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

harness/tests/test_corpus_pipeline.py

@@ -116,6 +116,29 @@ def pipeline(tmp_path, source, *, embedder=None, vectors=None, model="model-a",
     )
 
 
+def test_pipeline_routes_source_io_through_evidence_facade(tmp_path, monkeypatch):
+    import tht.evidence.acquisition as evidence_acquisition
+
+    calls = []
+
+    def discover(source):
+        calls.append(("discover", source))
+        return source.discover()
+
+    def acquire(source, source_item):
+        calls.append(("acquire", source_item.source_id))
+        return source.acquire(source_item)
+
+    monkeypatch.setattr(evidence_acquisition, "discover", discover)
+    monkeypatch.setattr(evidence_acquisition, "acquire", acquire)
+    source = Source([(item("one", "a"), "body")])
+
+    result = pipeline(tmp_path, source).run()
+
+    assert result.status == "succeeded"
+    assert calls == [("discover", source), ("acquire", "fs:one")]
+
+
 def test_retention_bounds_generations_and_purges_vectors_after_publish(tmp_path):
     vectors = Vectors()
     generations = []

harness/tests/test_evidence_facade_contract.py

@@ -1,5 +1,6 @@
 from datetime import UTC, datetime
 import hashlib
+import inspect
 from types import SimpleNamespace
 
 import pytest
@@ -10,7 +11,9 @@ from tht.decisions import DecisionRecord
 from tht.evidence import (
     acquire,
     active_searcher,
+    build_preprocessing_pipeline,
     build_retrieval_entries,
+    build_sources,
     discover,
     project_session,
     resolve_citation,
@@ -87,6 +90,67 @@ def test_acquisition_facade_preserves_classified_errors():
     assert captured.value.details == {"operation": "download"}
 
 
+def test_source_factory_preserves_legacy_first_order_and_filesystem_configuration(tmp_path):
+    from tht.adapters.factory import build_evidence_sources
+
+    legacy_root = tmp_path / "legacy"
+    configured_root = tmp_path / "configured"
+    (legacy_root / "evidence").mkdir(parents=True)
+    configured_root.mkdir()
+    cfg = SimpleNamespace(evidence=SimpleNamespace(
+        source_root=legacy_root,
+        evidence_dir="evidence",
+        sources=[SimpleNamespace(
+            type="filesystem",
+            root=configured_root,
+            patterns=("*.md",),
+            max_bytes=1024,
+        )],
+    ))
+
+    legacy = build_evidence_sources(cfg)
+    current = build_sources(cfg.evidence)
+
+    assert [type(source) for source in current] == [type(source) for source in legacy]
+    assert [source.root for source in current] == [
+        (legacy_root / "evidence").resolve(),
+        configured_root.resolve(),
+    ]
+    assert current[1].patterns == legacy[1].patterns == ("*.md",)
+    assert current[1].max_bytes == legacy[1].max_bytes == 1024
+
+
+def test_preprocessing_factory_forwards_only_evidence_pipeline_dependencies(monkeypatch):
+    captured = {}
+
+    class FakePipeline:
+        def __init__(self, **kwargs):
+            captured.update(kwargs)
+
+    monkeypatch.setattr("tht.corpus.pipeline.CorpusPipeline", FakePipeline)
+    dependencies = {
+        "store": object(),
+        "sources": [object()],
+        "embedder": object(),
+        "vector_store": object(),
+        "embedding_model": "model",
+        "embedding_dimensions": 3,
+        "chunk_policy": object(),
+        "pipeline_version": "evidence-v1",
+        "retain_published_generations": 2,
+        "workspace_id": None,
+    }
+
+    pipeline = build_preprocessing_pipeline(**dependencies)
+
+    assert isinstance(pipeline, FakePipeline)
+    assert captured == dependencies
+    assert all(
+        parameter.kind is not inspect.Parameter.VAR_KEYWORD
+        for parameter in inspect.signature(build_preprocessing_pipeline).parameters.values()
+    )
+
+
 def _active_config(tmp_path):
     store = CorpusStore(tmp_path / "corpus")
     generation = store.stage(

harness/tests/test_preprocess_cli.py

@@ -247,10 +247,13 @@ def test_run_from_config_uses_runtime_identity_workspace_id(monkeypatch, tmp_pat
             calls["run_as_job"] = kwargs
             return SimpleNamespace(model_dump=lambda mode=None: {"status": "succeeded"})
 
-    monkeypatch.setattr("tht.adapters.factory.build_evidence_sources", lambda cfg: [])
+    monkeypatch.setattr("tht.evidence.build_sources", lambda cfg: [])
+    monkeypatch.setattr(
+        "tht.evidence.build_preprocessing_pipeline",
+        lambda **kwargs: FakePipeline(**kwargs),
+    )
     monkeypatch.setattr("tht.adapters.factory.build_vector_store", lambda cfg, require_write: object())
     monkeypatch.setattr("tht.cli.vector_cmd.make_embedder", lambda cfg: object())
-    monkeypatch.setattr("tht.corpus.pipeline.CorpusPipeline", FakePipeline)
 
     command.run_from_config(config)