Spaces:
Running
Running
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,149 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from fasthtml.common import *
|
2 |
+
from datetime import datetime, timedelta
|
3 |
+
import requests
|
4 |
+
from datetime import datetime
|
5 |
+
import json
|
6 |
+
|
7 |
+
from dotenv import load_dotenv
|
8 |
+
|
9 |
+
loaded = load_dotenv("./.env", override=True)
|
10 |
+
print("Loaded .env file:", loaded)
|
11 |
+
|
12 |
+
API_URL = os.getenv("API_URL")
|
13 |
+
API_KEY = os.getenv("MS_SEARCH_KEY")
|
14 |
+
|
15 |
+
|
16 |
+
css_content = open("styles.css").read()
|
17 |
+
|
18 |
+
app, rt = fast_app(hdrs=(Style(css_content),))
|
19 |
+
|
20 |
+
def date_range_inputs(start_date, end_date):
|
21 |
+
return Div(
|
22 |
+
Input(type="date", name="start_date", value=start_date.strftime("%Y-%m-%d")),
|
23 |
+
Input(type="date", name="end_date", value=end_date.strftime("%Y-%m-%d")),
|
24 |
+
cls="date-range"
|
25 |
+
)
|
26 |
+
|
27 |
+
def search_form(start_date, end_date):
|
28 |
+
return Form(
|
29 |
+
Input(type="text", name="query", placeholder="Enter search query"),
|
30 |
+
date_range_inputs(start_date, end_date),
|
31 |
+
Button("Search", type="submit"),
|
32 |
+
hx_post="/search",
|
33 |
+
hx_target="#search-results",
|
34 |
+
hx_trigger="submit"
|
35 |
+
)
|
36 |
+
|
37 |
+
def iso_to_unix_timestamp(iso_string):
|
38 |
+
dt = datetime.fromisoformat(iso_string)
|
39 |
+
return int(dt.timestamp())
|
40 |
+
|
41 |
+
def unix_timestamp_to_iso(timestamp):
|
42 |
+
dt = datetime.fromtimestamp(timestamp)
|
43 |
+
return dt.isoformat()
|
44 |
+
|
45 |
+
def make_query(query, start_date, end_date, page=1, limit=10):
|
46 |
+
url = f"{API_URL}/indexes/comments/search"
|
47 |
+
headers = {
|
48 |
+
"Content-Type": "application/json",
|
49 |
+
"Authorization": f"Bearer {API_KEY}",
|
50 |
+
}
|
51 |
+
|
52 |
+
after_timestamp = iso_to_unix_timestamp(start_date)
|
53 |
+
before_timestamp = iso_to_unix_timestamp(end_date)
|
54 |
+
|
55 |
+
query = {
|
56 |
+
"q": query,
|
57 |
+
"limit": limit,
|
58 |
+
"offset": (page - 1) * limit,
|
59 |
+
"filter": f"comment_updatedAt_timestamp >= {after_timestamp} AND comment_updatedAt_timestamp < {before_timestamp}",
|
60 |
+
"attributesToCrop": ['comment_text'],
|
61 |
+
"cropLength": 30,
|
62 |
+
"attributesToHighlight": ["comment_text", "discussion_title"],
|
63 |
+
"highlightPreTag": "<span class=\"highlight\">",
|
64 |
+
"highlightPostTag": "</span>"
|
65 |
+
}
|
66 |
+
|
67 |
+
response = requests.post(url, headers=headers, json=query)
|
68 |
+
|
69 |
+
return response.json()
|
70 |
+
|
71 |
+
def search_results(query, start_date, end_date, page=1):
|
72 |
+
raw_results = make_query(query, start_date, end_date, page)
|
73 |
+
|
74 |
+
return Div(
|
75 |
+
make_results_bar(raw_results),
|
76 |
+
Div(*[make_card(r) for r in raw_results["hits"]]),
|
77 |
+
make_pagination(query, start_date, end_date, page, raw_results["estimatedTotalHits"]),
|
78 |
+
id="search-results"
|
79 |
+
)
|
80 |
+
|
81 |
+
def make_results_bar(results):
|
82 |
+
processing_time = results["processingTimeMs"]
|
83 |
+
estimated_hits = results["estimatedTotalHits"]
|
84 |
+
return Div(
|
85 |
+
Div(f"Processing time: {processing_time}ms"),
|
86 |
+
Div(f"Estimated total hits: {estimated_hits}"),
|
87 |
+
cls="results-bar"
|
88 |
+
)
|
89 |
+
|
90 |
+
def make_card(result):
|
91 |
+
result = result["_formatted"]
|
92 |
+
url = f"https://hf.co/{result['repo_id']}/discussions/{result['discussion_num']}"
|
93 |
+
date = unix_timestamp_to_iso(int(result["comment_updatedAt_timestamp"]))
|
94 |
+
|
95 |
+
return Div(
|
96 |
+
Div(
|
97 |
+
Strong(NotStr(result["discussion_title"])),
|
98 |
+
P(NotStr(result["comment_text"]), cls="comment-text"),
|
99 |
+
Span(date),
|
100 |
+
A(url, href=url, target="_blank"),
|
101 |
+
),
|
102 |
+
cls="card-item"
|
103 |
+
)
|
104 |
+
|
105 |
+
def make_pagination(query, start_date, end_date, current_page, total_hits, limit=10):
|
106 |
+
total_pages = -(-total_hits // limit) # Ceiling division
|
107 |
+
|
108 |
+
pagination = Div(cls="pagination")
|
109 |
+
|
110 |
+
if current_page > 1:
|
111 |
+
pagination.children += tuple(
|
112 |
+
Button("Previous",
|
113 |
+
hx_post=f"/search?page={current_page-1}",
|
114 |
+
hx_target="#search-results",
|
115 |
+
hx_include="[name='query'], [name='start_date'], [name='end_date']")
|
116 |
+
)
|
117 |
+
|
118 |
+
pagination.children += tuple(Span(f"Page {current_page} of {total_pages}"))
|
119 |
+
|
120 |
+
if current_page < total_pages:
|
121 |
+
pagination.children += tuple(
|
122 |
+
Button("Next",
|
123 |
+
hx_post=f"/search?page={current_page+1}",
|
124 |
+
hx_target="#search-results",
|
125 |
+
hx_include="[name='query'], [name='start_date'], [name='end_date']",
|
126 |
+
|
127 |
+
)
|
128 |
+
)
|
129 |
+
|
130 |
+
return pagination
|
131 |
+
|
132 |
+
@rt("/")
|
133 |
+
def get():
|
134 |
+
end_date = datetime.now()
|
135 |
+
start_date = end_date - timedelta(days=7)
|
136 |
+
return Titled(
|
137 |
+
"HF Discussion Search",
|
138 |
+
Div(
|
139 |
+
search_form(start_date, end_date),
|
140 |
+
Div(id="search-results"),
|
141 |
+
cls="container"
|
142 |
+
)
|
143 |
+
)
|
144 |
+
|
145 |
+
@rt("/search")
|
146 |
+
def post(query: str, start_date: str, end_date: str, page: int = 1):
|
147 |
+
return search_results(query, start_date, end_date, page)
|
148 |
+
|
149 |
+
serve()
|