[RPM M1] Assemble artifacts based on distribution not file extension

src/assemble_workflow/bundle.py

@@ -14,6 +14,7 @@
 
 from assemble_workflow.bundle_recorder import BundleRecorder
 from assemble_workflow.dist import Dist
+from assemble_workflow.dists import Dists
 from manifests.build_manifest import BuildComponent, BuildComponents, BuildManifest
 from paths.script_finder import ScriptFinder
 from system.temporary_directory import TemporaryDirectory
@@ -132,7 +133,8 @@ def __get_min_dist(self, build_components: BuildComponents) -> Dist:
         min_dist_path = self._copy_component(min_bundle, "dist")
         logging.info(f"Copied min bundle to {min_dist_path}.")
         min_path = f"{self.build.filename}-{self.build.version}".replace("-SNAPSHOT", "")
-        min_dist = Dist.from_path(min_bundle.name, min_dist_path, min_path)
+        logging.info(f"Start creating distribution {self.build.distribution} for {min_bundle.name}.")
+        min_dist = Dists.create_dist(min_bundle.name, min_dist_path, min_path, self.build.distribution)
         logging.info(f"Extracting dist into {self.tmp_dir.name}.")
         min_dist.extract(self.tmp_dir.name)
         logging.info(f"Extracted dist into {self.tmp_dir.name}.")

src/assemble_workflow/bundle_recorder.py

@@ -13,11 +13,17 @@
 
 
 class BundleRecorder:
+    EXTENSIONS = {
+        "tar": ".tar.gz",
+        "zip": ".zip",
+    }
+
     def __init__(self, build: BuildManifest.Build, output_dir: str, artifacts_dir: str, bundle_location: BundleLocation) -> None:
         self.output_dir = output_dir
         self.build_id = build.id
         self.bundle_location = bundle_location
         self.version = build.version
+        self.distribution = build.distribution
         self.package_name = self.__get_package_name(build)
         self.artifacts_dir = artifacts_dir
         self.architecture = build.architecture
@@ -27,6 +33,7 @@ def __init__(self, build: BuildManifest.Build, output_dir: str, artifacts_dir: s
             build.version,
             build.platform,
             build.architecture,
+            build.distribution,
             self.__get_package_location(),
         )
 
@@ -37,7 +44,8 @@ def __get_package_name(self, build: BuildManifest.Build) -> str:
             build.platform,
             build.architecture,
         ]
-        return "-".join(parts) + (".zip" if build.platform == "windows" else ".tar.gz")
+        extension = self.EXTENSIONS[self.distribution] if self.distribution else self.EXTENSIONS['tar']
+        return "-".join(parts) + extension
 
     # Assembled output are expected to be served from a separate "dist" folder
     # Example: https://ci.opensearch.org/ci/dbc/bundle-build/1.2.0/build-id/linux/x64/dist/
@@ -66,14 +74,15 @@ def write_manifest(self, folder: str) -> None:
         self.get_manifest().to_file(manifest_path)
 
     class BundleManifestBuilder:
-        def __init__(self, build_id: str, name: str, version: str, platform: str, architecture: str, location: str) -> None:
+        def __init__(self, build_id: str, name: str, version: str, platform: str, architecture: str, distribution: str, location: str) -> None:
             self.data: Dict[str, Any] = {}
             self.data["build"] = {}
             self.data["build"]["id"] = build_id
             self.data["build"]["name"] = name
             self.data["build"]["version"] = str(version)
             self.data["build"]["platform"] = platform
             self.data["build"]["architecture"] = architecture
+            self.data["build"]["distribution"] = distribution if distribution else "tar"
             self.data["build"]["location"] = location
             self.data["schema-version"] = "1.1"
             # We need to store components as a hash so that we can append artifacts by component name

src/assemble_workflow/dist.py

@@ -69,16 +69,6 @@ def build(self, name: str, dest: str) -> None:
         shutil.copyfile(name, path)
         logging.info(f"Published {path}.")
 
-    @classmethod
-    def from_path(cls, name: str, path: str, min_path: str) -> 'Dist':
-        ext = os.path.splitext(path)[1]
-        if ext == ".gz":
-            return DistTar(name, path, min_path)
-        elif ext == ".zip":
-            return DistZip(name, path, min_path)
-        else:
-            raise ValueError(f'Invalid min "dist" extension in input artifacts: {ext} ({path}).')
-
 
 class DistZip(Dist):
     def __extract__(self, dest: str) -> None:

src/assemble_workflow/dists.py

@@ -0,0 +1,24 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# The OpenSearch Contributors require contributions made to
+# this file be licensed under the Apache-2.0 license or a
+# compatible open source license.
+
+import logging
+
+from assemble_workflow.dist import Dist, DistTar, DistZip
+
+
+class Dists:
+    DISTRIBUTIONS_MAP = {
+        "tar": DistTar,
+        "zip": DistZip,
+    }
+
+    @classmethod
+    def create_dist(cls, name: str, path: str, min_path: str, distribution: str) -> Dist:
+        if distribution is None:
+            logging.info("Distribution not specified, default to tar")
+            distribution = 'tar'
+
+        return cls.DISTRIBUTIONS_MAP[distribution](name, path, min_path)

src/manifests/build_manifest.py

@@ -22,6 +22,8 @@
   version: string
   platform: linux, darwin or windows
   architecture: x64 or arm64
+  distribution: tar, zip, and rpm
+  id: build id
 components:
   - name: string
     repository: URL of git repository

src/manifests/bundle_manifest.py

@@ -23,6 +23,8 @@ class BundleManifest(ComponentManifest['BundleManifest', 'BundleComponents']):
           version: string
           platform: linux, darwin or windows
           architecture: x64 or arm64
+          distribution: tar, zip, and rpm
+          id: build id
           location: /relative/path/to/tarball
         components:
           - name: string
@@ -39,6 +41,7 @@ class BundleManifest(ComponentManifest['BundleManifest', 'BundleComponents']):
             "schema": {
                 "platform": {"required": True, "type": "string"},  # added in 1.1
                 "architecture": {"required": True, "type": "string"},
+                "distribution": {"type": "string"},
                 "id": {"required": True, "type": "string"},
                 "location": {"required": True, "type": "string"},
                 "name": {"required": True, "type": "string"},
@@ -80,6 +83,7 @@ def __init__(self, data: Dict[str, str]):
             self.version = data["version"]
             self.platform = data["platform"]
             self.architecture = data["architecture"]
+            self.distribution: str = data.get('distribution', None)
             self.location = data["location"]
             self.id = data["id"]
 
@@ -89,10 +93,15 @@ def __to_dict__(self) -> dict:
                 "version": self.version,
                 "platform": self.platform,
                 "architecture": self.architecture,
+                "distribution": self.distribution,
                 "location": self.location,
                 "id": self.id,
             }
 
+        @property
+        def filename(self) -> str:
+            return self.name.lower().replace(" ", "-")
+
 
 class BundleComponents(Components['BundleComponent']):
     @classmethod

tests/tests_assemble_workflow/data/artifacts/dist/opensearch-1.3.0/opensearch.txt

@@ -0,0 +1 @@
+This page intentionally left blank.