A Spell Checking Engine






4.88/5 (15 votes)
Jan 7, 2001

274926

7028
A free spell checking engine for use in your C++ applications. Includes the current US English dictionary
Description:
This project is an evolution of the spell checking engine project I submitted earlier. This project includes numerous enhancements to the core spelling engine plus the addition of a "check-as-you-type" edit control and the related support dialogs (see above).
This project is not complete, it is a work-in-progress. There are numerous issues with the current version which need to be addressed. My long term goal is to develop this to "commercial quality".
I am going to continue to improve this engine toward my goal. I will continue to post updates as I feel necessary.
Changes from previous version:
- Reorganized class architecture. Added
CFPSSpellCheckEngineOptions
,CFPSDictionary
to core engine. - Created
CFPSSpellingEditCtrl
CEdit
derived class to implement "check-as-you-type" edit control. - Created options property pages (see above)
- Created spelling dialog (see above)
- Created common-use dictionary. (Download available above)
- Updated US English dictionary w/improved word list + proper names.
- Incremental changes to the
MetaphoneEx
function. - Addition of the
EditDistance
function. - Added support for case-sensitive dictionary entries.
- Added file header to dictionaries.
To-do-list:
- Re-write
EditDistance
algorithm for performance. - Research compression options for dictionaries
- Create ATL ActiveX control from edit control
- Create COM based dictionary support for language independence
- Create Rich Edit "check-as-you-type" control.
- Add auto-correct capability.
- Add sentence begin recognition for automatic upper case decisions.
- Continue to improve US English dictionary.
- Implement a binary-search mechanism on dictionary look-ups.
- Continue to improve
MetaphoneEx
function. - Create C# (.net) version (when C# stabilizes)
- ETC...
Classes:
CFPSSpellCheckEngine |
This is the core spelling engine. It is intended to be language independent. (not currently). This engine encapsulates the functionality of managing dictionaries, making suggestions (through dictionaries), and maintaining spelling options. |
CFPSSpellCheckEngineOptions |
Support class for CFPSSpellCheckEngine which implements
support for storing, saving and loading spell checking options.
Currently, this uses a serialized file to store options, but could
easily be changed to INI file or registry. |
CFPSDictionary |
Base dictionary class. Defines a set of virtual functions generic to all dictionaries. Also, provides base implementation of all virtual functions based on current requirements. This class uses a defined file structure w/ a file header and any number of dictionary records. Future derivations of this class will provide language specific support. |
CDlgSpellChecker |
CDialog derived class which implements the spell checker
dialog. Currently the undo support is based on edit control undo
support, not spell checker undo. Need to improve this further. |
CPrShtSpellOptions |
CPropertySheet derived class which implements the spell
checking engines property sheet. |
CPrPgeSpellOptions_General |
CPropertyPage derived class implementing general options
panel. |
CPrPgeSpellOptions_User |
CPropertyPage derived class implementing user dictionary
options panel. |
CPrPgeSpellOptions_Common |
CPropertyPage derived class implementing common misspellings
options panel. |
Support functions of importance:
void CheckSpellingEdit (CFPSSpellCheckEngine* pEngine, CEdit* pEdit) |
This function is called when a
user presses the F7 (or configured) hot key from within the
"check-as-you-type" edit control. It displays the
CDlgSpellChecker dialog box. |
void CheckSpellingRich (CFPSSpellCheckEngine* pEngine, CRichEditCtrl* pEdit) |
This function is called when a
user presses the F7 (or configured) hot key from within the
"check-as-you-type" edit control. It displays the
CDlgSpellChecker dialog box.
NOTE: This function is not currently being used because the rich edit control is not complete. |
int EditDistance(const char *szWord1, const char *szWord2) |
This function is passed in 2 words and returns an approximation of the minimum number of changes a user would need to make to make the 2 words match. This function is not a true edit-distance algorithm, but is a customized algorithm for this spell checking application. |
void MetaphoneEx(const char *szInput, char *szOutput, int
iMaxLen) |
This function is passed a word
and it returns (through the szOutput parameter) a modified-metaphone
representation of the word. This is a variation on the algorithm
originally wrote by Lawrence
Philips. A newer version of his algorithm
(double-metaphone) is also available. I have tested this algorithm
with the spell checking engine and was not impressed with the
results. It does provide fast results and a high hit-rate, but it
also returns far too many results (on average). However, I am
considering using it in conjunction with the EditDistance
algorithm and will further review this. |
void SortMatches(LPCSTR lpszBadWord, CStringList &Matches) |
This function sorts a list of word suggestions based on the approximate edit-distance between the words in the list and the misspelled word based in as lpszBadWord. |
