"""Per-attempt record schema (machine-validated) and append-only storage."""
from __future__ import annotations
import json, uuid
from pathlib import Path
from jsonschema import Draft202012Validator
from .util import ROOT, utc_now, append_jsonl, sha256_json, git_commit

STATUSES = ["success", "invalid_output", "timeout", "oom", "incompatible", "implementation_error", "unavailable", "budget_exhausted"]

SCHEMA = {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "required": ["schema_version", "study_id", "experiment_id", "trial_id", "attempt_id", "run_group_id", "protocol_hash", "config_hash",
                 "git_commit", "timestamp_utc", "task_family", "document_id", "document_hash", "question_id", "split", "cluster_id",
                 "sender_model_revision", "receiver_model_revision", "tokenizer_hashes", "backend_versions", "hardware_id", "device_state",
                 "weight_precision", "method_id", "implementation_fidelity", "compression_config", "question_access", "topology",
                 "handoff_index", "reuse_count", "cache_hit", "cache_compatibility_hash", "random_seed", "source_tokens_sender",
                 "source_tokens_receiver", "query_tokens", "message_tokens", "message_utf8_bytes", "cache_tensor_bytes",
                 "serialized_payload_bytes", "side_information_bytes", "fetched_bytes", "transport_bytes_if_measured", "timing_spans",
                 "total_wall_ms", "ttft_ms", "input_usage", "output_usage", "provider_cached_usage", "process_memory", "backend_memory",
                 "peak_prefill_memory", "peak_decode_memory", "mapper_memory", "cost_with_source", "prediction_artifact", "evidence_artifact",
                 "reference_key_id", "task_scores", "status", "error_code", "retry_of", "log_artifact", "payload_hash", "output_hash"],
    "properties": {
        "schema_version": {"const": "wp30-attempt-1"},
        "status": {"enum": STATUSES},
        "question_access": {"enum": ["blind", "aware", "none"]},
        "implementation_fidelity": {"enum": ["official", "faithful_reimplementation", "adapted", "inspired", "custom", "n/a"]},
        "timing_spans": {"type": "object"},
        "task_scores": {"type": ["object", "null"]},
        "handoff_index": {"type": "integer"}, "reuse_count": {"type": "integer"},
    },
}
_validator = Draft202012Validator(SCHEMA)

def new_record(**kw) -> dict:
    rec = {k: None for k in SCHEMA["required"]}
    rec.update({"schema_version": "wp30-attempt-1", "study_id": "WP30", "attempt_id": uuid.uuid4().hex[:12], "timestamp_utc": utc_now(),
                "git_commit": git_commit(), "handoff_index": 0, "reuse_count": 1, "cache_hit": False, "timing_spans": {}, "fetched_bytes": 0,
                "side_information_bytes": 0, "transport_bytes_if_measured": None, "retry_of": None})
    rec.update(kw)
    return rec

def validate_record(rec: dict) -> list[str]:
    return [e.message for e in _validator.iter_errors(rec)]

def write_record(rec: dict, path: str | Path) -> None:
    errs = validate_record(rec)
    if errs:
        raise ValueError("invalid record: " + "; ".join(errs[:5]))
    append_jsonl(path, rec)

def trial_key(experiment_id: str, document_id: str, question_id: str, method_id: str, compression_config: dict, sender: str, receiver: str,
              question_access: str, protocol_hash: str, seed: int, topology: str = "single", handoff_index: int = 0, extra: dict | None = None) -> str:
    return sha256_json({"e": experiment_id, "d": document_id, "q": question_id, "m": method_id, "c": compression_config, "s": sender, "r": receiver,
                        "a": question_access, "p": protocol_hash, "seed": seed, "t": topology, "h": handoff_index, "x": extra or {}})[:24]