Improved SVG cleanup code

Aleksander Machniak

2016-01-16 ed1d212ae2daea5e4bd043417610177093e99f19

 program/lib/Roundcube/rcube_message.php

@@ -1,6 +1,6 @@
<?php

/*
/**
 +-----------------------------------------------------------------------+
 | This file is part of the Roundcube Webmail client                     |
 | Copyright (C) 2008-2014, The Roundcube Dev Team                       |
@@ -53,13 +53,13 @@
    public $uid;
    public $folder;
    public $headers;
    public $parts = array();
    public $mime_parts = array();
    public $sender;
    public $parts        = array();
    public $mime_parts   = array();
    public $inline_parts = array();
    public $attachments = array();
    public $subject = '';
    public $sender = null;
    public $is_safe = false;
    public $attachments  = array();
    public $subject      = '';
    public $is_safe      = false;

    const BODY_MAX_SIZE = 1048576; // 1MB

@@ -69,26 +69,27 @@
     *
     * Provide a uid, and parse message structure.
     *
     * @param string $uid    The message UID.
     * @param string $folder Folder name
     * @param string $uid     The message UID.
     * @param string $folder  Folder name
     * @param bool   $is_safe Security flag
     *
     * @see self::$app, self::$storage, self::$opt, self::$parts
     */
    function __construct($uid, $folder = null)
    function __construct($uid, $folder = null, $is_safe = false)
    {
        // decode combined UID-folder identifier
        if (preg_match('/^\d+-.+/', $uid)) {
            list($uid, $folder) = explode('-', $uid, 2);
        }

        $this->uid  = $uid;
        $this->app  = rcube::get_instance();
        $this->uid     = $uid;
        $this->app     = rcube::get_instance();
        $this->storage = $this->app->get_storage();
        $this->folder  = strlen($folder) ? $folder : $this->storage->get_folder();
        $this->storage->set_options(array('all_headers' => true));

        // Set current folder
        $this->storage->set_folder($this->folder);
        $this->storage->set_options(array('all_headers' => true));

        $this->headers = $this->storage->get_message($uid);

@@ -96,19 +97,18 @@
            return;
        }

        $this->mime = new rcube_mime($this->headers->charset);

        $this->mime    = new rcube_mime($this->headers->charset);
        $this->subject = $this->headers->get('subject');
        list(, $this->sender) = each($this->mime->decode_address_list($this->headers->from, 1));

        $this->set_safe((intval($_GET['_safe']) || $_SESSION['safe_messages'][$this->folder.':'.$uid]));
        $this->set_safe($is_safe || $_SESSION['safe_messages'][$this->folder.':'.$uid]);
        $this->opt = array(
            'safe' => $this->is_safe,
            'safe'        => $this->is_safe,
            'prefer_html' => $this->app->config->get('prefer_html'),
            'get_url' => $this->app->url(array(
                'action' => 'get',
                'mbox'   => $this->storage->get_folder(),
                'uid'    => $uid))
            'get_url'     => $this->app->url(array(
                    'action' => 'get',
                    'mbox'   => $this->folder,
                    'uid'    => $uid))
        );

        if (!empty($this->headers->structure)) {
@@ -122,7 +122,6 @@
        // notify plugins and let them analyze this structured message object
        $this->app->plugins->exec_hook('message_load', array('object' => $this));
    }


    /**
     * Return a (decoded) message header
@@ -140,7 +139,6 @@
        return $this->headers->get($name, !$raw);
    }


    /**
     * Set is_safe var and session data
     *
@@ -150,7 +148,6 @@
    {
        $_SESSION['safe_messages'][$this->folder.':'.$this->uid] = $this->is_safe = $safe;
    }


    /**
     * Compose a valid URL for getting a message part
@@ -166,7 +163,6 @@
        else
            return false;
    }


    /**
     * Get content of a specific part of this message
@@ -199,7 +195,6 @@
        }
    }


    /**
     * Get content of a specific part of this message
     *
@@ -217,11 +212,16 @@
            return;
        }

        // allow plugins to modify part body
        $plugin = $this->app->plugins->exec_hook('message_part_body',
            array('object' => $this, 'part' => $part));

        // only text parts can be formatted
        $formatted = $formatted && $part->ctype_primary == 'text';

        // part body not fetched yet... save in memory if it's small enough
        if ($part->body === null && is_numeric($mime_id) && $part->size < self::BODY_MAX_SIZE) {
            $this->storage->set_folder($this->folder);
            // Warning: body here should be always unformatted
            $part->body = $this->storage->get_message_part($this->uid, $mime_id, $part,
                null, null, true, 0, false);
@@ -263,20 +263,20 @@
        $this->storage->set_folder($this->folder);

