Add tests for materialize, from_dvc side-effect, dir manifest

- `Artifact.from_dvc()` for side-effect stages (no outs)
- `materialize()`: single run, skip-fresh, error-raises
- `find_parent_dvc_dir()`: basic lookup, nested paths, not-found
- `read_dir_manifest()`: JSON parsing, .dir suffix, missing manifest

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ryan-williams

tests/test_run_artifact.py

@@ -1,11 +1,12 @@
 """Tests for dvx.run.artifact module."""
 
+import os
 from pathlib import Path
 
 import pytest
 import yaml
 
-from dvx.run.artifact import Artifact, Computation, delayed, write_all_dvc
+from dvx.run.artifact import Artifact, Computation, delayed, materialize, write_all_dvc
 
 
 def test_artifact_basic(tmp_path):
@@ -297,3 +298,85 @@ def test_artifact_from_dvc_with_git_deps(tmp_path):
     # Check blob SHAs are stored as md5 on the Artifact objects
     git_dep_map = {d.path: d.md5 for d in artifact.computation.git_deps}
     assert git_dep_map == {"script.py": "aabbccdd", "lib.py": "eeff0011"}
+
+
+def test_artifact_from_dvc_side_effect(tmp_path):
+    """Test Artifact.from_dvc() for side-effect stages (no outs)."""
+    dvc_file = tmp_path / "deploy.dvc"
+    dvc_content = {
+        "meta": {
+            "computation": {
+                "cmd": "wrangler pages deploy dist",
+                "deps": {"dist/index.html": "aaa111"},
+            }
+        }
+    }
+    with open(dvc_file, "w") as f:
+        yaml.dump(dvc_content, f)
+
+    artifact = Artifact.from_dvc(tmp_path / "deploy")
+
+    assert artifact is not None
+    assert artifact.path == "deploy"
+    assert artifact.md5 is None
+    assert artifact.computation is not None
+    assert artifact.computation.cmd == "wrangler pages deploy dist"
+    assert len(artifact.computation.deps) == 1
+
+
+def test_materialize_single(tmp_path):
+    """Test materialize() runs a computation and updates the artifact."""
+    os.chdir(tmp_path)
+
+    output = tmp_path / "result.txt"
+    artifact = Artifact(
+        path=str(output),
+        computation=Computation(cmd=f"echo hello > {output}"),
+    )
+
+    computed = materialize([artifact], update_dvc=False)
+
+    assert len(computed) == 1
+    assert output.exists()
+    assert output.read_text().strip() == "hello"
+    assert computed[0].md5 is not None
+
+
+def test_materialize_skips_fresh(tmp_path):
+    """Test materialize() skips already-fresh artifacts (doesn't re-run cmd)."""
+    os.chdir(tmp_path)
+
+    output = tmp_path / "result.txt"
+    output.write_text("existing\n")
+
+    from dvx.run.hash import compute_md5
+    from dvx.run.dvc_files import write_dvc_file
+    md5 = compute_md5(output)
+
+    # Write .dvc so it's "fresh"
+    write_dvc_file(output_path=output, md5=md5, size=output.stat().st_size)
+
+    artifact = Artifact(
+        path=str(output),
+        md5=md5,
+        computation=Computation(cmd="echo should-not-run"),
+    )
+
+    materialize([artifact], update_dvc=False)
+
+    # File content should be unchanged (cmd was not executed)
+    assert output.read_text() == "existing\n"
+
+
+def test_materialize_error_raises(tmp_path):
+    """Test materialize() raises on command failure."""
+    os.chdir(tmp_path)
+
+    output = tmp_path / "result.txt"
+    artifact = Artifact(
+        path=str(output),
+        computation=Computation(cmd="false"),  # always fails
+    )
+
+    with pytest.raises(RuntimeError, match="Computation failed"):
+        materialize([artifact], update_dvc=False)

