Buckets:

download
raw
1.6 kB
from __future__ import annotations
from typing import TYPE_CHECKING, Any
if TYPE_CHECKING:
from chunker.types import CodeChunk
def _normalize(value: float, low: float, high: float) -> float:
if high <= low:
return 0.0
value = max(low, min(high, value))
return (value - low) / (high - low)
def compute_pack_hint(chunk: CodeChunk) -> float:
"""Compute a normalized [0,1] priority score for context packing.
Inputs considered:
- token_count (prefer smaller chunks)
- complexity (prefer higher complexity)
- degree (xref degree if present)
- recent_changes (optional recency weight)
The heuristic balances factors; callers can refine later.
"""
md: dict[str, Any] = chunk.metadata or {}
token_count = md.get("token_count") or 0
complexity = (md.get("complexity") or {}).get("cyclomatic") or 0
degree = md.get("degree") or 0
recent_changes = md.get("recent_changes") or 0
priority_bonus = md.get("priority") or 0
# Smaller token count preferred → invert
token_component = 1.0 - _normalize(float(token_count), 0.0, 2000.0)
complexity_component = _normalize(float(complexity), 0.0, 20.0)
degree_component = _normalize(float(degree), 0.0, 50.0)
recency_component = _normalize(float(recent_changes), 0.0, 100.0)
# Weighted sum
score = (
0.40 * token_component
+ 0.25 * complexity_component
+ 0.20 * degree_component
+ 0.10 * recency_component
+ 0.05 * _normalize(float(priority_bonus), 0.0, 1.0)
)
# Clamp
return max(0.0, min(1.0, score))

Xet Storage Details

Size:
1.6 kB
·
Xet hash:
cf37957589309a1b4b67dedf3c454fde49170a0c0d03c689fe5d355ea70f42cd

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.