Adding helper functions for index.db caching

iib/workers/tasks/oras_utils.py

@@ -8,7 +8,7 @@
 
 from iib.common.tracing import instrument_tracing
 from iib.exceptions import IIBError
-from iib.workers.tasks.utils import run_cmd, set_registry_auths
+from iib.workers.tasks.utils import run_cmd, set_registry_auths, get_image_digest
 
 log = logging.getLogger(__name__)
 
@@ -102,3 +102,81 @@ def push_oras_artifact(
             log.info('Successfully pushed OCI artifact to %s', artifact_ref)
         except Exception as e:
             raise IIBError(f'Failed to push OCI artifact to {artifact_ref}: {e}')
+
+
+def get_image_stream_digest(
+    tag: str,
+):
+    """
+    Retrieve the image digest from the OpenShift ImageStream.
+
+    This function queries the `index-db-cache` ImageStream to get the
+    SHA256 digest for a specific tag.
+
+    :param tag: The image tag to check.
+    :return: The image digest (e.g., "sha256:...").
+    :rtype: str
+    """
+    # This JSONPath expression navigates the ImageStream JSON structure to extract the image digest:
+    # - .status.tags: Access the 'tags' array within the 'status' object
+    # - [?(@.tag=="{tag}")]: Filter to find the tag object where the 'tag' field equals
+    #   the specified tag (@. refers to the current item in the array being filtered)
+    # - .items[0]: From the matched tag object, access the first item in its 'items' array
+    # - .image: Extract the 'image' field, which contains the SHA256 digest
+    jsonpath = f'\'{{.status.tags[?(@.tag=="{tag}")].items[0].image}}\''
+    return run_cmd(
+        ['oc', 'get', 'imagestream', 'index-db-cache', '-o', f'jsonpath={jsonpath}'],
+        exc_msg=f'Failed to get digest for ImageStream tag {tag}.',
+    )
+
+
+def verify_indexdb_cache_sync(tag: str) -> bool:
+    """
+    Compare the digest of the ImageStream with the digest of the image in repository.
+
+    This function verifies if the local ImageStream cache is up to date with
+    the latest image in the remote registry.
+
+    :param tag: The image tag to verify.
+    :return: True if the digests match (cache is synced), False otherwise.
+    :rtype: bool
+    """
+    # TODO - This is EXAMPLE location - final one should be loaded from config variable
+    repository = "quay.io/exd-guild-hello-operator/example-repository"
+
+    quay_digest = get_image_digest(f"{repository}:{tag}")
+    is_digest = get_image_stream_digest(tag)
+
+    return quay_digest == is_digest
+
+
+def refresh_indexdb_cache(
+    tag: str,
+    registry_auths: Optional[Dict[str, Any]] = None,
+) -> None:
+    """
+    Force a synchronization of the ImageStream with the remote registry.
+
+    This function imports the specified image from Quay.io into the `index-db-cache`
+    ImageStream, ensuring the local cache is up-to-date.
+
+    :param tag: The container image tag to refresh.
+    :param registry_auths: Optional authentication data for the registry.
+    """
+    log.info('Refreshing OCI artifact cache: %s', tag)
+
+    # TODO - This is EXAMPLE location - final one should be loaded from config variable
+    repository = "quay.io/exd-guild-hello-operator/example-repository"
+
+    # Use namespace-specific registry authentication if provided
+    with set_registry_auths(registry_auths, use_empty_config=True):
+        run_cmd(
+            [
+                'oc',
+                'import-image',
+                f'index-db-cache:{tag}',
+                f'--from={repository}:{tag}',
+                '--confirm',
+            ],
+            exc_msg=f'Failed to refresh OCI artifact {tag}.',
+        )

iib/workers/tasks/utils.py

@@ -503,6 +503,27 @@ def _get_container_image_name(pull_spec: str) -> str:
         return pull_spec.rsplit(':', 1)[0]
 
