from __future__ import annotations

import argparse
import json
import re
from pathlib import Path
from typing import Any


MARKER_RE = re.compile(r"^(\d+)-(\d+)$")


def read_json(path: Path) -> dict[str, Any]:
    return json.loads(path.read_text(encoding="utf-8"))


def as_int(value: Any) -> int | None:
    try:
        return int(value)
    except (TypeError, ValueError):
        return None


def tree_index(value: Any) -> int:
    text = str(value or "")
    if text.startswith("$."):
        text = text[2:]
    first = text.split(".", 1)[0]
    return as_int(first) if as_int(first) is not None else 10**9


def marker_rows_from_payload(payload: dict[str, Any]) -> list[dict[str, Any]]:
    if isinstance(payload.get("rows"), list):
        return [row for row in payload["rows"] if isinstance(row, dict)]
    probe = payload.get("probe") if isinstance(payload.get("probe"), dict) else {}
    rows = probe.get("marker_matrix") if isinstance(probe.get("marker_matrix"), list) else []
    result: list[dict[str, Any]] = []
    for row in rows:
        if not isinstance(row, dict):
            continue
        styles = row.get("styles") if isinstance(row.get("styles"), list) else []
        style = styles[0] if styles and isinstance(styles[0], dict) else {}
        next_record = style.get("next_moxel_record") if isinstance(style.get("next_moxel_record"), dict) else {}
        result.append(
            {
                "text": row.get("text"),
                "expected_row": row.get("expected_row"),
                "expected_col": row.get("expected_col"),
                "decoded_row": row.get("decoded_row"),
                "decoded_col": row.get("decoded_col"),
                "style_tree_position": style.get("tree_position"),
                "next_value": next_record.get("value") or next_record.get("head"),
                "preceding": style.get("last_7_preceding_values") or style.get("immediate_preceding_values"),
                "row_ok": row.get("row_ok"),
                "col_ok": row.get("col_ok"),
            }
        )
    return result


def last_preceding_int(row: dict[str, Any]) -> int | None:
    preceding = row.get("preceding") if isinstance(row.get("preceding"), list) else []
    for value in reversed(preceding):
        parsed = as_int(value)
        if parsed is not None:
            return parsed
    return None


def analyze_marker_matrix(payloads: list[dict[str, Any]]) -> dict[str, Any]:
    rows: list[dict[str, Any]] = []
    for payload in payloads:
        rows.extend(marker_rows_from_payload(payload))
    marker_rows = [row for row in rows if MARKER_RE.match(str(row.get("text") or ""))]
    total = len(marker_rows)
    row_ok = sum(1 for row in marker_rows if row.get("row_ok") is True or row.get("decoded_row") == row.get("expected_row"))
    col_ok = sum(1 for row in marker_rows if row.get("col_ok") is True or row.get("decoded_col") == row.get("expected_col"))

    last_preceding_predictions = []
    for row in marker_rows:
        expected_col = as_int(row.get("expected_col"))
        last = last_preceding_int(row)
        if expected_col is None or last is None:
            continue
        predicted = last + 1
        last_preceding_predictions.append(
            {
                "text": row.get("text"),
                "style_tree_position": row.get("style_tree_position"),
                "expected_col": expected_col,
                "predicted_col": predicted,
                "ok": predicted == expected_col,
                "preceding_last": last,
                "current_decoded_col": row.get("decoded_col"),
            }
        )

    prediction_total = len(last_preceding_predictions)
    prediction_ok = sum(1 for item in last_preceding_predictions if item.get("ok") is True)
    rows_by_expected_row: dict[int, list[dict[str, Any]]] = {}
    for row in marker_rows:
        expected_row = as_int(row.get("expected_row"))
        if expected_row is None:
            continue
        rows_by_expected_row.setdefault(expected_row, []).append(row)
    row_runs = []
    for expected_row, run_rows in sorted(rows_by_expected_row.items()):
        sorted_rows = sorted(run_rows, key=lambda item: tree_index(item.get("style_tree_position")))
        expected_cols = [as_int(row.get("expected_col")) for row in sorted_rows]
        predicted_cols = [(last_preceding_int(row) + 1) if last_preceding_int(row) is not None else None for row in sorted_rows]
        row_runs.append(
            {
                "expected_row": expected_row,
                "texts": [row.get("text") for row in sorted_rows],
                "style_positions": [row.get("style_tree_position") for row in sorted_rows],
                "expected_columns": expected_cols,
                "predicted_columns": predicted_cols,
                "sequence_ok": predicted_cols == expected_cols,
            }
        )

    confidence = "none"
    if prediction_total and prediction_ok == prediction_total and prediction_total >= 3:
        confidence = "high"
    elif prediction_total and prediction_ok / prediction_total >= 0.8:
        confidence = "medium"
    elif prediction_total:
        confidence = "low"