tests/test_run_dvc_files.py

@@ -8,9 +8,11 @@
 
 from dvx.run.dvc_files import (
     DVCFileInfo,
+    find_parent_dvc_dir,
     get_dvc_file_path,
     get_freshness_details,
     is_output_fresh,
+    read_dir_manifest,
     read_dvc_file,
     write_dvc_file,
 )
@@ -897,3 +899,96 @@ def test_is_fetch_due_naive_last_run():
     last = "2026-04-07T12:00:00"  # No timezone
     now = datetime(2026, 4, 8, 13, 0, 0, tzinfo=timezone.utc)
     assert is_fetch_due("daily", last, now=now) is True
+
+
+# =============================================================================
+# find_parent_dvc_dir / read_dir_manifest tests
+# =============================================================================
+
+
+def test_find_parent_dvc_dir_basic(tmp_path):
+    """find_parent_dvc_dir finds .dvc-tracked parent directory."""
+    # Create a directory tracked by DVC
+    data_dir = tmp_path / "data"
+    data_dir.mkdir()
+    (data_dir / "file1.txt").write_text("content1\n")
+    (data_dir / "sub").mkdir()
+    (data_dir / "sub" / "file2.txt").write_text("content2\n")
+
+    # Write .dvc file for the directory
+    dvc_file = tmp_path / "data.dvc"
+    dvc_content = {
+        "outs": [{"md5": "abc123.dir", "size": 1000, "nfiles": 2, "path": "data"}]
+    }
+    with open(dvc_file, "w") as f:
+        yaml.dump(dvc_content, f)
+
+    # Find parent for a file inside the directory
+    result = find_parent_dvc_dir(tmp_path / "data" / "file1.txt")
+    assert result is not None
+    parent_dir, relpath = result
+    assert parent_dir == tmp_path / "data"
+    assert relpath == "file1.txt"
+
+    # Find parent for a nested file
+    result2 = find_parent_dvc_dir(tmp_path / "data" / "sub" / "file2.txt")
+    assert result2 is not None
+    parent_dir2, relpath2 = result2
+    assert parent_dir2 == tmp_path / "data"
+    assert relpath2 == "sub/file2.txt"
+
+
+def test_find_parent_dvc_dir_not_found(tmp_path):
+    """find_parent_dvc_dir returns None when no parent .dvc exists."""
+    (tmp_path / "untracked.txt").write_text("hello\n")
+    result = find_parent_dvc_dir(tmp_path / "untracked.txt")
+    assert result is None
+
+
+def test_read_dir_manifest(tmp_path):
+    """read_dir_manifest reads .dir JSON manifest from cache."""
+    import json
+
+    # Create cache structure
+    cache_dir = tmp_path / "cache"
+    cache_dir.mkdir()
+    subdir = cache_dir / "ab"
+    subdir.mkdir()
+
+    # Write manifest file (hash = "abc123...", prefix "ab", rest "c123...")
+    manifest = [
+        {"md5": "111222333", "relpath": "file1.txt"},
+        {"md5": "444555666", "relpath": "sub/file2.txt"},
+    ]
+    manifest_file = subdir / "c123def456.dir"
+    manifest_file.write_text(json.dumps(manifest))
+
+    result = read_dir_manifest("abc123def456", cache_dir)
+
+    assert result == {"file1.txt": "111222333", "sub/file2.txt": "444555666"}
+
+
+def test_read_dir_manifest_with_dir_suffix(tmp_path):
+    """read_dir_manifest handles hash with .dir suffix."""
+    import json
+
+    cache_dir = tmp_path / "cache"
+    cache_dir.mkdir()
+    subdir = cache_dir / "ab"
+    subdir.mkdir()
+
+    manifest = [{"md5": "aaa", "relpath": "data.csv"}]
+    (subdir / "c123.dir").write_text(json.dumps(manifest))
+
+    # Pass hash with .dir suffix
+    result = read_dir_manifest("abc123.dir", cache_dir)
+    assert result == {"data.csv": "aaa"}
+
+
+def test_read_dir_manifest_missing(tmp_path):
+    """read_dir_manifest returns empty dict for missing manifest."""
+    cache_dir = tmp_path / "cache"
+    cache_dir.mkdir()
+
+    result = read_dir_manifest("nonexistent", cache_dir)
+    assert result == {}