change-wordpress-url-in-database

change-wordpress-url-in-database.php

#!/usr/bin/env php
<?php

// TODO - Add support for multisite
// Could start by just throwing an error when a multisite install is detected.

// TODO - Add support for urls inside serialized objects.
// This may be very easily to do by including PHP's serialized string
// syntax (including the length) in the list of strings to replace.

if(count($argv) != 4){
	die("You must specify an input file, new url, and output file!");
}

$inPath = $argv[1];
$newUrl = rtrim($argv[2], '/');
$outPath = $argv[3];

if(!file_exists($inPath)){
	die("The specified input file doesn't exist!");
}

if($inPath == $outPath){
	die("Input and output files cannot be the same!");
}

$oldUrl = getOldUrl($inPath);
replaceUrls($inPath, $outPath, $oldUrl, $newUrl);
return;

/*-------------------------------------------------------------------*/

function startsWith($haystack, $needle)
{
     $length = strlen($needle);
     return (substr($haystack, 0, $length) === $needle);
}

function getOldUrl($path){
	$file = fopen($path, "r");
	$siteUrl = null;

	while(!feof($file)){
	    $line = fgets($file);
	    if(startsWith($line, "INSERT INTO `wp_options`")){
	    	$siteUrlStart = strpos($line, '(') + 1;
	    	$siteUrlEnd = strpos($line, ')', $siteUrlStart);
	    	$homeStart = strpos($line, '(', $siteUrlEnd) + 1;
	    	$homeEnd = strpos($line, ')', $homeStart);

	    	$siteUrlParts = explode(',', substr($line, $siteUrlStart, $siteUrlEnd - $siteUrlStart));
	    	$homeParts = explode(',', substr($line, $homeStart, $homeEnd - $homeStart));

	    	$siteUrl = trim($siteUrlParts[2], "'");
	    	$home = trim($homeParts[2], "'");

	    	if($siteUrl != $home){
	    		throw new Exception("The siteurl ($siteUrl) and home ($home) are not the same!!!");
	    	}

	    	break;
	    }
	}

	fclose($file);

	if(empty($siteUrl)){
		throw new Exception("The siteurl could not be detected!");	
	}

	return rtrim($siteUrl, '/');
}

function getDomain($url){
	$parts = explode('/', $url);
	return $parts[2];
}

function replaceUrls($inPath, $outPath, $oldUrl, $newUrl){
	if(startsWith($newUrl, 'https')){
		$alternateOldUrl = str_replace('http://', 'https://', $oldUrl);
	}
	else{
		$alternateOldUrl = str_replace('https://', 'http://', $oldUrl);
	}

	$oldDomain = getDomain($oldUrl);
	$newDomain = getDomain($newUrl);

	$inFile = fopen($inPath, "r");
	$outFile = fopen($outPath, 'w');

	while(!feof($inFile)){
	    $line = fgets($inFile);

	    // $die = false;
	    // if(strpos($line, $oldUrl) !== FALSE){
	    // 	$die = true;
	    // 	echo "Before: $line\n";
	    // }

	    $line = replaceInSerializedArrays($oldUrl, $newUrl, $line);
	    $line = replaceInSerializedArrays($alternateOldUrl, $newUrl, $line);
	    $line = str_replace($oldUrl, $newUrl, $line);
	    $line = str_replace($alternateOldUrl, $newUrl, $line);

	    // Replace references to the domain only inside single quoted strings (like source_domain).
	    $line = str_replace("'$oldDomain'", "'$newDomain'", $line);

	    // if($die){
	    // 	echo "After:  $line\n";
	    // 	die();
	    // }

	    fwrite($outFile, $line);
	}

	fclose($inFile);
	fclose($outFile);
}

function replaceInSerializedArrays($find, $replace, $subject){
	$offset = 0;
	while(preg_match('/s:([0-9]+):\\\(")/', $subject, $matches, PREG_OFFSET_CAPTURE, $offset)){
		// print_r($matches);
		
		$start = $matches[0][1];
		$length = $matches[1][0];

		$handleChar = function($char, $nextChar){
			if($char == '\\'){
				if($nextChar == '\\'){
					return 1;
				}
				else{
					return 0;
				}
			}
			else{
				return 1;
			}
		};
		
		$contentStart = $matches[2][1]+1;
		$i = $contentStart;
		$charCount = 0;
		while($charCount < $length){	
			$charCount += $handleChar($subject[$i], $subject[$i+1]);
			$i++;
		}

		$end = $i+3;
		$endChars = substr($subject, $end-3, 3);
		$substr = substr($subject, $contentStart, $end - $contentStart);
		if($endChars != '\";'){
			throw new Exception("Error parsing serialized string.  Expected to find '\"' at offset $end, but found \"$endChars\" instead in the following string: $substr");
		}

		$rawLengthBefore = strlen($substr);
		$substr = str_replace($find, $replace, $substr);
		$length += strlen($substr) - $rawLengthBefore;

		$subject = substr($subject, 0, $start) . "s:$length:\\\"" . $substr . substr($subject, $end);

		$offset = $start+1;
	}

	return $subject;
}