"""Classify 1C Config/ConfigCAS payload parts without infobase-specific names."""

from __future__ import annotations

import base64
import hashlib
import re
from typing import Any

from .payload import collect_strings, decode_payload_lossless, encode_text, parse_brace_text, root_signature, scalar


BASE64_RE = re.compile(r"[A-Za-z0-9+/]{40,}={0,2}")
BSL_MARKERS = ("&На", "Процедура ", "Функция ", "#Область", "#КонецОбласти")
HTML_MARKERS = ("<!DOCTYPE", "<html", "<HTML", "<body", "<BODY")
STREAM_HEADER_RE = re.compile(rb"\r\n([0-9a-f]{8}) ([0-9a-f]{8}) 7fffffff \r\n")


def sha1_hex(data: bytes) -> str:
    return hashlib.sha1(data).hexdigest()


def normalized_text_sha1(text: str) -> str:
    normalized = str(text or "").replace("\r\n", "\n").replace("\r", "\n")
    return hashlib.sha1(normalized.encode("utf-8")).hexdigest()


def _text_with_line_ending(text: str, line_ending: str) -> str:
    """Normalize caller text first, then render it in a stream's convention."""
    return str(text or "").replace("\r\n", "\n").replace("\r", "\n").replace("\n", line_ending)


def decode_text(data: bytes) -> tuple[str | None, str | None]:
    if data.startswith(b"\xef\xbb\xbf"):
        try:
            return data.decode("utf-8-sig"), "utf-8-sig"
        except UnicodeDecodeError:
            pass
    candidates = ("utf-8-sig", "utf-8", "utf-16-le", "utf-16-be", "cp1251")
    best: tuple[str | None, str | None, int] = (None, None, -1)
    for encoding in candidates:
        try:
            text = data.decode(encoding)
        except UnicodeDecodeError:
            continue
        sample = text[:20000]
        marker_score = sum(500 for marker in (*BSL_MARKERS, *HTML_MARKERS) if marker in sample)
        printable = sum(1 for char in sample if char.isprintable() or char in "\r\n\t")
        score = printable + marker_score - sample.count("\x00") * 10
        if score > best[2]:
            best = (text, encoding, score)
    return best[0], best[1]


def payload_markers(payload: bytes) -> list[str]:
    markers = []
    if payload.startswith(b"MOXCEL"):
        markers.append("MOXCEL")
    if payload.startswith(b"\xef\xbb\xbf") or b"\xef\xbb\xbf" in payload[:256]:
        markers.append("utf8_bom")
    if STREAM_HEADER_RE.search(payload):
        markers.append("stream_headers")
    return markers


def extract_stream_blocks(payload: bytes, *, include_text: bool = False, limit: int = 100) -> list[dict[str, Any]]:
    blocks: list[dict[str, Any]] = []
    for match in STREAM_HEADER_RE.finditer(payload):
        declared_1 = int(match.group(1), 16)
        declared_2 = int(match.group(2), 16)
        start = match.end()
        size = declared_2
        if size <= 0 or start + size > len(payload):
            continue
        data = payload[start : start + size]
        text, encoding = decode_text(data)
        clean = (text or "").replace("\x00", "")
        item: dict[str, Any] = {
            "header_offset": match.start(),
            "data_offset": start,
            "declared_1": declared_1,
            "declared_2": declared_2,
            "bytes": len(data),
            "sha1": sha1_hex(data),
            "encoding": encoding,
            "text_preview": clean[:500],
            "has_bsl_marker": bool(text and any(marker in clean for marker in BSL_MARKERS)),
            "has_html_marker": bool(text and any(marker in clean for marker in HTML_MARKERS)),
        }
        if include_text:
            item["text"] = text
        blocks.append(item)
        if len(blocks) >= limit:
            break
    return blocks


def stream_blocks_with_data(payload: bytes, *, limit: int = 100) -> list[dict[str, Any]]:
    blocks: list[dict[str, Any]] = []
    for match in STREAM_HEADER_RE.finditer(payload):
        declared_1 = int(match.group(1), 16)
        declared_2 = int(match.group(2), 16)
        start = match.end()
        size = declared_2
        if size <= 0 or start + size > len(payload):
            continue
        data = payload[start : start + size]
        text, encoding = decode_text(data)
        blocks.append(
            {
                "header_offset": match.start(),
                "header_end": match.end(),
                "data_offset": start,
                "data_end": start + size,
                "declared_1": declared_1,
                "declared_2": declared_2,
                "bytes": len(data),
                "sha1": sha1_hex(data),
                "encoding": encoding,
                "text": text,
                "data": data,
            }
        )
        if len(blocks) >= limit:
            break
    return blocks


def structural_stream_blocks_with_data(payload: bytes, *, limit: int = 100) -> list[dict[str, Any]]:
    """Read one contiguous stream chain without matching headers inside data.

