#!/usr/bin/env python3
"""summarize.py — read every <app>/record.json of the sweep and print one review line per app: the measured facts and
a FIRST verdict hint. The hint is a filter for the reader, not the verdict: README.md carries the verdict after each
app's evidence was read."""
import glob, json, os, re, sys

EV = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
EN_IN_HU = re.compile(r"(?i)\b(the|and|your|you|please|click|with|this|password|settings|open|first|login)\b")


def hint(r):
    probs, broken = [], []
    inst = r.get("install") or {}
    if not inst.get("ok"):
        broken.append("install failed")
    elif not inst.get("front_door"):
        broken.append("front door never answered")
    if r.get("exception"):
        broken.append("cycle raised: " + r["exception"][:80])
    if r.get("fixture") and r.get("seed") is False:
        probs.append("seed failed")
    if r.get("seed") and r.get("seed_read_now") is False:
        probs.append("item not read back at once")
    back = r.get("seed_read_after_reinstall")
    if r.get("seed") and back is False and not r.get("seed_read_after_unit_restore"):
        probs.append("item NOT back after keep-data reinstall + unit restore")
    if r.get("seed_read_after_restore") is False:
        probs.append("item not back after the unit restore")
    if (r.get("kept") or {}).get("offered") and (r.get("kept") or {}).get("use_ok") is False:
        probs.append("use-my-kept-data failed")
    left = r.get("after_final_remove") or {}
    if any(left.get(k) for k in ("containers", "volumes", "appyaml", "appdata", "kept")):
        probs.append("leftovers after remove: " + json.dumps({k: v for k, v in left.items() if v and k != "subdomain"}))
    if left.get("subdomain") and not left["subdomain"].startswith("404"):
        probs.append("subdomain after remove: " + left["subdomain"][:40])
    for c, v in (r.get("containers") or {}).items():
        anon = (r.get("mem_anon_peak_mib") or {}).get(c)
        lim, peak = v.get("limit_mib"), anon if anon is not None else (r.get("mem_peak_mib") or {}).get(c)
        if lim and peak and peak > 0.9 * lim:
            probs.append(f"memory {c} {peak}/{lim} MiB ({'anon' if anon is not None else 'docker stats, may include cache'})")
        if v.get("restarts"):
            probs.append(f"restarts {c} {v['restarts']}")
        if v.get("health") == "unhealthy":
            probs.append(f"unhealthy {c}")
    errs = [l for l in (r.get("errors") or {}).get("sample", [])
            if not re.search(r"(?i)wrong (username|password)|invalid password|failed login|login fail|unauthori[sz]ed|status=40[13]|bad credentials", l)]
    n = (r.get("errors") or {}).get("count", 0) - (len((r.get("errors") or {}).get("sample", [])) - len(errs))
    if n > 0:   # the negative controls' own wrong-password logins are expected and not counted
        probs.append(f"error lines {n}")
    if r.get("http_port_80") and not r["http_port_80"].startswith(("301", "308")):
        probs.append("http :80 → " + r["http_port_80"][:30])
    for tab, fname in (("install_page", "install"), ("guide", "guide")):
        t = r.get(tab) or {}
        try:  # English OUTSIDE quotes: a quoted „Settings” is the app's own label, glossed in Hungarian — correct
            txt = open(os.path.join(EV, r["app"], f"page-{fname}-hu.txt")).read()
            bare = re.sub(r"„[^”]*”|“[^”]*”|\"[^\"]*\"|\([^)]*\)", " ", txt)
            words = sorted(set(m.lower() for m in EN_IN_HU.findall(bare)))
        except OSError:
            words = t.get("english_words_in_hu") or []
        if words:
            probs.append(f"{tab} EN words in HU {words[:5]}")
        if t.get("accented_in_en"):
            probs.append(f"{tab} HU letters in EN {t['accented_in_en']}")
        if t.get("same_text"):
            probs.append(f"{tab} same text HU/EN")
    for s in (r.get("shots_install") or []) + (r.get("shots_after") or []):
        if "hscroll=True" in s:
            probs.append("sideways scroll in a screenshot")
        if "STEP FAILED" in s:
            probs.append("a screenshot step failed")
    v = "broken" if broken else ("problem?" if probs else "works")
    return v, broken + probs


def main():
    rows = []
    for f in sorted(glob.glob(os.path.join(EV, "*", "record.json"))):
        r = json.load(open(f))
        v, why = hint(r)
        inst = r.get("install") or {}
        peak = sum((r.get("mem_peak_mib") or {}).values())
        lim = sum((c.get("limit_mib") or 0) for c in (r.get("containers") or {}).values())
        rows.append((r["app"], v, round(peak), lim, inst.get("seconds_to_front_door"), r.get("fixture") or "-",
                     r.get("stranger", "")[:40], "; ".join(why)))
    for row in rows:
        print("\t".join(str(x) for x in row))
    print(f"# {len(rows)} records; " + ", ".join(f"{k} {sum(1 for r in rows if r[1] == k)}" for k in ("works", "problem?", "broken")),
          file=sys.stderr)


if __name__ == "__main__":
    main()