ttsforge 0.1.0__py3-none-any.whl

ttsforge/vocab/__init__.py

@@ -0,0 +1,139 @@
+"""Vocabulary management for ttsforge tokenizer.
+
+This module provides a compatibility layer that wraps kokorog2p's vocabulary
+functions for backward compatibility with existing ttsforge code.
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+import kokorog2p
+from kokorog2p.vocab import (
+    get_vocab as _get_kokoro_vocab,
+    get_vocab_reverse as _get_vocab_reverse,
+    get_config as _get_kokoro_config,
+    N_TOKENS,
+    PAD_IDX,
+    encode,
+    decode,
+    validate_for_kokoro,
+    filter_for_kokoro,
+    phonemes_to_ids,
+    ids_to_phonemes,
+)
+
+if TYPE_CHECKING:
+    pass
+
+# Default version identifier (for compatibility)
+DEFAULT_VERSION = "v1.0"
+
+# Supported version strings (for backward compatibility)
+SUPPORTED_VERSIONS = {"v1.0"}
+
+
+def get_config_path() -> Path:
+    """Get the path to the config.json (compatibility function).
+
+    Returns:
+        Path to the embedded kokoro_config.json in kokorog2p.
+
+    Note:
+        This now returns the path to kokorog2p's embedded config,
+        not the downloaded config.json from onnx_backend.
+    """
+    import kokorog2p.data
+
+    return Path(kokorog2p.data.__file__).parent / "kokoro_config.json"
+
+
+def is_config_downloaded() -> bool:
+    """Check if config is available (always True with kokorog2p).
+
+    Returns:
+        True (kokorog2p embeds the vocabulary)
+    """
+    return True
+
+
+def load_vocab(config_path: Path | str | None = None) -> dict[str, int]:
+    """Load vocabulary from kokorog2p.
+
+    Args:
+        config_path: Ignored (kept for backward compatibility).
+            The vocabulary is now loaded from kokorog2p's embedded data.
+
+    Returns:
+        Dictionary mapping phoneme strings to token IDs.
+
+    Raises:
+        ValueError: If an unknown version string is provided.
+    """
+    # Handle backward compatibility with version strings
+    if isinstance(config_path, str):
+        if config_path in SUPPORTED_VERSIONS:
+            # Version string provided, use kokorog2p vocab
+            pass
+        elif config_path.startswith("v") and "." in config_path:
+            # Looks like a version string but not supported
+            raise ValueError(
+                f"Unknown vocabulary version: {config_path}. "
+                f"Supported versions: {', '.join(sorted(SUPPORTED_VERSIONS))}"
+            )
+        # Otherwise ignore and use kokorog2p vocab
+
+    return _get_kokoro_vocab()
+
+
+def get_vocab_info(config_path: Path | str | None = None) -> dict:
+    """Get metadata about the vocabulary.
+
+    Args:
+        config_path: Ignored (kept for backward compatibility).
+
+    Returns:
+        Dictionary with vocabulary metadata.
+    """
+    vocab = _get_kokoro_vocab()
+    return {
+        "version": DEFAULT_VERSION,
+        "path": str(get_config_path()),
+        "num_tokens": len(vocab),
+        "max_token_id": max(vocab.values()) if vocab else 0,
+        "n_tokens": N_TOKENS,
+        "downloaded": True,
+        "backend": "kokorog2p",
+    }
+
+
+def list_versions() -> list[str]:
+    """List all available vocabulary versions.
+
+    Returns:
+        List of version strings. Currently only "v1.0" is supported.
+    """
+    return [DEFAULT_VERSION]
+
+
+# Re-export kokorog2p vocabulary functions for convenience
+__all__ = [
+    # Compatibility functions
+    "DEFAULT_VERSION",
+    "SUPPORTED_VERSIONS",
+    "get_config_path",
+    "is_config_downloaded",
+    "load_vocab",
+    "get_vocab_info",
+    "list_versions",
+    # kokorog2p re-exports
+    "N_TOKENS",
+    "PAD_IDX",
+    "encode",
+    "decode",
+    "validate_for_kokoro",
+    "filter_for_kokoro",
+    "phonemes_to_ids",
+    "ids_to_phonemes",
+]