    Some 1C stream payloads legitimately contain the ASCII sequence used by a
    stream header inside a binary member.  ``stream_blocks_with_data`` remains
    a discovery heuristic for legacy readers; this function follows only the
    next header located exactly at the previous member's end and is suitable
    for evidence-bearing module decoding.
    """
    blocks: list[dict[str, Any]] = []
    first = STREAM_HEADER_RE.search(payload)
    if first is None:
        return blocks
    match = first
    while match is not None and len(blocks) < limit:
        declared_1 = int(match.group(1), 16)
        declared_2 = int(match.group(2), 16)
        data_offset = match.end()
        data_end = data_offset + declared_2
        if declared_2 <= 0 or data_end > len(payload):
            break
        data = payload[data_offset:data_end]
        text, encoding = decode_text(data)
        blocks.append(
            {
                "header_offset": match.start(),
                "header_end": match.end(),
                "data_offset": data_offset,
                "data_end": data_end,
                "declared_1": declared_1,
                "declared_2": declared_2,
                "bytes": len(data),
                "sha1": sha1_hex(data),
                "encoding": encoding,
                "text": text,
                "data": data,
                "structural": True,
            }
        )
        match = STREAM_HEADER_RE.match(payload, data_end)
    return blocks


def extract_structural_stream_blocks(payload: bytes, *, include_text: bool = False, limit: int = 100) -> list[dict[str, Any]]:
    """Public structural stream view with the same shape as discovery blocks."""
    result: list[dict[str, Any]] = []
    for block in structural_stream_blocks_with_data(payload, limit=limit):
        text = str(block.get("text") or "")
        clean = text.replace("\x00", "")
        item = {
            **{key: value for key, value in block.items() if key not in {"text", "data"}},
            "text_preview": clean[:500],
            "has_bsl_marker": bool(text and any(marker in clean for marker in BSL_MARKERS)),
            "has_html_marker": bool(text and any(marker in clean for marker in HTML_MARKERS)),
        }
        if include_text:
            item["text"] = block.get("text")
        result.append(item)
    return result


def decode_declared_utf8_bsl_prefix(payload: bytes, stream_index: int) -> dict[str, Any]:
    """Decode a BSL prefix whose byte length is declared by a stream header.

    Object-module containers observed in 1C keep the editable UTF-8 BSL bytes
    in the first ``declared_1`` bytes of a fixed-size member.  The remaining
    member bytes are opaque platform metadata, not source text.  This helper
    is read-only evidence; it intentionally does not construct replacements.
    """
    blocks = structural_stream_blocks_with_data(payload)
    if stream_index < 0 or stream_index >= len(blocks):
        return {"status": "not_found", "error": "stream_index_not_found"}
    block = blocks[stream_index]
    data = bytes(block["data"])
    prefix_bytes = int(block["declared_1"])
    if prefix_bytes <= 0 or prefix_bytes > len(data):
        return {
            "status": "unsupported",
            "error": "invalid_declared_bsl_prefix_length",
            "declared_1": prefix_bytes,
            "member_bytes": len(data),
        }
    prefix = data[:prefix_bytes]
    if not prefix.startswith(b"\xef\xbb\xbf"):
        return {
            "status": "unsupported",
            "error": "declared_bsl_prefix_not_utf8_bom",
            "declared_1": prefix_bytes,
            "member_bytes": len(data),
        }
    try:
        text = prefix.decode("utf-8-sig", errors="strict")
    except UnicodeDecodeError as exc:
        return {"status": "unsupported", "error": "declared_bsl_prefix_decode_error", "diagnostics": {"message": str(exc)}}
    return {
        "status": "ok",
        "text": text,
        "stream_index": stream_index,
        "header_offset": block["header_offset"],
        "data_offset": block["data_offset"],
        "bsl_prefix_bytes": prefix_bytes,
        "opaque_tail_bytes": len(data) - prefix_bytes,
        "member_bytes": len(data),
        "text_sha1": normalized_text_sha1(text),
        "structural": True,
    }


def replace_declared_utf8_bsl_prefix_same_width(
    payload: bytes,
    stream_index: int,
    *,
    text: str,
    expected_text_sha1: str | None = None,
) -> tuple[bytes, dict[str, Any]]:
    """Replace a proven fixed-width BSL prefix without touching its tail.

