src/domain/models/text_content.py

"""TextContent value object for representing text data with language and encoding validation."""

from dataclasses import dataclass
from typing import Optional
import re


@dataclass(frozen=True)
class TextContent:
    """Value object representing text content with language and encoding information."""

    text: str
    language: str
    encoding: str = 'utf-8'

    def __post_init__(self):
        """Validate text content after initialization."""
        self._validate()

    def _validate(self):
        """Validate text content properties."""
        if not isinstance(self.text, str):
            raise TypeError("Text must be a string")

        if not self.text.strip():
            raise ValueError("Text content cannot be empty or whitespace only")

        if len(self.text) > 50000:  # Reasonable limit for TTS processing
            raise ValueError("Text content too long (maximum 50,000 characters)")

        if not isinstance(self.language, str):
            raise TypeError("Language must be a string")

        if not self.language.strip():
            raise ValueError("Language cannot be empty")

        # Validate language code format (ISO 639-1 or ISO 639-3)
        if not re.match(r'^[a-z]{2,3}(-[A-Z]{2})?$', self.language):
            raise ValueError(f"Invalid language code format: {self.language}. Expected format: 'en', 'en-US', etc.")

        if not isinstance(self.encoding, str):
            raise TypeError("Encoding must be a string")

        if self.encoding not in ['utf-8', 'utf-16', 'ascii', 'latin-1']:
            raise ValueError(f"Unsupported encoding: {self.encoding}. Supported: utf-8, utf-16, ascii, latin-1")

        # Validate that text can be encoded with specified encoding
        try:
            self.text.encode(self.encoding)
        except UnicodeEncodeError:
            raise ValueError(f"Text cannot be encoded with {self.encoding} encoding")

    @property
    def word_count(self) -> int:
        """Get the approximate word count of the text."""
        return len(self.text.split())

    @property
    def character_count(self) -> int:
        """Get the character count of the text."""
        return len(self.text)

    @property
    def is_empty(self) -> bool:
        """Check if the text content is effectively empty."""
        return not self.text.strip()

    def truncate(self, max_length: int) -> 'TextContent':
        """Create a new TextContent with truncated text."""
        if max_length <= 0:
            raise ValueError("Max length must be positive")

        if len(self.text) <= max_length:
            return self

        truncated_text = self.text[:max_length].rstrip()
        return TextContent(
            text=truncated_text,
            language=self.language,
            encoding=self.encoding
        )