Phpstan level 10 reports an enormous number of errors. So does the excluded missingType.iterableValue. I do not plan to introduce either any time soon. But I will submit piecemeal changes from time to time. Changes will be mostly limited to phpdoc type declarations.
Phpstan hasn't identified any errors in Samples in a long time. But, as it turns out, one particular error is suppressed:
```
Variable $helper might not be defined.
```
This error would be generated by almost all samples, and the errors that it suppresses will become problematic if we move to Level 10. We are not yet committed to doing that, but it is pretty easy to write a script to change the samples so that error no longer happens, and there isn't really any reason to delay doing so. The results of that script constitute this PR.
Dealing mostly with mixed variable type. In a great many cases, problems can be resolved replacing `getValue` with `getValueString` or `getCalculatedValue` with `getCalculatedValueString`.
We currently use Phpstan Release 1. Release 2 has been available for some time, and we would like to stay current if possible. In order to get to Release 2, we first have to pass the "bleeding edge" tests for Release 1. I will do this by generating a new baseline using bleeding edge, then fix the baseline errors over several different PR's. This is the first of those.
Dependabot suggested some changes this month which required an unusual effort to implement successfully. With the elimination of Php 8.0 as a supported environment, it became possible to use Phpunit 10 rather than 9. Among other considerations, the configuration file for Phpunit is changed. I preserved the Phpunit 9 version under a different name. Aside from the configuration change, several other changes needed to be made to accommodate the change:
- Fix#3993. A peculiar problem indeed. One of the reporters said it had something to do with mocking, but I couldn't duplicate it. But Phpunit 10 revealed the problem in one test (Reader/Xlsx/AutoFilterTest), and that was sufficient for me to apply the trivial source code change to Worksheet/Worksheet.
- Some extra stringency on data providers required extra work in Calculation/CalculationFunctionListTest.
- There was a misplaced label in Calculation/ParseFormulaTest. Likewise in the data member CellGetRangeBoundaries.
- More stringency required changes to data members Shared/Trend/ExponentialBestFit and Shared/Trend/LinearBestFit.
- Issue3982Test testLoadAllRows seemed to go into a memory-acquiring loop in 10 that was not evident in 9. This particular test uses a lot of memory by design, but was included only to establish a base level for the other tests in that member. I feel it is acceptable to skip it for 10.
Phpstan errors with a new release are not unusual. One of the problems this time around was, however, unusual - it is fixed when Phpstan runs under Php8.3, but not for earlier Php releases. Our tools currently use Php 8.1. It is on my to-do list to get Phpstan and other test tools running under 8.3 before 8.1 goes EOL. There are also some configuration file changes needed for Phpstan.
Php-cs-fixer has been taking an increasingly long time to run. They've added an experimental option to permit it to run its checks in parallel. I've changed its configuration to use that option.
Change "mixed" declarations to more accurate types in test members; in particular, change those that would be flagged if we were to run Phpstan at level 9 (we currently run level 8). I may or may not follow up with source code (over 700 level-9 problems remain for src), but, as with strict typing, there is no reason to avoid the effort for test members.
It was necessary to update some doc blocks in src to accommodate this change. However, no executable code is touched.
* Automate Function-by-Name and Function-by-Category Docs
Examining whether it is possible to analyze the 2 Php scripts in the bin directory even though they lack a .php extension, I found that the docs generated by `generate-document` had not been updated in over a year even though support for a number of functions (e.g. ARRAYTOTEXT) had long ago been implemented. The files generated by the other script `generate-locales` are automatically generated as part of the unit test suite, and I can't see any reason not to do the same for `generate-document`.
It isn't totally clear that the 2 scripts in bin are needed any more now that each has an equivalent in the unit test suite. For now, I'm keeping them, and allowing them to be analyzed by phpcs, php-cs-fixer, and phpstan.
* Reflection Not Required
Remove it from bin scripts and DocumentGeneratorTest.
* Better Html/Css Display for Function Lists
* Redesign Features Cross Reference Doc
It was inaccurate, and pretty much impossible to read in entirety. This version should be easier to read, and to correct. That's important, because there are almost certainly new inaccuracies in this version.
* Add Html to Features Cross-Reference Readers
* Let Phpstan Run on Samples
Phpstan currently analyzes all source and test members. We already run phpcs and php-cs-fixer on samples as well. I would expect that samples are often used as templates for code in userland; it behooves us to be at least as careful with those members as for the others which are already being analyzed. Aside from 1300+ messages `Variable $helper might not be defined.`, which will be suppressed in phpstan.neon.dist, there are really only a few changes needed for sample members, so that part of the code base was already in good shape, and is now even better. No annotations were needed.
* Scrutinizer 2 out of 3
1 false positive, now suppressed; fix other 2.
* Remove Dead Code
* Very Minor Changes
* Add infra
And quite a bit more manual changes. The idea is that typing of our
tests can be a bit more loose, so we assume PHPDoc is mostly correct. If
that happens to be wrong, it should be caught by the tests themselves.
This is according to our formal, published, policy to only support
EOL PHP after 6 months.
