<?php

/**

* Cleans data from a file based on hard-coded limits.

*

* @param string $filePath The path to the file to clean.

* @return array|false An array of cleaned lines, or false on error.

*/

function cleanData(string $filePath): array|false

{

$cleanedData = [];

try {

$fileHandle = fopen($filePath, 'r');

if ($fileHandle === false) {

error_log("Error opening file: $filePath");

return false;

}

while (($line = fgets($fileHandle)) !== false) {

$cleanedLine = cleanLine($line);

if ($cleanedLine !== false) {

$cleanedData[] = $cleanedLine;

}

}

fclose($fileHandle);

return $cleanedData;

} catch (Exception $e) {

error_log("An unexpected error occurred: " . $e->getMessage());

return false;

}

}

/**

* Cleans a single line of data.

*

* @param string $line The line of data to clean.

* @return string|false The cleaned line, or false if the line is invalid.

*/

function cleanLine(string $line): string|false

{

// Limit line length to 256 characters.

if (strlen($line) > 256) {

$line = substr($line, 0, 256); // Truncate the line.

}

// Remove leading and trailing whitespace.

$line = trim($line);

// Remove control characters (ASCII 32-126)

$line = preg_replace("/[^a-zA-Z0-9\s]/", "", $line); //Keep alphanumeric and spaces

// Limit the number of spaces to 100

$line = preg_replace('/\s{101,}/', ' ', $line);

// Limit character set to ASCII

$line = preg_replace("/[^\x00-\x7F]/", "", $line);

if (empty($line)) {

return false; //Reject empty lines after cleaning

}

return $line;

}

// Example usage:

//$filePath = 'data.txt';

//$cleanedData = cleanData($filePath);

//if ($cleanedData !== false) {

// print_r($cleanedData);

//} else {

// echo "Data cleaning failed.\n";

//}

?>