#!/usr/bin/env python3
"""Absence — Appendix E, specification 6.

Establishes that something is NOT present. A harder claim than presence, and it needs
the search space stated explicitly. This script refuses to assert absence unless
coverage is complete: if any item could not be searched, the finding is reported as
INCONCLUSIVE rather than as absence.

  python3 absence.py --dir ./production --term "indemnification" --out absence.json

Exit codes: 0 absence established, 2 term present, 3 inconclusive (incomplete coverage).
"""
import argparse, json, os, platform, re, sys
from datetime import datetime, timezone

VERSION = "1.0"
TEXTLIKE = {".txt", ".csv", ".tsv", ".md", ".json", ".htm", ".html", ".xml", ".log", ".eml"}

def main():
    ap = argparse.ArgumentParser(description="Establish absence, with the search space stated.")
    ap.add_argument("--dir", required=True)
    ap.add_argument("--term", action="append", required=True)
    ap.add_argument("--regex", action="store_true")
    ap.add_argument("--case-sensitive", action="store_true")
    ap.add_argument("--out", required=True)
    a = ap.parse_args()

    if not os.path.isdir(a.dir):
        sys.exit(f"not a directory: {a.dir}")
    flags = 0 if a.case_sensitive else re.IGNORECASE
    pats = [(t, re.compile(t if a.regex else re.escape(t), flags)) for t in a.term]

    seen = searched = 0
    blocked, found = [], []
    for dp, dns, fns in os.walk(a.dir):
        dns.sort()
        for fn in sorted(fns):
            p = os.path.join(dp, fn); rel = os.path.relpath(p, a.dir).replace(os.sep, "/")
            seen += 1
            ext = os.path.splitext(fn)[1].lower()
            if ext not in TEXTLIKE:
                blocked.append({"item": rel, "reason": f"unsupported format {ext or '(none)'}"}); continue
            try:
                with open(p, encoding="utf-8", errors="strict") as f:
                    text = f.read()
            except (UnicodeDecodeError, OSError) as e:
                blocked.append({"item": rel, "reason": type(e).__name__}); continue
            if ext in {".htm", ".html", ".xml"}:
                text = re.sub(r"<[^>]+>", " ", text)
            searched += 1
            for term, rx in pats:
                if rx.search(text):
                    found.append({"item": rel, "term": term})

    complete = (searched == seen)
    if found:
        finding, code = "PRESENT", 2
    elif complete:
        finding, code = "ABSENT", 0
    else:
        finding, code = "INCONCLUSIVE", 3

    report = {
        "specification": "Absence (Appendix E, 6)",
        "script": os.path.basename(__file__), "script_version": VERSION,
        "run_utc": datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ"),
        "python": platform.python_version(),
        "finding": finding,
        "search_space": {"root": os.path.abspath(a.dir), "items_encountered": seen,
                         "items_searched": searched, "items_not_searched": len(blocked),
                         "coverage_complete": complete,
                         "formats_searchable": sorted(TEXTLIKE)},
        "terms": a.term,
        "occurrences": sorted(found, key=lambda h: (h["item"], h["term"])),
        "items_not_searched": blocked,
        "statement": {
            "ABSENT": "No occurrence was found, and every item in the stated search space was searched.",
            "PRESENT": "The term occurs. Absence cannot be asserted.",
            "INCONCLUSIVE": ("No occurrence was found in the items that could be searched, but some items "
                             "could not be. Absence is NOT established over the whole search space."),
        }[finding],
    }
    with open(a.out, "w", encoding="utf-8") as f:
        json.dump(report, f, indent=1); f.write("\n")
    print(f"{finding}: searched {searched}/{seen} -> {a.out}")
    print(report["statement"])
    sys.exit(code)

if __name__ == "__main__":
    main()