webandco/neos-dev-tools

Tools for Neos development

Installs: 891

Dependents: 0

Suggesters: 0

Security: 0

Stars: 0

Watchers: 4

Forks: 0

Open Issues: 0

Type:neos-webandco

1.0.1 2023-08-29 08:32 UTC

This package is auto-updated.

Last update: 2024-11-08 15:52:36 UTC


README

The package provides tools for Neos development

Installation

Install the package with composer. It is recommended to use the package only in development environments.

composer require webandco/neos-dev-tools --dev

Tools

nodePublished file tool

This tools create a file when something is published in Neos. The purpose for this, is to use the file eg in gulp watch for browser-sync and reload the frontend, when the file changes.

Configuration

Webandco:
  DevTools:
    nodePublished:
      use: true
      file: '%FLOW_PATH_ROOT%.WebandcoNeosDevToolsLastPublished'
Webandco.DevTools.nodePublished.use

true to enable writing nodePublished file, else false

Webandco.DevTools.nodePublished.file

path of the nodePublished file

FLOW commandline BASH autocomplete

To get autocomplete for the FLOW commandline you need to load the script

source Scripts/flow.complete.sh

in BASH. This might be handy in a server environment like Beach or custom docker containers.

Stopwatch

Sometimes it's quite handy to determine how many milliseconds an operation took to determine performance issues. This package makes use of Symfony's Stopwatch for this purpose.

The Stopwatch class is extended to provide some more features.

Instantiation

There are two ways to instantiate a stopwatch. Either directly using new Stopwatch() or via the provided StopwatchFactory.

The factory might be relevant if multiple stopwatches are needed in different code parts. Using the factory you could create some final log output for all used stopwatches.

Usage

The stopwatch can be used like (Symfony's Stopwatch)(https://symfony.com/doc/current/components/stopwatch.html).

Two additional features are available: Restart the stopwatch and estimations.

Restart

Given a stopwatch $s = new Stopwatch(); you can avoid consecutive calls to ->start('critical_section') and ->stop('critical_section') by using

$s->start('init');
....
$s->restart('workspace_live');
if ($workspace->getName() == 'live') {
    $s->restart('critical_section_1');
    ...
    $s->restart('critical_section_2');
    ...
    $s->restart('nodes');
    foreach($nodes as $node){
        ...
        $s->lap('nodes');
    }
}
$s->stop();

ETA and format duration

Let's assume you process a bunch of nodes in a loop and you know this takes a while. To get an estimate of how long it takes you can use the following:

$s = new Stopwatch();
...
$nodes = ....
$nodeCount = count($nodes);

$s->restart('nodes');
foreach($nodes as $node){
    $c = $stopwatch->countLaps('nodes');
    if($c % 10 == 0){  // print ETA on every 10'th node
        $this->systemLogger->debug("Duration: ".Stopwatch::format($s->getEvent('nodes')->getDuration())." ETA: ".Stopwatch::format($s->eta('nodes', $nodeCount)));
    }
    ...  // process a node
    $s->lap('nodes');
}
$s->stop('nodes');

Stopwatch duration

The duration in milliseconds can be retrieved for the whole stopwatch using $s->getDuration() or for a single event $s->getEvent('someeventname')->getDuration().

Stopwatch metadata

You can also add metadata to a stopwatch, which can be used later on.

public function processNode(Node $node){
    $s = new Stopwatch();
    $s->setMetadata('nodeIdentifier', $node->getIdentifier());
    $s->setMetadata('nodeType', $node->getNodeType()->getName());

    ...
    $s->restart('ciritical_section');
    $s->stop();

    // Log the stopwatch, e.g. using wLog($s); - see below
}

Stopwatch signals

The Stopwatch emits signals upon start/stop/openSection/stopSection. These signals can be used to create a tree of the stopwatch calls.

StopwatchTree

A StopwatchTree is provided which receives the signals emited by a Stopwatch and creates a calltree like structure.

The tree gives insight on which parts are slow in a method or multiple stacked methods.

For example, in Neos 4.3, one of the initial requests the neos backend issues is /neos/schema/node-type?version=.... This request depends on the number of nodetypes configured and the runtime depends mainly on Neos.Neos/Classes/Service/NodeTypeSchemaBuilder.php::generateConstraints().

