Files
SIA/sia/util.py
2024-10-24 08:39:31 +02:00

69 lines
2.3 KiB
Python

from typing import Iterator, TypeVar
import re
import xml.etree.ElementTree as ET
from .inference_result import InferenceResult
def get_valid_root_elements(schema: str) -> set:
"""
Extract valid root element names from the XML schema.
Args:
schema: XML schema string
Returns:
set: Set of valid root element names
"""
try:
schema = schema.strip()
ET.register_namespace('xs', 'http://www.w3.org/2001/XMLSchema')
root = ET.fromstring(schema)
ns = {'xs': 'http://www.w3.org/2001/XMLSchema'}
elements = root.findall(".//xs:element", ns)
return {elem.get('name') for elem in elements if elem.get('name')}
except ET.ParseError as e:
print(f"Error parsing schema: {e}")
return set()
def split_response(response: str, valid_elements: set) -> InferenceResult:
"""
Split the response into reasoning and actions based on valid XML elements.
Args:
response: Raw response string from the model
valid_elements: Set of valid root element names from the schema
Returns:
InferenceResult: Tuple containing reasoning and actions
"""
elements_pattern = '|'.join(map(re.escape, valid_elements))
pattern = f"<({elements_pattern})[^>]*>"
matches = list(re.finditer(pattern, response))
if not matches:
return InferenceResult(response.strip(), "")
last_match = matches[-1]
split_point = last_match.start()
reasoning = response[:split_point].strip()
actions = response[split_point:].strip()
return InferenceResult(reasoning, actions)
def stop_before_value(iterator: Iterator[str], stop_value: str) -> Iterator[str]:
"""
Creates an iterator that yields values from the input iterator
until it encounters the stop_value (exclusive).
Args:
iterator: The source iterator
stop_value: The value to stop before
Yields:
Values from the iterator until stop_value is encountered
If stop_value is part of an item, yields the part before stop_value
"""
for item in iterator:
if stop_value in item:
split_point = item.index(stop_value)
if split_point > 0:
yield item[:split_point]
break
yield item