-
Notifications
You must be signed in to change notification settings - Fork 1
/
app.py
287 lines (241 loc) · 7.88 KB
/
app.py
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
import os
import time
import urllib.parse
import json
import streamlit as st
import templates
import click
from backend import (
add_url,
predict,
setup_db,
list_glossary,
list_documents,
query_database,
get_query_return_nums,
)
IP = os.environ.get("PDF_FILE_SERVICE_HOST", "localhost")
def add_logo(width=300):
LOGO_URL = "https://superduperdb-public-demo.s3.amazonaws.com/superduper_logo.svg"
st.markdown(
f"""
<style>
.fixed-logo img {{
position: fixed;
top: 1rem;
left: 0rem;
z-index: 9999999;
}}
</style>
<div class="fixed-logo">
<img src="{LOGO_URL}" alt="Logo" style="width: {width}px; height: auto;">
</div>
""",
unsafe_allow_html=True
)
def set_session_state():
""" """
# default values
if "search" not in st.session_state:
st.session_state.search = None
if "tags" not in st.session_state:
st.session_state.tags = None
if "page" not in st.session_state:
st.session_state.page = 1
# get parameters in url
para = st.query_params
if "search" in para:
st.session_state.search = urllib.parse.unquote(para["search"][0])
if "tags" in para:
st.session_state.tags = para["tags"][0]
if "page" in para:
st.session_state.page = int(para["page"][0])
@click.command()
@click.option("--reset", is_flag=True, help="Reset the database.")
def main(reset):
st.set_page_config(page_title="SuperDuperDB Application", page_icon=":crystal_ball:")
add_logo()
st.markdown("## SuperDuper App Demo: <br> NVCA Legal Docs - RAG", unsafe_allow_html=True)
set_session_state()
db = st.cache_resource(setup_db)(reset=reset)
st.write(templates.load_css(), unsafe_allow_html=True)
[
tab_qa,
tab_glossary,
tab_documents,
tab_query,
] = st.tabs(
[
"Chat",
"Glossary",
"Documents",
"Query",
]
)
with tab_qa:
app_qa(db)
with tab_glossary:
app_glossary(db)
with tab_documents:
app_documents(db)
with tab_query:
app_query(db)
def app_add_url(db):
print("add url")
st.subheader("Add datas from S3 Urls")
url = st.text_input("Enter PDF S3 url:")
if not url.strip():
return
print(url)
if add_url(db, url):
st.write("##### Success!")
def app_qa(db):
"""search layout"""
# load css
page_size = 20
if st.session_state.search is None:
search = st.text_input("Enter search query:")
else:
search = st.text_input("Enter search query:", st.session_state.search)
if search:
print("search", search)
# reset tags when receive new search words
if search != st.session_state.search:
st.session_state.tags = None
# reset search word
st.session_state.search = None
from_i = (st.session_state.page - 1) * page_size
start_time = time.time()
answer, contexts = predict(db, search)
st.write("##### Response:")
st.write(answer)
st.divider()
cost_time = time.time() - start_time
total_hits = len(contexts)
if total_hits > 0:
# show number of results and time taken
st.write(
templates.number_of_results(total_hits, cost_time),
unsafe_allow_html=True,
)
render_contexts(search, contexts, from_i)
# pagination
if total_hits > page_size:
total_pages = (total_hits + page_size - 1) // page_size
pagination_html = templates.pagination(
total_pages,
search,
st.session_state.page,
st.session_state.tags,
)
st.write(pagination_html, unsafe_allow_html=True)
else:
# no result found
st.write(templates.no_result_html(), unsafe_allow_html=True)
def render_contexts(search, contexts, from_i=0):
# search results
# RESULT
i = 0
contexts = sorted(contexts, key=lambda x: x[-1], reverse=True)
for context in contexts:
page_number, url, text, dst, score = context
title = " ".join(text.split(" ")[:10])
title = title.replace("\n", " ")
res = {}
res["url"] = f"http://{IP}:8000/" + dst + f"#page={page_number}"
res["page_no"] = page_number
res["highlights"] = text
res["title"] = title
score = str(round(score, 2))
st.write(templates.search_result(i + from_i, **res), unsafe_allow_html=True)
i += 1
if i < 5:
tags = [score, "Used in AI Answer"]
else:
tags = [score, "Other Results"]
tags_html = templates.tag_boxes(search, tags, active_tag=score)
st.write(tags_html, unsafe_allow_html=True)
def app_glossary(db):
glossaries = list_glossary(db)
item2data = {data["item"]: data for data in glossaries}
item_list = sorted(item2data.keys())
option = st.selectbox(
"Choose a term mentioned in NVCA Legal Docs:",
item_list,
)
data = item2data[option]
st.write("**Definition:**")
st.write(data["definition"])
search = f"What is {data['item']}"
button = st.button("More details")
if button:
answer, contexts = predict(db, search)
st.write(answer)
render_contexts(search, contexts)
def app_documents(db):
documents = list_documents(db)
uris = [doc["uri"] for doc in documents]
selected_uri = st.selectbox("Choose a document to view:", uris)
selected_doc = [doc for doc in documents if doc["uri"] == selected_uri][0]
url = f"http://{IP}:8000/" + selected_doc["base_path"]
st.write(f"**[View document]({url})**")
st.markdown("#### Summary")
st.write(selected_doc["summary"])
# st.markdown("### Quickly Chat With AI On This Document")
# search = st.text_input("Input your question:")
# if search.strip():
# answer, contexts = predict(db, search)
#
# st.write("##### Response:")
# st.write(answer)
#
# render_contexts(search, contexts)
def reset_selected_index():
st.session_state["selected_index_str"] = None
def app_query(db):
collection_names = sorted(db.databackend.db.list_collection_names(), reverse=True)
selected_collection = st.selectbox(
"Choose a collection to search in:",
collection_names,
on_change=reset_selected_index,
)
query_string = st.text_area(
"Please input query (JSON format)",
"{}",
on_change=reset_selected_index,
)
query_button = st.button("Query", on_click=reset_selected_index)
if not (query_button or "selected_index_str" in st.session_state):
return
try:
query_dict = json.loads(query_string)
except json.JSONDecodeError:
st.error("Invalid JSON format. Please correct it and try again.")
return
num = get_query_return_nums(db, selected_collection, query_dict)
st.write(f"Number of results: {num}")
if num > 0:
col1, col2 = st.columns([5, 1])
result_indices = [str(i + 1) for i in range(min(num, 10))]
with col2:
current_selection = st.radio(
"Select result number",
result_indices,
key="selected_index",
)
selected_index = int(current_selection) - 1
st.session_state[
"selected_index_str"
] = current_selection
datas = query_database(
db, selected_collection, query_dict, skip=selected_index, limit=1
)
with col1:
if datas:
st.json(datas[0])
else:
st.write("No result found")
else:
st.write("No result found")
if __name__ == "__main__":
main()