📄 File: extract_pdf.php

Path: //home/outerorb/emp.outerorbittech.in/scripts/extract_pdf.php

Size: 1.19 KB

Permissions: 0666

<?php
$content = file_get_contents('C:/Users/hp/Downloads/amit kumar verma.pdf');
preg_match_all('/stream\r?\n(.*?)\r?\nendstream/s', $content, $matches);
foreach ($matches[1] as $i => $stream) {
    $decoded = @gzuncompress($stream);
    if ($decoded === false) $decoded = @gzinflate($stream);
    if ($decoded === false) $decoded = @gzinflate(substr($stream, 2));
    if ($decoded) {
        // Extract text from PDF stream (BT...ET blocks)
        preg_match_all('/BT(.*?)ET/s', $decoded, $blocks);
        foreach ($blocks[1] as $block) {
            preg_match_all('/\(([^\)\\\\]*(?:\\\\.[^\)\\\\]*)*)\)\s*Tj/', $block, $tj);
            foreach ($tj[1] as $t) {
                $t = stripcslashes($t);
                if (strlen(trim($t)) > 1) echo $t;
            }
            // Also try TJ operator
            preg_match_all('/\[(.*?)\]\s*TJ/s', $block, $tjArr);
            foreach ($tjArr[1] as $tj2) {
                preg_match_all('/\(([^\)]*)\)/', $tj2, $parts);
                $line = '';
                foreach ($parts[1] as $p) $line .= stripcslashes($p);
                if (strlen(trim($line)) > 1) echo $line;
            }
            echo "\n";
        }
    }
}

← Back to Directory Edit File 🔒 Chmod