discord-havoc/util/sing_util.py

import logging
import traceback

import aiohttp
import regex
import discord
from discord import Activity

logger = logging.getLogger(__name__)


class Utility:
    """Sing Utility"""

    def __init__(self) -> None:
        self.api_url: str = "http://127.0.0.1:52111/lyric/search"
        self.api_src: str = "DISC-HAVOC"

    def _smart_lyrics_wrap(self, lyrics: str, max_length: int = 1500, single_page: bool = False, max_verses: int = 100, max_lines: int = 150) -> list[str]:
        """
        Intelligently wrap lyrics to avoid breaking verses in the middle
        Prioritizes keeping verses intact over page length consistency

        Args:
            lyrics: Raw lyrics text
            max_length: Maximum character length per page (soft limit)
            single_page: If True, return only the first page

        Returns:
            List of lyrics pages
        """
        if not lyrics:
            return []

        # Strip markdown formatting from lyrics
        lyrics = discord.utils.escape_markdown(lyrics)
        verses = []
        current_verse: list[str] = []

        # Handle both regular newlines and zero-width space newlines
        lines = lyrics.replace("\u200b\n", "\n").split("\n")
        empty_line_count = 0

        for line in lines:
            stripped_line = line.strip()
            if not stripped_line or stripped_line in ["", "\u200b"]:
                empty_line_count += 1
                # One empty line indicates a section break (be more aggressive)
                if empty_line_count >= 1 and current_verse:
                    verses.append("\n".join(current_verse))
                    current_verse = []
                    empty_line_count = 0
            else:
                empty_line_count = 0
                current_verse.append(stripped_line)

        # Add the last verse if it exists
        if current_verse:
            verses.append("\n".join(current_verse))

        # If we have too few verses (verse detection failed), fallback to line-based splitting
        if len(verses) <= 1:
            all_lines = lyrics.split("\n")
            verses = []
            current_chunk = []

            for line in all_lines:
                current_chunk.append(line.strip())
                # Split every 8-10 lines to create artificial "verses"
                if len(current_chunk) >= 8:
                    verses.append("\n".join(current_chunk))
                    current_chunk = []

            # Add remaining lines
            if current_chunk:
                verses.append("\n".join(current_chunk))

        if not verses:
            return [lyrics[:max_length]]

        # If single page requested, return first verse or truncated
        if single_page:
            result = verses[0]
            if len(result) > max_length:
                # Try to fit at least part of the first verse
                lines_in_verse = result.split("\n")
                truncated_lines = []
                current_length = 0

                for line in lines_in_verse:
                    if current_length + len(line) + 1 <= max_length - 3:  # -3 for "..."
                        truncated_lines.append(line)
                        current_length += len(line) + 1
                    else:
                        break

                result = "\n".join(truncated_lines) + "..." if truncated_lines else verses[0][:max_length-3] + "..."
            return [result]

        # Group complete verses into pages, never breaking a verse
        # Limit by character count, verse count, AND line count for visual appeal
        max_verses_per_page = max_verses
        max_lines_per_page = max_lines
        pages = []
        current_page_verses: list[str] = []
        current_page_length = 0
        current_page_lines = 0

        for verse in verses:
            verse_length = len(verse)
            # Count lines properly - handle both regular newlines and zero-width space newlines
            verse_line_count = verse.count("\n") + verse.count("\u200b\n") + 1

            # Calculate totals if we add this verse (including separator)
            separator_length = 3 if current_page_verses else 0  # "\n\n\n" between verses
            separator_lines = 3 if current_page_verses else 0   # 3 empty lines between verses
            total_length_with_verse = current_page_length + separator_length + verse_length
            total_lines_with_verse = current_page_lines + separator_lines + verse_line_count

            # Check all three limits: character, verse count, and line count
            exceeds_length = total_length_with_verse > max_length
            exceeds_verse_count = len(current_page_verses) >= max_verses_per_page
            exceeds_line_count = total_lines_with_verse > max_lines_per_page

            # If adding this verse would exceed any limit AND we already have verses on the page
            if (exceeds_length or exceeds_verse_count or exceeds_line_count) and current_page_verses:
                # Finish current page with existing verses
                pages.append("\n\n".join(current_page_verses))
                current_page_verses = [verse]
                current_page_length = verse_length
                current_page_lines = verse_line_count
            else:
                # Add verse to current page
                current_page_verses.append(verse)
                current_page_length = total_length_with_verse
                current_page_lines = total_lines_with_verse

        # Add the last page if it has content
        if current_page_verses:
            pages.append("\n\n".join(current_page_verses))

        return pages if pages else [lyrics[:max_length]]

    def parse_song_input(
        self, song: str | None = None, activity: Activity | None = None,
    ) -> bool | tuple:
        """
        Parse Song (Sing Command) Input

        Args:
            song (Optional[str]): Song to search
            activity (Optional[discord.Activity]): Discord activity, used to attempt lookup if no song is provided

