CROSS REFERENCE TO RELATED UNITED STATES APPLICATIONS
BACKGROUND OF THE INVENTION
This application claims priority from “METHOD AND SYSTEM FOR SEAMLESS INTEGRATION OF PREPROCESSING AND POSTPROCESSING FUNCTIONS WITH AN EXISTING APPLICATION PROGRAM”, U.S. Provisional Patent Application No. 60/204,261, filed May 15, 2000 by Goodman, et al., the contents of which are incorporated herein by reference.
The present invention relates to a method and system for performing a preprocessing function on a file after it is opened and before it is operated on by an existing program, and to a method and system for performing a postprocessing function on a file after it is closed by the existing program.
Application programs such as the Microsoft Word word processing program, the Microsoft Excel spreadsheet program, and the Microsoft PowerPoint presentation program, all are designed to open a file (or create a new one), perform certain functions on the file, and then close the file when the user has finished his tasks. These application programs all share many common traits and functionality, and in fact are designed to be part of a suite of tools, which is referred to as Microsoft Office (other vendors feature similar suites with different names, but the same intended functionality). In the case of Word, the program typically opens a file from storage, edits the file for example by adding text, optionally prints the file, and then saves and closes the file. Since these application programs are obtained in a compiled version, it is not possible for users to be able to modify the functionality of the application, and the user can only perform the functions provided by the vendor.
It is desired to be able to give a user an opportunity to perform certain “value-added” operations on a file that are not specifically related to the application program. For example, it is often desired to encrypt a data file for security purposes. Encryption techniques are of course well known, and several discrete application programs exist that allow the user to execute encryption and decryption steps on a data file. That is, the user would have to run the encryption program on the file on disk, and then run the decryption program prior to running the word processing application so that the word processing program may operate on the file “in the clear”. The user would again have to run the encryption program on the file in order to ensure its security.
Likewise, other preprocessing and postprocessing operations may need to be performed on a file, such as compression, language translation, file backup, anti-virus analysis, and debugging. While certain programs exist for some of these applications, they all suffer from the same problem of having to be separately executed by a user. Since a user may forget to do these operations, this paradigm is undesirable.
- SUMMARY OF THE INVENTION
Thus, the inventors have recognized the need for being able to seamlessly integrate one or several preprocessing and postprocessing functions into an existing application so that they are automatically performed on a file without requiring a user to affirmatively execute a separate program. In particular, it is desired to be able to add this seamless functionality to existing programs such as the Microsoft Office programs, which have a large installed user base. Since the user will have the Office application program only in a complied version, there exist no solution today that solves the problems described above.
Disclosed herein is a method and program for adding functionality to an existing, compiled application program such as Microsoft Word, or a compiled, dynamically linked library, so that preprocessing and postprocessing functions can be performed on the files without requiring any user intervention. This program for adding functionality to an existing executable program in effect impersonates the application by replacing calls to operating system interface functions with calls to functions that are part of the system of the invention. This enables the system of the invention to intercept messages from the operating system to the application program, or from one function of the application to another function of the application. It is this ability to impersonate or masquerade as the application that enables the program of the invention to seamlessly add functionality to the application. The program and method of the invention will be referred to herein as the impersonation method and system.
It is an element of the impersonation method of the invention that all file activity happening through API calls is related to windows representing graphic displays of that or related files in these different interfaces.
In another aspect of the invention, provided is a method for intercepting software calls to subroutines and functions contained in program executables and dynamically linked libraries (“DLLs”).
In a graphical operating system such as Microsoft Windows, the relationship between file activity on a storage medium such as a hard disk, and the display of the file on screen is completely arbitrary and defined by the application. There is currently no operating system provided method for an external application to determine this relationship. It is a method of the invention that such a relationship can be created. The purpose of the invention is to provide third-party programs with the opportunity to act on files before they are opened and displayed by the primary application, and to take action on files when the application is finished.
As an exemplary application of the preprocessing and postprocessing functionality of the present invention, also disclosed herein is an invention that can be used, in a preferred embodiment, to encrypt and decrypt files used in any of the Microsoft Office applications (Word, PowerPoint, Excel, etc.). This file encryption/decryption process of the invention is referred to herein as file locking, and the system for carrying out this process is referred to as the file locking system. File locking provides application-based, rather than file-based security. It will be immediately apparent to the skilled artisan, however, that program impersonation system and method presented herein has many other applications besides file locking.
