app/services/commit_signal_patterns.py

"""Deterministic phrase-pattern matchers used to corroborate LLM-proposed
attribution and scope for Commit memory proposals. An LLM claiming
attribution="explicit_student" or scope="global" on its own is not
sufficient — these patterns are checked against the verified quote text
before either claim is trusted. See docs/commit-memory-adaptation-architecture.md.
"""
from __future__ import annotations

import re

_EXPLICIT_SIGNAL_PATTERNS = (
    re.compile(r"\bi prefer\b", re.IGNORECASE),
    re.compile(r"\bi learn better\b", re.IGNORECASE),
    re.compile(r"\bi already know\b", re.IGNORECASE),
    re.compile(r"\bi struggle with\b", re.IGNORECASE),
    re.compile(r"\bi do not understand\b", re.IGNORECASE),
    re.compile(r"\bi don't understand\b", re.IGNORECASE),
    re.compile(r"\bdo not give me\b", re.IGNORECASE),
    re.compile(r"\bdon't give me\b", re.IGNORECASE),
    re.compile(r"\bmy goal is\b", re.IGNORECASE),
    re.compile(r"\bremember that\b", re.IGNORECASE),
    re.compile(r"\bfrom now on\b", re.IGNORECASE),
)

_GLOBAL_SIGNAL_PATTERNS = (
    re.compile(r"\bi generally prefer\b", re.IGNORECASE),
    re.compile(r"\bi always prefer\b", re.IGNORECASE),
    re.compile(r"\bi usually prefer\b", re.IGNORECASE),
    re.compile(r"\bi learn better when\b", re.IGNORECASE),
    re.compile(r"\bfrom now on\b", re.IGNORECASE),
    re.compile(r"\balways\b", re.IGNORECASE),
    re.compile(r"\bnever\b", re.IGNORECASE),
)


def is_explicit_student_signal(text: str) -> bool:
    return any(pattern.search(text) for pattern in _EXPLICIT_SIGNAL_PATTERNS)


def is_explicit_global_signal(text: str) -> bool:
    return any(pattern.search(text) for pattern in _GLOBAL_SIGNAL_PATTERNS)