[Serializer] Ignore comments when decoding XML #26445
Conversation
Previously, if the first line of XML was a comment, that would be used as the root node of the decoded XML. This work strips comments from decoded XML by default, but also allows for customizing which XML node types are ignored during decoding.
| */ | ||
| public function __construct(string $rootNodeName = 'response', int $loadOptions = null) | ||
| public function __construct(string $rootNodeName = 'response', int $loadOptions = null, array $ignoredNodeTypes = null) |
There was a problem hiding this comment.
, array $ignoredNodeTypes = array(XML_PI_NODE, XML_COMMENT_NODE)then you can get rid of ternary
There was a problem hiding this comment.
I wondered about that. I was following the pattern I saw for $loadOptions, but I'm happy to switch to that.
dec179a to
c6b294c
Compare
c6b294c to
b55454e
Compare
|
I realized I forgot to update the CHANGELOG.md with some changes I made to the function, so I've updated that now. |
| */ | ||
| public function __construct(string $rootNodeName = 'response', int $loadOptions = null) | ||
| public function __construct(string $rootNodeName = 'response', int $loadOptions = null, array $ignoredNodeTypes = array(XML_PI_NODE, XML_COMMENT_NODE)) |
There was a problem hiding this comment.
the default should be BC -safe, thus XML_COMMENT_NODE should not be listed, isn't it?
There was a problem hiding this comment.
Listing XML_COMMENT_NODE is for fixing a bug tho. I don't think anybody relies on old behaviour.
There was a problem hiding this comment.
I had originally put BC breaks to yes, but it looks like that was changed at some point. Personally, I think comments should be stripped by default, as a comment at the top of the XML to decode ends up becoming the XML root node. See the tests for an example.
There was a problem hiding this comment.
ok, can you add an entry in the UPGRADE file when, please?
There was a problem hiding this comment.
@nicolas-grekas since this is in the 4.1 milestone, I assume you are referring to the UPGRADE-4.1.md file? If so, I pushed a commit to note the change in this PR.
| @@ -11,6 +11,8 @@ CHANGELOG | |||
| * added optional `bool $escapeFormulas = false` argument to `CsvEncoder::__construct` | |||
| * added `AbstractObjectNormalizer::setMaxDepthHandler` to set a handler to call when the configured | |||
| maximum depth is reached | |||
| * added optional `int[] $ignoredNodeTypes` argument to `XmlEncoder::__construct`. Xml decoding now | |||
|
Thank you @q0rban. |
Previously, if the first line of XML was a comment, that would be used as the root node of the decoded XML. This work strips comments from decoded XML by default, but also allows for customizing which XML node types are ignored during decoding. The first two commits in this PR contain tests only to prove the existence of this "bug".