The StopwatchTree can be used to get a better view on the method:

....
    /**
     * @Flow\Inject
     * @var StopwatchFactoryInterface
     */
    protected $stopwatchFactory;

    /**
     * @Flow\Inject
     * @var StopwatchTreeInterface
     */
    protected $stopwatchTree;
....
    /**
     * Generate the list of allowed sub-node-types per parent-node-type and child-node-name.
     *
     * @return array constraints
     */
    protected function generateConstraints()
    {
        $s = $this->stopwatchFactory->create();

        //$s = new Stopwatch();
        $s->start('generateConstraints');
        $constraints = [];
        $nodeTypes = $this->nodeTypeManager->getNodeTypes(true);
        $s->restart('outer-nodeTypes');
        /** @var NodeType $nodeType */
        foreach ($nodeTypes as $nodeTypeName => $nodeType) {
            $constraints[$nodeTypeName] = [
                'nodeTypes' => [],
                'childNodes' => []
            ];
            $s->start('inner-nodeTypes');
            foreach ($nodeTypes as $innerNodeTypeName => $innerNodeType) {
                $s->start('allowsChildNodeType');
                if ($nodeType->allowsChildNodeType($innerNodeType)) {
                    $constraints[$nodeTypeName]['nodeTypes'][$innerNodeTypeName] = true;
                }
                $s->stop('allowsChildNodeType');

                $s->lap('inner-nodeTypes');
            }

            $s->restart('inner-autocreatedChildNodes');

            foreach ($nodeType->getAutoCreatedChildNodes() as $key => $_x) {
                $s->start('inner-nodeTypes-2');
                foreach ($nodeTypes as $innerNodeTypeName => $innerNodeType) {
                    $s->start('allowsGrandchildNodeType');
                    if ($nodeType->allowsGrandchildNodeType($key, $innerNodeType)) {
                        $constraints[$nodeTypeName]['childNodes'][$key]['nodeTypes'][$innerNodeTypeName] = true;
                    }
                    $s->stop('allowsGrandchildNodeType');

                    $s->lap('inner-nodeTypes-2');
                }
                $s->stop('inner-nodeTypes-2');

                $s->lap('inner-autocreatedChildNodes');
            }
            $s->stop('inner-autocreatedChildNodes');

            $s->lap('outer-nodeTypes');
        }
        $s->stop('outer-nodeTypes');

        wLog($s, "\n".$this->stopwatchTree->getTreeString());

        return $constraints;
    }

The final wLog() logs the signal itself and the call tree:

20-04-21 10:56:28 26         DEBUG                          Webandco\DevTools\Domain\Model\Dto\Stopwatch 00:01:07.763
  Events:
    generateConstraints 00:00:00.128 : 126.00 MiB 
    outer-nodeTypes 00:00:23.012 : 244.02 MiB 
    inner-nodeTypes 00:00:14.205 : 244.02 MiB 
    allowsChildNodeType 00:00:07.032 : 244.02 MiB 
    inner-autocreatedChildNodes 00:00:08.525 : 244.02 MiB 
    inner-nodeTypes-2 00:00:08.442 : 242.02 MiB 
    allowsGrandchildNodeType 00:00:06.419 : 242.02 MiB 
 
generateConstraints : 00:00:00.128
outer-nodeTypes : 00:00:23.012 (laps 546)
  inner-nodeTypes : 00:00:14.205 (laps 297570)
    allowsChildNodeType : 00:00:07.032 (laps 297025)
  inner-autocreatedChildNodes : 00:00:08.525 (laps 688)
    inner-nodeTypes-2 : 00:00:08.442 (laps 78078)
      allowsGrandchildNodeType : 00:00:06.419 (laps 77935)

Keep in mind, that such heavy use of Stopwatch and Signals triples the runtime. The method getConstraints() took around 23 seconds, which splits in 14 seconds for inner-nodeTypes and 8 seconds for inner-autocreatedChildNodes.

Without the Stopwatch this method took around 6 seconds in the tested project.

Performance

From a few tests it seems that excessive use of Stopwatch can double the original runtime without signals. Using signals seems to triple the original runtime compared to without a stopwatch.

Logging

This package also provides a quick and dirty way to log to the SystemLogger. Keep in mind that this should not be used in production!

The goal is to get a fast way to log to the system log and have some convenience alike console.log known from browsers.

In Package.php a function wLog() is declared in the global namespace. This function makes use of the LogService.

After the package is loaded you can use the function wLog() without the need to inject the systemLogger by simply writing

wLog("Something to log", 4711, null, true);

somewhere in your code, which creates an output like

20-04-10 13:41:11 2865       DEBUG                          Something to log here 4711 NULL true

You can also write multiple lines using the syntax

wLog("First line")->wLog("Second line");

Configuration

See Configuration/Settings.yaml.

  • enabled: To enable or disable logging to SystemLogger. The global method wLog() is created always.
  • pretty: In case a complex object is given as an argument, the object is logged using json_encode. If pretty is true the option JSON_PRETTY_PRINT is used for those arguments.
  • color: If set to true every new log message is printed in a new color. Thus you get a rainbow colored system log and maybe some issues or patterns might be easier to spot. The color can also be fixed using any of the names none, bold, dark, italic, underline, blink, concealed, black, red, green, yellow, blue, magenta,cyan, white, bg_black, bg_red, bg_green, bg_yellow, bg_blue, bg_magenta, bg_cyan, bg_white.
  • level: The log level to use for logging. Can be any of emergency, alert, critical, error, warning, notice, info, debug
  • renderer: For custom objects you can provide a custom renderer which implements the LogRendererInterface.php
  • signal: If enabled the emitted signals and their corresponding slots are logged using the DevTools log facility (aka wLog())
Overrule config

