Optimize database queries and file handling in content serializers

[noissue]

pulp_deb/app/serializers/content_serializers.py

@@ -488,12 +488,12 @@ def to822(self, component=""):
             artifact = RemoteArtifact.objects.filter(sha256=self.instance.sha256).first()
 
         if artifact:
-            if artifact.md5:
-                ret["MD5sum"] = artifact.md5
-            if artifact.sha1:
-                ret["SHA1"] = artifact.sha1
-            ret["SHA256"] = artifact.sha256
-            ret["Size"] = str(artifact.size)
+            ret.update({
+                "SHA256": artifact.sha256,
+                "Size": str(artifact.size),
+                **({"MD5sum": artifact.md5} if artifact.md5 else {}),
+                **({"SHA1": artifact.sha1} if artifact.sha1 else {}),
+            })
 
         ret["Filename"] = self.instance.filename(component)
 
@@ -600,42 +600,37 @@ def deferred_validate(self, data):
         data = super().deferred_validate(data)
 
         try:
-            package_paragraph = debfile.DebFile(fileobj=data["artifact"].file).debcontrol()
+            with debfile.DebFile(fileobj=data["artifact"].file) as deb_file:
+                package_paragraph = deb_file.debcontrol()
         except debfile.DebError as e:
-            if "[Errno 2] No such file or directory: 'unzstd'" in "{}".format(e):
-                message = (
-                    "The package file provided uses zstd compression, but the unzstd binary is not "
-                    "available! Make sure the zstd package (depending on your package manager) is "
-                    "installed."
-                )
-            else:
-                message = (
-                    "python-debian was unable to read the provided package file! The error is '{}'."
+            error_message = str(e)
+            if "unzstd" in error_message:
+                raise ValidationError(
+                    _("The package file used zstd compression, but 'unzstd' is not available!")
                 )
-            raise ValidationError(_(message).format(e))
+            raise ValidationError(_("Failed to read the package file: {}").format(error_message))
 
         from822_serializer = self.Meta.from822_serializer.from822(data=package_paragraph)
         from822_serializer.is_valid(raise_exception=True)
-        package_data = from822_serializer.validated_data
-        data.update(package_data)
+        data.update(from822_serializer.validated_data)
+
         data["sha256"] = data["artifact"].sha256
+        model_instance = self.Meta.model(**data)
+        relative_path = data.get("relative_path", model_instance.filename())
 
-        if "relative_path" not in data:
-            data["relative_path"] = self.Meta.model(**package_data).filename()
-        elif not os.path.basename(data["relative_path"]) == "{}.{}".format(
-            self.Meta.model(**package_data).name, self.Meta.model.SUFFIX
-        ):
-            data["artifact"].touch()  # Orphan cleanup protection so the user can try again!
-            raise ValidationError(_("Invalid relative_path provided, filename does not match."))
+        if os.path.basename(relative_path) != model_instance.filename():
+            data["artifact"].touch()  # Protect the artifact from orphan cleanup.
+            raise ValidationError(_("Invalide relative_path provided, filename does not match."))
 
+        data["relative_path"] = relative_path
         return data
 
     def retrieve(self, validated_data):
         content = self.Meta.model.objects.filter(
             sha256=validated_data["sha256"], relative_path=validated_data["relative_path"]
-        )
+        ).first()
 
-        return content.first()
+        return content
 
     class Meta:
         fields = (
@@ -1218,8 +1213,8 @@ def deferred_validate(self, data):
 
         artifacts = {data["relative_path"]: data["artifact"]}
         for source in data["checksums_sha256"]:
-            content = Artifact.objects.filter(sha256=source["sha256"], size=source["size"])
-            if not content.exists():
+            content = Artifact.objects.get(sha256=source["sha256"], size=source["size"])
+            if not content:
                 raise ValidationError(
                     _(
                         "A source file is listed in the DSC file but is not yet available '{name}'"
@@ -1237,7 +1232,9 @@ def retrieve(self, data):
         """
         If the Source Package already exists, retrieve it
         """
-        return SourcePackage.objects.filter(source=data["source"], version=data["version"]).first()
+        return SourcePackage.objects.only("source", "version").filter(
+            source=data["source"], version=data["version"]
+        ).first()
 
     class Meta:
         fields = MultipleArtifactContentSerializer.Meta.fields + (