The file locking system is a data security application that integrates seamlessly with the entire family of Microsoft Office applications. The file locking system is designed to provide fully automatic protection. Once installed, it is impossible to identify where Office ends and the file locking system begins. It becomes part of Microsoft Office, automatically decrypting files when they're opened and encrypting them again when they're closed. Users never need to remember to follow security procedures—the file locking system remembers for them.
The file locking system encrypts and decrypts files seamlessly and automatically. A user logs on to the file locking system when they start Windows. Subsequently, when any file within any protected application is closed, the file locking system encrypts it automatically. When an encrypted file is opened by an authenticated user, it is automatically decrypted. It doesn't matter whether the file is a Microsoft Word document, an Excel spreadsheet, or what folder the file is saved in. The file locking system knows when files are being opened or closed because it scans the memory space of protected applications and monitors all traffic.
The file locking system uses a secret key to encrypt and decrypt files. Every installation of the file locking system generates its own secret key at installation time. This key uses random data and is unique to the computer on which the file locking system is installed. The secret key is encrypted on the protected PC. When a user logs on, the file locking system decrypts the key, holds it in protected memory, and uses the decrypted key to encrypt data. If an attacker manages to copy the key and some encrypted files to his own computer, the encrypted files cannot be decrypted on another PC because the attacker does not know the logon password to decrypt the key.
- Encryption Overview
The file locking system is different from other encryption products (PGP and Your Eyes Only, for example) in that these other products require that users save files to a specific protected location. In other words, in order for a file to be secure, it must be saved in a folder that has been designated as protected. The problem with this is that many users don't know or care where their files are; they save files “in Word”, or “in Excel”. It is unrealistic to demand that all users understand the file system. The file locking system avoids this problem by providing fully automatic encryption and decryption of Windows documents through seamless integration with protected applications. Once the file locking system is installed and a protected application is running, it is impossible to tell where the Windows application ends and the file locking system begins. Instead of protecting locations on disk, the file locking system protects everything that a protected application creates, wherever it creates it and wherever a user chooses to send it. Users can utilize their Windows applications the way they always do. There is nothing new to learn or to remember.
The “Open Source Code” encryption engine used by the file locking system is highly regarded by leading cryptographers. Its source code has long been publicly available on the Internet for peer review by the cryptographic community. Use of open source code unambiguously demonstrates that the file locking system's encryption methods can withstand close scrutiny.
- File Headers
“Closed Source Code” is the path that most publishers of security software follow; they do not make the source code, and therefore the soundness, of their encryption engines readily available for public review and critique.
Every Office file has a file header, which includes information about the file including the file type, location, size, the dates the file was created, last modified, last accessed, etc. The ability to search header information for files with particular properties can be a mission-critical data management tool. Other encryption programs encrypt the entire file, including header data, effectively disabling this powerful search tool. Because the file locking system is seamlessly integrated into Microsoft Office, it can encrypt file data without encrypting file headers. Users can retain the ability to search file header data without compromising file security.
Other programs encrypt the entire file so that if an unauthorized individual opens it, they see a garbled stream of data. Either by accident or maliciously, someone could insert a character into the encrypted file and corrupt it so that it could not be decrypted. Instead of displaying the encrypted file, the file locking system displays meaningful alternative information. The information is write-protected so that there is no way for an accidental recipient or viewer to modify and corrupt the file. A viewer who accidentally opens a file locked, encrypted, document knows that they have opened a file that they are not authorized to view.
File shredding is necessary to ensure data security. After an unwanted file is deleted “normally”, both the file and its data can still be easily recovered if an attacker has the right tools. File shredding removes this risk by thoroughly wiping the data from the hard disk at the sector level.
The file locking system file shredding conforms to the rigorous data security standards set forth in the U.S. National Industrial Security Program Operating Manual (NISPOM; DoD 5220.22-M). The Manual prescribes requirements, restrictions, and other safeguards that are necessary to prevent unauthorized disclosure of classified information. These guidelines were formulated pursuant to Executive Order 12829, and with the concurrence of the Secretary of Energy, the Chairman of the Nuclear Regulatory Commission, and the Director of Central Intelligence.
BRIEF DESCRIPTION OF THE DRAWINGS
Shredding plays an important role in the file locking system operation. To the file locking system user, encryption is a transparent, automatic process; a user closes a file and it is automatically encrypted. However, the actual procedure is more complex. When a file is closed within a protected application, the file locking system creates an unencrypted copy of the file and stores it in a recovery folder. This ensures that if a power failure or other disruption occurs during encryption, the file can be retrieved. Then the file locking system makes an encrypted copy. Once the encrypted copy is successfully created, both the original file and the clear copy in the recovery folder are shredded and deleted so that they cannot be recovered. Similar risk management procedures are in place for decryption, ensuring that a file can be recovered if something goes wrong during the decryption process. Should a problem occur during encryption or decryption, the next time the user logs on to the file locking system he or she is prompted to recover the file. The file locking system can automatically shred both data files and temporary files.
