Merge pull request #12 from Emilgardis/include-only

implement filter-tags

README.md

@@ -199,6 +199,22 @@ Restrict deletions to images without specific tags, if specified.
 
 Supports Unix-shell style wildcards, i.e 'v*' to match all tags starting with 'v'.
 
+## filter-tags
+
+* **Required**: `No`
+* **Example**: `sha-*`
+
+Comma-separated list of tags to consider for deletion.
+
+Supports Unix-shell style wildcards, i.e 'sha-*' to match all tags starting with 'sha-'.
+
+## filter-include-untagged
+
+* **Required**: `No`
+* **Default**: `true`
+
+Whether to consider untagged images for deletion.
+
 # Nice to knows
 
 * The GitHub API restricts us to fetching 100 image versions per image name, so if your registry isn't 100% clean after

action.yml

@@ -35,6 +35,12 @@ inputs:
     description: 'How many images to keep no matter what. Defaults to 0 which means you might delete everything'
     required: false
     default: '0'
+  filter-tags:
+    description: "Comma-separated list of tags to consider for deletion. Supports Unix-shell style wildcards"
+    required: false
+  filter-include-untagged:
+    description: "Whether to consider untagged images for deletion."
+    required: false
 
 runs:
   using: 'docker'
@@ -49,3 +55,5 @@ runs:
     - ${{ inputs.untagged-only }}
     - ${{ inputs.skip-tags }}
     - ${{ inputs.keep-at-least }}
+    - ${{ inputs.filter-tags }}
+    - ${{ inputs.filter-include-untagged }}

main.py

@@ -5,11 +5,11 @@
 from dataclasses import dataclass
 from distutils.util import strtobool
 from enum import Enum
+from fnmatch import fnmatch
 from functools import partial
 from sys import argv
 from typing import TYPE_CHECKING
 from urllib.parse import quote_from_bytes
-from fnmatch import fnmatch
 
 from dateparser import parse
 from httpx import AsyncClient
@@ -122,6 +122,8 @@ class Inputs:
     untagged_only: bool
     skip_tags: list[str]
     keep_at_least: int
+    filter_tags: list[str]
+    filter_include_untagged: bool
     org_name: Optional[str] = None
 
     def __post_init__(self) -> None:
@@ -186,7 +188,18 @@ async def get_and_delete_old_versions(image_name: ImageName, inputs: Inputs, htt
             # Skipping because no tagged images should be deleted
             continue
 
+        # if untagged and we don't include untagged images, skip this image.
+        if not image_tags and not inputs.filter_include_untagged:
+            continue
+
         skip = False
+        for filter_tag in inputs.filter_tags:
+            if not any(fnmatch(tag, filter_tag) for tag in image_tags):
+                # no image tag matches the filter tag, skip this image
+                skip = True
+        if skip:
+            break
+
         for skip_tag in inputs.skip_tags:
             if any(fnmatch(tag, skip_tag) for tag in image_tags):
                 # Skipping because this image version is tagged with a protected tag
@@ -210,6 +223,8 @@ def validate_inputs(
     untagged_only: Union[bool, str],
     skip_tags: Optional[str],
     keep_at_least: Optional[str],
+    filter_tags: Optional[str],
+    filter_include_untagged: Union[bool, str],
 ) -> Inputs:
     """
     Perform basic validation on the incoming parameters and return an Inputs instance.
@@ -241,6 +256,16 @@ def validate_inputs(
         if keep_at_least_ < 0:
             raise ValueError('keep-at-least must be 0 or positive')
 
+    if filter_tags is None:
+        filter_tags_ = []
+    else:
+        filter_tags_ = [i.strip() for i in filter_tags.split(',')]
+
+    if isinstance(filter_include_untagged, str):
+        filter_include_untagged_ = strtobool(filter_include_untagged) == 1
+    else:
+        filter_include_untagged_ = filter_include_untagged
+
     return Inputs(
         parsed_cutoff=parsed_cutoff,
         timestamp_type=TimestampType(timestamp_type),
@@ -249,6 +274,8 @@ def validate_inputs(
         untagged_only=untagged_only_,
         skip_tags=skip_tags_,
         keep_at_least=keep_at_least_,
+        filter_tags=filter_tags_,
+        filter_include_untagged=filter_include_untagged_,
     )
 
