on January 13, 2020. in Programming, Development. A 5 minute read.
I’m working on a small side project to gather stats about my blog. Posts published, words written, number of code examples. Good for practicing TDD. I want to share how I made one part of it better.
This part is a post collector that goes through a folder and collects post files. The first version of the PostCollector iterates over a path with DirectoryIterator. It finds all markdown files and creates a Post from them:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use DirectoryIterator;
use Generator;
use SplFileInfo;
class PostCollector
{
public function collectFrom(string $path): Generator
{
$directory = new DirectoryIterator($path);
foreach($directory as $file) {
if ($this->isAPostFile($file)) {
yield Post::fromFile($file);
}
}
}
private function isAPostFile(SplFileInfo $file): bool
{
if (! $file->isFile() || ! $file->isReadable()) {
return false;
}
if ($file->getExtension() !== 'md') {
return false; }
return true;
}
}
The PostCollector works and the tests are green, but it can be better. It does too many things, breaks the Single Responsibility Principle:
$path argument,The Post that the PostCollector creates does nothing:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use SplFileInfo;
class Post
{
private SplFileInfo $file;
private function __construct(SplFileInfo $file)
{
$this->file = $file;
}
public static function fromFile(SplFileInfo $file): self
{
return new self($file);
}
}
Let’s improve the PostCollector. We remove the internal DirectoryIterator. Instead, we pass in a ContentDirectory which better describes the type we are working with:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use Generator;
class PostCollector
{
public function collectFrom(ContentDirectory $contentDirectory): Generator
{
foreach($contentDirectory as $file) {
if ($this->isAPostFile($file)) {
yield Post::fromFile($file);
}
}
}
private function isAPostFile(SplFileInfo $file): bool
{
// No changes here from the previous example...
}
The PostContent has one responsibility less. Progress!
ContentDirectory is a DirectoryIterator:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use DirectoryIterator;
class ContentDirectory extends DirectoryIterator
{
}
I feel like the ContentDirectory wants to be an interface though. Let’s try that:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
interface ContentDirectory
{
}
As we are getting the post files from a local content directory, let’s have that implement our content directory interface:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use DirectoryIterator;
class LocalContentDirectory extends DirectoryIterator
implements ContentDirectory
{
}
If we keep in mind the “composition over inheritance” rule, then the extends DirectoryIterator part stands out. We can make the LocalContentDirectory a wrapper around DirectoryIterator:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use DirectoryIterator;
class LocalContentDirectory implements ContentDirectory
{
private DirectoryIterator $directory;
public function __construct(DirectoryIterator $directory)
{
$this->directory = $directory;
}
}
If we remember, we loop over the contents our of ContentDirectory in the PostCollector. A way to do it is to have our LocalContentDirectory be an Iterator as well. The iterator methods will call the methods on our internal DirectoryIterator:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use DirectoryIterator;
use Iterator;
class LocalContentDirectory implements ContentDirectory, Iterator
{
private DirectoryIterator $directory;
public function __construct(DirectoryIterator $directory)
{
$this->directory = $directory;
}
public function current()
{
return $this->directory->current();
}
public function next()
{
$this->directory->next();
}
public function key()
{
return $this->directory->key();
}
public function valid()
{
return $this->directory->valid();
}
public function rewind()
{
$this->directory->rewind();
}
}
The code looks better. We have a content directory type marked by the ContentDirectory interface. We have our local content directory LocalContentDirectory that implements the ContentDirectory interface. Internally it uses a DirectoryIterator and itself is a type of Iterator which makes it possible to loop over it.
As a final step, let’s tackle the fact that the Post does nothing, only holds a reference to the SplFileInfo object. One thing I didn’t like about the PostCollector is that it was deciding whether a file is a post or not. This is the job for the Post object. Let’s have it decide for its own whether a file is a Post or not:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use SplFileInfo;
class Post
{
private SplFileInfo $file;
private function __construct(SplFileInfo $file)
{
if (! $file->isFile()) {
throw new \RuntimeException('File is not a file');
}
if (! $file->isReadable()) {
throw new \RuntimeException('File is not readable');
}
if (! $this->fileHasMarkdownExtension($file)) {
throw new \RuntimeException('File does not have a markdown extension');
}
$this->file = $file;
}
public static function fromFile(SplFileInfo $file): self
{
return new self($file);
}
private function fileHasMarkdownExtension(SplFileInfo $file): bool
{
return $file->getExtension() == 'md';
}
}
And now our PostCollector becomes what we wanted it to be — a collector of post files:
<?php declare(strict_types=1);
namespace RobertBasic\HugoStats;
use Generator;
class PostCollector
{
public function collectFrom(ContentDirectory $directory): Generator
{
foreach($directory as $file) {
try {
yield Post::fromFile($file);
} catch (\RuntimeException $runtimeException) {
continue;
}
}
}
}
For every file in our content directory, we try to create a Post from it. If we don’t succeed, then we carry on as if nothing happened, otherwise we yield the Post file for further use.
From a PostCollector that did too many things, we now have 3 things each doing their job. A LocalContentDirectory responsible for getting the files from a folder. A Post responsible for deciding is a SplFileInfo a post file or not. PostCollector iterating over the files and creating Posts out of them.
What would you do differently? Do you see room for more improvement? Let me know on Twitter or drop me an email.
Happy hackin’!
If you'd like, you can get the next article I write to your email inbox.