"""Replace records of a base JSONL with re-captured records (matched by id), keeping the base order.
Image paths of the replacements are rewritten relative to the output file's directory.

usage: python scripts/swap_records.py base.jsonl replacements.jsonl out.jsonl [--only-substr settings_destructive]
"""
import json, os, sys


def load(p):
    with open(p, encoding="utf-8") as f:
        return [json.loads(l) for l in f if l.strip()]


def main(base, repl, dst, only=None):
    rep = {r["id"]: r for r in load(repl)}
    rdir, odir = os.path.dirname(os.path.abspath(repl)), os.path.dirname(os.path.abspath(dst))
    n = swapped = 0
    used, unmatched = set(), []
    with open(dst, "w", encoding="utf-8", newline="\n") as out:
        for r in load(base):
            n += 1
            if r["id"] in rep:
                new = json.loads(json.dumps(rep[r["id"]]))
                new["images"] = [os.path.relpath(os.path.join(rdir, im), odir).replace("\\", "/") for im in new["images"]]
                missing = [im for im in new["images"] if not os.path.exists(os.path.join(odir, im))]
                if missing:
                    sys.exit(f"missing image {missing}")
                new["meta"]["swapped_from"] = os.path.relpath(repl, odir).replace("\\", "/")
                r = new
                used.add(r["id"])
                swapped += 1
            elif only and only in r["id"]:
                unmatched.append(r["id"])
            out.write(json.dumps(r, ensure_ascii=False) + "\n")
    print(f"{n} records, swapped={swapped}, replacements unused={len(rep) - len(used)}, "
          f"base '{only}' not swapped={len(unmatched)} {unmatched[:5]} -> {dst}")


if __name__ == "__main__":
    a = sys.argv[1:]
    only = None
    if "--only-substr" in a:
        i = a.index("--only-substr"); only = a[i + 1]; del a[i:i + 2]
    main(*a, only=only)