    This is deliberately narrower than ``replace_stream_block``.  The report
    object-module carrier has a fixed-size stream member whose first declared
    bytes are UTF-8 source and whose remaining bytes are opaque.  A shorter
    source is right-padded with spaces *inside the declared source field*;
    longer source is rejected.  Consequently the member, every following
    stream, and the opaque tail stay byte-for-byte identical.

    It does not attempt to synthesize the platform's independent version
    atoms.  The caller remains responsible for the proven paired
    ``__configinfo`` SHA-1 update.
    """
    decoded = decode_declared_utf8_bsl_prefix(payload, stream_index)
    if decoded.get("status") != "ok":
        raise ValueError(str(decoded.get("error") or "declared_bsl_prefix_unavailable"))
    old_text = str(decoded["text"])
    if not is_declared_utf8_bsl_source(old_text):
        raise ValueError("declared_bsl_prefix_is_not_bsl_source")
    old_sha1 = normalized_text_sha1(old_text)
    if expected_text_sha1 and expected_text_sha1.lower() != old_sha1:
        raise ValueError("expected_text_sha1 does not match declared BSL prefix")
    line_ending = "\r\n" if "\r\n" in old_text else "\r" if "\r" in old_text else "\n"
    rendered = _text_with_line_ending(text, line_ending)
    encoded = b"\xef\xbb\xbf" + rendered.encode("utf-8")
    prefix_bytes = int(decoded["bsl_prefix_bytes"])
    if len(encoded) > prefix_bytes:
        raise ValueError("replacement_declared_bsl_prefix_exceeds_fixed_width")
    # BSL whitespace outside string literals is semantically inert.  Padding
    # is restricted to the fixed source field and is observable in readback.
    padded = encoded + (b" " * (prefix_bytes - len(encoded)))
    if len(padded) != prefix_bytes:
        raise AssertionError("declared BSL prefix width changed")
    data_offset = int(decoded["data_offset"])
    new_payload = payload[:data_offset] + padded + payload[data_offset + prefix_bytes :]
    if payload[data_offset + prefix_bytes :] != new_payload[data_offset + prefix_bytes :]:
        raise AssertionError("opaque member tail changed")
    return new_payload, {
        "stream_index": stream_index,
        "mode": "declared_utf8_bsl_prefix_same_width",
        "old_text_sha1": old_sha1,
        "new_text_sha1": normalized_text_sha1(rendered),
        "old_bsl_prefix_bytes": prefix_bytes,
        "new_bsl_source_bytes": len(encoded),
        "padding_bytes": prefix_bytes - len(encoded),
        "opaque_tail_bytes": int(decoded["opaque_tail_bytes"]),
        "opaque_tail_preserved": True,
        "old_text_preview": old_text[:500],
        "new_text_preview": rendered[:500],
    }


def is_declared_utf8_bsl_source(text: str) -> bool:
    """Recognize source evidence in a declared UTF-8 stream prefix.

    A module may legitimately consist solely of comments, while other stream
    members can also have a UTF-8 prefix (for example a brace descriptor).
    The prefix is BSL evidence only when it has a normal BSL marker or every
    nonblank source line is a BSL line comment.
    """
    source = str(text or "").lstrip("\ufeff")
    if any(marker in source for marker in BSL_MARKERS):
        return True
    lines = [line.strip() for line in source.replace("\r\n", "\n").replace("\r", "\n").split("\n") if line.strip()]
    return bool(lines) and all(line.startswith("//") for line in lines)


def stream_header(size: int) -> bytes:
    if size < 0 or size > 0xFFFFFFFF:
        raise ValueError("stream size is outside 8-hex header range")
    encoded = f"{size:08x}"
    return f"\r\n{encoded} {encoded} 7fffffff \r\n".encode("ascii")