        $body = $this->storage->get_message_part($this->uid, $mime_id, $part,
            $mode === -1, is_resource($mode) ? $mode : null, !$formatted, $max_bytes, $formatted);

        if (!$mode && $body && $formatted) {
            $body = self::format_part_body($body, $part, $this->headers->charset);
        }
            $mode === -1, is_resource($mode) ? $mode : null,
            !($mode && $formatted), $max_bytes, $mode && $formatted);

        if (is_resource($mode)) {
            rewind($mode);
            return $body !== false;
        }

        if (!$mode && $body && $formatted) {
            $body = self::format_part_body($body, $part, $this->headers->charset);
        }

        return $body;
    }


    /**
     * Format text message part for display
@@ -318,16 +318,16 @@
        return $body;
    }


    /**
     * Determine if the message contains a HTML part. This must to be
     * a real part not an attachment (or its part)
     *
     * @param bool $enriched Enables checking for text/enriched parts too
     * @param bool               $enriched Enables checking for text/enriched parts too
     * @param rcube_message_part &$part    Reference to the part if found
     *
     * @return bool True if a HTML is available, False if not
     */
    function has_html_part($enriched = false)
    public function has_html_part($enriched = false, &$part = null)
    {
        // check all message parts
        foreach ($this->mime_parts as $part) {
@@ -361,17 +361,20 @@
            }
        }

        $part = null;

        return false;
    }


    /**
     * Determine if the message contains a text/plain part. This must to be
     * a real part not an attachment (or its part)
     *
     * @param rcube_message_part &$part Reference to the part if found
     *
     * @return bool True if a plain text part is available, False if not
     */
    function has_text_part()
    public function has_text_part(&$part = null)
    {
        // check all message parts
        foreach ($this->mime_parts as $part) {
@@ -401,56 +404,59 @@
            }
        }

        $part = null;

        return false;
    }


    /**
     * Return the first HTML part of this message
     *
     * @param rcube_message_part &$part    Reference to the part if found
     * @param bool               $enriched Enables checking for text/enriched parts too
     *
     * @return string HTML message part content
     */
    function first_html_part()
    public function first_html_part(&$part = null, $enriched = false)
    {
        // check all message parts
        foreach ($this->mime_parts as $pid => $part) {
            if ($part->mimetype == 'text/html') {
                return $this->get_part_body($pid, true);
        if ($this->has_html_part($enriched, $part)) {
            $body = $this->get_part_body($part->mime_id, true);

            if ($part->mimetype == 'text/enriched') {
                $body = rcube_enriched::to_html($body);
            }

            return $body;
        }
    }


    /**
     * Return the first text part of this message
     * Return the first text part of this message.
     * If there's no text/plain part but $strict=true and text/html part
     * exists, it will be returned in text/plain format.
     *
     * @param rcube_message_part $part Reference to the part if found
     * @param rcube_message_part &$part  Reference to the part if found
     * @param bool               $strict Check only text/plain parts
     *
     * @return string Plain text message/part content
     */
    function first_text_part(&$part=null)
    public function first_text_part(&$part = null, $strict = false)
    {
        // no message structure, return complete body
        if (empty($this->parts))
        if (empty($this->parts)) {
            return $this->body;

        // check all message parts
        foreach ($this->mime_parts as $mime_id => $part) {
            if ($part->mimetype == 'text/plain') {
                return $this->get_part_body($mime_id, true);
            }
            else if ($part->mimetype == 'text/html') {
                $out = $this->get_part_body($mime_id, true);

                // create instance of html2text class
                $txt = new rcube_html2text($out);
                return $txt->get_text();
            }
        }

        $part = null;
        return null;
    }
        if ($this->has_text_part($part)) {
            return $this->get_part_body($part->mime_id, true);
        }

        if (!$strict && ($body = $this->first_html_part($part, true))) {
            // create instance of html2text class
            $h2t  = new rcube_html2text($body);
            return $h2t->get_text();
        }
    }

    /**
     * Checks if part of the message is an attachment (or part of it)
@@ -477,6 +483,27 @@
        return false;
    }

    /**
     * In a multipart/encrypted encrypted message,
     * find the encrypted message payload part.
     *
     * @return rcube_message_part
     */
    public function get_multipart_encrypted_part()
    {
        foreach ($this->mime_parts as $mime_id => $mpart) {
            if ($mpart->mimetype == 'multipart/encrypted') {
                $this->pgp_mime = true;
            }
            if ($this->pgp_mime && ($mpart->mimetype == 'application/octet-stream' ||
                    (!empty($mpart->filename) && $mpart->filename != 'version.txt'))) {
                $this->encrypted_part = $mime_id;
                return $mpart;
            }
        }

        return false;
    }

    /**
     * Read the message structure returend by the IMAP server
@@ -497,8 +524,9 @@
                $structure->headers = rcube_mime::parse_headers($headers);
            }
        }
        else
        else {
            $mimetype = $structure->mimetype;
        }

