NormalizeNamespaceByPSR4ComposerAutoloadRector: Replace references correctly #2644
Conversation
e2e/psr-4/rector.php
Outdated
| ]); | ||
|
|
||
| $rectorConfig->rule(NormalizeNamespaceByPSR4ComposerAutoloadRector::class); | ||
| $rectorConfig->rule(ClassRenamingPostRector::class); |
There was a problem hiding this comment.
Not sure if this is needed but it does not work either way – it does not look like the ClassRenamingPostRector does anything to fully-qualified name references.
There was a problem hiding this comment.
Apparently, it is not necessary. And it is skipping the processing here:
rector-src/rules/Renaming/NodeManipulator/ClassRenamer.php
Lines 164 to 166 in 02cec7e
There was a problem hiding this comment.
Apparently, the $parentNode is null. Possibly FullyQualifyStmtsAnalyzer is removing the attribute?
There was a problem hiding this comment.
FullyQualifyStmtsAnalyzer->process() produce create new FullyQualified instance, the attribute may need to be mirrored:
return new FullyQualified($name, $node->getAttributes());There was a problem hiding this comment.
Even then, it still does not seem to work properly – only fixes references in one of the files (weird, they are structurally identical). Will need to do more digging.
There was a problem hiding this comment.
Possibly, you can re-use RenamedClassesDataCollector. Add old -> new names there and it should be picked up in ClassRenamingPostRector and resolve this for you.
The post-Rector are run after all other rules, to post-process the final printer file code.
There was a problem hiding this comment.
I am using RenamedClassesDataCollector now and it still updates the references only in one of the files correctly but not the other. Possibly by that time the namespace of the referenced file is changed so the rename no longer finds it.
jtojnar
force-pushed
the
better-psr4
branch
2 times, most recently
from
ad3d9ce
to
0b961d2
Compare
rules/PSR4/Rector/FileWithoutNamespace/NormalizeNamespaceByPSR4ComposerAutoloadRector.php
Outdated
Show resolved
Hide resolved
|
Do you need help with this one? It seems almost finished 🙂 |
|
I have yet to find time to investigate why it still does not work. Not sure when I will be able to work on this. |
jtojnar
force-pushed
the
better-psr4
branch
6 times, most recently
from
67ff423
to
a399935
Compare
It will be needed in `NormalizeNamespaceByPSR4ComposerAutoloadRector` next. Additionally, make it use `ClassAnalyzer::isAnonymousClass` to cover more cases.
…rrectly Previously, `NormalizeNamespaceByPSR4ComposerAutoloadRector` did not update references to other normalized classes properly. This change fixes that by recording the renames using `RenamedClassesDataCollector`.
jtojnar
force-pushed
the
better-psr4
branch
from
a399935
to
021123a
Compare
|
@jtojnar We'd love to have this feature in Rector :) What is missing here to make CI pass? |
|
I would expect the code to work as is so there is something wrong with my assumptions. Unfortunately, I still have not found a time to dive into Rector innards to learn how rules are interleaved, and how |
|
Closing as the buggy rule is now removed to avoid causing harm. |
Previously,
NormalizeNamespaceByPSR4ComposerAutoloadRectordid not update references to other normalized classes properly. This change fixes that by recording the renames usingRenamedClassesDataCollector.Supersedes #2482