Spaces:
Sleeping
Sleeping
byaldi
Browse files
app.py
CHANGED
@@ -4,11 +4,20 @@ from datetime import datetime
|
|
4 |
|
5 |
import dotenv
|
6 |
import lancedb
|
|
|
7 |
from datasets import load_dataset
|
8 |
from fasthtml.common import * # noqa
|
9 |
from huggingface_hub import login, whoami
|
10 |
from rerankers import Reranker
|
11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
dotenv.load_dotenv()
|
13 |
login(token=os.environ.get("HF_TOKEN"))
|
14 |
|
@@ -77,7 +86,7 @@ def _format_results(results):
|
|
77 |
return ret
|
78 |
|
79 |
|
80 |
-
def retrieve_and_rerank(query, k=
|
81 |
# retrieve ---
|
82 |
n_fetch = 25
|
83 |
|
@@ -163,6 +172,21 @@ style = Style("""
|
|
163 |
margin-right: 10px;
|
164 |
vertical-align: middle;
|
165 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
166 |
""")
|
167 |
|
168 |
# get the fast app and route
|
@@ -229,6 +253,13 @@ def SearchResult(result):
|
|
229 |
)
|
230 |
|
231 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
232 |
def log_query_and_results(query, results):
|
233 |
log_entry = {
|
234 |
"timestamp": datetime.now(),
|
@@ -241,9 +272,19 @@ def log_query_and_results(query, results):
|
|
241 |
@rt("/search")
|
242 |
async def post(query: str):
|
243 |
results = retrieve_and_rerank(query)
|
|
|
|
|
244 |
log_query_and_results(query, results)
|
245 |
|
246 |
-
return Div(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
247 |
|
248 |
|
249 |
def LogEntry(entry):
|
|
|
4 |
|
5 |
import dotenv
|
6 |
import lancedb
|
7 |
+
import requests
|
8 |
from datasets import load_dataset
|
9 |
from fasthtml.common import * # noqa
|
10 |
from huggingface_hub import login, whoami
|
11 |
from rerankers import Reranker
|
12 |
|
13 |
+
|
14 |
+
def get_images(query: str):
|
15 |
+
url = "http://147.189.194.113:80/get_pages"
|
16 |
+
|
17 |
+
response = requests.get(url, params={"query": query})
|
18 |
+
return response.json()
|
19 |
+
|
20 |
+
|
21 |
dotenv.load_dotenv()
|
22 |
login(token=os.environ.get("HF_TOKEN"))
|
23 |
|
|
|
86 |
return ret
|
87 |
|
88 |
|
89 |
+
def retrieve_and_rerank(query, k=3):
|
90 |
# retrieve ---
|
91 |
n_fetch = 25
|
92 |
|
|
|
172 |
margin-right: 10px;
|
173 |
vertical-align: middle;
|
174 |
}
|
175 |
+
.image-results {
|
176 |
+
display: flex;
|
177 |
+
flex-wrap: wrap;
|
178 |
+
gap: 10px;
|
179 |
+
margin-top: 20px;
|
180 |
+
}
|
181 |
+
.image-result {
|
182 |
+
width: calc(20% - 10px);
|
183 |
+
text-align: center;
|
184 |
+
}
|
185 |
+
.image-result img {
|
186 |
+
max-width: 100%;
|
187 |
+
height: auto;
|
188 |
+
border-radius: 5px;
|
189 |
+
}
|
190 |
""")
|
191 |
|
192 |
# get the fast app and route
|
|
|
253 |
)
|
254 |
|
255 |
|
256 |
+
def ImageResult(image):
|
257 |
+
return Div(
|
258 |
+
Img(src=f"data:image/jpeg;base64,{image}", alt="arxiv image"),
|
259 |
+
cls="image-result",
|
260 |
+
)
|
261 |
+
|
262 |
+
|
263 |
def log_query_and_results(query, results):
|
264 |
log_entry = {
|
265 |
"timestamp": datetime.now(),
|
|
|
272 |
@rt("/search")
|
273 |
async def post(query: str):
|
274 |
results = retrieve_and_rerank(query)
|
275 |
+
image_results = get_images(query)
|
276 |
+
|
277 |
log_query_and_results(query, results)
|
278 |
|
279 |
+
return Div(
|
280 |
+
H3("Byaldi Results"),
|
281 |
+
Div(*[ImageResult(img) for img in image_results], cls="image-results"),
|
282 |
+
H3("Text Results"),
|
283 |
+
Div(*[SearchResult(r) for r in results], id="text-results"),
|
284 |
+
id="search-results",
|
285 |
+
)
|
286 |
+
|
287 |
+
# return Div(*[SearchResult(r) for r in results], id="search-results")
|
288 |
|
289 |
|
290 |
def LogEntry(entry):
|