source: trunk/scripts/check-attachments @ 1184

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

require_once(dirname($argv[0]) . "/utilities.php");
require_once(IA_ROOT_DIR . "common/attachment.php");
require_once(IA_ROOT_DIR . "common/db/attachment.php");
db_connect();

log_print("Checking attachment dir...");
log_warn("This script consumes tons of memory so errors might occur");

$res = db_query("SELECT `id`, `name`, `page`, `size` FROM ia_file");

$extra_files = array_flip(glob(IA_ROOT_DIR . "attach/*"));
$extra_atts = array();
$total_files = count($extra_files);
$total_atts = 0;
$mismatched_sizes = array();

// Check attachments.
while ($att = db_next_row($res)) {
    ++$total_atts;
    $fname = attachment_get_filepath($att);
    if (array_key_exists($fname, $extra_files)) {
        unset($extra_files[$fname]);
    } else {
        $extra_atts[] = $att;
        continue;
    }
    if (filesize($fname) != $att['size']) {
        $att['file_size'] = filesize($fname);
        $mismatched_sizes[] = $att;
    }
}

// FIXME: check cache.
$extra_files = array_keys($extra_files);

if (count($extra_files) == 0 && count($extra_atts) == 0) {
    log_print("Database and attach dir are in perfect sync, $total_files total files.");
} else {
    log_print("There are ".count($extra_files)." out of ".$total_files." disk files with no db entry.");
    if (count($extra_files)) {
        if (read_bool("Do you want to see a list?", false)) {
            foreach ($extra_files as $file) {
                log_print($file);
            }
        }
        if (read_bool("Want to delete them?", false)) {
            foreach ($extra_files as $file) {
                unlink($file);
            }
        }
    }

    log_print("There are ".count($extra_atts)." out of ".$total_atts." db files with no disk file.");
    if (count($extra_atts)) {
        if (read_bool("Do you want to see a list?", false)) {
            foreach ($extra_atts as $att) {
                log_print("page ".$att['page']." name ".$att['name']);
            }
        }
        if (read_bool("Want to delete them?", false)) {
            foreach ($extra_atts as $att) {
                attachment_delete_by_id($att['id']);
            }
        }
    }
}

unset($extra_atts);
unset($extra_files);

if (count($mismatched_sizes) != 0) {
    log_print('There are ' . count($mismatched_sizes) . ' files with ' .
              'mismatching sizes on disk and in the db');
    if (read_bool('Do you want to see a list?', false)) {
        foreach ($mismatched_sizes as $att) {
            log_print(sprintf('page %s name %s file_size %s db_size %s',
                              $att['page'], $att['name'],
                              $att['file_size'], $att['size']));
        }
    }
    if (read_bool('Do you want to try to fix this?', false)) {
        foreach ($mismatched_sizes as $att) {
            if ($att['file_size'] == 0 && $att['size'] != 0) {
                // File was likely uploaded when disk was out of space.
                // Nothing we can do, delete it.
                log_print(sprintf('Deleting attachment page %s name %s',
                                  $att['page'], $att['name']));
                attachment_delete_by_id($att['id']);
                continue;
            }

            $safe_update = false;
            if (starts_with($att['name'], 'grader_test') &&
                $att['file_size'] == $att['size'] + 1) {
                // A script was run that added missing trailing endlines
                // to grader tests.
                $safe_update = true;
            }
            if ($att['file_size'] != 0 && $att['size'] == 0) {
                $safe_update = true;
            }

            if ($safe_update) {
                $new_att = attachment_get_by_id($att['id']);
                $new_att['size'] = $att['file_size'];
                attachment_update($new_att['id'], $new_att['name'],
                                  $new_att['size'], $new_att['mime_type'],
                                  $new_att['page'], $new_att['user_id'],
                                  $new_att['remote_ip_info']);
                continue;
            }
            log_print(sprintf('I don\'t know how to handle attachment page ' .
                              $att['page'] . ' name ' . $att['name']));
        }
    }
} else {
    log_print('Stored file sizes match those of the files on disk.');
}

if (read_bool("Check mimetypes (could take a long time)?", false)) {
    $res = db_query("SELECT `id`, `name`, `page`, `mime_type` FROM ia_file");
    $mismatched_mimetypes = array();
    while ($att = db_next_row($res)) {
        $fname = attachment_get_filepath($att);
        if (!is_file($fname)) {
            continue;
        }
        if (get_mime_type($fname) != $att['mime_type']) {
            $att['file_mime_type'] = get_mime_type($fname);
            $mismatched_mimetypes[] = $att;
        }
    }

    if (count($mismatched_mimetypes) != 0) {
        log_print('There are ' . count($mismatched_mimetypes) . ' files with ' .
                  'mismatching mimetypes on disk and in the db');
        if (read_bool('Do you want to see a list?', false)) {
            foreach ($mismatched_mimetypes as $att) {
                log_print(
                    sprintf('page %s name %s file_mime_type %s db_mime_type %s',
                            $att['page'], $att['name'],
                            $att['file_mime_type'], $att['mime_type']));
            }
        }
        if (read_bool('Do you want to update them?', false)) {
            foreach ($mismatched_mimetypes as $att) {
                $new_att = attachment_get_by_id($att['id']);
                $new_att['mime_type'] = $att['file_mime_type'];
                attachment_update($new_att['id'], $new_att['name'],
                                  $new_att['size'], $new_att['mime_type'],
                                  $new_att['page'], $new_att['user_id'],
                                  $new_att['remote_ip_info']);
            }
        }
    } else {
        log_print('Stored mime_types match those of the files on disk.');
    }
}

if (read_bool("Enforce permissions?", false)) {
    $defowner = null;
    if (preg_match('/\/home\/([^\/]*)/', IA_ROOT_DIR, $matches)) {
        $defowner = $matches[1];
    }
    $userown = read_line("Owning user?", $defowner);
    $groupown = read_line("Owning group?", $defowner);
    $errs = false;
    $files = glob(IA_ROOT_DIR . "attach/*");
    foreach ($files as $file) {
        $err = array();
        if (@chmod($file, 0640) == false) {
            $err[] = "chmod";
        }
        if (@chown($file, $userown) == false) {
            $err[] = "chown";
        }
        if (@chgrp($file, $groupown) == false) {
            $err[] = "chgrp";
        }
        $errs |= $err;
        if ($err) {
            log_print("Failed ".implode(', ', $err)." on $file");
        }
    }
    if ($errs) {
        log_print("There were some errors. You should run this as root.");
    }
}

?>