ForcePilot/backend/package/yuxi/channels/adapters/signal/normalize.py

from __future__ import annotations

import re
import time
from datetime import datetime, UTC

from yuxi.channels.models import (
    Attachment,
    ChannelIdentity,
    ChannelMessage,
    ChannelType,
    ChatType,
    EventType,
    MentionsInfo,
    MessageType,
)

E164_PATTERN = re.compile(r"^\+\d{7,15}$")
UUID_PREFIX = "uuid:"
GROUP_PREFIX = "group:"

DEDUP_TTL = 60
_dedup_store: dict[str, float] = {}


def is_own_message(data: dict, account_number: str, account_uuid: str | None = None) -> bool:
    envelope = data.get("envelope", {})
    source = envelope.get("source", "")
    if normalize_target(source) == normalize_target(account_number):
        return True
    if account_uuid:
        source_uuid = envelope.get("sourceUuid", "")
        if source_uuid and source_uuid == account_uuid:
            return True
    return False


def is_sync_message(data: dict) -> bool:
    return "syncMessage" in data


DEBOUNCE_TTL = 5
_debounce_store: dict[str, float] = {}


def check_debounce(conversation_key: str, interval_ms: int = 0) -> bool:
    if interval_ms <= 0:
        return False
    now = time.monotonic()
    _prune_debounce_store(now)
    if conversation_key in _debounce_store:
        last = _debounce_store[conversation_key]
        if (now - last) * 1000 < interval_ms:
            return True
    _debounce_store[conversation_key] = now
    return False


def _prune_debounce_store(now: float) -> None:
    expired = [k for k, ts in _debounce_store.items() if now - ts > DEBOUNCE_TTL]
    for k in expired:
        _debounce_store.pop(k, None)


def build_dedup_key(data: dict, account_id: str) -> str | None:
    envelope = data.get("envelope", {})
    source = envelope.get("source", "")
    timestamp = (
        data.get("dataMessage", {}).get("timestamp")
        or data.get("reaction", {}).get("targetSentTimestamp")
        or data.get("deleteMessage", {}).get("targetSentTimestamp")
        or data.get("receiptMessage", {}).get("timestamps", [None])[0]
    )
    if not source or not timestamp:
        return None
    group_id = data.get("dataMessage", {}).get("groupInfo", {}).get("groupId") or ""
    conversation = group_id or source
    return f"signal:{account_id}:{conversation}:{source}:{timestamp}"


def check_and_add_dedup(key: str) -> bool:
    now = time.monotonic()
    _prune_dedup_store(now)
    if key in _dedup_store:
        return True
    _dedup_store[key] = now
    return False


def _prune_dedup_store(now: float) -> None:
    expired = [k for k, ts in _dedup_store.items() if now - ts > DEDUP_TTL]
    for k in expired:
        _dedup_store.pop(k, None)


def normalize_target(raw: str) -> str:
    stripped = raw.strip()

    if stripped.startswith(GROUP_PREFIX):
        return stripped
    if stripped.startswith(UUID_PREFIX):
        return stripped
    if E164_PATTERN.match(stripped):
        return stripped

    digits = re.sub(r"[^\d]", "", stripped)
    if 7 <= len(digits) <= 15:
        return f"+{digits}"

    raise ValueError(f"Unable to normalize Signal target: {raw}")


def normalize_e164(raw: str) -> str:
    stripped = raw.strip().removeprefix("+")
    return re.sub(r"[^\d]", "", stripped)


def parse_signal_message(data: dict, channel_id: str = "signal") -> ChannelMessage | None:
    envelope = data.get("envelope", {})
    data_msg = data.get("dataMessage", {})

    if not data_msg:
        return None

    source = envelope.get("source", "unknown")
    source_name = envelope.get("sourceName", "")
    group_info = data_msg.get("groupInfo", {})
    group_id = group_info.get("groupId")
    chat_id = f"group:{group_id}" if group_id else source
    timestamp_raw = data_msg.get("timestamp", 0)
    ts = datetime.fromtimestamp(timestamp_raw / 1000.0, tz=UTC) if timestamp_raw else datetime.now(tz=UTC)

