feat(TextIterator): support SplFileObject, remove method setFlags(), …

…by default trim line and skip empty lines
h4kuna · Jul 15, 2024 · 4823101 · 4823101
1 parent 7ad6cbe
commit 4823101
Show file tree

Hide file tree

Showing 12 changed files with 174 additions and 140 deletions.
diff --git a/phpstan.neon b/phpstan.neon
@@ -3,6 +3,11 @@ parameters:
 	paths:
 		- src
 		- tests/src
+	ignoreErrors:
+		-
+			message: "#^Generator expects value type TValue of array\\<int, string\\|null\\>\\|string, array\\<int, string\\|null\\>\\|string given\\.$#"
+			count: 1
+			path: src/Iterators/TextIterator.php
 
 includes:
 	- vendor/phpstan/phpstan-strict-rules/rules.neon
diff --git a/src/Iterators/README.md b/src/Iterators/README.md
@@ -10,7 +10,7 @@ bar
 joe";
 
 $textIterator = new TextIterator($incomingString);
-$textIterator->setFlags($textIterator::SKIP_EMPTY_LINE | $textIterator::TRIM_LINE);
+// $textIterator->enableCsv();
 foreach($textIterator as $line) {
     echo $line;
 }

diff --git a/src/Iterators/TextIterator.php b/src/Iterators/TextIterator.php
@@ -2,137 +2,120 @@
 
 namespace h4kuna\DataType\Iterators;
 
+use Generator;
 use h4kuna\DataType\Basic\BitwiseOperations;
+use h4kuna\DataType\Exceptions\InvalidStateException;
+use IteratorAggregate;
 use Nette\Utils\Strings;
+use SplFileObject;
 
 /**
  * Iterate via line
- * @template TValue
- * @extends \ArrayIterator<int, TValue>
+ * @template TValue of array<int, string|null>|string|non-empty-string
+ * @implements IteratorAggregate<int, TValue>
  */
-class TextIterator extends \ArrayIterator
+final class TextIterator implements IteratorAggregate
 {
-	public const SKIP_EMPTY_LINE = 1048576; // 2^20
-	public const CSV_MODE = 2097152; // 2^21
-	public const SKIP_FIRST_LINE = 4194304; // 2^22
-	public const TRIM_LINE = 8388608; // 2^23
-	public const SKIP_EMPTY_AND_TRIM_LINE = self::TRIM_LINE | self::SKIP_EMPTY_LINE;
-
-	private string $_current = '';
-
-	private int $flags = 0;
+	public const NoSetup = 0;
+	public const KeepEmptyLine = 1;
+	public const CsvMode = 2;
+	public const SkipFirstLine = 4;
+	public const SkipTrimLine = 8;
+	private const KeyDelimiter = 'delimiter';
+	private const KeyEnclosure = 'enclosure';
+	private const KeyEscape = 'escape';
 
 	/** @var array{delimiter: string, enclosure: string, escape: string} */
 	private array $csv = [
-		'delimiter' => ',',
-		'enclosure' => '"',
-		'escape' => '\\',
+		self::KeyDelimiter => ',',
+		self::KeyEnclosure => '"',
+		self::KeyEscape => '\\',
 	];
 
+	/** @var array<scalar> */
+	private array|SplFileObject $data;
+
 
 	/**
-	 * @param array<string|null>|string $text
+	 * @param array<scalar>|string $text
+	 * @param int $flags, trim line and skip empty line by default
 	 */
-	public function __construct(array|string $text)
+	public function __construct(
+		array|string|SplFileObject $text,
+		private int $flags = self::NoSetup
+	)
 	{
-		parent::__construct(is_array($text) ? $text : self::text2Array($text));
+		$this->data = is_string($text) ? self::text2Array($text) : $text;
 	}
 
