<?php

​

// Set the absolute path to the image folder

$imageFolder = __DIR__ . '/tempImage';

​

// Set the absolute path to the text folder

$textFolder = __DIR__ . '/tempText';

​

// Set the Tesseract executable path

$tesseractPath = 'C:\\Program Files\\Tesseract-OCR\\tesseract.exe';

​

// Set the output file base name

$outputFileBase = 'box';

​

// Initialize a counter for auto-incrementing filenames

$fileCounter = 12;

​

// Loop through each file in the folder

foreach (glob($imageFolder . '/*.png') as $imagePath) {

    // Sanitize the file name by removing spaces

    $sanitizedImageName = str_replace(' ', '', $imagePath);

​

    // Execute Tesseract OCR

    $outputFilePath = "$textFolder/$outputFileBase$fileCounter";

    $command = "\"$tesseractPath\" \"$sanitizedImageName\" \"$outputFilePath\"";

    exec($command);

​

    // Output the Tesseract command for debugging

    echo "Tesseract command: $command\n";

​

    // Set the output file path with the correct extension and auto-increment

    $outputFile = $outputFileBase . $fileCounter . '.txt';

​

    // Check if the output file exists

    if (file_exists($outputFilePath . '.txt')) {

        // Read the contents of the output file line by line

        $lines = file($outputFilePath . '.txt', FILE_IGNORE_NEW_LINES);

​

        // Output the extracted text

        echo "Extracted Text for $imagePath:\n<br>" . implode("<br>", $lines) . "\n<br>";

​

        // Create a new line in box.txt for separation

        echo "<br><br><br>";

​

        // Increment the file counter

        $fileCounter++;

    } else {

        echo "Error: Output file not created for $imagePath.<br>";

    }

}

​

echo "<br>All text extracted from images in tempImage folder.<br>";

​

?>