Alex Galarce
674add9a3e
Feature/black and isort * black and isort constants * black and isort codes * black and isort crosswalks * black and isort investment * black and isort prompts * isort testbed * isort tracking * black and isort utils * poetry and black tests Approved-by: Katon Minhas
103 lines
3.8 KiB
Python
103 lines
3.8 KiB
Python
import logging
|
|
import re
|
|
|
|
import src.investment.prompt_calls as prompt_calls
|
|
from src.utils import string_utils
|
|
|
|
|
|
def has_explicit_lesser_of_carveout(reimb_term: str) -> bool:
|
|
"""Check for explicit 'lesser of' carve-out language in the reimbursement term.
|
|
|
|
This function uses a list of keywords to identify if the reimbursement term contains explicit language indicating a 'lesser of' carve-out.
|
|
|
|
Args:
|
|
reimb_term (str): The reimbursement term to check.
|
|
|
|
Returns:
|
|
bool: True if 'lesser of' carve-out language is found, False otherwise.
|
|
"""
|
|
|
|
carveout_patterns = ["not subject to.*lesser", "not bound by.*lesser"]
|
|
reimb_lower = reimb_term.lower()
|
|
return any(re.search(pattern, reimb_lower) for pattern in carveout_patterns)
|
|
|
|
|
|
def create_exhibit_lesser_of_row(
|
|
original_row: dict, exhibit_lesser_of: str, filename: str
|
|
) -> dict:
|
|
"""Create enhanced row with exhibit lesser-of constraint applied, avoiding redundancy with substring checks.
|
|
|
|
Args:
|
|
original_row (dict): Original row dictionary containing reimbursement details.
|
|
exhibit_lesser_of (str): Exhibit lesser-of statement to apply.
|
|
filename (str): The name of the file being processed, for logging.
|
|
|
|
Returns:
|
|
dict: Enhanced row dictionary with exhibit lesser-of constraint applied.
|
|
"""
|
|
enhanced_row = original_row.copy()
|
|
|
|
# Get original REIMB_TERM
|
|
original_reimb = original_row.get("REIMB_TERM", "")
|
|
|
|
# Skip if exhibit_lesser_of is empty or N/A
|
|
if not exhibit_lesser_of or string_utils.is_empty(exhibit_lesser_of):
|
|
return enhanced_row
|
|
|
|
# FIRST-ORDER DEDUPLICATION: Check if the exhibit constraint is already present (case-insensitive)
|
|
if exhibit_lesser_of.lower() in original_reimb.lower():
|
|
logging.debug(
|
|
f"Exhibit constraint '{exhibit_lesser_of}' already present in REIMB_TERM: {original_reimb}"
|
|
)
|
|
return enhanced_row # No need to modify if already present
|
|
|
|
# SECOND-ORDER DEDUPLICATION: Use LLM to check if the exhibit constraint is redundant
|
|
if prompt_calls.is_exhibit_lesser_of_redundant(
|
|
original_reimb, exhibit_lesser_of, filename
|
|
):
|
|
logging.debug(
|
|
f"LLM determined that exhibit constraint '{exhibit_lesser_of}' is semantically redundant with REIMB_TERM: {original_reimb}"
|
|
)
|
|
return enhanced_row # No need to modify if redundant
|
|
|
|
# Apply the exhibit lesser-of constraint
|
|
enhanced_reimb = f"lesser of ({original_reimb}) or ({exhibit_lesser_of})"
|
|
enhanced_row["REIMB_TERM"] = enhanced_reimb
|
|
|
|
logging.debug(
|
|
f"Enhanced REIMB_TERM with exhibit lesser-of: {enhanced_row['REIMB_TERM']}"
|
|
)
|
|
return enhanced_row
|
|
|
|
|
|
def apply_exhibit_lesser_of(
|
|
rows: list[dict], exhibit_lesser_of: str, filename: str
|
|
) -> list[dict]:
|
|
"""Apply exhibit lesser-of constraint to rows that don't explicitly have a lesser-of carve-out.
|
|
|
|
Args:
|
|
rows (list[dict]): The rows to which the constraint should be applied.
|
|
exhibit_lesser_of (str): The exhibit lesser-of statement to apply.
|
|
filename (str): The name of the file being processed, for logging.
|
|
|
|
Returns:
|
|
list[dict]: The rows with the exhibit lesser-of constraint applied.
|
|
"""
|
|
enhanced_rows = []
|
|
|
|
for row in rows:
|
|
reimb_term = row.get("REIMB_TERM", "")
|
|
|
|
# Check if row explicitly carves out from lesser-of
|
|
if has_explicit_lesser_of_carveout(reimb_term):
|
|
logging.debug(
|
|
f"Skipping exhibit lesser-of application for row with explicit carve-out: {row}"
|
|
)
|
|
enhanced_rows.append(row) # Keep as-is
|
|
else:
|
|
# Apply exhibit lesser-of to rows without explicit carve-out
|
|
row = create_exhibit_lesser_of_row(row, exhibit_lesser_of, filename)
|
|
enhanced_rows.append(row)
|
|
|
|
return enhanced_rows
|