You can overrule the config using

wLog()->pretty(true)->color('red')->level('critical')->wLog("Something gone wrong, see this Exception", $e);

Enable logging the call depth given by count(debug_backtrace(false)):

wLog("log message")->withCallDepth();
Colored log messages

The messages written to the systemlogger are wrapped by bash color formating. It is possible to log messages in bold, italic, blinking, underlined or with a background color.

wLog("very important")->color('red')->background('green')->italic()->bold()->blink()->underline();
Conditional logging

To reduce the amount of log messages a condition can be added

wLog("problem node", $node)->condition($node->getName() == 'node-og6r6je5wpwnd');
Timing

To determine how long a method is running and how often it is called

function someMethod(){
$wLog = wLog(__METHOD__, __LINE__, $interestingArgument)->withTiming(__METHOD__, __LINE__);
....
}

Example output

22-03-22 15:14:50 28         DEBUG                          [⏰ 0.0244s = 1.477% of 1.653s] Webco\Test\Service\TestService_Original::someMethod 371
Hex dump

A modified version of https://stackoverflow.com/a/34279537 allows to log a string in hex dump like format

wLog()->hexDump($fusionValueWithCacheMarkers);

with an output like this

hexdump.png

By default, binary is inverted in the console. In this example the fusion cache markers can be seen.

Hexdump has the following configuration options:

  • highlightBinary: How to highlight binary. Default is to use the "color" invert
  • lineSeparator: The lineseparator to use. Default "\n"
  • bytesPerLine: How many bytes should be shown per line. Default 48
  • paddingCharacter: The chatacter to use for binary. Default .
Caller

By default the caller of wLog() is determined and written to the log

22-03-22 16:05:06 35         DEBUG                          Webco\Test\Service\TestService:testMethod:433 some log message

It takes around 0.2 ms to determine the caller. This can be disabled via Webandco.DevTools.log.caller.enable in Settings.yaml.

Custom Log Renderer

Complex objects are rendered using json_encode. Often this doesn't expose the needed information, thus you can create a custom renderer to transform a complex object to simpler log message parts. For an example, have a look at ThrowableLogRenderer.php and the Settings.yaml.

Eel Helper

An Eel Helper is added to make use of wLog via fusion. The helper provides two methods

  • wLog() which just forwards the arguments to the actual function wLog()
  • rwLog() which uses the first argument as the return value. The other arguments are forwarded to the actual function wLog() again.

For example

@process.log = ${value + Webandco.DevTools.rwLog(Date.format('now', 'Y-m-d H:i:s'), this)}

will add the current date and time to the current fusion value and writes the fusion object in the log. An AbstractFusionObjectRenderer is also provided to format the given AbstractFusionObjects.

Determine Caller

Since flow creates proxy classes it is more complicated to get the caller of a method from the backtrace.

The method BacktraceService::getCaller() returns caller file, line, class and method for a given method and class and uses Neos\Flow\Error\Debugger.

Log format mixing in one line

It is possible to write log lines with changing format:

wLog()->bold(LogService::FORMAT_ON)->wLog('This is bold')->bold(LogService::FORMAT_OFF)
      ->italic(LogService::FORMAT_ON)->wLog('This is italic text')->italic(LogService::FORMAT_OFF);

This line results in a single log line that looks like: This is bold This is italic

Force log message

By default the logs are written during __destruct() of the LogService. If you inject the LogService you can call eol() to force writing the log message:

wLog('Whats wrong with the node:', $nodeData)->eol();
LogService injection

To avoid the global wLog() method you can also inject the LogService

    /**
     * @Flow\Inject
     * @var LogService
     */
    protected $logService;

The injected LogService can then be used to write log messages using the eol() method

...
// first log line
$this->logService->wLog('Some log message')->eol();
...
// second log line
$this->logService->wLog('Another log message')->eol();
....
Signal logging

For debugging certain requests or implementing a feature it can be useful to know which signals are emitted and who catches them.
This can be enabled in the Settings.yaml via Webco.DevTools.log.signal.enabled. A regex can be specified to only show matching signals matching, e.g. /.*nodePropertyChanged$/i.

Log to stdout in case of cli

In Settings.yaml you can set the configuration logger depending on the php sapi name:

Webandco:
  DevTools:
    log:
      sapiLogger:
        cli: 'webcoDevStdoutLogger'

In the above configuration, webcoDevStdoutLogger is used to log to stdout in case of CLI commands.

FusionRenderingAspect

Via the configuration

Webandco:
  DevTools:
    fusion:
      enableRenderStats: false

an aspect is enabled which collects stats about rendered fusion objects. The result is written to the System log via wLog(). This can be handy in determining which fusion prototypes should be cached. The list printed via wLog() is ordered by the least relevant prototype to the most relevant which would benefit the most if it is cached.