def replace_stream_block(
    payload: bytes,
    stream_index: int,
    *,
    text: str | None = None,
    data: bytes | None = None,
    replace: dict[str, Any] | None = None,
    routine: dict[str, Any] | None = None,
    expected_contains: str | None = None,
    expected_text_sha1: str | None = None,
) -> tuple[bytes, dict[str, Any]]:
    blocks = stream_blocks_with_data(payload)
    if stream_index < 0 or stream_index >= len(blocks):
        raise IndexError(f"stream_index {stream_index} is outside {len(blocks)} stream blocks")
    block = blocks[stream_index]
    old_data = bytes(block["data"])
    old_text = block.get("text")
    encoding = block.get("encoding")
    if expected_contains and (old_text is None or expected_contains not in old_text):
        raise ValueError("expected_contains was not found in stream text")
    old_text_sha1 = normalized_text_sha1(old_text or "") if old_text is not None else None
    if expected_text_sha1 and (old_text_sha1 is None or expected_text_sha1.lower() != old_text_sha1):
        raise ValueError("expected_text_sha1 does not match current stream text")
    if replace is not None:
        if old_text is None:
            raise ValueError("stream text is not decodable")
        old = str(replace.get("old") or "")
        new = str(replace.get("new") or "")
        if not old:
            raise ValueError("replace.old is required")
        count = int(replace.get("count") or 1)
        # Public code.read normalizes BSL to LF while streams often retain
        # CRLF.  Treat that representation difference as irrelevant, but do
        # not loosen matching of any other character (spaces/tabs remain
        # exact).  The replacement is rendered back in the stream's original
        # line-ending convention to avoid unrelated formatting churn.
        line_ending = "\r\n" if "\r\n" in old_text else "\r" if "\r" in old_text else "\n"
        source_old = old
        source_new = _text_with_line_ending(new, line_ending) if ("\n" in new or "\r" in new) else new
        if source_old not in old_text:
            source_old = _text_with_line_ending(old, line_ending)
            source_new = _text_with_line_ending(new, line_ending)
        if source_old not in old_text:
            raise ValueError("replace.old was not found in stream text")
        text = old_text.replace(source_old, source_new, count)
    routine_edit = None
    if routine is not None:
        if old_text is None:
            raise ValueError("stream text is not decodable")
        from .bsl_validation import replace_routine_text

        text, routine_edit = replace_routine_text(
            old_text,
            str(routine.get("text") or ""),
            operation=str(routine.get("operation") or "replace"),
            name=str(routine.get("name")) if routine.get("name") else None,
            expected_old_sha1=str(routine.get("expected_old_sha1")) if routine.get("expected_old_sha1") else None,
            expected_old_contains=str(routine.get("expected_old_contains")) if routine.get("expected_old_contains") else None,
        )
    if data is None:
        if text is None:
            raise ValueError("text, data, replace, or routine is required")
        data = encode_text(text, encoding)
    header = stream_header(len(data))
    new_payload = payload[: block["header_offset"]] + header + data + payload[block["data_end"] :]
    return new_payload, {
        "stream_index": stream_index,
        "encoding": encoding,
        "old_sha1": sha1_hex(old_data),
        "new_sha1": sha1_hex(data),
        "old_text_sha1": old_text_sha1,
        "new_text_sha1": normalized_text_sha1(text) if text is not None else None,
        "old_bytes": len(old_data),
        "new_bytes": len(data),
        "old_text_preview": (old_text or "")[:500],
        "new_text_preview": (text or "")[:500] if text is not None else None,
        **({"routine": routine_edit} if routine_edit else {}),
    }


def collect_base64_blocks(value: Any) -> list[str]:
    blocks: list[str] = []

    def walk(node: Any) -> None:
        if isinstance(node, dict) and node.get("type") == "list":
            items = node.get("items") or []
            if items and scalar(items[0]) == "#base64":
                chunks = [scalar(item) for item in items[1:] if BASE64_RE.fullmatch(scalar(item))]
                if chunks:
                    blocks.append("".join(chunks))
            for child in items:
                walk(child)

    walk(value)
    return blocks


def decode_base64_blocks(blocks: list[str], *, include_text: bool = False, limit: int = 50) -> list[dict[str, Any]]:
    result: list[dict[str, Any]] = []
    for value in blocks[:limit]:
        try:
            data = base64.b64decode(value, validate=True)
        except Exception:
            continue
        text, encoding = decode_text(data)
        clean = (text or "").replace("\x00", "")
        item: dict[str, Any] = {
            "block_length": len(value),
            "bytes": len(data),
            "sha1": sha1_hex(data),
            "encoding": encoding,
            "text_preview": clean[:500],
            "has_bsl_marker": bool(text and any(marker in clean for marker in BSL_MARKERS)),
            "has_html_marker": bool(text and any(marker in clean for marker in HTML_MARKERS)),
        }
        if include_text:
            item["text"] = text
        result.append(item)
    return result


