"""Benchmark discovery pool.

A copy of ~/prog/fin/benchmarks.txt (the user's curated benchmark universe),
augmented with a few common index ETFs that exist in the data universe.
The ORIGINAL file outside this repository is never modified.

Format: section header lines (ending in ':' or free-standing), then
'symbol<tab or 2+ spaces>label' lines.
"""

from __future__ import annotations

import re
from dataclasses import dataclass
from pathlib import Path

POOL_FILE = Path(__file__).parent / "pool" / "benchmarks.txt"

_ENTRY = re.compile(r"^(\S{1,6})[\t ]{2,}(.+)$")


@dataclass(frozen=True)
class PoolEntry:
    section: str
    symbol: str      # lowercase
    label: str


def load_pool(path: Path = POOL_FILE) -> list[PoolEntry]:
    entries: list[PoolEntry] = []
    section = ""
    for line in path.read_text().splitlines():
        line = line.rstrip()
        if not line.strip():
            continue
        m = _ENTRY.match(line)
        if m:
            entries.append(PoolEntry(section, m.group(1).lower(),
                                     m.group(2).strip()))
        else:
            section = line.strip().rstrip(":")
    return entries


def pool_labels(entries: list[PoolEntry] | None = None) -> dict[str, str]:
    """symbol -> 'section: label' (for keyword matching and display)."""
    if entries is None:
        entries = load_pool()
    return {e.symbol: f"{e.section}: {e.label}" for e in entries if e.section}