FIG. 1 is a flow chart of the steps involved in loading an application and linking to a dynamically linked library.
FIG. 2 a is a block diagram of the memory space of an application program executable.
FIG. 2 b is a block diagram of the memory space of an application program executable with an impersonation program DLL.
FIG. 3 depicts a flow chart of how the impersonation system starts processing and the actions it performs when an application starts.
FIG. 4 depicts a flow chart of the actions performed by the impersonation system when a document window is created.
FIG. 5 depicts a flow chart of the actions performed by the impersonation system when a file is opened or closed.
FIG. 6 depicts a flow chart of the actions performed by the impersonation system when a document window is destroyed.
DETAILED DESCRIPTION OF THE INVENTION
FIG. 7 depicts a flow chart of the steps performed when the file locking system encrypts a document.
The Windows operating system uses two primary types of window interfaces to represent documents to the user on screen, and two hybrid interfaces that contain elements of both. The first is called Single Document Interface (SDI), and is used to display a single application window for each instance of the program that is run, containing only one file. Additional files that are opened are represented by an additional application instance and window created for the additional files. Another method of display is referred to as a Multiple Document Interface (MDI). With MDI, a single application window is created, and an MDI parent window is created that is a child of the application window. All document representations are created as children of the parent MDI window. Starting with the application Office 2000, Microsoft has adopted two additional hybrid interfaces that contain elements of both of these primary types. The first is an MDI interface that associates hidden application windows with each of the documents in the MDI. This allows the user to perceive each document as if it where an SDI, with the associated features of an application window such as buttons on the task bar, and the listing of the document in the Task List. This is a description of the hybrid used by programs such as Microsoft Excel 2000. Microsoft Word 2000 uses a second hybrid MDI interface. It continues to display each document in an MDI format with the document represented as a child of the MDI parent; however, each subsequent document is displayed in a additional application windows containing a parent MDI and child document window.
Additionally, many applications make use of multiple thread technology, where actions are taken on files in a background thread that relate to the document displayed; however, these background threads contain no direct connection to the window representing the document. For example, the AutoSave feature of Microsoft Office applications will respond to timer input and automatically save a copy of open files in a background thread. These AutoSave files do not have a direct relationship to the windows displayed, and in fact are prohibited by the nature of multiple threads from sending messages to windows not created by that thread; however, the user perceives these actions as happening to windows displayed on the screen.
An executable program compiled for the Win32 (Windows' 32-bit) environment contains a feature known as an import table. The import table is integral to the format of executable programs that can run on many platforms. The import table contains the name of dynamically linked libraries (“DLLs”) and the name or ordinal number of functions and subroutines in those DLLs that are used by the executable. These functions are collectively referred to as the Application Program Interfaces (“APIs”). The import table also contains placeholder's for the addresses of these functions in the address space allocated by the operating system when these DLLs are loaded. A DLL can also have an import table and import functions from other DLLs referenced therein.
Referring now to FIG. 1, when loading an executable application program, the Windows program loader takes the following actions. First, at step 101, the loader reads the import table to find the DLLs needed by the executable. Next, at step 102, the referenced DLLs are loaded into memory. Then, at step 103, the loader maps the physical memory address of these DLLs in memory into the address space of the application. The loader then at step 104 reads in the import table the names of the subroutines or functions in the DLLs referenced by the executable, and finds at step 105 the addresses of these subroutines and functions in the loaded DLLs. Finally, at step 106, the loader writes the addresses of these subroutines and functions into the placeholders in the import table as they are loaded into the address space allocated for the executable application. The step of writing function addresses into the placeholder of the import table is referred to in the art as “fixing-up” the import table.
- Hooking the APIs
It is a feature of many current programs to provide higher level API calls that encapsulate many lower-level functions. For example, Microsoft's Object Linking and Embedding (OLE) format, now also known as Component Object format (COM), might use one function imported from a DLL such as StgOpenStorage, and return a memory pointer to a structure known as a Virtual Import Table (“Vtbl”). This table, which is an array of pointers to functions, serves the same purpose as the import table in a executable or DLL. The impersonation process described herein can call the high-level API function, inspect the return value, and insert substitute function references into the Vtbl. This enables the impersonation process to intercept executable calls to these high level functions. The impersonation system function can then optionally call the replaced function.
