from pathlib import Path
import re

ROOT = Path(__file__).resolve().parents[2]
SCAN = [
    ROOT / "index.html",
    ROOT / "research",
    ROOT / "report",
    ROOT / "deck",
]
# Allow mentions that explicitly retire/forbid the figure
ALLOW = re.compile(
    r"(retir(e|ed|es|ing)|wrong|invalid|never|not .*895|mixed|anti-pattern|RETIRED)",
    re.I,
)
BAD = re.compile(
    r"895\s*MW(?!\s*\(retir)|delivering 895|895 MW of firm|portfolio.*895|−895 MW|-\s*895 MW",
    re.I,
)


def test_no_active_895_firm_claims():
    offenders = []
    for base in SCAN:
        if not base.exists():
            continue
        if base.is_file():
            paths = [base]
        else:
            paths = list(base.rglob("*.md")) + list(base.rglob("*.html"))
        for path in paths:
            # master.md/.html are intermediates the Typst build writes on its way
            # to the PDF. They re-wrap lines, which can separate a retired figure
            # from the word retiring it and trip this line-based check. The .typ
            # source and the published PDF are the surfaces that matter.
            if path.name in {"master.html", "master.md"}:
                continue
            if "__pycache__" in path.parts or "rendered" in path.parts or "dist" in path.parts:
                continue
            text = path.read_text(errors="ignore")
            for i, line in enumerate(text.splitlines(), 1):
                if BAD.search(line) and not ALLOW.search(line):
                    offenders.append(f"{path.relative_to(ROOT)}:{i}:{line.strip()[:120]}")
    assert offenders == [], "Retired 895 MW firm claims still active:\n" + "\n".join(offenders)