michael@0: /*
michael@0:  ***************************************************************************
michael@0:  * Copyright (C) 2008-2009, International Business Machines Corporation
michael@0:  * and others. All Rights Reserved.
michael@0:  ***************************************************************************
michael@0:  *   file name:  uspoof_build.cpp
michael@0:  *   encoding:   US-ASCII
michael@0:  *   tab size:   8 (not used)
michael@0:  *   indentation:4
michael@0:  *
michael@0:  *   created on: 2008 Dec 8
michael@0:  *   created by: Andy Heninger
michael@0:  *
michael@0:  *   Unicode Spoof Detection Data Builder
michael@0:  *   Builder-related functions are kept in separate files so that applications not needing
michael@0:  *   the builder can more easily exclude them, typically by means of static linking.
michael@0:  *
michael@0:  *   There are three relatively independent sets of Spoof data,
michael@0:  *      Confusables,
michael@0:  *      Whole Script Confusables
michael@0:  *      ID character extensions.
michael@0:  *
michael@0:  *   The data tables for each are built separately, each from its own definitions
michael@0:  */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uspoof.h"
michael@0: #include "unicode/unorm.h"
michael@0: #include "unicode/uregex.h"
michael@0: #include "unicode/ustring.h"
michael@0: #include "cmemory.h"
michael@0: #include "uspoof_impl.h"
michael@0: #include "uhash.h"
michael@0: #include "uvector.h"
michael@0: #include "uassert.h"
michael@0: #include "uarrsort.h"
michael@0: #include "uspoof_conf.h"
michael@0: #include "uspoof_wsconf.h"
michael@0: 
michael@0: #if !UCONFIG_NO_NORMALIZATION
michael@0: 
michael@0: U_NAMESPACE_USE
michael@0: 
michael@0: 
michael@0: // The main data building function
michael@0: 
michael@0: U_CAPI USpoofChecker * U_EXPORT2
michael@0: uspoof_openFromSource(const char *confusables,  int32_t confusablesLen,
michael@0:                       const char *confusablesWholeScript, int32_t confusablesWholeScriptLen,
michael@0:                       int32_t *errorType, UParseError *pe, UErrorCode *status) {
michael@0: 
michael@0:     if (U_FAILURE(*status)) {
michael@0:         return NULL;
michael@0:     }
michael@0: #if UCONFIG_NO_REGULAR_EXPRESSIONS 
michael@0:     *status = U_UNSUPPORTED_ERROR;      
michael@0:     return NULL;
michael@0: #else
michael@0:     if (errorType!=NULL) {
michael@0:         *errorType = 0;
michael@0:     }
michael@0:     if (pe != NULL) {
michael@0:         pe->line = 0;
michael@0:         pe->offset = 0;
michael@0:         pe->preContext[0] = 0;
michael@0:         pe->postContext[0] = 0;
michael@0:     }
michael@0: 
michael@0:     // Set up a shell of a spoof detector, with empty data.
michael@0:     SpoofData *newSpoofData = new SpoofData(*status);
michael@0:     SpoofImpl *This = new SpoofImpl(newSpoofData, *status);
michael@0: 
michael@0:     // Compile the binary data from the source (text) format.
michael@0:     ConfusabledataBuilder::buildConfusableData(This, confusables, confusablesLen, errorType, pe, *status);
michael@0:     buildWSConfusableData(This, confusablesWholeScript, confusablesWholeScriptLen, pe, *status);
michael@0:     
michael@0:     if (U_FAILURE(*status)) {
michael@0:         delete This;
michael@0:         This = NULL;
michael@0:     }
michael@0:     return (USpoofChecker *)This;
michael@0: #endif // UCONFIG_NO_REGULAR_EXPRESSIONS 
michael@0: }
michael@0: 
michael@0: #endif