    content = data_msg.get("message", "") or data_msg.get("body", "")
    chat_type = ChatType.GROUP if group_id else ChatType.DIRECT

    message_type = MessageType.TEXT
    attachments: list[Attachment] = []
    if data_msg.get("attachments"):
        raw_attachments = data_msg["attachments"] if isinstance(data_msg["attachments"], list) else []
        for att in raw_attachments:
            content_type = att.get("contentType", "")
            attachments.append(
                Attachment(
                    type=_attachment_type(content_type),
                    file_id=att.get("id", ""),
                    filename=att.get("filename"),
                    mime_type=content_type,
                    size_bytes=att.get("size"),
                    metadata={"digest": att.get("digest", "")},
                )
            )
        if attachments:
            message_type = _attachment_message_type(attachments[0].mime_type or "")

    reply_to_id = None
    quote = data_msg.get("quote")
    if quote:
        reply_to_id = str(quote.get("id", ""))

    mentions = _extract_mentions(data_msg, content)
    extracted_urls = _extract_urls(data_msg, content)

    metadata: dict = {}
    if data_msg.get("expiresInSeconds"):
        metadata["expires_in_seconds"] = data_msg["expiresInSeconds"]
    if data_msg.get("viewOnce"):
        metadata["view_once"] = True
    if data_msg.get("sticker"):
        metadata["sticker_pack_id"] = data_msg["sticker"].get("packId", "")
        metadata["sticker_id"] = data_msg["sticker"].get("stickerId", "")
    if data_msg.get("forwarded"):
        metadata["forwarded"] = True
    if data_msg.get("contacts"):
        contacts = data_msg["contacts"]
        metadata["contacts"] = contacts if isinstance(contacts, list) else [contacts]
    if data_msg.get("location"):
        loc = data_msg["location"]
        metadata["location"] = {
            "latitude": loc.get("latitude"),
            "longitude": loc.get("longitude"),
            "label": loc.get("label", ""),
        }
    if source_name:
        metadata["sender_display_name"] = source_name

    if data_msg.get("sticker"):
        message_type = MessageType.STICKER
    elif data_msg.get("location"):
        message_type = MessageType.LOCATION
    elif attachments:
        message_type = _attachment_message_type(attachments[0].mime_type or "")

    return ChannelMessage(
        identity=ChannelIdentity(
            channel_id=channel_id,
            channel_type=ChannelType.SIGNAL,
            channel_user_id=source,
            channel_chat_id=chat_id,
            channel_message_id=str(timestamp_raw),
        ),
        event_type=EventType.MESSAGE_RECEIVED,
        message_type=message_type,
        chat_type=chat_type,
        content=content,
        reply_to_message_id=reply_to_id,
        attachments=attachments,
        mentions=mentions,
        extracted_urls=extracted_urls,
        metadata=metadata,
        timestamp=ts,
    )


def parse_signal_reaction(data: dict, channel_id: str = "signal") -> ChannelMessage | None:
    envelope = data.get("envelope", {})
    reaction = data.get("reaction", {})

    if not reaction:
        return None

    source = envelope.get("source", "unknown")
    target_author = reaction.get("targetAuthor", source)
    emoji = reaction.get("emoji", "")
    is_remove = reaction.get("remove", False)
    target_timestamp = reaction.get("targetSentTimestamp", 0)
    group_info = reaction.get("groupInfo", {})
    group_id = group_info.get("groupId")
    chat_id = f"group:{group_id}" if group_id else source

    return ChannelMessage(
        identity=ChannelIdentity(
            channel_id=channel_id,
            channel_type=ChannelType.SIGNAL,
            channel_user_id=source,
            channel_chat_id=chat_id,
            channel_message_id=str(target_timestamp),
        ),
        event_type=EventType.MESSAGE_UPDATED,
        message_type=MessageType.TEXT,
        chat_type=ChatType.GROUP if group_id else ChatType.DIRECT,
        content="(reaction_removed)" if is_remove else f"(reacted: {emoji})",
        metadata={
            "reaction_emoji": emoji,
            "reaction_removed": is_remove,
            "reaction_target_author": target_author,
        },
    )