Architecture:
CORE ENGINE
The core spell checking engine consists of the three classes:
CFPSSpellCheckEngine
,CFPSSpellCheckEngineOptions
andCFPSDictionary
. These classes provide support for dictionary related functions such as add a word, remove a word, ignore a word, load dictionary, save dictionary, Is a word in the dictionary, suggest possible matches, etc.The core engine is implemented as a strict back-end engine. It has no user-interface components. Most of the functions exposed by these classes where an error might occur return an
int
return code. These return codes are defined in 1) FPSSpellCheckerInclude.h and 2) the header file for a given class. The return codes should always be examined to determine the completion status of these functions.Special care has been taken to insure that these classes are very stable and robust. Also, performance considerations weigh heavy on the implementation of these classes. Very little MFC code is used in these classes and functions.
CHECK-AS-YOU-TYPE EDIT CONTROL
The check-as-you-type edit control is contained in the
CFPSSpellingEditCtrl
class. It is derived off ofCEdit
and works by subclassing an existing edit control through theAttachEdit
function.To improve performance, this control implements a timer and whenever there is no user activity (typing, mouse clicking, scrolling, etc) checks the spelling of the displayed portion of the edit control. The function
RedrawSpellingErrors
is called to perform the checking. It checks only the displayed portion of the edit control and callsDrawSpellingError
for each displayed word. If a word is not found in the dictionary, this function callsDrawSquiglyI
to draw the squigly underline for the word.DrawSquigly
creates a structure of typeFPSSPELLEDIT_ERRORS
and adds it to the m_SpellingErrors member list.The
OnRButtonDown
function checks the m_SpellingErrors to determine when to display the normal popup menu and when to display the spell check popup menu. Suggestions returned from the core engine are sorted using theSortMatches
function to display them in order of edit-distance.The
PreTranslateMessage
checks for a hot key (defaults to F7). This can be customized by calling theSetHotKey
static member function. When the hot key is pressed theCheckSpellingEdit
function is called to display the spell checking dialog box.SPELL CHECK DIALOG BOX
The spell checking dialog box is implemented in the
CDlgSpellChecker
class. This is a standardCDialog
derived class based on theIDD_SPELL_CHECK
dialog resource.The spell checking dialog is modelled after the Microsoft Word implementation of spell checking. It is laid out the same and functions (for the most part) the same. This dialog searches an edit control (or rich edit control) for sentences misspelled words and displays the sentence with the misspelled word highlighted.
Suggestions returned from the core engine are sorted using the
SortMatches
function to display them in order of edit-distance.
How to use the demo:
- Unzip the provided file into a directory (be sure to extract the sub directories.)
- Make sure that the USMain.dic file is in the \Release directory.
- Make sure that the USCommon.dic file is in the \Release directory.
- Execute the FPSSpellChecker.exe from the \Release directory.
How to incorporate the spell checker into an application:
- In your applications
InitInstance
function, add a call toCFPSSpellingEditCtrl::InitSpellingEngine(NULL)
static member function; OR, instead ofNULL
, pass in a string containing a fully qualified path to a spell checking engine options file. - In your applications
ExitInstance
function, add a call toCFPSSpellingEditCtrl::Terminate
static member function - Add the following files to your
project.
DlgSpellChecker.cpp DlgSpellChecker.h DlgSpellingEditCtrl.cpp DlgSpellingEditCtrl.h FPSDictionary.cpp FPSDictionary.h FPSSpellCheckEngine.cpp FPSSpellCheckEngine.h FPSSpellCheckEngineOptions.cpp FPSSpellCheckEngineOptions.h FPSSpellCheckerInclude.cpp FPSSpellCheckerInclude.h FPSSpellingEditCtrl.cpp FPSSpellingEditCtrl.h PrPgeSpellOptions_Common.cpp PrPgeSpellOptions_Common.h PrPgeSpellOptions_General.cpp PrPgeSpellOptions_General.h PrPgeSpellOptions_User.cpp PrPgeSpellOptions_User.h PrShtSpellOptions.cpp PrShtSpellOptions.h - Copy the following resource items to your
project.
IDD_SPELL_CHECK IDD_SPELL_OPTION_COMMON IDD_SPELL_OPTION_GENERAL IDD_SPELL_OPTION_USER - Include the "FPSSpellCheckerInclude.h"
file in your stdafx.h file.
#include "FPSSpellCheckerInclude.h" - Place a standard edit control on a form or
dialog resource and give it a unique control id (ie.
ID_TEST_EDIT
) - Add a member variable of type
CFPSSpellingEditCtrl
to the dialog/form class file (ie. m_editTest) - In the
OnInitDialog
function, call theAttachEdit
member function ofCFPSSpellingEditCtrl
(ie.m_editTest.AttachEdit(this, ID_TEST_EDIT);
Known Issues
- Performance is still not as good as it needs to be.
- Language support is limited to US English.
- The
EditDistance
function needs work. - The
MetaphoneEx
function needs work. - There is a painting problem with the edit control when scrolling the control while the spelling error "squigly" lines are displayed.
- No complete support for rich edit control.