Hooks are locations in executable code which can be changed to a call to an outside function or subroutine. The impersonation process of the invention can not only hook the lowest level file operations, but also any higher level API's that reference these functions—especially those in related DLLs. For example, the impersonation process can not only hook the “CreateFile” API from the Windows kernel DLL directly, but can also find the import table of the OLE32.dll, and hook the kernel functions the OLE32.DLL import table. It is an additional element that before an application can open a file, the impersonation system opens the file itself and increments a file reference count by one. The impersonation system maintains an internal memory table to store the reference count of open operations performed on the file not only by the application, but by any additional DLLs referenced by the application. The impersonation system also keeps track of the close operations performed by the application and associated DLLs referenced by the application. These close operations are matched up to the reference count in the internal memory structure. When the reference count reaches 1, all associated file operations in the program have released their references to the file, and impersonation system can then release access to the file.
In addition, this hooking of API functions is directly associated with the document presentation visible on the screen.
The preferred embodiment of the impersonation process obtains a handle/memory address of an application executable from the operating system. In the Win32 environment, the handle is also a mapped memory address of the executable code. Using the published specifications of the executable format and the structures contained therein, the impersonation system can locate the list of the DLLs, functions and subroutines, and “fixed-up” address locations of the placeholders for those functions. The impersonation process saves the original address information in an internal memory structure for future reference, and then substitutes the address of functions or subroutines in the import table with its own functions.
For example, the Windows standard accessory Notepad.exe, could contain references to the CreateFile function in Kernel32.dll. The Windows program loader can fix-up the placeholder with the address of CreateFile, e.g., 0x80000A32. This is depicted in FIG. 2 a. The impersonation process can find this information, and load its own DLL containing a substitute function, such as “myCreateFile”, as shown in FIG. 2 b. For example, the impersonation system DLL can be loaded at address 0x20000000, and the function myCreateFile can start at address 0x200000D8. The impersonation process then substitutes this address, 0x200000D8, in the placeholder of the import table for the address of CreateFile, 0x80000A32. When the Notepad executable's code calls the CreateFile function it expected to be provided by Kernel32, it instead finds the address of the substitute function, 0x200000D8, and execution of code is passed to “myCreateFile”. Optionally, the substitute function could look up the address of the original function or subroutine and call that code.
- Window to File Relationships
Dynamic Link Libraries (DLLs) can also have an import table and import the functions from other DLLs. It is an element of the impersonation process that it can also hook the import table of associated DLLs. By searching for a specific DLL of interest or reading the import table of the main executable program, it can get the module handle/memory location of associated DLLs. The impersonation process can then hook the import table of these DLLs in the same manner.
The impersonation process establishes a relationship between the file activity of an application and the windows graphically displaying a representation of the file (and related referenced files) on screen. This is done by using the messaging features inherent in Windows for manipulating graphical windows, and creating the impersonation process's own window procedures to track the relationship between file activity and screen display.
This relationship is tracked by several methods.
Referring now to FIG. 3, a primary impersonation system executable program is run at step 302 by Windows on start up at step 301. This primary executable establishes at step 303 a system-wide window hook using a standard operating system API, SetWindowsHookEx. The hook is associated by the API with code contained in a first impersonation system DLL. This tells Windows that the impersonation system wants notification of all windows that are opened or closed. Windows is provided with the address of the first impersonation system DLL and the function to call.
When an application is started at step 305 and a new application window is created, the operating system loads at step 306 the code of the first impersonation system DLL into the memory space of the application, fixes the import table of the application, and sends a message to the hook function that an application window has been created. At this point, the first DLL dynamically loads a second impersonation system DLL that contains code to determine if this is an application that the impersonation system is interested in (such as a document handler or Microsoft Office application). If the application is of interest, the import table of the application is read and fixed-up with addresses of the impersonation system functions and DLLs at step 307. These other impersonation system DLLs are loaded into the memory space of the application. The impersonation system also substitutes the application's main window function with an impersonation system function at step 308. This function substitution is referred to as sub-classing. Application window sub-classing allows the impersonation system to monitor the creation of document and MDI parent windows. If a MDI parent window is created, the impersonation system subclasses that and monitors it for creation of document child windows. The impersonation system also changes the address that Windows would have used to send messages to the application to a substituted address provided by the impersonation system at step 309. Finally, the impersonation system creates a window referred to as the “key window” at step 310. The key window serves a file association function that will be explained below.
