Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -5,47 +5,20 @@ import json
|
|
| 5 |
|
| 6 |
def wayback(website):
|
| 7 |
if not website:
|
| 8 |
-
return '
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
<style>
|
| 12 |
-
table {
|
| 13 |
-
width: 100%;
|
| 14 |
-
border-collapse: collapse;
|
| 15 |
-
}
|
| 16 |
-
th, td {
|
| 17 |
-
border: 1px solid #dddddd;
|
| 18 |
-
text-align: left;
|
| 19 |
-
padding: 8px;
|
| 20 |
-
}
|
| 21 |
-
th {
|
| 22 |
-
background-color: #f2f2f2;
|
| 23 |
-
}
|
| 24 |
-
tr:nth-child(even) {
|
| 25 |
-
background-color: #f9f9f9;
|
| 26 |
-
}
|
| 27 |
-
a {
|
| 28 |
-
color: #0645ad;
|
| 29 |
-
text-decoration: none;
|
| 30 |
-
}
|
| 31 |
-
a:hover {
|
| 32 |
-
text-decoration: underline;
|
| 33 |
-
}
|
| 34 |
-
</style>
|
| 35 |
-
"""
|
| 36 |
-
# Intentar primero con la Wayback CDX Server API
|
| 37 |
end_date = datetime.now()
|
| 38 |
-
start_date = end_date - timedelta(days=
|
| 39 |
datefrom = start_date.strftime('%Y%m%d')
|
| 40 |
dateto = end_date.strftime('%Y%m%d')
|
| 41 |
-
cdx_api_url = f"http://web.archive.org/cdx/search/cdx?url={website}&output=json&from={datefrom}&to={dateto}&limit=
|
| 42 |
|
| 43 |
try:
|
| 44 |
response = requests.get(cdx_api_url)
|
| 45 |
if response.status_code == 200:
|
| 46 |
content = json.loads(response.text)
|
| 47 |
if len(content) > 1:
|
| 48 |
-
# Procesar y mostrar los resultados
|
| 49 |
results = []
|
| 50 |
for row in content[1:]:
|
| 51 |
date, page, status = [row[i] for i in [1, 2, 4]]
|
|
@@ -53,8 +26,8 @@ def wayback(website):
|
|
| 53 |
formatted_wayback_url = f"https://web.archive.org/web/{date}/{page}"
|
| 54 |
results.append({'date': date, 'formatted_date': formatted_date, 'link': formatted_wayback_url})
|
| 55 |
results.sort(key=lambda x: x['date'], reverse=True)
|
| 56 |
-
formatted_results = [f"
|
| 57 |
-
return "<
|
| 58 |
except Exception as e:
|
| 59 |
pass # Falla silenciosa, intentar con el siguiente método
|
| 60 |
|
|
@@ -70,7 +43,7 @@ def wayback(website):
|
|
| 70 |
snapshot_url = closest_snapshot["url"]
|
| 71 |
timestamp = closest_snapshot["timestamp"]
|
| 72 |
formatted_date = datetime.strptime(timestamp, '%Y%m%d%H%M%S').strftime('%d/%m/%Y')
|
| 73 |
-
return f
|
| 74 |
except Exception as e:
|
| 75 |
return f"<p>😭 Error: {e}</p>"
|
| 76 |
|
|
|
|
| 5 |
|
| 6 |
def wayback(website):
|
| 7 |
if not website:
|
| 8 |
+
return '😭 Error: introduce una url correcta'
|
| 9 |
+
|
| 10 |
+
# Intenta primero con la Wayback CDX Server API
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
end_date = datetime.now()
|
| 12 |
+
start_date = end_date - timedelta(days=365)
|
| 13 |
datefrom = start_date.strftime('%Y%m%d')
|
| 14 |
dateto = end_date.strftime('%Y%m%d')
|
| 15 |
+
cdx_api_url = f"http://web.archive.org/cdx/search/cdx?url={website}&output=json&from={datefrom}&to={dateto}&limit=3000"
|
| 16 |
|
| 17 |
try:
|
| 18 |
response = requests.get(cdx_api_url)
|
| 19 |
if response.status_code == 200:
|
| 20 |
content = json.loads(response.text)
|
| 21 |
if len(content) > 1:
|
|
|
|
| 22 |
results = []
|
| 23 |
for row in content[1:]:
|
| 24 |
date, page, status = [row[i] for i in [1, 2, 4]]
|
|
|
|
| 26 |
formatted_wayback_url = f"https://web.archive.org/web/{date}/{page}"
|
| 27 |
results.append({'date': date, 'formatted_date': formatted_date, 'link': formatted_wayback_url})
|
| 28 |
results.sort(key=lambda x: x['date'], reverse=True)
|
| 29 |
+
formatted_results = [f"👓 {item['formatted_date']} <a href='{item['link']}' target='_blank'>{item['link']}</a>" for item in results]
|
| 30 |
+
return "<br>".join(formatted_results)
|
| 31 |
except Exception as e:
|
| 32 |
pass # Falla silenciosa, intentar con el siguiente método
|
| 33 |
|
|
|
|
| 43 |
snapshot_url = closest_snapshot["url"]
|
| 44 |
timestamp = closest_snapshot["timestamp"]
|
| 45 |
formatted_date = datetime.strptime(timestamp, '%Y%m%d%H%M%S').strftime('%d/%m/%Y')
|
| 46 |
+
return f"👓 Instantánea más reciente: {formatted_date} <a href='{snapshot_url}' target='_blank'>{snapshot_url}</a>"
|
| 47 |
except Exception as e:
|
| 48 |
return f"<p>😭 Error: {e}</p>"
|
| 49 |
|