Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -2,15 +2,15 @@ import streamlit as st
|
|
2 |
import pandas as pd
|
3 |
import xml.etree.ElementTree as ET
|
4 |
import requests
|
5 |
-
from datetime import datetime
|
6 |
-
import
|
|
|
7 |
import hmac
|
8 |
import os
|
9 |
import time
|
10 |
from PIL import Image
|
11 |
-
from
|
12 |
|
13 |
-
import re
|
14 |
|
15 |
TREND_TOPICS = {
|
16 |
1: "Autos and Vehicles",
|
@@ -34,12 +34,29 @@ TREND_TOPICS = {
|
|
34 |
19: "Travel and Transportation"
|
35 |
}
|
36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
def parse_url(url):
|
38 |
response = requests.get(url)
|
39 |
|
40 |
root = ET.fromstring(response.content)
|
41 |
return root
|
42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
def convert_into_dict(req_json):
|
44 |
result = {}
|
45 |
|
@@ -117,69 +134,15 @@ def find_details2(req_json):
|
|
117 |
if 'reset' not in st.session_state:
|
118 |
st.session_state.reset = False
|
119 |
|
120 |
-
def display_trends_from_yesterday():
|
121 |
-
checkbox_statuses = {}
|
122 |
-
urls = []
|
123 |
-
|
124 |
-
timezone = 'Europe/Vienna'
|
125 |
-
today = datetime.now(pytz.timezone(timezone)).date()
|
126 |
-
feed = parse_url(feed_url1)
|
127 |
-
entries = []
|
128 |
-
ns = {'ht': 'https://trends.google.de/trends/trendingsearches/daily'} # Define namespace
|
129 |
-
for item in feed.findall('.//item'):
|
130 |
-
pubDate = datetime.strptime(item.find('pubDate').text, '%a, %d %b %Y %H:%M:%S %z').date()
|
131 |
-
# Filter: Überspringe, wenn pubDate heute ist
|
132 |
-
if pubDate == today:
|
133 |
-
continue
|
134 |
-
entry = {
|
135 |
-
'title': item.find('title').text,
|
136 |
-
'pubDate': item.find('pubDate').text,
|
137 |
-
'approx_traffic': item.find('ht:approx_traffic', ns).text if item.find('ht:approx_traffic', ns) is not None else None,
|
138 |
-
'news_items': []
|
139 |
-
}
|
140 |
-
for news_item in item.findall('ht:news_item', ns):
|
141 |
-
news_details = {
|
142 |
-
'title': news_item.find('ht:news_item_title', ns).text,
|
143 |
-
'snippet': news_item.find('ht:news_item_snippet', ns).text,
|
144 |
-
'url': news_item.find('ht:news_item_url', ns).text,
|
145 |
-
'source': news_item.find('ht:news_item_source', ns).text
|
146 |
-
}
|
147 |
-
entry['news_items'].append(news_details)
|
148 |
-
entries.append(entry)
|
149 |
-
|
150 |
-
count = 1
|
151 |
-
for entry in entries:
|
152 |
-
with st.expander(f"{count}• {entry['title']} | Generated Traffic: {entry['approx_traffic']}"):
|
153 |
-
st.write(f"Veröffentlichungsdatum : {entry['pubDate']}")
|
154 |
-
for count2, link in enumerate(entry['news_items'], start=1):
|
155 |
-
checkbox_label = f"yesterday_{count}_{count2}"
|
156 |
-
if st.session_state.reset:
|
157 |
-
st.session_state[checkbox_label] = False
|
158 |
-
checkbox_statuses[checkbox_label] = st.session_state.get(checkbox_label, False)
|
159 |
-
checkbox_statuses[checkbox_label] = st.checkbox(
|
160 |
-
f"{count2}• {link['title']} | {link['source']} | [Go To →]({link['url']})",
|
161 |
-
value=checkbox_statuses[checkbox_label],
|
162 |
-
key=checkbox_label
|
163 |
-
)
|
164 |
-
if checkbox_statuses[checkbox_label]:
|
165 |
-
urls.append(link['url'])
|
166 |
-
|
167 |
-
# Button am Ende des Expanders
|