        Returns:
            Union[bool, tuple]
        """
        try:
            if (not song or len(song) < 2) and not activity:
                return False
            if not song and activity:
                if not activity.name:
                    return False  # No valid activity found
                match activity.name.lower():
                    case "codey toons" | "cider" | "sonixd":
                        search_artist: str = " ".join(
                            str(activity.state).strip().split(" ")[1:],
                        )
                        search_artist = regex.sub(
                            r"(\s{0,})(\[(spotify|tidal|sonixd|browser|yt music)])$",
                            "",
                            search_artist.strip(),
                            flags=regex.IGNORECASE,
                        )
                        search_song = str(activity.details)
                        song = f"{search_artist} : {search_song}"
                    case "tidal hi-fi":
                        search_artist = str(activity.state)
                        search_song = str(activity.details)
                        song = f"{search_artist} : {search_song}"
                    case "spotify":
                        if not activity.title or not activity.artist:  # type: ignore[attr-defined]
                            """
                            Attributes exist, but mypy does not recognize them.  Ignored.
                            """
                            return False
                        search_artist = str(activity.artist)  # type: ignore[attr-defined]
                        search_song = str(activity.title)  # type: ignore[attr-defined]
                        song = f"{search_artist} : {search_song}"
                    case "serious.fm" | "cocks.fm" | "something":
                        if not activity.details:
                            song = str(activity.state)
                        else:
                            search_artist = str(activity.state).rsplit("[", maxsplit=1)[
                                0
                            ]  # Strip genre
                            search_song = str(activity.details)
                            song = f"{search_artist} : {search_song}"
                    case _:
                        return False  # Unsupported activity detected

            search_split_by: str = (
                ":" if not (song) or len(song.split(":")) > 1 else "-"
            )  # Support either : or - to separate artist/track
            if not song:
                return False
            search_artist = song.split(search_split_by)[0].strip()
            search_song = "".join(song.split(search_split_by)[1:]).strip()
            search_subsearch: str | None = None
            if (
                search_split_by == ":" and len(song.split(":")) > 2
            ):  # Support sub-search if : is used (per instructions)
                search_song = song.split(
                    search_split_by,
                )[
                    1
                ].strip()  # Reduce search_song to only the 2nd split of : [the rest is meant to be lyric text]
                search_subsearch = "".join(
                    song.split(search_split_by)[2:],
                )  # Lyric text from split index 2 and beyond
            return (search_artist, search_song, search_subsearch)
        except Exception as e:
            logger.debug("Exception: %s", str(e))
            traceback.print_exc()
            return False

    async def lyric_search(
        self, artist: str, song: str, sub: str | None = None, is_spam_channel: bool = True,
    ) -> list | None:
        """
        Lyric Search

        Args:
            artist (str): Artist to search
            song (str): Song to search
            sub (Optional[str]): Lyrics for subsearch
        Returns:
            Optional[list]
        """
        try:
            if not artist or not song:
                return [("FAIL! Artist/Song not provided",)]

            search_obj: dict = {
                "a": artist.strip(),
                "s": song.strip(),
                "extra": True,
                "src": self.api_src,
            }

            if len(song.strip()) < 1:
                search_obj.pop("a")
                search_obj.pop("s")
                search_obj["t"] = artist.strip()  # Parse failed, try title without sep

            if sub and len(sub) >= 2:
                search_obj["sub"] = sub.strip()

            async with aiohttp.ClientSession() as session:
                async with await session.post(
                    self.api_url,
                    json=search_obj,
                    timeout=aiohttp.ClientTimeout(connect=5, sock_read=10),
                ) as request:
                    request.raise_for_status()
                    response: dict = await request.json()
                    if response.get("err"):
                        return [(f"ERR: {response.get('errorText')}",)]

                    out_lyrics = regex.sub(
                        r"<br>", "\u200b\n", response.get("lyrics", ""),
                    )
                    response_obj: dict = {
                        "artist": response.get("artist"),
                        "song": response.get("song"),
                        "lyrics": out_lyrics,
                        "src": response.get("src"),
                        "confidence": float(response.get("confidence", 0.0)),
                        "time": float(response.get("time", -1.0)),
                    }

                    lyrics = response_obj.get("lyrics")
                    if not lyrics:
                        return None
                    # Use different limits based on channel type
                    if is_spam_channel:
                        # Spam channels: higher limits for more content per page
                        response_obj["lyrics"] = self._smart_lyrics_wrap(lyrics.strip(), max_length=8000, max_verses=100, max_lines=150)
                    else:
                        # Non-spam channels: much shorter limits for better UX in regular channels
                        response_obj["lyrics"] = self._smart_lyrics_wrap(lyrics.strip(), max_length=2000, max_verses=15, max_lines=25)
                    
                    response_obj["lyrics_short"] = self._smart_lyrics_wrap(lyrics.strip(), max_length=500, single_page=True)

                    return [
                        (
                            response_obj.get("artist"),
                            response_obj.get("song"),
                            response_obj.get("src"),
                            f"{int(response_obj.get('confidence', -1.0))}%",
                            f"{response_obj.get('time', -666.0):.4f}s",
                        ),
                        response_obj.get("lyrics"),
                        response_obj.get("lyrics_short"),
                    ]
        except Exception as e:
            traceback.print_exc()
            return [f"Retrieval failed: {e!s}"]