 
+def get_image_digest(pull_spec: str) -> str:
+    """
+    Get the digest of the image defined by pull_spec.
+
+    :param str pull_spec: the pull specification of the container image
+    :return: the digest of the image
+    :rtype: str
+    """
+    skopeo_output = skopeo_inspect(f'docker://{pull_spec}', '--raw', return_json=False)
+    if json.loads(skopeo_output).get('schemaVersion') == 2:
+        raw_digest = hashlib.sha256(skopeo_output.encode('utf-8')).hexdigest()
+        return f'sha256:{raw_digest}'
+
+    # Schema 1 is not a stable format. The contents of the manifest may change slightly
+    # between requests causing a different digest to be computed. Instead, let's leverage
+    # skopeo's own logic for determining the digest in this case. In the future, we
+    # may want to use skopeo in all cases, but this will have significant performance
+    # issues until https://github.com/containers/skopeo/issues/785
+    return skopeo_inspect(f'docker://{pull_spec}')['Digest']
+
+
 def get_resolved_image(pull_spec: str) -> str:
     """
     Get the pull specification of the container image using its digest.
@@ -513,17 +534,7 @@ def get_resolved_image(pull_spec: str) -> str:
     """
     log.debug('Resolving %s', pull_spec)
     name = _get_container_image_name(pull_spec)
-    skopeo_output = skopeo_inspect(f'docker://{pull_spec}', '--raw', return_json=False)
-    if json.loads(skopeo_output).get('schemaVersion') == 2:
-        raw_digest = hashlib.sha256(skopeo_output.encode('utf-8')).hexdigest()
-        digest = f'sha256:{raw_digest}'
-    else:
-        # Schema 1 is not a stable format. The contents of the manifest may change slightly
-        # between requests causing a different digest to be computed. Instead, let's leverage
-        # skopeo's own logic for determining the digest in this case. In the future, we
-        # may want to use skopeo in all cases, but this will have significant performance
-        # issues until https://github.com/containers/skopeo/issues/785
-        digest = skopeo_inspect(f'docker://{pull_spec}')['Digest']
+    digest = get_image_digest(pull_spec)
     pull_spec_resolved = f'{name}@{digest}'
     log.debug('%s resolved to %s', pull_spec, pull_spec_resolved)
     return pull_spec_resolved

tests/test_workers/test_tasks/test_oras_utils.py

@@ -8,6 +8,9 @@
 from iib.workers.tasks.oras_utils import (
     get_oras_artifact,
     push_oras_artifact,
+    verify_indexdb_cache_sync,
+    get_image_stream_digest,
+    refresh_indexdb_cache,
 )
 