        // show message headers
        if ($recursive && is_array($structure->headers) &&
@@ -514,11 +542,15 @@
            array('object' => $this, 'structure' => $structure,
                'mimetype' => $mimetype, 'recursive' => $recursive));

        if ($plugin['abort'])
        if ($plugin['abort']) {
            return;
        }

        $structure = $plugin['structure'];
        list($message_ctype_primary, $message_ctype_secondary) = explode('/', $plugin['mimetype']);
        $mimetype  = $plugin['mimetype'];
        $recursive = $plugin['recursive'];

        list($message_ctype_primary, $message_ctype_secondary) = explode('/', $mimetype);

        // print body if message doesn't have multiple parts
        if ($message_ctype_primary == 'text' && !$recursive) {
@@ -548,12 +580,6 @@
        else if ($mimetype == 'multipart/alternative'
            && is_array($structure->parts) && count($structure->parts) > 1
        ) {
            $plain_part   = null;
            $html_part    = null;
            $print_part   = null;
            $related_part = null;
            $attach_part  = null;

            // get html/plaintext parts, other add to attachments list
            foreach ($structure->parts as $p => $sub_part) {
                $sub_mimetype = $sub_part->mimetype;
@@ -574,8 +600,10 @@
                    $related_part = $p;
                else if ($sub_mimetype == 'text/plain' && !$plain_part)
                    $plain_part = $p;
                else if ($sub_mimetype == 'text/html' && !$html_part)
                else if ($sub_mimetype == 'text/html' && !$html_part) {
                    $html_part = $p;
                    $this->got_html_part = true;
                }
                else if ($sub_mimetype == 'text/enriched' && !$enriched_part)
                    $enriched_part = $p;
                else {
@@ -631,8 +659,19 @@
            $p->ctype_secondary = 'plain';
            $p->mimetype        = 'text/plain';
            $p->realtype        = 'multipart/encrypted';
            $p->mime_id         = $structure->mime_id;

            $this->parts[] = $p;

            // add encrypted payload part as attachment
            if (is_array($structure->parts)) {
                for ($i=0; $i < count($structure->parts); $i++) {
                    $subpart = $structure->parts[$i];
                    if ($subpart->mimetype == 'application/octet-stream' || !empty($subpart->filename)) {
                        $this->attachments[] = $subpart;
                    }
                }
            }
        }
        // this is an S/MIME ecrypted message -> create a plaintext body with the according message
        else if ($mimetype == 'application/pkcs7-mime') {
@@ -642,8 +681,13 @@
            $p->ctype_secondary = 'plain';
            $p->mimetype        = 'text/plain';
            $p->realtype        = 'application/pkcs7-mime';
            $p->mime_id         = $structure->mime_id;

            $this->parts[] = $p;

            if (!empty($structure->filename)) {
                $this->attachments[] = $structure;
            }
        }
        // message contains multiple parts
        else if (is_array($structure->parts) && !empty($structure->parts)) {
@@ -652,24 +696,16 @@
                $mail_part      = &$structure->parts[$i];
                $primary_type   = $mail_part->ctype_primary;
                $secondary_type = $mail_part->ctype_secondary;
                $part_mimetype  = $mail_part->mimetype;

                // real content-type of message/rfc822
                if ($mail_part->real_mimetype) {
                    $part_orig_mimetype = $mail_part->mimetype;
                    $part_mimetype = $mail_part->real_mimetype;
                    list($primary_type, $secondary_type) = explode('/', $part_mimetype);
                }
                else {
                    $part_mimetype = $part_orig_mimetype = $mail_part->mimetype;
                  }

                // multipart/alternative
                if ($primary_type == 'multipart') {
                // multipart/alternative or message/rfc822
                if ($primary_type == 'multipart' || $part_mimetype == 'message/rfc822') {
                    $this->parse_structure($mail_part, true);

                    // list message/rfc822 as attachment as well (mostly .eml)
                    if ($part_orig_mimetype == 'message/rfc822' && !empty($mail_part->filename))
                    if ($primary_type == 'message' && !empty($mail_part->filename)) {
                        $this->attachments[] = $mail_part;
                    }
                }
                // part text/[plain|html] or delivery status
                else if ((($part_mimetype == 'text/plain' || $part_mimetype == 'text/html') && $mail_part->disposition != 'attachment') ||
@@ -680,11 +716,12 @@
                        array('object' => $this, 'structure' => $mail_part,
                            'mimetype' => $part_mimetype, 'recursive' => true));

                    if ($plugin['abort'])
                    if ($plugin['abort']) {
                        continue;
                    }

                    if ($part_mimetype == 'text/html' && $mail_part->size) {
                        $got_html_part = true;
                        $this->got_html_part = true;
                    }

