Spaces:
Sleeping
Sleeping
File size: 11,708 Bytes
ef513a5 0b474cc 79adaa2 39f39ce 79adaa2 7b68202 ef513a5 f09334e 39f313e 0b474cc 79adaa2 39f313e 13a7929 79adaa2 74f60fc 39f313e 74f60fc 39f313e 74f60fc 13a7929 74f60fc 9ea268c 74f60fc 13a7929 9ea268c 13a7929 9ea268c 13a7929 9ea268c 74f60fc 39f313e 79adaa2 9ea268c 39f313e 74f60fc f09334e 39f313e 13a7929 74f60fc f09334e 79adaa2 73ca4a0 74f60fc 13a7929 74f60fc ef513a5 79adaa2 74f60fc 9ea268c 56e4e5f 9ea268c 56e4e5f 9ea268c 56e4e5f 9ea268c 56e4e5f 9ea268c 74f60fc bfaaaee 74f60fc bfaaaee 79adaa2 73ca4a0 79adaa2 73ca4a0 bfaaaee 73ca4a0 79adaa2 c6eeec6 bfaaaee 74f60fc bfaaaee 79adaa2 c6eeec6 79adaa2 73ca4a0 79adaa2 e546bbb 74f60fc a14f7cc 79adaa2 74f60fc 73ca4a0 79adaa2 74f60fc bfaaaee 74f60fc bfaaaee 79adaa2 f09334e 39f39ce 79adaa2 74f60fc bfaaaee 74f60fc bfaaaee 74f60fc bfaaaee 74f60fc bfaaaee 74f60fc bfaaaee 39f39ce 79adaa2 74f60fc 39f39ce 74f60fc 39f313e 79adaa2 d71c08c 79adaa2 73ca4a0 d71c08c b62f4f4 1a8ac84 b62f4f4 79adaa2 74f60fc b62f4f4 5f4804b 79adaa2 5f4804b 79adaa2 74f60fc a859b2e a14f7cc a859b2e 74f60fc bd85152 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 | import streamlit as st
import os
from huggingface_hub import HfApi, hf_hub_download
import time
# --- IMPORT OUR NEW MODULES ---
from src.database import DatabaseManager
from src.search import SearchEngine
from src.parsers import process_file, chunk_text
from src.llm_client import ask_llm
# --- CONFIGURATION ---
DATASET_REPO_ID = "NavyDevilDoc/navy-policy-index"
HF_TOKEN = os.environ.get("HF_TOKEN")
INDEX_FILE = "navy_index.faiss"
META_FILE = "navy_metadata.pkl"
st.set_page_config(page_title="Navy Policy Architect", layout="wide", page_icon="β")
# --- CLOUD SYNC MANAGER (FIXED) ---
class SyncManager:
"""Handles downloading/uploading the Database & Index to Hugging Face"""
@staticmethod
def get_remote_dbs():
if not HF_TOKEN: return []
try:
api = HfApi(token=HF_TOKEN)
# This worked because we specified repo_type="dataset" here
files = api.list_repo_files(repo_id=DATASET_REPO_ID, repo_type="dataset")
dbs = [f for f in files if f.endswith(".db")]
return dbs
except Exception as e:
return []
@staticmethod
def pull_data(db_filename):
if not HF_TOKEN:
st.error("HF_TOKEN missing.")
