Convert CSV of HTML paths to PHP redirects

redirect-files.php

<?php
error_reporting( E_ALL );
ini_set( "auto_detect_line_endings", true );

//---------- SETUP OPTIONS FOR SCRIPT ---------------//

// Primary folder containing the redirects
$base_path = dirname( __FILE__ ) . '/_redirects';

// The file name to grab redirects from
$csv = dirname( __FILE__ ) . '/module_1.csv';

// Delimiter used in processing redirects. Use a comma for CSV files. Spaces are most common for txt files
$delimiter = ',';

// The file extension we're looking to replace
$extension = '.html';

// Whether this is a dry run, or a live run. A dry run allows you to see what folders are created, but no PHP files are generated.
$dry_run = ( isset( $argv ) && in_array( 'live', $argv ) ) ? 0 : 1;

//---------- END SETUP OPTIONS FOR SCRIPT -----------//



// Setup the file pointer
$file = file( $csv, FILE_IGNORE_NEW_LINES );

echo ( $dry_run ? 'Dry' : 'Live' ) . " Run\n";

// Dry Run informational output
if ( $dry_run ) {
    echo "<pre>\n"; print_r( $file ); echo "</pre>\n";
}

// Initiate variables used in looping through the file
$data = array();
$count = 0;

// Now process the file
foreach ( $file as $string ) {
    // Separate based on the delimiter for each line of the file
    $pieces = explode( $delimiter, $string );
    
    // Make sure we're not matching string with query args
    if ( preg_match( "/\\$extension\?/", $pieces[0] ) ) {
        $count++;
        continue;
    }
    
    // Assumes the first string ends with ".html", and replaces it with .php
    $pieces[0] = str_replace( $extension, '.php', $pieces[0] );
    
    // Breaks off the domain name of the old path if it exists
    // The regex needs improved, because it also strips off any path between the domain and file name
    //$pieces[0] = preg_replace( '#.*/([^/]+\.php)$#', '$1', $pieces[0] );
    
    // Stored as an array of $old => $new keys/values
    $data[ $pieces[0] ] = $pieces[1];
}

// Dry Run informational output
if ( $dry_run ) {
    echo "<pre>\n"; print_r( $data ); echo "</pre>\n";
}

// Take the processed version of the file, and create the redirect files and directory structure
foreach ( $data as $old => $new ) {
    $path = $base_path . '/' . ltrim( $old, ' /' );
    $created = true;
    
    // Attempt to create new directories regardless of whether this is a dry run or not
    if ( ! is_dir( dirname( $path ) ) ) {
        $created = mkdir( dirname( $path ), 0775, true );
    }
    
    // If not a dry run, then attempt to actually create the files.
    if ( ! $dry_run ) {
        $new_file = fopen( $path, 'w' );
        fwrite( $new_file, "<?php\n\nheader( 'HTTP/1.1 301 Moved Permanently' );\nheader( 'Location: $new' );" );
        fclose( $new_file );
    }
    echo ( $dry_run ? 'Dry Run: ' : '' ) . ( $created ? '' : 'Error ' ) . "Creating $path\n";
}

// Let us know if we need to go back over any skipped rules manually
echo "\n\nSkipped $count rule(s) due to query args found.\n\n";

redirect-paths.php

<?php

error_reporting( E_ALL );
ini_set( "auto_detect_line_endings", true );

//---------- SETUP OPTIONS FOR SCRIPT ---------------//
// The file name to grab redirects from
$import_file = dirname( __FILE__ ) . '/redirects.txt';

// New file for JSON output
$json_output_file = dirname( __FILE__ ) . '/redirects.json';

// Delimiter used in processing redirects. Use a comma for CSV files. Spaces are most common for txt files
$delimiter = "\t";

// Whether this is a dry run, or a live run. A dry run allows you to see what folders are created, but no PHP files are generated.
$dry_run = ( isset( $argv ) && in_array( 'live', $argv ) ) ? 0 : 1;

//---------- END SETUP OPTIONS FOR SCRIPT -----------//

// Setup the file pointers
$import_pointer = file( $import_file, FILE_IGNORE_NEW_LINES );
$json_file_pointer = fopen( $json_output_file, 'w' );

echo ( $dry_run ? 'Dry' : 'Live' ) . " Run\n";

// Dry Run informational output
if ( $dry_run ) {
	echo "<pre>\n";
	print_r( $import_pointer );
	echo "</pre>\n";
}

// Initiate variables used in looping through the file
$data = array();

// Now process the file
foreach ( $import_pointer as $string ) {
	// Possibly remove "Redirect 301" from the beginning of the string, if it's there
	$string = preg_replace( '#^Redirect\s301\s#i', '', $string );

	if ( $dry_run ) {
		echo "$string\n";
	}

	// Separate based on the delimiter for each line of the file
	$pieces = explode( $delimiter, $string );

	if ( $dry_run ) {
		print_r( $pieces );
		echo "\n";
	}

	$matches = array( );

	// Break up the string into appropriate components
	// TODO: Include regex to grab query args
	preg_match( '#^(https?://)?([^/]+)?(/.*)#', $pieces[0], &$matches );

	if ( $dry_run ) {
		print_r( $matches );
	}

	// Stored as an array of data for rewrites
	$data[] = array(
		'name'			 => '',
		'pattern'		 => '^' . $matches[3],
		'replacement'	 => $pieces[1],
		'type'			 => 'permanent',
		'domain'		 => $matches[2],
		'argmatch'		 => '',
	);
}

// Dry Run informational output
if ( $dry_run ) {
	echo "<pre>\n";
	print_r( $data );
	echo "</pre>\n";
}

// Write to the file
$success = fwrite( $json_file_pointer, json_encode( $data ) );

// Close the file
fclose( $json_file_pointer );

if ( $success ) {
	echo "$success bytes written to file.\n";
} else {
	echo "Writing to file failed!\n";
}