ThothII/harness/tests/test_config_resources.py

import pytest

from tht.config import (
    ConfigError,
    PgvectorDirectConfig,
    PostgresDwhConfig,
    ThothRestDwhConfig,
    ThothVectorHttpConfig,
    workspace_id_for_config,
    load_config,
)
from tht.adapters.evidence import FilesystemEvidenceSource, HttpManifestEvidenceSource
from tht.adapters.factory import build_evidence_sources


def test_direct_vector_passwords_load_from_file_references(monkeypatch, tmp_path):
    reader = tmp_path / "reader"
    writer = tmp_path / "writer"
    reader.write_text("reader-secret")
    writer.write_text("writer-secret")
    monkeypatch.setenv("READER_FILE", str(reader))
    monkeypatch.setenv("WRITER_FILE", str(writer))
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text("""
dwh:
  type: postgres_direct
  connection: {database: d, schema: public, user: u, password: p}
vectors:
  type: pgvector_direct
  reader: {database: d, schema: vectors, user: r, password_file: '${READER_FILE}'}
  writer: {database: d, schema: vectors, user: w, password_file: '${WRITER_FILE}'}
""")
    config = load_config(workspace)
    assert config.vectors.reader.password == "reader-secret"
    assert config.vectors.writer.password == "writer-secret"


def test_direct_vector_secret_file_rejects_whitespace(tmp_path):
    secret = tmp_path / "reader"
    secret.write_text("bad secret")
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text(f"""
dwh:
  type: postgres_direct
  connection: {{database: d, schema: public, user: u, password: p}}
vectors:
  type: pgvector_direct
  reader: {{database: d, schema: vectors, user: r, password_file: {secret}}}
""")
    with pytest.raises(ConfigError, match="secret file"):
        load_config(workspace)


def test_loads_discriminated_dwh_and_vector_resources(tmp_path):
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text(
        """
dwh:
  type: thoth_rest
  database:
    database: analytics
    schema: mart
  endpoint:
    base_url: https://dwh.example.test/
    api_key: dwh-reader
vectors:
  type: thoth_vector_http
  reader:
    base_url: https://vectors.example.test/
    api_key: vector-reader
  writer:
    base_url: https://vectors.example.test/
    api_key: vector-writer
roots:
  artifacts: build/artifacts
  indexes: build/indexes
  sessions: build/sessions
"""
    )

    cfg = load_config(workspace)

    assert isinstance(cfg.dwh, ThothRestDwhConfig)
    assert cfg.dwh.database.db_schema == "mart"
    assert isinstance(cfg.vectors, ThothVectorHttpConfig)
    assert cfg.vectors.writer.api_key == "vector-writer"
    assert cfg.roots.sessions.as_posix() == "build/sessions"


def test_runtime_handoff_preserves_canonical_identity_and_durable_roots(monkeypatch, tmp_path):
    data_root = tmp_path / "data"
    runtime_root = data_root / "sessions" / "psd-clinical"
    workspace = tmp_path / "runtime-random-uuid.yaml"
    workspace.write_text(f"""
runtime_identity:
  workspace_id: psd-clinical
  workspace_revision: {'a' * 40}
dwh:
  type: postgres_direct
  connection: {{database: analytics, schema: mart, user: reader, password: secret}}
vectors:
  type: pgvector_direct
  connection: {{database: analytics, schema: vectors, user: vector, password: secret}}
roots:
  sessions: {runtime_root / 'sessions'}
  artifacts: {runtime_root / 'artifacts'}
  indexes: {runtime_root / 'indexes'}
embeddings: {{base_url: http://embedding.invalid, model: embed, dim: 768}}
""")
    monkeypatch.setenv("THT_DATA_ROOT", str(data_root))

    cfg = load_config(workspace)

    assert cfg._workspace_id == "psd-clinical"
    assert cfg._workspace_revision == "a" * 40
    assert cfg.paths.sessions == runtime_root / "sessions"
    assert cfg.paths.artifacts == runtime_root / "artifacts"
    assert cfg.paths.indexes == runtime_root / "indexes"


def test_runtime_identity_is_authoritative_over_runtime_filename(tmp_path):
    workspace = tmp_path / "runtime-random-uuid.yaml"
    workspace.write_text(f"""
runtime_identity:
  workspace_id: psd-clinical
  workspace_revision: {'a' * 40}
dwh:
  type: postgres_direct
  connection: {{database: analytics, schema: mart, user: reader, password: secret}}
roots: {{sessions: sessions, artifacts: artifacts, indexes: indexes}}
""")

    cfg = load_config(workspace)

    assert workspace_id_for_config(cfg, workspace) == "psd-clinical"


def test_load_config_rejects_executable_yaml_tags_without_running_them(tmp_path):
    marker = tmp_path / "must-not-exist"
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text(f"dwh: !command touch {marker}\n")

    with pytest.raises(ConfigError, match="YAML|configurazione"):
        load_config(workspace)

    assert not marker.exists()


def test_loads_direct_discriminated_resources(tmp_path):
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text(
        """
dwh:
  type: postgres_direct
  connection: &database
    host: db
    database: analytics
    schema: mart
    user: reader
    password: secret
vectors:
  type: pgvector_direct
  connection:
    <<: *database
    schema: vectors
"""
    )

    cfg = load_config(workspace)

    assert isinstance(cfg.dwh, PostgresDwhConfig)
    assert cfg.database.transport == "direct"
    assert isinstance(cfg.vectors, PgvectorDirectConfig)
    assert cfg.vector_db.db_schema == "vectors"


def test_loads_writer_only_http_vector_resource(tmp_path):
    workspace = tmp_path / "workspace.yaml"
    workspace.write_text(
        """
dwh:
  type: thoth_rest
  database: {database: analytics, schema: mart}
  endpoint: {base_url: https://dwh.test/, api_key: reader}
vectors:
  type: thoth_vector_http
  writer: {base_url: https://vectors.test/, api_key: writer}
embeddings: {base_url: http://ollama:11434, dim: 768}
"""
    )

    cfg = load_config(workspace)
    assert cfg.vectors.reader is None
    assert cfg.vectors.writer.api_key == "writer"


def test_builds_typed_evidence_sources_and_keeps_legacy_compatible(tmp_path):
    common = """
dwh:
  type: postgres_direct
  connection: {database: d, schema: public, user: u, password: p}
"""
    modern = tmp_path / "modern.yaml"
    modern.write_text(common + f"""
evidence:
  sources:
    - type: filesystem
      root: {tmp_path}
      max_bytes: 123
    - type: http
      urls: ['https://example.test/doc.md?token=transport-only']
""")
    cfg = load_config(modern)
    assert "transport-only" not in repr(cfg.evidence)
    assert "transport-only" not in cfg.evidence.model_dump_json()
    assert cfg.evidence.sources[1].allow_private_hosts is False
    sources = build_evidence_sources(cfg)
    assert isinstance(sources[0], FilesystemEvidenceSource)
    assert isinstance(sources[1], HttpManifestEvidenceSource)
    assert "transport-only" not in repr(sources[1])

    legacy = tmp_path / "legacy.yaml"
    (tmp_path / "curated").mkdir()
    legacy.write_text(common + f"""
evidence:
  source_root: {tmp_path}
  evidence_dir: curated
""")
    legacy_source = build_evidence_sources(load_config(legacy))[0]
    assert isinstance(legacy_source, FilesystemEvidenceSource)
    assert legacy_source.root == (tmp_path / "curated").resolve()