 
@@ -326,3 +329,135 @@ def test_get_oras_artifact_with_base_dir_wont_leak_credentials(
     all_messages = ' '.join(caplog.messages)
     assert 'dXNlcjpwYXNz' not in all_messages  # base64 encoded credentials
     assert 'user:pass' not in all_messages  # decoded credentials
+
+
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd')
+def test_get_image_stream_digest(mock_run_cmd):
+    """Test successful retrieval of image digest from ImageStream."""
+    mock_run_cmd.return_value = 'sha256:12345'
+    tag = 'test-tag'
+
+    digest = get_image_stream_digest(tag)
+
+    assert digest == 'sha256:12345'
+    mock_run_cmd.assert_called_once_with(
+        [
+            'oc',
+            'get',
+            'imagestream',
+            'index-db-cache',
+            '-o',
+            'jsonpath=\'{.status.tags[?(@.tag=="test-tag")].items[0].image}\'',
+        ],
+        exc_msg='Failed to get digest for ImageStream tag test-tag.',
+    )
+
+
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd')
+def test_get_image_stream_digest_empty_string(mock_run_cmd):
+    """Test get_image_stream_digest with empty string output."""
+    mock_run_cmd.return_value = ''
+    tag = 'test-tag'
+    digest = get_image_stream_digest(tag)
+
+    assert digest is None or digest == '', "Expected None or empty digest for empty string output"
+
+
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd')
+def test_get_image_stream_digest_invalid_format(mock_run_cmd):
+    """Test get_image_stream_digest with non-digest output."""
+    mock_run_cmd.return_value = 'not-a-digest'
+    tag = 'test-tag'
+    digest = get_image_stream_digest(tag)
+
+    assert (
+        digest is None or digest == 'not-a-digest'
+    ), "Expected None or raw output for invalid digest format"
+
+
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd', side_effect=IIBError('cmd failed'))
+def test_get_image_stream_digest_failure(mock_run_cmd):
+    """Test failure during retrieval of image digest from ImageStream."""
+    with pytest.raises(IIBError, match='cmd failed'):
+        get_image_stream_digest('test-tag')
+
+
+@mock.patch('iib.workers.tasks.oras_utils.get_image_stream_digest')
+@mock.patch('iib.workers.tasks.oras_utils.get_image_digest')
+def test_verify_indexdb_cache_sync_match(mock_get_image_digest, mock_get_is_digest):
+    """Test successful verification when digests match."""
+    mock_get_image_digest.return_value = 'sha256:abc'
+    mock_get_is_digest.return_value = 'sha256:abc'
+    tag = 'test-tag'
+
+    result = verify_indexdb_cache_sync(tag)
+
+    assert result is True
+    mock_get_image_digest.assert_called_once_with(
+        'quay.io/exd-guild-hello-operator/example-repository:test-tag'
+    )
+    mock_get_is_digest.assert_called_once_with(tag)
+
+
+@mock.patch('iib.workers.tasks.oras_utils.get_image_stream_digest')
+@mock.patch('iib.workers.tasks.oras_utils.get_image_digest')
+def test_verify_indexdb_cache_sync_no_match(mock_get_image_digest, mock_get_is_digest):
+    """Test successful verification when digests don't match."""
+    mock_get_image_digest.return_value = 'sha256:abc'
+    mock_get_is_digest.return_value = 'sha256:xyz'
+    tag = 'test-tag'
+
+    result = verify_indexdb_cache_sync(tag)
+
+    assert result is False
+    mock_get_image_digest.assert_called_once_with(
+        'quay.io/exd-guild-hello-operator/example-repository:test-tag'
+    )
+    mock_get_is_digest.assert_called_once_with(tag)
+
+
+@mock.patch('iib.workers.tasks.oras_utils.set_registry_auths')
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd')
+def test_refresh_indexdb_cache_success(mock_run_cmd, mock_auth, registry_auths):
+    """Test successful cache refresh."""
+    tag = 'test-tag'
+
+    refresh_indexdb_cache(tag, registry_auths)
+
+    mock_auth.assert_called_once_with(registry_auths, use_empty_config=True)
+    mock_run_cmd.assert_called_once_with(
+        [
+            'oc',
+            'import-image',
+            'index-db-cache:test-tag',
+            '--from=quay.io/exd-guild-hello-operator/example-repository:test-tag',
+            '--confirm',
+        ],
+        exc_msg='Failed to refresh OCI artifact test-tag.',
+    )
+
+
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd', side_effect=IIBError('refresh failed'))
+def test_refresh_indexdb_cache_failure(mock_run_cmd):
+    """Test cache refresh failure."""
+    tag = 'test-tag'
+
+    with pytest.raises(IIBError, match='refresh failed'):
+        refresh_indexdb_cache(tag)
+
+
+@mock.patch('iib.workers.tasks.oras_utils.set_registry_auths')
+@mock.patch('iib.workers.tasks.oras_utils.run_cmd')
+def test_refresh_indexdb_cache_with_empty_registry_auths(mock_run_cmd, mock_auth):
+    """Test that refresh_indexdb_cache works correctly when registry_auths is an empty dict."""
+    tag = 'v4.15'
+    empty_auths = {}
+
+    # Call the function with empty registry_auths
+    refresh_indexdb_cache(tag, registry_auths=empty_auths)
+
+    # Verify set_registry_auths was called with empty dict as argument
+    mock_auth.assert_called_once_with(empty_auths, use_empty_config=True)
+
+    # Verify the oc command was executed
+    mock_run_cmd.assert_called_once()

tests/test_workers/test_tasks/test_utils.py

@@ -1,4 +1,6 @@
 # SPDX-License-Identifier: GPL-3.0-or-later
+import hashlib
+import json
 import logging
 import os
 import stat
@@ -528,6 +530,103 @@ def mock_handler(spam, eggs, request_id, bacon):
     assert not logs_dir.listdir()
 