168 |
-
base_url = os.getenv("url")
|
169 |
-
query_params = "&".join([f"article-links[]={url}" for url in urls])
|
170 |
-
full_url = f"{base_url}{query_params}"
|
171 |
-
st.link_button("Open All Links" , url= full_url)
|
172 |
-
count += 1
|
173 |
-
|
174 |
-
|
175 |
|
176 |
# Function to display articles for a specific category
|
177 |
def display_articles_for_category(category):
|
178 |
checkbox_statuses = {}
|
179 |
urls = []
|
180 |
|
181 |
-
trending_data = st.session_state["real_trending_searches"][selected_country][
|
182 |
#st.write(trending_data)
|
|
|
183 |
if st.session_state.get("reset", False):
|
184 |
for idx, (topic, data) in enumerate(trending_data.items()):
|
185 |
for article_index, _ in enumerate(data["articles"]):
|
@@ -188,8 +151,12 @@ def display_articles_for_category(category):
|
|
188 |
|
189 |
for idx, (topic, data) in enumerate(trending_data.items()):
|
190 |
|
|
|
191 |
|
192 |
-
|
|
|
|
|
|
|
193 |
|
194 |
for article_index, article in enumerate(data["articles"], start=1):
|
195 |
checkbox_label = f"{category}_{idx}_{article_index}"
|
@@ -201,8 +168,12 @@ def display_articles_for_category(category):
|
|
201 |
|
202 |
disabled = (not current_value) and (sum(checkbox_statuses.values()) >= MAX_CHECKED)
|
203 |
|
|
|
|
|
|
|
|
|
204 |
checkbox_statuses[checkbox_label] = st.checkbox(
|
205 |
-
f"{
|
206 |
value=current_value,
|
207 |
key=checkbox_label,
|
208 |
disabled=disabled
|
@@ -216,39 +187,6 @@ def display_articles_for_category(category):
|
|
216 |
full_url = f"{base_url}{query_params}"
|
217 |
st.link_button("Open All Links", url=full_url)
|
218 |
|
219 |
-
# Funktion zum Rendern von Artikeln für heute
|
220 |
-
def display_articles_for_today(count, index):
|
221 |
-
checkbox_statuses = {}
|
222 |
-
urls = []
|
223 |
-
# Dictionary zur Verwaltung des Status jeder Checkbox
|
224 |
-
for count2, url in enumerate(index['articles'], start=1):
|
225 |
-
checkbox_label = f"today_{count}_{count2}"
|
226 |
-
if st.session_state.reset:
|
227 |
-
st.session_state[checkbox_label] = False
|
228 |
-
checkbox_statuses[checkbox_label] = st.session_state.get(checkbox_label, False)
|
229 |
-
|
230 |
-
|
231 |
-
with st.expander(f"{count+1}• {index['title']['query']} | Generated Traffic: {index['formattedTraffic']}"):
|
232 |
-
articles = find_details2(index['articles'])
|
233 |
-
for count2, url in enumerate(articles, start=1):
|
234 |
-
checkbox_label = f"today_{count}_{count2}"
|
235 |
-
disabled = not checkbox_statuses[checkbox_label] and sum(checkbox_statuses.values()) >= MAX_CHECKED
|
236 |
-
checkbox_statuses[checkbox_label] = st.checkbox(
|
237 |
-
f"{count2}• {url['articleTitle']} | {url['source']} | [Go To →]({url['url']})",
|
238 |
-
value=checkbox_statuses[checkbox_label],
|
239 |
-
key=checkbox_label,
|
240 |
-
disabled=disabled
|
241 |
-
)
|
242 |
-
if checkbox_statuses[checkbox_label]:
|
243 |
-
urls.append(url['url'])
|
244 |
-
|
245 |
-
# Button am Ende des Expanders
|
246 |
-
base_url = os.getenv("url")
|
247 |
-
query_params = "&".join([f"article-links[]={url}" for url in urls])
|
248 |
-
full_url = f"{base_url}{query_params}"
|
249 |
-
st.link_button("Open All Links" , url= full_url)
|
250 |
-
|
251 |
-
|
252 |
country_list = {