 
@@ -274,6 +301,8 @@ async def main(
     untagged_only: Union[bool, str] = False,
     skip_tags: Optional[str] = None,
     keep_at_least: Optional[str] = None,
+    filter_tags: Optional[str] = None,
+    filter_include_untagged: Union[bool, str] = True,
 ) -> None:
     """
     Delete old image versions.
@@ -289,11 +318,17 @@ async def main(
                             Must contain a reference to the timezone.
     :param token: The personal access token to authenticate with.
     :param untagged_only: Whether to only delete untagged images.
-    :param skip_tags: Comma-separated list of tags to not delete. Supports wildcard '*', '?', '[seq]' and '[!seq]' via Unix shell-style wildcards
+    :param skip_tags: Comma-separated list of tags to not delete.
+        Supports wildcard '*', '?', '[seq]' and '[!seq]' via Unix shell-style wildcards
     :param keep_at_least: Number of images to always keep
+    :param filter_tags: Comma-separated list of tags to consider for deletion.
+        Supports wildcard '*', '?', '[seq]' and '[!seq]' via Unix shell-style wildcards
+    :param filter_include_untagged: Whether to consider untagged images for deletion.
     """
     parsed_image_names: list[ImageName] = parse_image_names(image_names)
-    inputs: Inputs = validate_inputs(account_type, org_name, timestamp_type, cut_off, untagged_only, skip_tags, keep_at_least)
+    inputs: Inputs = validate_inputs(
+        account_type, org_name, timestamp_type, cut_off, untagged_only, skip_tags, keep_at_least, filter_tags, filter_include_untagged
+    )
     headers = {'accept': 'application/vnd.github.v3+json', 'Authorization': f'Bearer {token}'}
 
     async with AsyncClient(headers=headers) as http_client:

main_tests.py

@@ -73,6 +73,8 @@ def test_inputs_dataclass():
         untagged_only=False,
         skip_tags=[],
         keep_at_least=0,
+        filter_tags=[],
+        filter_include_untagged=True,
     )
     assert personal.is_org is False
     assert personal.list_package_versions == list_package_versions
@@ -86,6 +88,8 @@ def test_inputs_dataclass():
         untagged_only=False,
         skip_tags=[],
         keep_at_least=0,
+        filter_tags=[],
+        filter_include_untagged=True,
     )
     assert org.is_org is True
     assert isinstance(org.list_package_versions, partial)
@@ -132,6 +136,8 @@ class TestGetAndDeleteOldVersions:
         'untagged_only': False,
         'skip_tags': [],
         'keep_at_least': '0',
+        'filter_tags': [],
+        'filter_include_untagged': True,
     }
 
     @staticmethod
@@ -241,6 +247,19 @@ async def test_untagged_only(self, capsys):
         captured = capsys.readouterr()
         assert captured.out == 'No more versions to delete for a\n'
 
+    @pytest.mark.asyncio
+    async def test_filter_tags(self, capsys):
+        data = deepcopy(self.valid_data)
+        data[0]['metadata'] = {'container': {'tags': ['sha-deadbeef', 'edge']}}
+
+        Inputs.list_package_versions = partial(self._mock_list_package_versions, data)
+        inputs = Inputs(**self.valid_inputs | {'filter_tags': ['sha-*']})
+
+        await get_and_delete_old_versions(image_name=ImageName('a', 'a'), inputs=inputs, http_client=mock_http_client)
+
+        captured = capsys.readouterr()
+        assert captured.out == 'Deleted old image: a:1234567\n'
+
 
 def test_inputs_bad_account_type():
     defaults = {
@@ -251,6 +270,8 @@ def test_inputs_bad_account_type():
         'untagged_only': False,
         'skip_tags': None,
         'keep_at_least': 0,
+        'filter_tags': None,
+        'filter_include_untagged': True,
     }
 
     # Account type
@@ -294,6 +315,18 @@ def test_inputs_bad_account_type():
     with pytest.raises(ValueError, match='keep-at-least must be 0 or positive'):
         validate_inputs(**defaults | {'keep_at_least': '-1'})
 
+    # Filter tags
+    assert validate_inputs(**defaults | {'filter_tags': 'a'}).filter_tags == ['a']
+    assert validate_inputs(**defaults | {'filter_tags': 'sha-*,latest'}).filter_tags == ['sha-*', 'latest']
+    assert validate_inputs(**defaults | {'filter_tags': 'sha-* , latest'}).filter_tags == ['sha-*', 'latest']
+
+    # Filter include untagged
+    for i in ['true', 'True', '1']:
+        assert validate_inputs(**defaults | {'filter_include_untagged': i}).filter_include_untagged is True
+    for j in ['False', 'false', '0']:
+        assert validate_inputs(**defaults | {'filter_include_untagged': j}).filter_include_untagged is False
+    assert validate_inputs(**defaults | {'filter_include_untagged': False}).filter_include_untagged is False
+
 
 def test_parse_image_names():
     assert parse_image_names('a') == [ImageName('a', 'a')]

pyproject.toml

@@ -17,6 +17,7 @@ pytest-mock = "^3.6.1"
 pytest-asyncio = "^0.15.1"
 pytest-cov = "^2.12.1"
 coverage = {extras = ["toml"], version = "^6.0.2"}
+black = {version = "^21.12b0", allow-prereleases = true}
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]