def classify_role(root_marker: str | None, markers: list[str], text: str | None, stream_blocks: list[dict[str, Any]], base64_blocks: list[dict[str, Any]]) -> str:
    clean = (text or "").replace("\x00", "")
    if root_marker == "1":
        return "metadata_payload"
    if root_marker == "4":
        return "form_payload"
    if root_marker == "5" or any(block.get("has_html_marker") for block in base64_blocks):
        return "help_or_html_payload"
    if root_marker == "8" or "MOXCEL" in markers:
        return "template_payload"
    if any(block.get("has_bsl_marker") for block in stream_blocks) or any(marker in clean for marker in BSL_MARKERS):
        return "bsl_module_payload"
    if "stream_headers" in markers:
        return "stream_container"
    if root_marker:
        return "brace_payload"
    return "binary_or_unknown_payload"


def classify_payload(data: bytes, *, include_text: bool = False, include_tree: bool = False) -> dict[str, Any]:
    decoded = decode_payload_lossless(data)
    payload = decoded.get("payload") if isinstance(decoded.get("payload"), (bytes, bytearray)) else b""
    markers = payload_markers(bytes(payload))
    # Prefer proven contiguous boundaries for normal container decoding.  Keep
    # the regex scan only as a discovery fallback for legacy irregular blobs.
    stream_blocks = extract_structural_stream_blocks(bytes(payload), include_text=include_text)
    if not stream_blocks and "stream_headers" in markers:
        stream_blocks = extract_stream_blocks(bytes(payload), include_text=include_text)
    # A report object module observed in ConfigCAS stores source in the
    # declared UTF-8 prefix of a fixed-size stream member.  The remainder is
    # opaque platform state and must never be exposed as BSL.  Keep this as
    # read-only evidence: replacement still requires a separately proven
    # reverse codec for that carrier.
    if stream_blocks:
        for stream_index, stream in enumerate(stream_blocks):
            declared_prefix = decode_declared_utf8_bsl_prefix(bytes(payload), stream_index)
            if declared_prefix.get("status") != "ok" or not is_declared_utf8_bsl_source(str(declared_prefix.get("text") or "")):
                continue
            stream["declared_utf8_bsl_prefix"] = {
                key: declared_prefix[key]
                for key in ("bsl_prefix_bytes", "opaque_tail_bytes", "member_bytes", "text_sha1", "structural")
                if key in declared_prefix
            }
            if include_text:
                stream["text"] = declared_prefix["text"]
                stream["text_preview"] = str(declared_prefix["text"] or "").replace("\x00", "")[:500]
    text = decoded.get("text")
    tree = None
    root = None
    base64_decoded: list[dict[str, Any]] = []
    strings_sample: list[str] = []
    if text and "{" in text and "stream_headers" not in markers:
        try:
            tree = parse_brace_text(text)
            root = root_signature(tree)
            strings_sample = collect_strings(tree, limit=80)
            base64_decoded = decode_base64_blocks(collect_base64_blocks(tree), include_text=include_text)
        except Exception:
            tree = None
    root_marker = root.get("root_marker") if isinstance(root, dict) else None
    result: dict[str, Any] = {
        "status": "ok" if payload else "undecodable",
        "compression": decoded.get("compression"),
        "encoding": decoded.get("encoding"),
        "raw_bytes": decoded.get("raw_bytes"),
        "payload_bytes": decoded.get("payload_bytes"),
        "sha1": sha1_hex(data),
        "payload_sha1": sha1_hex(bytes(payload)),
        "markers": markers,
        "root": root,
        "role": classify_role(root_marker, markers, text, stream_blocks, base64_decoded),
        "strings_sample": strings_sample,
        "stream_blocks": stream_blocks,
        "base64_blocks": base64_decoded,
        "counts": {
            "stream_blocks": len(stream_blocks),
            "base64_blocks": len(base64_decoded),
            "strings_sample": len(strings_sample),
        },
    }
    if include_text:
        result["text"] = text
    if include_tree:
        result["tree"] = tree
    return result