    return {
        "marker_rows": total,
        "current_decoder": {
            "row_accuracy": {"ok": row_ok, "total": total},
            "column_accuracy": {"ok": col_ok, "total": total},
        },
        "candidate_rules": [
            {
                "id": "inline_text_column_from_last_preceding_scalar_plus_one",
                "target": "moxel.inline_text_cell.column",
                "expression": "one_based_column = int(last_numeric(preceding_scalars)) + 1",
                "confidence": confidence,
                "evidence": {"ok": prediction_ok, "total": prediction_total},
                "samples": last_preceding_predictions[:20],
            }
        ],
        "row_runs": row_runs,
    }


def raw_scalar_diffs(before: list[Any], after: list[Any]) -> list[int]:
    return [index for index, pair in enumerate(zip(before, after)) if pair[0] != pair[1]]


def infer_named_range_rules(history_matrix: dict[str, Any], property_candidates: dict[str, Any]) -> dict[str, Any]:
    coordinate_fields: dict[str, set[int]] = {"left": set(), "right": set(), "top": set(), "bottom": set()}
    evidence: list[dict[str, Any]] = []
    transitions: list[dict[str, Any]] = []
    transitions.extend([item for item in history_matrix.get("transitions") or [] if isinstance(item, dict)])
    transitions.extend([item for item in property_candidates.get("classified_transitions") or [] if isinstance(item, dict)])

    for transition in transitions:
        named = transition.get("named_range_changes") if isinstance(transition.get("named_range_changes"), dict) else {}
        if "named_range_diff" in transition and isinstance(transition["named_range_diff"], dict):
            named = {
                "one_based": {
                    "before": transition["named_range_diff"].get("before_coords") or {},
                    "after": transition["named_range_diff"].get("after_coords") or {},
                },
                "raw_scalars": {
                    "before": transition["named_range_diff"].get("before_raw") or [],
                    "after": transition["named_range_diff"].get("after_raw") or [],
                },
            }
        coords = named.get("one_based") if isinstance(named.get("one_based"), dict) else {}
        raw = named.get("raw_scalars") if isinstance(named.get("raw_scalars"), dict) else {}
        before_coords = coords.get("before") if isinstance(coords.get("before"), dict) else {}
        after_coords = coords.get("after") if isinstance(coords.get("after"), dict) else {}
        before_raw = raw.get("before") if isinstance(raw.get("before"), list) else []
        after_raw = raw.get("after") if isinstance(raw.get("after"), list) else []
        changed = raw_scalar_diffs(before_raw, after_raw)
        if not changed:
            continue
        changed_coord_names = [
            name
            for name in ("left", "right", "top", "bottom", "row_start", "row_end", "column_start", "column_end")
            if before_coords.get(name) != after_coords.get(name)
        ]
        for name in changed_coord_names:
            if name in {"left", "column_start"}:
                coordinate_fields["left"].update(changed)
            elif name in {"right", "column_end"}:
                coordinate_fields["right"].update(changed)
            elif name in {"top", "row_start"}:
                coordinate_fields["top"].update(changed)
            elif name in {"bottom", "row_end"}:
                coordinate_fields["bottom"].update(changed)
        evidence.append(
            {
                "from_file": transition.get("from_file"),
                "to_file": transition.get("to_file"),
                "changed_raw_indexes": changed,
                "changed_coordinates": changed_coord_names,
            }
        )

    rules = []
    for target, indexes in coordinate_fields.items():
        if not indexes:
            continue
        rules.append(
            {
                "target": f"moxel.named_range.{target}",
                "raw_scalar_indexes": sorted(indexes),
                "expression": "one_based = int(raw_scalar) + 1",
                "confidence": "high" if len(indexes) == 1 else "medium",
            }
        )
    return {"rules": rules, "evidence": evidence}


def summarize_property_candidates(property_candidates: dict[str, Any]) -> dict[str, Any]:
    classified = [item for item in property_candidates.get("classified_transitions") or [] if isinstance(item, dict)]
    label_counts: dict[str, int] = {}
    for item in classified:
        for label in item.get("labels") or []:
            label_counts[str(label)] = label_counts.get(str(label), 0) + 1
    unmapped = [
        {
            "from_file": item.get("from_file"),
            "to_file": item.get("to_file"),
            "labels": item.get("labels") or [],
            "reason": "No stable property field yet; capture one-property-per-save snapshots on the same cell.",
        }
        for item in classified
        if "uncategorized" in (item.get("labels") or []) or "tracked_signature_unchanged" in (item.get("labels") or [])
    ]
    return {
        "label_counts": [{"label": key, "count": label_counts[key]} for key in sorted(label_counts)],
        "unmapped_transitions": unmapped,
        "recommended_next_steps": property_candidates.get("recommended_next_steps") or [],
    }