 
 	/**
 	 * Active csv parser.
+	 * @return static<TValue>
 	 */
-	public function setCsv(string $delimiter = ',', string $enclosure = '"', string $escape = '\\'): static
+	public function enableCsv(string $delimiter = ',', string $enclosure = '"', string $escape = '\\'): self
 	{
-		$this->setFlags($this->getFlags() | self::SKIP_EMPTY_LINE | self::CSV_MODE | self::TRIM_LINE);
+		$this->flags |= self::CsvMode;
 		$this->csv = [
-			'delimiter' => $delimiter,
-			'enclosure' => $enclosure,
-			'escape' => $escape,
+			self::KeyDelimiter => $delimiter,
+			self::KeyEnclosure => $enclosure,
+			self::KeyEscape => $escape,
 		];
 
 		return $this;
 	}
 
 
 	/**
-	 * @return array<string>
-	 */
-	private static function text2Array(string $text): array
-	{
-		$existsNewMethod = method_exists(Strings::class, 'unixNewLines'); // @phpstan-ignore-line
-		return explode("\n", $existsNewMethod
-			? Strings::unixNewLines($text)
-			: Strings::normalizeNewLines($text)
-		);
-	}
-
-
-	/**
-	 * Change API behavior *****************************************************
-	 * *************************************************************************
+	 * @return Generator<TValue>
 	 */
-
-	public function setFlags(int $flags): void
-	{
-		parent::setFlags($flags);
-		$this->flags = $flags;
-	}
-
-
-	public function getFlags(): int
+	public function getIterator(): Generator
 	{
-		return parent::getFlags() | $this->flags;
-	}
-
+		$isTrimEnabled = BitwiseOperations::check($this->flags, self::SkipTrimLine) === false;
+		$skipEmptyLine = BitwiseOperations::check($this->flags, self::KeepEmptyLine) === false;
+		$isCsvMode = BitwiseOperations::check($this->flags, self::CsvMode);
 
-	/** @return TValue */
-	public function current(): mixed
-	{
-		$content = $this->_current;
-		if (BitwiseOperations::check($this->getFlags(), self::CSV_MODE)) {
-			$result = str_getcsv($content, $this->csv['delimiter'], $this->csv['enclosure'], $this->csv['escape']);
-
-			return $result;
+		if ($this->data instanceof SplFileObject) {
+			if (BitwiseOperations::check($this->data->getFlags(), SplFileObject::READ_CSV)) {
+				throw new InvalidStateException('SplFileObject is in csv read mode. Let\'s disable it and use self::enableCsv() instead.');
+			}
+			$first = 0;
+		} else {
+			$first = array_key_first($this->data);
 		}
 
-		return $content;
-	}
-
-
-	public function rewind(): void
-	{
-		parent::rewind();
-		if ($this->valid() && BitwiseOperations::check($this->getFlags(), self::SKIP_FIRST_LINE)) {
-			$this->next();
-		}
-	}
+		foreach ($this->data as $key => $row) {
+			if ($first === $key && BitwiseOperations::check($this->flags, self::SkipFirstLine)) {
+				continue;
+			}
+			assert(is_array($row) === false);
 
+			$strRow = (string) $row;
 
-	public function valid(): bool
-	{
-		$flags = $this->getFlags();
-		do {
-			if (parent::valid() === false) {
-				return false;
+			if ($isTrimEnabled) {
+				$strRow = trim($strRow);
 			}
-			$current = parent::current();
-			assert(is_string($current));
-			$this->_current = $current;
-			if (BitwiseOperations::check($flags, self::TRIM_LINE)) {
-				$this->_current = trim($this->_current);
+
+			if ($skipEmptyLine && $strRow === '') {
+				continue;
 			}
-		} while (BitwiseOperations::check($flags, self::SKIP_EMPTY_LINE) && $this->_current === '' && $this->moveInternalPointer());
 
-		return true;
+			yield $key => $isCsvMode
+				? str_getcsv($strRow, $this->csv[self::KeyDelimiter], $this->csv[self::KeyEnclosure], $this->csv[self::KeyEscape])
+				: $strRow;
+		}
 	}
 
