Enhance Ollama model management and performance tuning

- Updated the `provision_llm` function to utilize the `/api/tags` endpoint for verifying available models, improving accuracy in model management.
- Introduced a new `already_have_ollama_tag` function to ensure exact tag matching, preventing mismatches during model checks.
- Enhanced the `pull_stream` function to require a successful status from the API before proceeding, ensuring reliable model downloads.
- Added logic to handle unwritten blob files, improving the robustness of the model pulling process.
- Updated documentation and tests to reflect these changes, ensuring clarity and reliability in Ollama model operations.

tests/test_ollama_pull.py

@@ -21,7 +21,95 @@ def test_exact_tag_only():
     assert already_have(have, "qwen2.5:3b")
 
 
-def test_latest_alias():
-    assert already_have({"foo:latest"}, "foo")
-    assert already_have({"foo"}, "foo:latest")
-    assert not already_have({"foo:3b"}, "foo")
+def test_pull_stream_requires_success_then_tags(monkeypatch):
+    class FakeResp:
+        def __init__(self, lines: list[str]):
+            self._lines = [ln.encode() for ln in lines]
+            self._i = 0
+
+        def readline(self) -> bytes:
+            if self._i >= len(self._lines):
+                return b""
+            row = self._lines[self._i]
+            self._i += 1
+            return row
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *a):
+            return False
+
+    lines = [
+        '{"status":"pulling manifest"}\n',
+        '{"status":"downloading","total":100,"completed":100}\n',
+    ]
+    monkeypatch.setattr(
+        _mod.urllib.request, "urlopen", lambda *a, **k: FakeResp(lines)
+    )
+    cli = {"n": 0}
+
+    def fake_cli(*_a, **_k):
+        cli["n"] += 1
+
+    monkeypatch.setattr(_mod, "pull_cli", fake_cli)
+    monkeypatch.setattr(_mod, "api_tags", lambda: set())
+    monkeypatch.setattr(_mod, "listed", lambda: set())
+    monkeypatch.setattr(_mod.time, "sleep", lambda *_a, **_k: None)
+    try:
+        _mod.pull_stream("foo:7b", "x", tags_wait=0)
+        assert False, "expected RuntimeError"
+    except RuntimeError as exc:
+        assert "/api/tags" in str(exc)
+    assert cli["n"] == 2
+
+
+def test_pull_stream_ok_when_success_and_tags(monkeypatch):
+    class FakeResp:
+        def __init__(self, lines: list[str]):
+            self._lines = [ln.encode() for ln in lines]
+            self._i = 0
+
+        def readline(self) -> bytes:
+            if self._i >= len(self._lines):
+                return b""
+            row = self._lines[self._i]
+            self._i += 1
+            return row
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *a):
+            return False
+
+    lines = ['{"status":"success"}\n']
+    monkeypatch.setattr(
+        _mod.urllib.request, "urlopen", lambda *a, **k: FakeResp(lines)
+    )
+    monkeypatch.setattr(_mod, "api_tags", lambda: {"foo:7b"})
+    monkeypatch.setattr(_mod, "listed", lambda: set())
+    _mod.pull_stream("foo:7b", "x", tags_wait=0)
+
+
+def test_unwritten_blob_is_all_nuls(tmp_path):
+    z = tmp_path / "zeros"
+    z.write_bytes(b"\x00" * 23)
+    real = tmp_path / "gguf"
+    real.write_bytes(b"GGUF" + b"\x00" * 12)
+    empty = tmp_path / "empty"
+    empty.write_bytes(b"")
+    assert _mod.is_unwritten_blob(z)
+    assert not _mod.is_unwritten_blob(real)
+    assert not _mod.is_unwritten_blob(empty)
+
+
+def test_purge_nul_blobs_keeps_gguf(tmp_path):
+    blobs = tmp_path / "blobs"
+    blobs.mkdir()
+    (blobs / "sha256-dead").write_bytes(b"\x00" * 28)
+    (blobs / "sha256-gguf").write_bytes(b"GGUF\x03\x00\x00\x00")
+    n = _mod.purge_nul_blobs(tmp_path)
+    assert n == 1
+    assert not (blobs / "sha256-dead").exists()
+    assert (blobs / "sha256-gguf").exists()