Whenever a document window is created by the application, the impersonation system creates a child window of that document window that receives messages from the operating system and other impersonation system functions. This is depicted in steps 401 and 402 of FIG. 4. This child window is referred to as the notify window. The notify window receives messages from the operating system at step 404 whenever its parent document window is activated by the program or user action. These activation messages are used by the impersonation system to keep track of the currently active document, so as to associate file activity with the correct document window. When a child document window is created in an MDI application, the impersonation system creates a notify window that is a child of that child document window.
Referring now to FIG. 5, when a new file is opened by the application at step 501, the file can be opened, its contents read into memory, and the file closed before any visible reference to the new file appears on screen. The impersonation process can intercept at step 502 an operating system message that a file is being opened and closed, and send a message to the key window indicating file activity. Because the key window does not get any processing time from the operating system until after the file open/close activity is complete, the key window receives this open notification after the document has been displayed and the file has been closed. The key window then looks for a topmost notify window associated with the active document window at step 503 and creates at step 504 a child notify window to store information about the file that was just opened, such as its name, path and attributes. This child notify window is also not visible. The child notify window contains all the information necessary to act upon the file when the document on screen is closed. This enables the impersonation system to distinguish between a true file close and the sequence of operations performed when a file is saved, which involves closing, updating, and reopening while the document window remains unchanged. This is different from a true file close, in which the document window is also closed along the actual file and all temporary files associated with the document window.
In addition, the impersonation system can perform an action on a file as it is being opened, based on rules established by the program indicating how to handle files of that type. For example, the program could decrypt a file that has been encrypted.
- Encryption and File Formats
When the application destroys the document window visible on screen, depicted at step 601 of FIG. 6, the operating system sends at step 602 a WM_DESTROY message to the child windows before the document is removed from the screen. The impersonation system notify window is a child of this document window. It responds to this message by checking at step 603 for the existence of all other child notify windows in order to obtain the names of files that were opened in association with this document. For example, opening a file called “test.doc” would have caused a child notify window to be created by the impersonation system for that file name; however, associated files such as temporary files, backup files, AutoSave files, and linked documents including templates, could also have been opened and closed while the document was on screen. Each of these items can be associated with a child notify window. Thus, when the document window is destroyed, the notify window contains child notify windows with the names and attributes of each of these files. Following rules established by the program indicating how to handle files of the type, the impersonation system can perform an action at step 604 on each file. For example, the user could be prompted to encrypt the file, or the file could be automatically encrypted. For the case of encryption, the file locking system acts upon the files listed in the child notify windows and associated with the document window prior to returning from the WM_DESTROY message. Thus, when the document is removed from the screen, all files associated with that document have been encrypted or appropriately handled (e.g., encrypted, shredded, etc.) by the impersonation system.
The basic premise of the impersonation system as applied to the file locking system program is that when a file is opened, its contents are checked to see if it is encrypted, and if so, assuming the user has been authenticated, it is decrypted before allowing the application program access to the contents. When the file is closed, the contents are encrypted.
The file locking system uses publicly available encryption methods to encrypt its data; however, it is an element of the process that the data is further protected from corruption by the methodology used to save the data in its physical format on disk.
Microsoft Office products use the OLE Docfile format to save file information. The Docfile format consists of header information, and one or more named streams representing elements of the document and additional information such as document summary information. When the file locking system encrypts a Docfile, the entire contents are encrypted, as depicted at step 701 of FIG. 7. A template of the same format, e.g. a Word document, is copied to a new file at step 702. The template contains visual information representing that the file is encrypted. The document summary information from the encrypted file is copied in readable form over the information contained in the new file at step 703. The encrypted data is then added at step 704 as an additional named stream in the file. There are two results with this method. (1) When the file is opened by an application that does not have the file locking system running, it only opens the named streams that are referenced in the document information contained in the main stream. This substitute stream contains no references to the actual encrypted data. The substitute stream is also coded with application specific data that makes the document read-only or password protected, preventing the user from saving the file and potentially corrupting or changing the contained encrypted data. (2) The summary information for the original document is still contained in the encrypted document in readable form. This allows applications to search for documents based on author or keywords, even though the document is encrypted.
When the file locking system opens a Docfile, it looks for the named data stream with the encrypted data, and decrypts it. The decrypted data is renamed to the original filename, and the containing encrypted document is destroyed and wiped from the disk (see disk wiping).
