Update app.py
Browse files
app.py
CHANGED
|
@@ -305,22 +305,23 @@ def find_all(purpose,task,history, url):
|
|
| 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 |
-
|
| 312 |
-
c=0
|
| 313 |
-
out = str(out)
|
| 314 |
-
rl = len(out)
|
| 315 |
-
print(f'rl:: {rl}')
|
| 316 |
-
#for ea in out:
|
| 317 |
-
for i in str(out):
|
| 318 |
-
|
| 319 |
-
|
| 320 |
-
print (f'c:: {c}')
|
|
|
|
| 321 |
if rl > MAX_DATA:
|
| 322 |
print("compressing...")
|
| 323 |
-
rawp = compress_data(
|
| 324 |
print (rawp)
|
| 325 |
print (f'out:: {out}')
|
| 326 |
history += "observation: the search results are:\n {}\n".format(out)
|
|
|
|
| 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([{p.name:p.string,"parent":p.parent.name,"previous":p.previous,"first-child":[b.name for b in p.children],"content":p}])
|
| 312 |
+
#c=0
|
| 313 |
+
#out = str(out)
|
| 314 |
+
#rl = len(out)
|
| 315 |
+
#print(f'rl:: {rl}')
|
| 316 |
+
##for ea in out:
|
| 317 |
+
#for i in str(out):
|
| 318 |
+
# if i == " " or i=="," or i=="\n":
|
| 319 |
+
# c +=1
|
| 320 |
+
#print (f'c:: {c}')
|
| 321 |
+
rl=len(rawp)
|
| 322 |
if rl > MAX_DATA:
|
| 323 |
print("compressing...")
|
| 324 |
+
rawp = compress_data(rl,purpose,task,out)
|
| 325 |
print (rawp)
|
| 326 |
print (f'out:: {out}')
|
| 327 |
history += "observation: the search results are:\n {}\n".format(out)
|