DU / llm_parsing.py
Dmitry Kirsanov
locally debuggable, lots of development
43e002c
Raw History Blame Contribute Delete
1.01 kB
import re
def parse_list(ret, what):
lines = ret.splitlines()
inlist = False
retlist = []
for line in lines:
line = line.strip()
if line == "":
inlist = False
elif line[-1] == ":":
if line[:-1].upper() == what.upper():
inlist = True
else:
inlist = False
elif inlist:
if line.startswith("- "):
item = line[2:].strip()
else:
item = line
retlist.append(item)
print(f"===Parsed list {what}: {retlist}")
return retlist
def extract_document_from_response(ret):
r = re.sub(r'^<derivative-document>(.*)</derivative-document>$', r'\1', ret, flags=re.DOTALL)
r = re.sub(r'^<revised-document>(.*)</revised-document>$', r'\1', r, flags=re.DOTALL)
r = re.sub(r'^<document>(.*)</document>$', r'\1', r, flags=re.DOTALL)
return r