- Temp Files and Disk Wiping
Documents that don't use the OLE Docfile format such as text files contain a header with text warning the user not to edit the document or the encrypted data may become unusable. It is a backup feature of the file locking system that the header includes indicators that allow for the searching of the start of the encrypted data, and that indicate the length of the encrypted data. So, even if a user inadvertently adds data to the start or end of the document, the original data is still recoverable.
Most operating systems delete a file by deleting the first character of its name from its file allocation table. It is a feature of several disk utilities to recover deleted files by restoring this first character. The references in the table to the location on the disk of the original data remain until reused by the system. Most Windows applications make heavy use of temp files in editing, changing and saving a file. These temporary files often contain clear text of confidential information that the user is totally unaware of. Also, the encryption process itself must use a clear text version of the file as source for encryption. Many times it is possible using disk utilities or other forensic software technology to find significant portions of clear text of files that were encrypted or deleted. Even though the user believes his file to be encrypted or deleted, recoverable clear text fragments of this file still exist on the disk. It is an element of the file locking process that because it knows of all temp files that were associated with a document (because of the notify windows described above), and when the document is closed and deleted (because of the child notify windows created by the key window), the file locking system can take action to protect this information. The government's NISPOM standards specify a method for destroying classified data on a disk. The file locking system uses this method on all temporary files that are created and deleted by the application. In addition, to further frustrate recovery, the file locking system renames the temporary file to a single letter name before deleting it from the file allocation table. Thus when the temporary file is deleted, there are no other characters in the file allocation table to indicate the file name or purpose.
- Special Purpose Header Information
Not only are the temporary files created by the application wiped in this manner, but the user is also given the option of deleting other existing files using this feature.
The format of encrypted data stored on disk is integral to the system. Encrypted information and flags in the header are necessary to decrypt the document successfully. The flags contained in this header also indicate the type of encryption used. For example, files can be encrypted with the user's master key, with a passphrase, with a shared group key, or with one of many public key encryption formats.
The header also can contain information with special handling procedures associated with it. For example, the file locking system can contain code signifying it should immediately delete a file after it is closed. Timestamp related functions are possible, so that files can only be decrypted within a specified number of days. Other flags can indicate that the file locking system program should not allow the system to copy to the clipboard, copy the file to another name, send the file using a mail program, or print the file.
- Block Format of Sensitive Information
These special purpose flags are possible because the file locking system has sub-classed the appropriate window procedure, and can intercept associated operating system messages to take these actions.
It is the nature of cryptographic functions, that sensitive data must be passed from one function to another to accomplish encryption or decryption. For example, the file locking system uses a user interface to authenticate the user via logon information. This logon information is then used to decrypt the master key. The master key is then used by the file locking system to encrypt or decrypt each of the files opened and closed in each protected application. The clear text data of the masterkey is highly sensitive information. If it were intercepted by an attacker using a “data sniffing” program embodied in a Trojan, worm, or virus, then the file locking system security would be completely compromised.
Rather than pass such sensitive information as the clear text of the decrypted masterkey to one of the file locking system's functions, the file locking system instead passes a pointer to a shared block of memory that contains this information. In a preferred embodiment, the block is at least 1024 bytes long, and contains other random data visually indistinguishable from the sensitive data. The sensitive data is located at a constantly changing offset in the block known only to the programs. It is an element of the file locking process that the sensitive data is stored at a different offset in the block each time the file locking system is run.
Operating systems routinely ‘page’ memory to temporary disk files. It is a common mistake of many cryptographic systems to ignore this system behavior. The file locking system further protects system security because of the temp file wiping mentioned above. Even if sensitive data were ‘paged’ to a disk file, this data is wiped from the disk drive when the file locking system handles the deletion of temporary files.
The program impersonation system of the invention can also be used to add to existing applications many more capabilities besides encryption/decryption. These added capabilities include:
2. Language Translation, e.g., English to French, French to
3. Format Translation, e.g., Microsoft Word format to WordPerfect format.
4. Virus Checking.
5. Document Security—e.g., adding a password.
6. Automatic Backup—Duplicate of document stored at another location
7. Version Control: each save of the document can cause a copy of the current file to be stored with a different name or in a different stream.
8. Remote Storage: a reference to the document is stored locally, but the document contents are stored at a remote location.
9. Spell checking and/or grammar checking a document.
10. Changing the actual text of a document.
This list is for illustrative purposes. Any other added functionality that utilizes the impersonation method of the invention is also within the scope of the invention. The invention is defined by the appended claims.