|
253 |
"Germamy" : "DE",
|
254 |
"Austria" : "AT"
|
@@ -259,16 +197,24 @@ if 'base_load_finished' not in st.session_state:
|
|
259 |
st.session_state["base_data"] = {}
|
260 |
st.session_state["pn"] = "AT"
|
261 |
if 'base_load_finished' not in st.session_state or st.session_state.reset:
|
|
|
262 |
with st.spinner("Loading Trends"):
|
263 |
st.session_state["today"] = {}
|
264 |
st.session_state["base"] = {}
|
265 |
for country_name, pn_option in country_list.items():
|
266 |
st.session_state["base_data"][pn_option] = {}
|
267 |
st.session_state["real_trending_searches"][pn_option] = {}
|
268 |
-
|
269 |
-
|
270 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
271 |
st.session_state["base_load_finished"]= True
|
|
|
272 |
|
273 |
|
274 |
MAX_CHECKED = 3
|
@@ -328,42 +274,60 @@ fixed_order = [
|
|
328 |
if 'selected_option' not in st.session_state:
|
329 |
st.session_state['selected_option'] = "default_value"
|
330 |
|
|
|
331 |
img = Image.open(r"heute_tensora.png")
|
332 |
st.sidebar.image(img)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
333 |
|
334 |
# Selectbox to choose a country
|
335 |
selected_country = st.sidebar.selectbox("Choose a Country", ["AT", "DE"])
|
336 |
-
feed_url1 = f'https://trends.google.de/trends/trendingsearches/daily/rss?geo={selected_country}'
|
337 |
|
338 |
-
# Button to trigger actions
|
339 |
-
if st.sidebar.button("Change Country"):
|
340 |
-
if selected_country == "AT":
|
341 |
-
st.session_state["pn"] = selected_country
|
342 |
-
elif selected_country == "DE":
|
343 |
-
st.session_state["pn"] = selected_country
|
344 |
|
345 |
-
selected_option = st.sidebar.radio("Choose an option", ["Realzeit Anfragen", "Trends von Gestern"])#, "Tagesaktuelle Anfragen"
|
346 |
st.warning("Die aufgelisteten Keywörter für erhöhte Reichweite in den Überschriften verwenden")
|
347 |
|
348 |
-
|
349 |
-
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
-
|
354 |
-
|
355 |
-
|
356 |
-
|
357 |
-
|
358 |
-
|
359 |
-
|
360 |
-
|
361 |
-
|
362 |
-
|
363 |
-
|
364 |
-
|
365 |
-
|
366 |
-
|
|
|
|
|
367 |
|
368 |
if st.session_state.reset:
|
369 |
st.session_state["reset"] = False
|
|
|
2 |
import pandas as pd
|
3 |
import xml.etree.ElementTree as ET
|
4 |
import requests
|
5 |
+
from datetime import datetime, timedelta
|
6 |
+
from streamlit_autorefresh import st_autorefresh
|
7 |
+
import json
|
8 |
import hmac
|
9 |
import os
|
10 |
import time
|
11 |
from PIL import Image
|
12 |
+
from azure.cosmos import CosmosClient
|
13 |
|
|
|
14 |
|
15 |
TREND_TOPICS = {
|
16 |
1: "Autos and Vehicles",
|
|
|
34 |
19: "Travel and Transportation"
|
35 |
}
|
36 |
|
37 |
+
|
38 |
+
client = CosmosClient(os.getenv("COSMOS_ENDPOINT"), os.getenv("COSMOS_KEY"))
|
39 |
+
database = client.get_database_client(os.getenv("DATABASE_NAME"))
|
40 |
+
container = database.get_container_client(os.getenv("CONTAINER_NAME"))
|
41 |
+
|
42 |
+
|
43 |
+
|
44 |
def parse_url(url):
|
45 |
response = requests.get(url)
|
46 |
|
47 |
root = ET.fromstring(response.content)
|
48 |
return root
|
49 |
|
50 |
+
|
51 |
+
def get_latest_trend(country: str):
|
52 |
+
"""
|
53 |
+
Gibt das neueste Dokument für das angegebene Land zurück,
|
54 |
+
basierend auf dem timestamp-Feld.
