Detecting a password-protected document

Solution 1:

I have created an utility method that tries to detect if a given office document is protected by a password or not. Here are the list of advantages:

• supports Word, Excel and PowerPoint documents, both legacy (doc, xls, ppt) and new OpenXml version (docx, xlsx, pptx)
• does not depend on COM or any other library
• requires only System, System.IO and System.Text namespaces
• pretty fast and reliable detection (respects legacy .doc, .ppt and .xls file formats)
• low memory usage (maximum 64KB)

Here is the code, hope someone will find it useful:

public static class MsOfficeHelper
{
    /// <summary>
    /// Detects if a given office document is protected by a password or not.
    /// Supported formats: Word, Excel and PowerPoint (both legacy and OpenXml).
    /// </summary>
    /// <param name="fileName">Path to an office document.</param>
    /// <returns>True if document is protected by a password, false otherwise.</returns>
    public static bool IsPasswordProtected(string fileName)
    {
        using (var stream = File.OpenRead(fileName))
            return IsPasswordProtected(stream);
    }

    /// <summary>
    /// Detects if a given office document is protected by a password or not.
    /// Supported formats: Word, Excel and PowerPoint (both legacy and OpenXml).
    /// </summary>
    /// <param name="stream">Office document stream.</param>
    /// <returns>True if document is protected by a password, false otherwise.</returns>
    public static bool IsPasswordProtected(Stream stream)
    {
        // minimum file size for office file is 4k
        if (stream.Length < 4096)
            return false;

        // read file header
        stream.Seek(0, SeekOrigin.Begin);
        var compObjHeader = new byte[0x20];
        ReadFromStream(stream, compObjHeader);

        // check if we have plain zip file
        if (compObjHeader[0] == 'P' && compObjHeader[1] == 'K')
        {
            // this is a plain OpenXml document (not encrypted)
            return false;
        }

        // check compound object magic bytes
        if (compObjHeader[0] != 0xD0 || compObjHeader[1] != 0xCF)
        {
            // unknown document format
            return false;
        }

        int sectionSizePower = compObjHeader[0x1E];
        if (sectionSizePower < 8 || sectionSizePower > 16)
        {
            // invalid section size
            return false;
        }
        int sectionSize = 2 << (sectionSizePower - 1);

        const int defaultScanLength = 32768;
        long scanLength = Math.Min(defaultScanLength, stream.Length);

        // read header part for scan
        stream.Seek(0, SeekOrigin.Begin);
        var header = new byte[scanLength];
        ReadFromStream(stream, header);

        // check if we detected password protection
        if (ScanForPassword(stream, header, sectionSize))
            return true;

        // if not, try to scan footer as well

        // read footer part for scan
        stream.Seek(-scanLength, SeekOrigin.End);
        var footer = new byte[scanLength];
        ReadFromStream(stream, footer);

        // finally return the result
        return ScanForPassword(stream, footer, sectionSize);
    }

    static void ReadFromStream(Stream stream, byte[] buffer)
    {
        int bytesRead, count = buffer.Length;
        while (count > 0 && (bytesRead = stream.Read(buffer, 0, count)) > 0)
            count -= bytesRead;
        if (count > 0) throw new EndOfStreamException();
    }

    static bool ScanForPassword(Stream stream, byte[] buffer, int sectionSize)
    {
        const string afterNamePadding = "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0";

        try
        {
            string bufferString = Encoding.ASCII.GetString(buffer, 0, buffer.Length);

            // try to detect password protection used in new OpenXml documents
            // by searching for "EncryptedPackage" or "EncryptedSummary" streams
            const string encryptedPackageName = "E\0n\0c\0r\0y\0p\0t\0e\0d\0P\0a\0c\0k\0a\0g\0e" + afterNamePadding;
            const string encryptedSummaryName = "E\0n\0c\0r\0y\0p\0t\0e\0d\0S\0u\0m\0m\0a\0r\0y" + afterNamePadding;
            if (bufferString.Contains(encryptedPackageName) ||
                bufferString.Contains(encryptedSummaryName))
                return true;

            // try to detect password protection for legacy Office documents
            const int coBaseOffset = 0x200;
            const int sectionIdOffset = 0x74;

            // check for Word header
            const string wordDocumentName = "W\0o\0r\0d\0D\0o\0c\0u\0m\0e\0n\0t" + afterNamePadding;
            int headerOffset = bufferString.IndexOf(wordDocumentName, StringComparison.InvariantCulture);
            int sectionId;
            if (headerOffset >= 0)
            {
                sectionId = BitConverter.ToInt32(buffer, headerOffset + sectionIdOffset);
                int sectionOffset = coBaseOffset + sectionId * sectionSize;
                const int fibScanSize = 0x10;
                if (sectionOffset < 0 || sectionOffset + fibScanSize > stream.Length)
                    return false; // invalid document
                var fibHeader = new byte[fibScanSize];
                stream.Seek(sectionOffset, SeekOrigin.Begin);
                ReadFromStream(stream, fibHeader);
                short properties = BitConverter.ToInt16(fibHeader, 0x0A);
                // check for fEncrypted FIB bit
                const short fEncryptedBit = 0x0100;
                return (properties & fEncryptedBit) == fEncryptedBit;
            }

