nextcloud_groupfolder_filesystem_snapshots/lib/Manager/DiffManager.php

<?php

namespace OCA\GroupfolderFilesystemSnapshots\Manager;

class DiffManager {

    public function __construct(){
    }

    private function seperateFilesFromFolders($parentDir, $items) {
        $files = [];
        $folders = [];

        foreach($items as $item) {
            if(is_dir($parentDir . DIRECTORY_SEPARATOR . $item)) {
                $folders[] = $item;
            } else {
                $files[] = $item;
            }
        }

        return array($files, $folders);
    }

    private function getFilesAndFolders($dir) {
        $scan = array_diff(scandir($dir), array('..', '.'));

        return $this->seperateFilesFromFolders($dir, $scan);
    }

    private function getFilesizesOfFiles($prefix, array $files) {
        $result = array();
        foreach($files as $index=>$file) {
            $result[$index] = filesize($prefix . DIRECTORY_SEPARATOR . $file);
        }

        return $result;
    }

    function diffDirectories($dir1, $dir2, $prefix = "") {
        $diff = [];

        $scan1files = [];
        $scan1folders = [];
        if(file_exists($dir1) && is_dir($dir1)) {
            list($scan1files, $scan1folders) = $this->getFilesAndFolders($dir1);
        }
        
        $scan2files = [];
        $scan2folders = [];
        if(file_exists($dir2) && is_dir($dir2)) {
            list($scan2files, $scan2folders) = $this->getFilesAndFolders($dir2);
        }

        $fileCreations = array_diff($scan2files, $scan1files);
        $fileCreationsFilesizes = $this->getFilesizesOfFiles($dir2, $fileCreations);

        $fileDeletions = array_diff($scan1files, $scan2files);
        $fileDeletionsFilesizes = $this->getFilesizesOfFiles($dir1, $fileDeletions);

        $filePossibleEdits = array_intersect($scan1files, $scan2files);

        $allSubfolders = array_unique(array_merge($scan1folders, $scan2folders));

        /*$diff[] = [
            "type" => "DEBUG",
            "prefix" => $prefix,
            "fileCreations" => $fileCreations,
            "fileCreationsFilesizes" => $fileCreationsFilesizes,
            "fileDeletions" => $fileDeletions,
            "fileDeletionsFilesizes" => $fileDeletionsFilesizes,
            //"folderCreations" => $folderCreations,
            //"folderDeletions" => $folderDeletions,
            "allSubfolders" => $allSubfolders,
        ];*/

        // search for creations and deletions, that are actually renames
        foreach($fileCreations as $creationIndex=>$creation) {
            $creationPath = $dir2 . DIRECTORY_SEPARATOR . $creation;
            $creationSize = $fileCreationsFilesizes[$creationIndex];
            
            $renameContenders = array_keys($fileDeletionsFilesizes, $creationSize);

            if(sizeof($renameContenders) != 0) {
                /*$diff[] = [
                    "type" => "DEBUG",
                    "comparing" => [
                        "creation" => $creationIndex,
                        "deletions" => $renameContenders,
                    ],
                ];*/

                $creationSHA = sha1_file($creationPath);
                foreach($renameContenders as $contender) {
                    $deletion = $fileDeletions[$contender];
                    $deletionPath = $dir1 . DIRECTORY_SEPARATOR . $deletion;
                    $deletionSHA = sha1_file($deletionPath);

                    if($deletionSHA == $creationSHA) {
                        $diff[] = [
                            "type" => "RENAME",
                            "before" => [
                                "exists" => True,
                                "path" => $prefix . DIRECTORY_SEPARATOR . $deletion,
                                "size" => $creationSize,
                            ],
                            "afterwards" => [
                                "exists" => True,
                                "path" => $prefix . DIRECTORY_SEPARATOR . $creation,
                                "size" => $creationSize,
                            ]
                            
                        ];

                        unset($fileCreations[$creationIndex]);
                        unset($fileDeletions[$contender]);

                        break;
                    }
                }
            }
        }

        foreach($fileCreations as $index=>$creation) {
            $diff[] = [
                "type" => "CREATION",
                "before" => [
                    "exists" => False,
                ],
                "afterwards" => [
                    "exists" => True,
                    "path" => $prefix . DIRECTORY_SEPARATOR . $creation,
                    "size" => $fileCreationsFilesizes[$index],
                ]
            ];
        }

        foreach($fileDeletions as $index=>$deletion) {
            $diff[] = [
                "type" => "DELETION",
                "before" => [
                    "exists" => True,
                    "path" => $prefix . DIRECTORY_SEPARATOR . $deletion,
                    "size" => $fileDeletionsFilesizes[$index],
                ],
                "afterwards" => [
                    "exists" => False,
                ],
            ];
        }

        foreach($filePossibleEdits as $possibleEdit) {
            $file1 = $dir1 . DIRECTORY_SEPARATOR . $possibleEdit;
            $file2 = $dir2 . DIRECTORY_SEPARATOR . $possibleEdit;
            $file1Size = filesize($file1);
            $file2Size = filesize($file2);

            if(filemtime($file1) == filemtime($file2)) {
                //not different because same mtime
                continue;
            } else {
                // mtime different, but could just have gotten touched without modifications
                if($file1Size == $file2Size) {
                    // if filesize is the same check for binary differences
                    $handle1 = fopen($file1, 'rb');
                    $handle2 = fopen($file2, 'rb');

                    $filesdifferent = false;

                    while(!feof($handle1)) {
                        if(fread($handle1, 8192) != fread($handle2, 8192)) {
                            // files are different
                            $filesdifferent = true;
                            break;
                        }
                    }

                    fclose($handle1);
                    fclose($handle2);

                    if(!$filesdifferent) {
                        continue;
                    }
                }
            }

            
            $diff[] = [
                "type" => "EDIT",
                "before" => [
                    "path" => $prefix . DIRECTORY_SEPARATOR . $possibleEdit,
                    "size" => $file1Size,
                ],
                "afterwards" => [
                    "path" => $prefix . DIRECTORY_SEPARATOR . $possibleEdit,
                    "size" => $file2Size,
                ]
                
            ];
        }

        foreach($allSubfolders as $folder) {
            array_push($diff, ...($this->diffDirectories($dir1 . DIRECTORY_SEPARATOR . $folder, $dir2 . DIRECTORY_SEPARATOR . $folder, $prefix . DIRECTORY_SEPARATOR . $folder)));
        }

        return $diff;
    }
}