|
55 |
+
"""
|
56 |
+
query = f"SELECT * FROM c WHERE c.country = '{country}' ORDER BY c.timestamp DESC"
|
57 |
+
items = list(container.query_items(query=query, enable_cross_partition_query=True))
|
58 |
+
return items[0] if items else None
|
59 |
+
|
60 |
def convert_into_dict(req_json):
|
61 |
result = {}
|
62 |
|
|
|
134 |
if 'reset' not in st.session_state:
|
135 |
st.session_state.reset = False
|
136 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
137 |
|
138 |
# Function to display articles for a specific category
|
139 |
def display_articles_for_category(category):
|
140 |
checkbox_statuses = {}
|
141 |
urls = []
|
142 |
|
143 |
+
trending_data = st.session_state["real_trending_searches"][selected_country][category]
|
144 |
#st.write(trending_data)
|
145 |
+
active_string = " | :chart_with_upwards_trend: :green[Aktiver Trend]"
|
146 |
if st.session_state.get("reset", False):
|
147 |
for idx, (topic, data) in enumerate(trending_data.items()):
|
148 |
for article_index, _ in enumerate(data["articles"]):
|
|
|
151 |
|
152 |
for idx, (topic, data) in enumerate(trending_data.items()):
|
153 |
|
154 |
+
with st.expander(f"{idx + 1}• {topic}" + (active_string if not data["is_trend_finished"] else "")):
|
155 |
|
156 |
+
if data["related_queries"]:
|
157 |
+
st.markdown("**Related Keywords:**")
|
158 |
+
related_queries_string = f"{', '.join(data['related_queries'])}"
|
159 |
+
st.markdown(related_queries_string)
|
160 |
|
161 |
for article_index, article in enumerate(data["articles"], start=1):
|
162 |
checkbox_label = f"{category}_{idx}_{article_index}"
|
|
|
168 |
|
169 |
disabled = (not current_value) and (sum(checkbox_statuses.values()) >= MAX_CHECKED)
|
170 |
|
171 |
+
dt = datetime.fromisoformat(article["time"])
|
172 |
+
formatted_dt = dt.strftime("%d.%m.%Y %H:%M")
|
173 |
+
st.markdown(f"{article_index}• {article['source']} / {formatted_dt}:")
|
174 |
+
|
175 |
checkbox_statuses[checkbox_label] = st.checkbox(
|
176 |
+
f"{article['title']} | [Go To →]({article['href']})",
|
177 |
value=current_value,
|
178 |
key=checkbox_label,
|
179 |
disabled=disabled
|
|
|
187 |
full_url = f"{base_url}{query_params}"
|
188 |
st.link_button("Open All Links", url=full_url)
|
189 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
190 |
country_list = {
|
191 |
"Germamy" : "DE",
|
192 |
"Austria" : "AT"
|
|
|
197 |
st.session_state["base_data"] = {}
|
198 |
st.session_state["pn"] = "AT"
|
199 |
if 'base_load_finished' not in st.session_state or st.session_state.reset:
|
200 |
+
|
201 |
with st.spinner("Loading Trends"):
|
202 |
st.session_state["today"] = {}
|
203 |
st.session_state["base"] = {}
|
204 |
for country_name, pn_option in country_list.items():
|
205 |
st.session_state["base_data"][pn_option] = {}
|
206 |
st.session_state["real_trending_searches"][pn_option] = {}
|
207 |
+