            // check for Excel header
            const string workbookName = "W\0o\0r\0k\0b\0o\0o\0k" + afterNamePadding;
            headerOffset = bufferString.IndexOf(workbookName, StringComparison.InvariantCulture);
            if (headerOffset >= 0)
            {
                sectionId = BitConverter.ToInt32(buffer, headerOffset + sectionIdOffset);
                int sectionOffset = coBaseOffset + sectionId * sectionSize;
                const int streamScanSize = 0x100;
                if (sectionOffset < 0 || sectionOffset + streamScanSize > stream.Length)
                    return false; // invalid document
                var workbookStream = new byte[streamScanSize];
                stream.Seek(sectionOffset, SeekOrigin.Begin);
                ReadFromStream(stream, workbookStream);
                short record = BitConverter.ToInt16(workbookStream, 0);
                short recordSize = BitConverter.ToInt16(workbookStream, sizeof(short));
                const short bofMagic = 0x0809;
                const short eofMagic = 0x000A;
                const short filePassMagic = 0x002F;
                if (record != bofMagic)
                    return false; // invalid BOF
                // scan for FILEPASS record until the end of the buffer
                int offset = sizeof(short) * 2 + recordSize;
                int recordsLeft = 16; // simple infinite loop check just in case
                do
                {
                    record = BitConverter.ToInt16(workbookStream, offset);
                    if (record == filePassMagic)
                        return true;
                    recordSize = BitConverter.ToInt16(workbookStream, sizeof(short) + offset);
                    offset += sizeof(short) * 2 + recordSize;
                    recordsLeft--;
                } while (record != eofMagic && recordsLeft > 0);
            }

            // check for PowerPoint user header
            const string currentUserName = "C\0u\0r\0r\0e\0n\0t\0 \0U\0s\0e\0r" + afterNamePadding;
            headerOffset = bufferString.IndexOf(currentUserName, StringComparison.InvariantCulture);
            if (headerOffset >= 0)
            {
                sectionId = BitConverter.ToInt32(buffer, headerOffset + sectionIdOffset);
                int sectionOffset = coBaseOffset + sectionId * sectionSize;
                const int userAtomScanSize = 0x10;
                if (sectionOffset < 0 || sectionOffset + userAtomScanSize > stream.Length)
                    return false; // invalid document
                var userAtom = new byte[userAtomScanSize];
                stream.Seek(sectionOffset, SeekOrigin.Begin);
                ReadFromStream(stream, userAtom);
                const int headerTokenOffset = 0x0C;
                uint headerToken = BitConverter.ToUInt32(userAtom, headerTokenOffset);
                // check for headerToken
                const uint encryptedToken = 0xF3D1C4DF;
                return headerToken == encryptedToken;
            }
        }
        catch (Exception ex)
        {
            // BitConverter exceptions may be related to document format problems
            // so we just treat them as "password not detected" result
            if (ex is ArgumentException)
                return false;
            // respect all the rest exceptions
            throw;
        }

        return false;
    }
}

Solution 2:

Here is a crude version of a password detecter i made. Does not need to open any Office objects.

    public static bool IsPassworded(string file) {
        var bytes = File.ReadAllBytes(file);
        return IsPassworded(bytes);
        return false;
    }
    public static bool IsPassworded(byte[] bytes) {
        var prefix = Encoding.Default.GetString(bytes.Take(2).ToArray());
        if (prefix == "PK") {
            //ZIP and not password protected
            return false;
        }
        if (prefix == "ÐÏ") {
            //Office format.

            //Flagged with password
            if (bytes.Skip(0x20c).Take(1).ToArray()[0] == 0x2f) return true; //XLS 2003
            if (bytes.Skip(0x214).Take(1).ToArray()[0] == 0x2f) return true; //XLS 2005
            if (bytes.Skip(0x20B).Take(1).ToArray()[0] == 0x13) return true; //DOC 2005

            if (bytes.Length < 2000) return false; //Guessing false
            var start = Encoding.Default.GetString(bytes.Take(2000).ToArray()); //DOC/XLS 2007+
            start = start.Replace("\0", " ");
            if (start.Contains("E n c r y p t e d P a c k a g e")) return true;
            return false;
        }

        //Unknown.
        return false;
    }

It might not be 100%. The flags I found by comparing several Excel and Word documents with and without password. To add for PowerPoint just do the same.