ThothII/harness/tht/adapters/factory.py

"""Central construction of deployment-specific adapters."""

from tht.adapters.dwh import PostgresDwhAdapter, ThothRestDwhAdapter
from tht.adapters.evidence import FilesystemEvidenceSource, HttpManifestEvidenceSource
from tht.adapters.evidence.s3 import S3EvidenceSource
from tht.adapters.vector import PgVectorStore, QdrantVectorStore, ThothHttpVectorStore
from tht.config import Config, ConfigError
from tht.db.connection import make_engine
from tht.ports.dwh import DwhAdapter
from tht.ports.vector import VectorStore
from tht.vectorstore.rest_client import VectorRestClient


def build_dwh(cfg: Config) -> DwhAdapter:
    """Build the DWH adapter selected by the validated workspace resource."""
    resource = cfg.dwh
    match resource.type:
        case "postgres_direct":
            return PostgresDwhAdapter(
                resource.connection,
                statement_timeout_ms=cfg.execution.statement_timeout_ms,
            )
        case "thoth_rest":
            return ThothRestDwhAdapter(resource.database, resource.endpoint)
        case other:  # pragma: no cover - Pydantic's discriminator rejects this first.
            raise ConfigError(f"Adapter DWH non supportato: {other}")


def build_vector_store(cfg: Config, *, require_write: bool = False) -> VectorStore:
    """Build the vector adapter, optionally requiring write capability."""
    resource = cfg.vectors
    if resource is None:
        raise ConfigError("Risorsa vectors non configurata")

    match resource.type:
        case "pgvector_direct":
            reader = resource.reader or resource.connection
            # Legacy server workspaces use one RW `vector_db` connection. Keep
            # that deployment contract without turning a workstation's legacy
            # compatibility connection into an implicit writer.
            writer = resource.writer or (
                resource.connection if cfg.profile == "server" else None
            )
            if require_write and writer is None:
                raise ConfigError("Vector writer non configurato per pgvector_direct")
            return PgVectorStore(
                reader,
                writer,
                expected_dimension=cfg.embeddings.dim if cfg.embeddings is not None else None,
            )
        case "thoth_vector_http":
            if require_write and resource.writer is None:
                raise ConfigError("Vector writer non configurato")
            return ThothHttpVectorStore(
                VectorRestClient(resource.reader) if resource.reader is not None else None,
                VectorRestClient(resource.writer) if resource.writer is not None else None,
                expected_dimension=cfg.embeddings.dim if cfg.embeddings is not None else None,
            )
        case "qdrant":
            return QdrantVectorStore(
                base_url=resource.base_url,
                collection=resource.collection,
                workspace_id=cfg._workspace_id,
                workspace_revision=cfg._workspace_revision,
                expected_dimension=cfg.embeddings.dim if cfg.embeddings is not None else None,
            )
        case other:  # pragma: no cover - Pydantic's discriminator rejects this first.
            raise ConfigError(f"Adapter vector non supportato: {other}")


def build_vector_loader(cfg: Config, collection: str):
    """Compatibility construction for legacy collection sync commands."""
    resource = cfg.vectors
    if resource is None:
        raise ConfigError("Risorsa vectors non configurata")
    if cfg.embeddings is None:
        raise ConfigError("Embeddings non configurati")

    if (
        resource.type == "thoth_vector_http"
        and resource.writer is not None
        and (cfg.profile == "workstation" or resource.direct is None)
    ):
        from tht.vectorstore.rest_writer import RestVectorWriter

        return RestVectorWriter(VectorRestClient(resource.writer), table=collection)

    connection = (
        resource.writer or resource.connection
        if resource.type == "pgvector_direct"
        else resource.direct
    )
    if connection is None:
        raise ConfigError("Vector writer non configurato")
    from tht.vectorstore.store import VectorStore as TableVectorStore

    return TableVectorStore(
        make_engine(connection),
        schema=connection.db_schema,
        table=collection,
        dim=cfg.embeddings.dim,
    )


def build_evidence_sources(cfg: Config):
    """Build configured Evidence sources, including the legacy curated filesystem tree."""
    evidence = cfg.evidence
    if evidence is None:
        return []
    sources = []
    if evidence.source_root is not None:
        sources.append(FilesystemEvidenceSource(evidence.source_root / evidence.evidence_dir))
    for resource in evidence.sources:
        match resource.type:
            case "filesystem":
                sources.append(
                    FilesystemEvidenceSource(
                        resource.root,
                        patterns=resource.patterns,
                        max_bytes=resource.max_bytes,
                    )
                )
            case "http":
                sources.append(
                    HttpManifestEvidenceSource(
                        [url.get_secret_value() for url in resource.urls],
                        connect_timeout=resource.connect_timeout,
                        read_timeout=resource.read_timeout,
                        max_bytes=resource.max_bytes,
                        max_redirects=resource.max_redirects,
                        allow_private_hosts=resource.allow_private_hosts,
                        max_cache_bytes=resource.max_cache_bytes,
                    )
                )
            case "s3":
                def secret(value):
                    return value.get_secret_value() if value is not None else None

                sources.append(S3EvidenceSource(
                    bucket=resource.bucket, prefix=resource.prefix,
                    endpoint_url=resource.endpoint_url, region=resource.region,
                    access_key=secret(resource.access_key), secret_key=secret(resource.secret_key),
                    session_token=secret(resource.session_token),
                    trusted_endpoint=resource.trusted_endpoint,
                    allow_private_endpoint=resource.allow_private_endpoint,
                    allow_insecure_endpoint=resource.allow_insecure_endpoint,
                    max_bytes=resource.max_bytes, max_objects=resource.max_objects,
                    max_pages=resource.max_pages, page_size=resource.page_size,
                ))
            case other:  # pragma: no cover - Pydantic rejects unsupported discriminators.
                raise ConfigError(f"Adapter evidence non supportato: {other}")
    return sources


__all__ = ["build_dwh", "build_evidence_sources", "build_vector_loader", "build_vector_store"]