index: make build_entry raise on FileNotFound

Pre-requisite for fixing iterative/dvc#9541

src/dvc_data/index/build.py

@@ -21,10 +21,7 @@ def build_entry(
     hash_name: str = DEFAULT_ALGORITHM,
 ):
     if info is None:
-        try:
-            info = fs.info(path)
-        except FileNotFoundError:
-            return DataIndexEntry()
+        info = fs.info(path)
 
     if compute_hash and info["type"] != "directory":
         meta, hash_info = hash_file(
@@ -62,13 +59,16 @@ def build_entries(
             root_key = fs.path.relparts(root, path)
 
         for name in chain(dirs, files):
-            entry = build_entry(
-                fs.path.join(root, name),
-                fs,
-                compute_hash=compute_hash,
-                state=state,
-                hash_name=hash_name,
-            )
+            try:
+                entry = build_entry(
+                    fs.path.join(root, name),
+                    fs,
+                    compute_hash=compute_hash,
+                    state=state,
+                    hash_name=hash_name,
+                )
+            except FileNotFoundError:
+                entry = DataIndexEntry()
             entry.key = (*root_key, name)
             yield entry
 

tests/index/test_build.py

@@ -0,0 +1,44 @@
+import pytest
+
+from dvc_data.index.build import DataIndexEntry, build, build_entry
+
+
+def test_build_entry(tmp_upath, as_filesystem):
+    (tmp_upath / "foo").write_bytes(b"foo\n")
+
+    fs = as_filesystem(tmp_upath.fs)
+
+    entry = build_entry(str(tmp_upath / "foo"), fs)
+    assert isinstance(entry, DataIndexEntry)
+    assert entry.meta.size == 4
+    assert entry.key is None
+    assert entry.hash_info is None
+
+    with pytest.raises(FileNotFoundError):
+        build_entry(str(tmp_upath / "missing"), fs)
+
+
+def test_build(tmp_upath, as_filesystem):
+    (tmp_upath / "foo").write_bytes(b"foo\n")
+    (tmp_upath / "data").mkdir()
+    (tmp_upath / "data" / "bar").write_bytes(b"bar\n")
+    (tmp_upath / "data" / "baz").write_bytes(b"baz\n")
+
+    fs = as_filesystem(tmp_upath.fs)
+    index = build(str(tmp_upath), fs)
+    assert index[("foo",)].meta.size == 4
+    assert index.storage_map.get_data(index[("foo",)]) == (
+        fs,
+        str(tmp_upath / "foo"),
+    )
+    assert index[("data",)].meta.isdir
+    assert index[("data", "bar")].meta.size == 4
+    assert index.storage_map.get_data(index[("data", "bar")]) == (
+        fs,
+        str(tmp_upath / "data" / "bar"),
+    )
+    assert index[("data", "baz")].meta.size == 4
+    assert index.storage_map.get_data(index[("data", "baz")]) == (
+        fs,
+        str(tmp_upath / "data" / "baz"),
+    )

tests/index/test_index.py

@@ -146,32 +146,6 @@ def test_save(tmp_upath, odb, as_filesystem):
     assert odb.exists("258622b1688250cb619f3c9ccaefb7eb")
 
 
-def test_build(tmp_upath, as_filesystem):
-    (tmp_upath / "foo").write_bytes(b"foo\n")
-    (tmp_upath / "data").mkdir()
-    (tmp_upath / "data" / "bar").write_bytes(b"bar\n")
-    (tmp_upath / "data" / "baz").write_bytes(b"baz\n")
-
-    fs = as_filesystem(tmp_upath.fs)
-    index = build(str(tmp_upath), fs)
-    assert index[("foo",)].meta.size == 4
-    assert index.storage_map.get_data(index[("foo",)]) == (
-        fs,
-        str(tmp_upath / "foo"),
-    )
-    assert index[("data",)].meta.isdir
-    assert index[("data", "bar")].meta.size == 4
-    assert index.storage_map.get_data(index[("data", "bar")]) == (
-        fs,
-        str(tmp_upath / "data" / "bar"),
-    )
-    assert index[("data", "baz")].meta.size == 4
-    assert index.storage_map.get_data(index[("data", "baz")]) == (
-        fs,
-        str(tmp_upath / "data" / "baz"),
-    )
-
-
 def test_add(tmp_upath, as_filesystem):
     (tmp_upath / "foo").write_bytes(b"foo\n")
     (tmp_upath / "data").mkdir()