return False
try:
# FIX: Added repo_type="dataset"
hf_hub_download(
repo_id=DATASET_REPO_ID,
filename=db_filename,
local_dir=".",
token=HF_TOKEN,
repo_type="dataset", # <--- THE MISSING LINK
force_download=False
)
# Download Index (Best effort)
try:
hf_hub_download(
repo_id=DATASET_REPO_ID,
filename=INDEX_FILE,
local_dir=".",
token=HF_TOKEN,
repo_type="dataset" # <--- Added here too
)
hf_hub_download(
repo_id=DATASET_REPO_ID,
filename=META_FILE,
local_dir=".",
token=HF_TOKEN,
repo_type="dataset" # <--- And here
)
except:
pass
return True
except Exception as e:
# We return the actual error message so the UI can show it permanently
return str(e)
@staticmethod
def push_data(db_filename):
if not HF_TOKEN: return
api = HfApi(token=HF_TOKEN)
try:
# This was already working because we had repo_type="dataset"
api.upload_file(path_or_fileobj=db_filename, path_in_repo=db_filename, repo_id=DATASET_REPO_ID, repo_type="dataset")
api.upload_file(path_or_fileobj=INDEX_FILE, path_in_repo=INDEX_FILE, repo_id=DATASET_REPO_ID, repo_type="dataset")
api.upload_file(path_or_fileobj=META_FILE, path_in_repo=META_FILE, repo_id=DATASET_REPO_ID, repo_type="dataset")
st.toast("Cloud Sync Complete!", icon="βοΈ")
except Exception as e:
st.error(f"Sync Error (Push): {e}")
# --- SIDEBAR: KNOWLEDGE BASE SELECTOR ---
with st.sidebar:
st.header("ποΈ Knowledge Base")
# 1. Database Selector
# We fetch available DBs from the cloud to populate the dropdown
if "available_dbs" not in st.session_state:
st.session_state.available_dbs = SyncManager.get_remote_dbs()
if not st.session_state.available_dbs:
st.session_state.available_dbs = ["navy_docs.db"] # Default if empty
selected_db = st.selectbox("Select Database:", st.session_state.available_dbs)
# 2. Create New Database Option
with st.expander("β Create New Database"):
new_db_name = st.text_input("Name (e.g., 'Medical.db')")
if st.button("Create"):
if not new_db_name.endswith(".db"):
new_db_name += ".db"
st.session_state.available_dbs.append(new_db_name)
# Force reload to switch to this new DB
st.rerun()
# --- INITIALIZATION (Dynamic based on selection) ---
# If the DB has changed or isn't loaded, load it now
if 'current_db_name' not in st.session_state or st.session_state.current_db_name != selected_db:
# We use an empty container to hold messages
msg_container = st.empty()
with st.spinner(f"Syncing {selected_db}..."):
# 1. Attempt the Pull
result = SyncManager.pull_data(selected_db)
# 2. Check the Result
if result is True:
# Success! Cloud file found.
msg_container.success(f"Loaded {selected_db} from Cloud.")
else:
# Failure! (File deleted or new setup)
# INSTEAD OF STOPPING, we warn and create a fresh local DB.
msg_container.warning(f"Could not find {selected_db} in cloud. Creating new local database.")
# We do NOT run st.stop() here anymore.
# 3. Initialize the Database Manager (Either with the downloaded file or a new blank one)
try:
st.session_state.db = DatabaseManager(selected_db)
st.session_state.search_engine = SearchEngine()
st.session_state.current_db_name = selected_db
# We intentionally do NOT rerun immediately here to let the warning show
except Exception as e:
st.error(f"Failed to initialize database: {e}")
st.stop()
# 3. Upload Section
if "uploader_key" not in st.session_state:
st.session_state.uploader_key = 0
uploaded_files = st.file_uploader(
f"Upload to {selected_db}",
accept_multiple_files=True,
type=['pdf', 'docx', 'txt', 'csv', 'xlsx'],
key=f"uploader_{st.session_state.uploader_key}"
)
if uploaded_files and st.button("Ingest Documents"):
progress_bar = st.progress(0)
status = st.empty()
existing_files = st.session_state.db.get_all_filenames()
for i, f in enumerate(uploaded_files):
status.text(f"Processing: {f.name}...")
if f.name in existing_files:
st.toast(f"β»οΈ Updating: {f.name}")
st.session_state.db.delete_document(f.name)
text, filename, method = process_file(f)
if "Error" in method:
st.error(f"Failed {filename}: {method}")
continue
chunks, doc_id = chunk_text(text, filename)
# Generate Abstract
abstract = "No summary generated."
if len(text) > 500:
with st.spinner(f"Writing abstract for {filename}..."):
abstract = ask_llm(
query="Generate Abstract",
context=text[:30000],
mode="Abstract Generator",
model_provider="Gemini"
)
st.session_state.db.add_document(doc_id, filename, text, abstract=abstract)
st.session_state.search_engine.add_features(chunks)
progress_bar.progress((i + 1) / len(uploaded_files))
status.text("Syncing to Cloud...")