def parse_signal_receipt(data: dict, channel_id: str = "signal") -> ChannelMessage | None:
    envelope = data.get("envelope", {})
    receipt = data.get("receiptMessage", {})

    if not receipt:
        return None

    source = envelope.get("source", "unknown")
    receipt_type = receipt.get("type", "UNKNOWN")
    timestamps = receipt.get("timestamps", [])

    return ChannelMessage(
        identity=ChannelIdentity(
            channel_id=channel_id,
            channel_type=ChannelType.SIGNAL,
            channel_user_id=source,
            channel_chat_id=source,
        ),
        event_type=EventType.READ_RECEIPT,
        message_type=MessageType.TEXT,
        chat_type=ChatType.DIRECT,
        content=f"(receipt: {receipt_type})",
        metadata={
            "receipt_type": receipt_type,
            "receipt_timestamps": timestamps,
        },
    )


def _attachment_type(content_type: str) -> str:
    if content_type.startswith("image/"):
        return "image"
    if content_type.startswith("video/"):
        return "video"
    if content_type.startswith("audio/"):
        return "audio"
    return "file"


def _attachment_message_type(mime_type: str) -> MessageType:
    if mime_type.startswith("image/"):
        return MessageType.IMAGE
    if mime_type.startswith("video/"):
        return MessageType.VIDEO
    if mime_type.startswith("audio/"):
        return MessageType.AUDIO
    return MessageType.FILE


URL_PATTERN = re.compile(r"https?://\S+")


def _extract_mentions(data_msg: dict, content: str, account_uuid: str | None = None) -> MentionsInfo | None:
    body_ranges = data_msg.get("bodyRanges")
    if not body_ranges or not isinstance(body_ranges, list):
        return None

    mentioned_ids: list[str] = []
    is_bot_mentioned = False
    for br in body_ranges:
        mention_uuid = br.get("mentionUuid")
        if mention_uuid:
            mentioned_ids.append(mention_uuid)
            if account_uuid and mention_uuid == account_uuid:
                is_bot_mentioned = True
        elif br.get("startsWith") and content:
            start = br.get("start", 0)
            length = br.get("length", 0)
            if 0 <= start < len(content):
                mention_text = content[start : start + length]
                mentioned_ids.append(mention_text)

    if not mentioned_ids:
        return None

    return MentionsInfo(
        mentioned_user_ids=mentioned_ids,
        is_bot_mentioned=is_bot_mentioned,
        raw_text=content,
    )


def _extract_urls(data_msg: dict, content: str) -> list[str]:
    previews = data_msg.get("previews")
    if previews and isinstance(previews, list):
        return [p.get("url", "") for p in previews if p.get("url")]

    return URL_PATTERN.findall(content) if content else []


def parse_signal_delete(data: dict, channel_id: str = "signal") -> ChannelMessage | None:
    envelope = data.get("envelope", {})
    delete_msg = data.get("deleteMessage")

    if not delete_msg:
        return None

    source = envelope.get("source", "unknown")
    target_timestamp = delete_msg.get("targetSentTimestamp", 0)
    group_info = delete_msg.get("groupInfo", {})
    group_id = group_info.get("groupId")
    chat_id = f"group:{group_id}" if group_id else source

    return ChannelMessage(
        identity=ChannelIdentity(
            channel_id=channel_id,
            channel_type=ChannelType.SIGNAL,
            channel_user_id=source,
            channel_chat_id=chat_id,
            channel_message_id=str(target_timestamp),
        ),
        event_type=EventType.MESSAGE_DELETED,
        message_type=MessageType.TEXT,
        chat_type=ChatType.GROUP if group_id else ChatType.DIRECT,
        content="(message deleted)",
        metadata={"deleted_timestamp": target_timestamp},
    )


def looks_like_signal_target_id(raw: str) -> bool:
    stripped = raw.strip()
    if stripped.startswith(GROUP_PREFIX) or stripped.startswith(UUID_PREFIX):
        return True
    if E164_PATTERN.match(stripped):
        return True
    digits = re.sub(r"[^\d]", "", stripped)
    return 7 <= len(digits) <= 15