See https://phpspreadsheet.readthedocs.io/en/latest/#php-version-support
Also share the exact same dev deps across all PHP version for GitHub
Actions so runs are faster, much most importantly they are stable and
predictable. And we decide manually when we want to migrate to PHPUnit
10.
Fixes#3634Closes#3710
Fix#3624. Use the same logic as elsewhere in the same module to invoke `imagesavealpha` when appropriate. (I confess that I do not understand the use case where you would not use imagesavealpha.) The fix was easy; writing a test was not. Google to the rescue.
Zipstream's newest production release requires Php8.1. This is not necessarily a problem for us as we are locked to an earlier release. However, psr/simple-cache became a real mess for us as people wanted to install PhpSpreadsheet alongside other products which required a version of simple-cache that could not run in some of the versions of Php which we support, and I would like to avoid having to go through that again. This ticket is a contingency to put us ahead of the curve in case such a problem should arise with Zipstream. There is reason to believe that Zipstream is not so widely used as simple-cache, so we do not need to rush this change in. However, should that belief prove incorrect, the change should be ready to go when needed.
For one of the Phpstan upgrades, some message text had changed so drastically that the only practical solution at the time was to move the messages from phpstan-baseline.neon to phpstan.neon.dist. This was not ideal, but it allowed us time to move on and study the errors, which I have now done. At one point, Parser is expecting a variable to be an array, and that was not clear from the code. If not an array, the code will error out (which was Phpstan's concern); I have changed it to throw an exception instead. This satisfies Phpstan, and I can get the message out of neon.dist (without needing to restore it to baseline). Unsurprisingly, the exception was never thrown in the existing test suite, although I added a couple of tests to exercise that code.
In Helper/Dimension, Phpstan flagged a statement inappropriately. I suppressed the message using an annotation and filed a bug report https://github.com/phpstan/phpstan/issues/7563. A fix for the problem was merged yesterday, which is good, but it puts us in a tenuous position. The annotation is needed now, but, when the fix is inevitably pushed to the version we use, the no-longer-needed annotation will trigger a different message. Recode so that neither the current nor the future versions will issue a message, eliminating the annotation in the process.
* Replace Dev jpgraph/jpgraph with mitoteam/jpgraph
PR #2979 added support for mitoteam/jpgraph as an alternative to jpgraph/jpgraph. The package jpgraph/jpgraph is abandoned in composer, and the version loaded with composer has been unusable for some time. This PR removes the dev requirement for jpgraph/jpgraph, and adds a dev requirement for mitoteam/jpgraph in its place.
With a usable graph library, a number of tests and samples that had been disabled are now re-enabled. A lot of new functionality has been added to Charts recently. Some of that new code has exposed bugs in JpgraphRendererBase. I have fixed those where I could. A handful of exceptions remain; I will investigate, and hopefully fix, those over time, but I don't feel it is necessary to fix them all before installing this PR - we are already way ahead of the game with the graphs that are working.
Three members had been ignoring code coverage in whole or in part because of the unavailability of a usable graph libray. Code coverage is restored in them. I am relieved to report that, although they aren't completely covered, adding them did not reduce code coverage by much - it is still over 90.4%.
I took a look at JpgraphRendererBase and Phpstan. Phpstan reports 128 problems. When I added some docblocks to correct some of those, the number increased to 284. Sigh. I will investigate over time, but, for now, we will still suppress Phpstan for JpgraphRendererBase.
I do not find a License file for mitoteam. However, there also wasn't one for jpgraph in the first place. Based on that and the discussion in #2996 (mitoteam will be used in exactly the same manner as mpdf), I don't think this is a problem. IANAL.
* PHP 8.2 Problems
Tons of "cannot create dynamic property" deprecations in jpgraph. Disable the test with most of those for now; leave the two with only a handful of messages enabled.
* Correct Failures in 2 Stock Charts
Down to 6 templates on which Render fails.
This one class consumes a lot of space in Phpstan baseline. The problem is that it is an interface to Jpgraph, which is not maintained in Composer. This means that we have to disable tests involving this module, since we are dealing with very old code in our test suite. This means that we are very unlikely to do any work on this member, so the code error reports are more of a distraction than anything else. Remove them for now, restoring them if we ever solve this problem.
As configured, Phpstan running under Php7 reports no errors. However, running under Php8, it reports 218 (!) errors. The vast majority of these are due to two reasons:
- renaming parameters in Php builtin functions in preparation for named parameters.
- using the new class `GdImage` rather than type `resource` as the argument type for many image-based functions.
Regardless of the cause, this will be a problem sooner or later. This PR is an attempt to get ahead of that problem. It tweaks only the Phpstan configuration files, without changing any PhpSpreadsheet code. Php7 continues to report no errors, and the number of errors for Php8 is reduced to 13. Although those 13 can also be tweaked, there is a case to be made that code corrections might be better in at least some of those instances. I will attend to those as one or more PRs after this one.
Because this involves no changes to code, and because Phpstan baseline is a common cause of merge difficulties, I will probably merge this in a day or two, more quickly than I customarily do.
oleibman
Adrien Crivelli
Mikhail Oleynik
Mark Baker
oleibman