# Push SPECIFICALLY the active database
SyncManager.push_data(selected_db)
st.success(f"Ingested {len(uploaded_files)} docs into {selected_db}!")
time.sleep(1)
st.session_state.uploader_key += 1
st.rerun()
st.divider()
# 4. Document Library
st.subheader(f"Files in {selected_db}")
all_files = st.session_state.db.get_all_filenames()
if all_files:
with st.expander("View File List", expanded=False):
for f in all_files:
st.text(f"β’ {f}")
file_to_del = st.selectbox("Delete File:", [""] + all_files)
if file_to_del and st.button("ποΈ Delete Selected"):
deleted_id = st.session_state.db.delete_document(file_to_del)
st.toast(f"Removed {file_to_del}")
SyncManager.push_data(selected_db)
time.sleep(1)
st.rerun()
if st.button("β οΈ Nuke Database", type="primary"):
for f in all_files:
st.session_state.db.delete_document(f)
st.session_state.search_engine.reset_index()
SyncManager.push_data(selected_db)
st.success("Database wiped.")
time.sleep(1)
st.rerun()
else:
st.info("Library is empty.")
# --- MAIN UI: SEARCH ---
st.title("β Navy Policy Architect")
st.caption(f"Connected to Knowledge Base: {st.session_state.current_db_name}")
query = st.text_input("Enter your query...", placeholder="Search...")
if query:
results = st.session_state.search_engine.search(query, top_k=5)
if not results:
st.info("No matching documents found.")
else:
top_match = results[0]
full_doc_text = st.session_state.db.get_doc_text(top_match['doc_id'])
with st.container():
st.markdown("### π€ Intelligence Hub")
col1, col2 = st.columns(2)
with col1:
analysis_mode = st.selectbox(
"Select Analysis Type:",
["Executive Summary", "Action Plan", "Risk Assessment", "Socratic Review", "Instructor Mode"]
)
with col2:
model_choice = st.selectbox(
"Select Model:",
["Gemini (Cloud - Smartest)", "Granite (Private Space)"]
)
provider = "Gemini" if "Gemini" in model_choice else "Granite"
if st.button("β¨ Generate Assessment"):
with st.spinner(f"Consulting {provider}..."):
response = ask_llm(query, full_doc_text, mode=analysis_mode, model_provider=provider)
st.markdown("---")
st.markdown(response)
st.markdown("---")
# --- SEARCH RESULTS SECTION (FIXED HTML) ---
with st.expander("π Reference Documents", expanded=True):
for res in results:
score = res['score']
color = "#09ab3b" if score > 2 else "#ffbd45" if score > 0 else "#ff4b4b"
doc_abstract = st.session_state.db.get_doc_abstract(res['doc_id'])
# IMPORTANT: Left-aligned HTML string to prevent Code Block rendering
html_content = f"""
<div style="
border-left: 5px solid {color};
padding: 15px;
background-color: #f0f2f6;
margin-bottom: 15px;
border-radius: 5px;
color: #1f1f1f;
">
<div style="display: flex; justify-content: space-between; align-items: center; margin-bottom: 10px;">
<h4 style="margin:0; color: #0e1117;">π {res['source']}</h4>
<span style="font-size: 0.8em; color: #555; background: #ddd; padding: 2px 8px; border-radius: 4px;">Relevance: {score:.2f}</span>
</div>
<div style="background: #e3e6ea; padding: 10px; border-radius: 5px; margin-bottom: 10px;">
<p style="margin: 0; font-size: 0.9em; color: #333;"><strong>π€ Abstract:</strong> {doc_abstract}</p>
</div>
<p style="margin: 0; font-style: italic; font-size: 0.85em; color: #555;">
"Matching Chunk: ...{res['snippet']}..."
</p>
</div>
"""
st.markdown(html_content, unsafe_allow_html=True) |