                    $mail_part = $plugin['structure'];
@@ -704,23 +741,22 @@
                        $this->attachments[] = $mail_part;
                    }
                }
                // part message/*
                else if ($primary_type == 'message') {
                    $this->parse_structure($mail_part, true);

                    // list as attachment as well (mostly .eml)
                    if (!empty($mail_part->filename))
                        $this->attachments[] = $mail_part;
                }
                // ignore "virtual" protocol parts
                else if ($primary_type == 'protocol') {
                    continue;
                }
                // part is Microsoft Outlook TNEF (winmail.dat)
                else if ($part_mimetype == 'application/ms-tnef') {
                    foreach ((array)$this->tnef_decode($mail_part) as $tpart) {
                    $tnef_parts = (array) $this->tnef_decode($mail_part);
                    foreach ($tnef_parts as $tpart) {
                        $this->mime_parts[$tpart->mime_id] = $tpart;
                        $this->attachments[] = $tpart;
                    }

                    // add winmail.dat to the list if it's content is unknown
                    if (empty($tnef_parts) && !empty($mail_part->filename)) {
                        $this->mime_parts[$mail_part->mime_id] = $mail_part;
                        $this->attachments[] = $mail_part;
                    }
                }
                // part is a file/attachment
@@ -735,21 +771,14 @@

                    // part belongs to a related message and is linked
                    if (preg_match('/^multipart\/(related|relative)/', $mimetype)
                        && ($mail_part->headers['content-id'] || $mail_part->headers['content-location'])) {
                        && ($mail_part->headers['content-id'] || $mail_part->headers['content-location'])
                    ) {
                        if ($mail_part->headers['content-id'])
                            $mail_part->content_id = preg_replace(array('/^</', '/>$/'), '', $mail_part->headers['content-id']);
                        if ($mail_part->headers['content-location'])
                            $mail_part->content_location = $mail_part->headers['content-base'] . $mail_part->headers['content-location'];

                        $this->inline_parts[] = $mail_part;
                    }
                    // attachment encapsulated within message/rfc822 part needs further decoding (#1486743)
                    else if ($part_orig_mimetype == 'message/rfc822') {
                        $this->parse_structure($mail_part, true);

                        // list as attachment as well (mostly .eml)
                        if (!empty($mail_part->filename))
                            $this->attachments[] = $mail_part;
                    }
                    // regular attachment with valid content type
                    // (content-type name regexp according to RFC4288.4.2)
@@ -766,9 +795,13 @@
                        $this->attachments[] = $mail_part;
                    }
                }
                // attachment part as message/rfc822 (#1488026)
                else if ($mail_part->mimetype == 'message/rfc822') {
                    $this->parse_structure($mail_part);
                // calendar part not marked as attachment (#1490325)
                else if ($part_mimetype == 'text/calendar') {
                    if (!$mail_part->filename) {
                        $mail_part->filename = 'calendar.ics';
                    }

                    $this->attachments[] = $mail_part;
                }
            }

@@ -789,7 +822,7 @@
                        // MS Outlook sends sometimes non-related attachments as related
                        // In this case multipart/related message has only one text part
                        // We'll add all such attachments to the attachments list
                        if (!isset($got_html_part) && empty($inline_object->content_id)) {
                        if (!isset($this->got_html_part)) {
                            $this->attachments[] = $inline_object;
                        }
                        // MS Outlook sometimes also adds non-image attachments as related
@@ -822,7 +855,6 @@
        }
    }


    /**
     * Fill aflat array with references to all parts, indexed by part numbers
     *
@@ -838,7 +870,6 @@
                $this->get_mime_numbers($part->parts[$i]);
    }


    /**
     * Decode a Microsoft Outlook TNEF part (winmail.dat)
     *
@@ -849,7 +880,7 @@
    {
        // @TODO: attachment may be huge, handle body via file
        $body     = $this->get_part_body($part->mime_id);
        $tnef     = new tnef_decoder;
        $tnef     = new rcube_tnef_decoder;
        $tnef_arr = $tnef->decompress($body);
        $parts    = array();

@@ -873,7 +904,6 @@

        return $parts;
    }


    /**
     * Parse message body for UUencoded attachments bodies
@@ -1000,5 +1030,4 @@
    {
        return rcube_mime::format_flowed($text, $length);
    }

}