refactor(evidence): migrate acquisition and preprocessing (#30)

harness/tht/adapters/factory.py

@@ -1,8 +1,6 @@
 """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 QdrantVectorStore
 from tht.config import Config, ConfigError
 from tht.ports.dwh import DwhAdapter
@@ -45,53 +43,10 @@ def build_vector_store(cfg: Config, *, require_write: bool = False) -> VectorSto
 
 
 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(
-                        resource.transport_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
+    """Compatibility shim for callers not yet migrated to ``tht.evidence``."""
+    from tht.evidence import build_sources
 
-                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
+    return build_sources(cfg.evidence)
 
 
 __all__ = ["build_dwh", "build_evidence_sources", "build_vector_store"]