with open(f"data_{pn_option}.json", "r") as file:
|
208 |
+
data = json.load(file)
|
209 |
+
try:
|
210 |
+
st.session_state["real_trending_searches"][pn_option] = get_latest_trend(pn_option).get("trends")
|
211 |
+
except:
|
212 |
+
st.warning("No data found in the Trendcrawl"
|
213 |
+
)
|
214 |
+
st.stop()
|
215 |
+
st.session_state["_"] = 0
|
216 |
st.session_state["base_load_finished"]= True
|
217 |
+
st.session_state["start_time"] = datetime.now()
|
218 |
|
219 |
|
220 |
MAX_CHECKED = 3
|
|
|
274 |
if 'selected_option' not in st.session_state:
|
275 |
st.session_state['selected_option'] = "default_value"
|
276 |
|
277 |
+
|
278 |
img = Image.open(r"heute_tensora.png")
|
279 |
st.sidebar.image(img)
|
280 |
+
elapsed_time = datetime.now() - st.session_state["start_time"]
|
281 |
+
if elapsed_time > timedelta(minutes=10):
|
282 |
+
st.markdown(
|
283 |
+
"""
|
284 |
+
<style>
|
285 |
+
/* Zentriert den Button und vergrößert ihn */
|
286 |
+
div.stButton > button {
|
287 |
+
width: 300px;
|
288 |
+
height: 70px;
|
289 |
+
font-size: 20px;
|
290 |
+
margin: auto;
|
291 |
+
display: block;
|
292 |
+
}
|
293 |
+
</style>
|
294 |
+
""",
|
295 |
+
unsafe_allow_html=True
|
296 |
+
)
|
297 |
+
|
298 |
+
if st.button(f"Letzte Aktualisierung {st.session_state['start_time'].strftime('%H:%M')}\n- Für Update Hier Klicken"):
|
299 |
+
st.session_state["reset"] = True
|
300 |
+
st.session_state["base_load_finished"] = False
|
301 |
+
st.rerun()
|
302 |
+
|
303 |
|
304 |
# Selectbox to choose a country
|
305 |
selected_country = st.sidebar.selectbox("Choose a Country", ["AT", "DE"])
|
|
|
306 |
|
|
|
|
|
|
|
|
|
|
|
|
|
307 |
|
|
|
308 |
st.warning("Die aufgelisteten Keywörter für erhöhte Reichweite in den Überschriften verwenden")
|
309 |
|
310 |
+
st.warning("Der Hinweis „Aktiver Trend“ signalisiert, dass zurzeit signifikant mehr Suchanfragen zu diesem Thema erfolgen. Bitte beachten Sie, dass auch andere Themen weiterhin von Relevanz sind.")
|
311 |
+
|
312 |
+
available_topics = st.session_state["real_trending_searches"][selected_country].keys()
|
313 |
+
filtered_topics = [topic for topic in TREND_TOPICS.values() if topic in available_topics]
|
314 |
+
all_topics = ["All categories"] + filtered_topics
|
315 |
+
auswahl = st.selectbox("Select Ressort", all_topics, index=0)
|
316 |
+
|
317 |
+
display_articles_for_category(auswahl)
|
318 |
+
|
319 |
+
if st.session_state["base_load_finished"]:
|
320 |
+
st.session_state["_"] = st_autorefresh(interval=600000, limit=5, key=f"autorefresh_{st.session_state['start_time']}")
|
321 |
+
st.markdown(
|
322 |
+
"""
|
323 |
+
<style>
|
324 |
+
[data-testid="stToolbar"] {
|
325 |
+
display: none;
|
326 |
+
}
|
327 |
+
</style>
|
328 |
+
""",
|
329 |
+
unsafe_allow_html=True
|
330 |
+
)
|
331 |
|
332 |
if st.session_state.reset:
|
333 |
st.session_state["reset"] = False
|