 
+@mock.patch('iib.workers.tasks.utils.skopeo_inspect')
+def test_get_image_digest_schema_2(mock_skopeo_inspect):
+    """
+    Tests the get_image_digest function for an image with schemaVersion 2.
+
+    Verifies that the function correctly calculates the SHA256 digest from the raw JSON output.
+    """
+    # Simulate raw JSON output from skopeo inspect for schema v2
+    raw_output = textwrap.dedent(
+        """
+        {
+           "schemaVersion": 2,
+           "mediaType": "application/vnd.docker.distribution.manifest.v2+json",
+           "config": {
+              "mediaType": "application/vnd.docker.container.image.v1+json",
+              "size": 5545,
+              "digest": "sha256:720713e1a4410985aacd7008719efd13d8a32e76d08d34fca202a60ff43e516d"
+           }
+        }
+        """
+    ).strip()
+
+    # Configure the mock object to return the raw output
+    mock_skopeo_inspect.return_value = raw_output
+
+    # Expected digest (SHA256 hash of the raw_output)
+    expected_digest = f"sha256:{hashlib.sha256(raw_output.encode('utf-8')).hexdigest()}"
+
+    # Run the function under test
+    actual_digest = utils.get_image_digest("some-image:latest")
+
+    # Verify that skopeo_inspect was called with the correct parameters
+    mock_skopeo_inspect.assert_called_once_with(
+        "docker://some-image:latest", "--raw", return_json=False
+    )
+
+    # Verify the result
+    assert actual_digest == expected_digest
+
+
+@mock.patch('iib.workers.tasks.utils.skopeo_inspect')
+def test_get_image_digest_schema_1(mock_skopeo_inspect):
+    """
+    Tests the get_image_digest function for an image with schemaVersion 1.
+
+    Verifies that the function correctly returns the digest from the parsed JSON output
+    (the second skopeo_inspect call).
+    """
+    # Simulate two different skopeo_inspect calls with different return values
+    mock_skopeo_inspect.side_effect = [
+        # First call (with --raw) to check the schema
+        '{"schemaVersion": 1, "name": "repository/name"}',
+        # Second call (without --raw) to get the digest
+        {
+            "Name": "registry.example.com/repository/name",
+            "Digest": "sha256:expected-digest-from-skopeo-output",
+        },
+    ]
+
+    # Expected digest from the second mock call
+    expected_digest = "sha256:expected-digest-from-skopeo-output"
+
+    # Run the function under test
+    actual_digest = utils.get_image_digest("registry.example.com/repository/name:1.0.0")
+
+    # Verify that the skopeo_inspect function was called twice with the correct parameters
+    mock_skopeo_inspect.assert_has_calls(
+        [
+            mock.call(
+                "docker://registry.example.com/repository/name:1.0.0", "--raw", return_json=False
+            ),
+            mock.call("docker://registry.example.com/repository/name:1.0.0"),
+        ]
+    )
+
+    # Verify the result
+    assert actual_digest == expected_digest
+
+
+@mock.patch('iib.workers.tasks.utils.skopeo_inspect')
+def test_get_image_digest_invalid_json(mock_skopeo_inspect):
+    """
+    Tests get_image_digest for invalid JSON output.
+
+    Verifies that the function correctly raises an error if the skopeo output is invalid.
+    """
+    # Simulate invalid JSON output
+    mock_skopeo_inspect.return_value = "This is not valid JSON"
+
+    # We expect the function to raise a ValueError (from json.loads)
+    with pytest.raises(json.JSONDecodeError):
+        utils.get_image_digest("invalid-image:latest")
+
+    # Verify that skopeo_inspect was called
+    mock_skopeo_inspect.assert_called_once()
+
+
 @pytest.mark.parametrize(
     'pull_spec, expected',
     (