def build_registry(
    marker_payloads: list[dict[str, Any]],
    history_matrix: dict[str, Any],
    property_candidates: dict[str, Any],
    sources: dict[str, list[str] | str | None],
) -> dict[str, Any]:
    marker_analysis = analyze_marker_matrix(marker_payloads)
    named_range = infer_named_range_rules(history_matrix, property_candidates)
    properties = summarize_property_candidates(property_candidates)
    rules = []
    rules.extend(marker_analysis.get("candidate_rules") or [])
    rules.extend(named_range.get("rules") or [])
    return {
        "schema": "codex_1c_moxel_schema_discovery.v1",
        "sources": sources,
        "status": "ok",
        "rules": rules,
        "analysis": {
            "inline_text_cells": marker_analysis,
            "named_ranges": named_range,
            "properties": properties,
        },
        "next_automation": [
            "Capture one-property-per-save probes for alignment, font, border, fill, protection, wrapping, merge, row height, and column width.",
            "Promote high-confidence rules into the adapter decoder only after fixture tests and a disposable 1C validation run.",
            "For write support, require a separate round-trip proof for every scalar path before enabling mutation.",
        ],
    }


def render_markdown(payload: dict[str, Any]) -> str:
    lines: list[str] = []
    lines.append("# 1C MOXCEL schema discovery")
    lines.append("")
    lines.append(f"- Status: `{payload.get('status')}`")
    lines.append(f"- Rules: `{len(payload.get('rules') or [])}`")
    inline = ((payload.get("analysis") or {}).get("inline_text_cells") or {})
    current = inline.get("current_decoder") or {}
    row_acc = current.get("row_accuracy") or {}
    col_acc = current.get("column_accuracy") or {}
    lines.append(f"- Inline row accuracy: `{row_acc.get('ok')}/{row_acc.get('total')}`")
    lines.append(f"- Inline column accuracy: `{col_acc.get('ok')}/{col_acc.get('total')}`")
    lines.append("")
    lines.append("## Rules")
    lines.append("")
    lines.append("| Rule | Target | Confidence | Evidence |")
    lines.append("| --- | --- | --- | --- |")
    for rule in payload.get("rules") or []:
        evidence = rule.get("evidence") or {}
        evidence_text = f"{evidence.get('ok')}/{evidence.get('total')}" if evidence else ", ".join(map(str, rule.get("raw_scalar_indexes") or []))
        lines.append(f"| `{rule.get('id') or rule.get('target')}` | `{rule.get('target')}` | `{rule.get('confidence')}` | `{evidence_text}` |")
    lines.append("")
    lines.append("## Property Gaps")
    lines.append("")
    properties = ((payload.get("analysis") or {}).get("properties") or {})
    for item in properties.get("unmapped_transitions") or []:
        lines.append(f"- `{item.get('from_file')}` -> `{item.get('to_file')}`: `{', '.join(item.get('labels') or [])}`")
    if not properties.get("unmapped_transitions"):
        lines.append("- none")
    lines.append("")
    lines.append("## Next Automation")
    lines.append("")
    for step in payload.get("next_automation") or []:
        lines.append(f"- {step}")
    lines.append("")
    return "\n".join(lines)


def main() -> int:
    parser = argparse.ArgumentParser(description="Infer 1C MOXCEL decoder/property rules from captured template probes.")
    parser.add_argument("--marker-matrix", action="append", default=[], help="Marker matrix JSON. Repeatable.")
    parser.add_argument("--probe-snapshot", action="append", default=[], help="Probe snapshot JSON from capture_1c_template_probe.py. Repeatable.")
    parser.add_argument("--history-matrix", help="History matrix JSON from analyze_1c_template_history_matrix.py.")
    parser.add_argument("--property-candidates", help="Property candidate JSON from infer_1c_template_property_candidates.py.")
    parser.add_argument("--output-json", default="reports/1c-template-baselines/Primer3_moxel_schema_discovery.json")
    parser.add_argument("--output-markdown", default="reports/1c-template-baselines/Primer3_moxel_schema_discovery.md")
    args = parser.parse_args()

    marker_paths = [Path(path) for path in args.marker_matrix]
    probe_paths = [Path(path) for path in args.probe_snapshot]
    marker_payloads = [read_json(path) for path in [*marker_paths, *probe_paths]]
    history_matrix = read_json(Path(args.history_matrix)) if args.history_matrix else {}
    property_candidates = read_json(Path(args.property_candidates)) if args.property_candidates else {}
    payload = build_registry(
        marker_payloads,
        history_matrix,
        property_candidates,
        {
            "marker_matrix": [str(path) for path in marker_paths],
            "probe_snapshot": [str(path) for path in probe_paths],
            "history_matrix": args.history_matrix,
            "property_candidates": args.property_candidates,
        },
    )

    json_path = Path(args.output_json)
    md_path = Path(args.output_markdown)
    json_path.parent.mkdir(parents=True, exist_ok=True)
    md_path.parent.mkdir(parents=True, exist_ok=True)
    json_path.write_text(json.dumps(payload, ensure_ascii=False, indent=2), encoding="utf-8")
    md_path.write_text(render_markdown(payload), encoding="utf-8")
    print(json.dumps({"status": "ok", "json": str(json_path), "markdown": str(md_path), "rules": len(payload["rules"])}, ensure_ascii=False, indent=2))
    return 0


if __name__ == "__main__":
    raise SystemExit(main())