: 85%;
|
| 122 |
text-align: right;
|
| 123 |
}
|
| 124 |
|
| 125 |
.chat-bot {
|
| 126 |
+
background: var(--primary);
|
| 127 |
+
padding: 14px;
|
| 128 |
+
border-radius: 12px 12px 12px 2px;
|
| 129 |
margin: 6px auto 6px 0;
|
| 130 |
max-width: 85%;
|
| 131 |
}
|
| 132 |
|
| 133 |
+
/* Scrollbars */
|
| 134 |
+
*::-webkit-scrollbar { width: 8px; }
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 135 |
*::-webkit-scrollbar-thumb {
|
| 136 |
background: #2d3748;
|
| 137 |
border-radius: 4px;
|
| 138 |
}
|
| 139 |
+
|
| 140 |
+
/* Hide uploaded filename */
|
| 141 |
+
[data-testid="stFileUploaderFile"] { display: none; }
|
| 142 |
</style>
|
| 143 |
""", unsafe_allow_html=True)
|
| 144 |
|
|
|
|
| 147 |
<div class="fixed-header">
|
| 148 |
<div class="title-text">π PDF Assistant</div>
|
| 149 |
<div class="creator-text">
|
| 150 |
+
by <a href="https://www.linkedin.com/in/abhishek-iitr/" target="_blank" style="color:#4da6ff">
|
| 151 |
Abhishek Saxena</a>
|
| 152 |
</div>
|
| 153 |
</div>
|
|
|
|
| 183 |
tmp.write(uploaded.getvalue())
|
| 184 |
path = tmp.name
|
| 185 |
|
| 186 |
+
docs = PyPDFLoader(path).load()
|
| 187 |
+
chunks = RecursiveCharacterTextSplitter(
|
| 188 |
+
chunk_size=800, chunk_overlap=60
|
| 189 |
+
).split_documents(docs)
|
|
|
|
| 190 |
|
| 191 |
embeddings = HuggingFaceEmbeddings(
|
| 192 |
model_name="sentence-transformers/all-MiniLM-L6-v2",
|
|
|
|
| 193 |
encode_kwargs={"normalize_embeddings": True}
|
| 194 |
)
|
| 195 |
|
| 196 |
+
vs = Chroma.from_documents(chunks, embeddings)
|
| 197 |
+
st.session_state.retriever = vs.as_retriever(search_kwargs={"k": 5})
|
| 198 |
os.unlink(path)
|
| 199 |
|
| 200 |
+
def ask(q):
|
| 201 |
+
docs = st.session_state.retriever.invoke(q)
|
| 202 |
context = "\n\n".join(d.page_content for d in docs)
|
| 203 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
res = client.chat.completions.create(
|
| 205 |
model=GROQ_MODEL,
|
| 206 |
+
messages=[{
|
| 207 |
+
"role": "user",
|
| 208 |
+
"content": f"Use ONLY the PDF content.\n\n{context}\n\nQuestion: {q}"
|
| 209 |
+
}],
|
| 210 |
temperature=0.1
|
| 211 |
)
|
| 212 |
return res.choices[0].message.content.strip()
|
| 213 |
|
| 214 |
# ---------------- SIDEBAR ----------------
|
| 215 |
with st.sidebar:
|
| 216 |
+
st.button("ποΈ Clear Chat History", on_click=clear_chat, use_container_width=True)
|
| 217 |
+
st.button("π₯ Clear PDF Memory", on_click=clear_memory, use_container_width=True)
|
|
|
|
|
|
|
|
|
|
| 218 |
|
| 219 |
st.markdown("---")
|
| 220 |
|
|
|
|
| 238 |
# ---------------- INPUT ----------------
|
| 239 |
disabled = not st.session_state.uploaded_file_name or not client
|
| 240 |
|
| 241 |
+
with st.form("chat_form", clear_on_submit=True):
|
| 242 |
+
col1, col2 = st.columns([0.9, 0.1])
|
| 243 |
with col1:
|
| 244 |
question = st.text_input(
|
| 245 |
"",
|
|
|
|
| 252 |
if send and question:
|
| 253 |
st.session_state.messages.append(("user", question))
|
| 254 |
with st.spinner("Thinking..."):
|
| 255 |
+
st.session_state.messages.append(("bot", ask(question)))
|
|
|
|
| 256 |
st.rerun()
|
| 257 |
|
| 258 |
+
# ---------------- CHAT ----------------
|
| 259 |
for role, msg in reversed(st.session_state.messages):
|
| 260 |
+
css = "chat-user" if role == "user" else "chat-bot"
|
| 261 |
+
st.markdown(f"<div class='{css}'>{msg}</div>", unsafe_allow_html=True)
|
|
|
|
|
|