 
-	private function moveInternalPointer(): bool
+	/**
+	 * @return array<string>
+	 */
+	private static function text2Array(string $text): array
 	{
-		$this->next();
-
-		return true;
+		$existsNewMethod = method_exists(Strings::class, 'unixNewLines'); // @phpstan-ignore-line
+		return explode("\n", $existsNewMethod
+			? Strings::unixNewLines($text)
+			: Strings::normalizeNewLines($text)
+		);
 	}
-
 }
diff --git a/tests/Fixtures/csvWithHead.csv → tests/Fixtures/csvNoSetUp.csv b/tests/Fixtures/csvWithHead.csv → tests/Fixtures/csvNoSetUp.csv
@@ -2,4 +2,4 @@ a:4:{i:0;s:6:"1Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:7:"Windows";}
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:7:"Solaris";}
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:5:"Linux";}
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:3:"Mac";}
-a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:4:"amet";}
+a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:4:"amet";}
diff --git a/tests/Fixtures/csv.csv → tests/Fixtures/csvSkipFirstLine.csv b/tests/Fixtures/csv.csv → tests/Fixtures/csvSkipFirstLine.csv
@@ -1,4 +1,4 @@
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:7:"Solaris";}
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:5:"Linux";}
 a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:3:"Mac";}
-a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:4:"amet";}
+a:4:{i:0;s:5:"Lorem";i:1;s:5:"ipsum";i:2;s:9:"dolor sit";i:3;s:4:"amet";}
diff --git a/tests/Fixtures/fileObject.csv b/tests/Fixtures/fileObject.csv
@@ -0,0 +1,4 @@
+a:1:{i:0;s:30:"1Lorem;ipsum;dolor sit;Windows";}
+a:1:{i:0;s:59:"Lorem;ipsum;dolor sit;Solaris  Lorem;ipsum;dolor sit;Linux";}
+a:1:{i:0;s:25:"Lorem;ipsum;dolor sit;Mac";}
+a:1:{i:0;s:26:"Lorem;ipsum;dolor sit;amet";}

diff --git a/tests/Fixtures/trimEmptyLine.csv → tests/Fixtures/keepEmptyLine.csv b/tests/Fixtures/trimEmptyLine.csv → tests/Fixtures/keepEmptyLine.csv
@@ -2,4 +2,7 @@
 Lorem;ipsum;dolor sit;Solaris
 Lorem;ipsum;dolor sit;Linux
 Lorem;ipsum;dolor sit;Mac
-Lorem;ipsum;dolor sit;amet
+
+
+Lorem;ipsum;dolor sit;amet
+
diff --git a/tests/Fixtures/emptyLine.csv → tests/Fixtures/keepEmptyLineAndSkipTrim.csv b/tests/Fixtures/emptyLine.csv → tests/Fixtures/keepEmptyLineAndSkipTrim.csv
@@ -3,4 +3,6 @@
  Lorem;ipsum;dolor sit;Linux
 Lorem;ipsum;dolor sit;Mac 
 
-Lorem;ipsum;dolor sit;amet
+
+Lorem;ipsum;dolor sit;amet
+
diff --git a/tests/Fixtures/noSetup.csv b/tests/Fixtures/noSetup.csv
@@ -1,8 +1,5 @@
-
 1Lorem;ipsum;dolor sit;Windows
- Lorem;ipsum;dolor sit;Solaris 
- Lorem;ipsum;dolor sit;Linux
-Lorem;ipsum;dolor sit;Mac 
-
-
+Lorem;ipsum;dolor sit;Solaris
+Lorem;ipsum;dolor sit;Linux
+Lorem;ipsum;dolor sit;Mac
 Lorem;ipsum;dolor sit;amet
diff --git a/tests/Fixtures/original.csv b/tests/Fixtures/original.csv
@@ -0,0 +1,5 @@
+1Lorem;ipsum;dolor sit;Windows
+ Lorem;ipsum;dolor sit;Solaris  Lorem;ipsum;dolor sit;Linux
+Lorem;ipsum;dolor sit;Mac    
+
+Lorem;ipsum;dolor sit;amet