Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -299,16 +299,17 @@ def find_all(purpose,task,history, url):
|
|
| 299 |
# get attributes:
|
| 300 |
print(soup.title.name)
|
| 301 |
# get values:
|
| 302 |
-
print(soup.title.string)
|
| 303 |
# beginning navigation:
|
| 304 |
-
print(soup.title.parent.name)
|
| 305 |
#rawp.append([tag.name for tag in soup.find_all()] )
|
| 306 |
-
print([tag.name for tag in soup.find_all()])
|
| 307 |
rawp=(f'RAW TEXT RETURNED: {soup.text}')
|
| 308 |
out.append(rawp)
|
|
|
|
| 309 |
q=("a","p","span","content","article")
|
| 310 |
for p in soup.find_all(q):
|
| 311 |
-
out.append([{
|
| 312 |
c=0
|
| 313 |
out = str(out)
|
| 314 |
rl = len(out)
|
|
|
|
| 299 |
# get attributes:
|
| 300 |
print(soup.title.name)
|
| 301 |
# get values:
|
| 302 |
+
#print(soup.title.string)
|
| 303 |
# beginning navigation:
|
| 304 |
+
#print(soup.title.parent.name)
|
| 305 |
#rawp.append([tag.name for tag in soup.find_all()] )
|
| 306 |
+
#print([tag.name for tag in soup.find_all()])
|
| 307 |
rawp=(f'RAW TEXT RETURNED: {soup.text}')
|
| 308 |
out.append(rawp)
|
| 309 |
+
out.append("HTML fragments: ")
|
| 310 |
q=("a","p","span","content","article")
|
| 311 |
for p in soup.find_all(q):
|
| 312 |
+
out.append([{"string":p.string,"content":p])
|
| 313 |
c=0
|
| 314 |
out = str(out)
|
| 315 |
rl = len(out)
|