185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho/* 285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho*************************************************************************** 38393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius* Copyright (C) 2008-2013, International Business Machines Corporation 485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* and others. All Rights Reserved. 585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho*************************************************************************** 685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* 785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* uspoof_impl.h 885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* 985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* Implemenation header for spoof detection 1085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho* 1185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho*/ 1285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 1385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#ifndef USPOOFIM_H 1485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOFIM_H 1585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 1685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#include "unicode/utypes.h" 1785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#include "unicode/uspoof.h" 1885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#include "unicode/uscript.h" 1985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#include "unicode/udata.h" 2085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 218393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius#include "utrie2.h" 2285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 2385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#if !UCONFIG_NO_NORMALIZATION 2485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 25103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius#ifdef __cplusplus 2627f654740f2a26ad62a5c155af9199af9e69b889claireho 2785bf2e2fbc60a9f938064abc8127d61da7d19882Claire HoU_NAMESPACE_BEGIN 2885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 2985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The maximium length (in UTF-16 UChars) of the skeleton replacement string resulting from 3085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// a single input code point. This is function of the unicode.org data. 3185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_MAX_SKELETON_EXPANSION 20 3285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 3385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The default stack buffer size for copies or conversions or normalizations 3485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// of input strings being checked. (Used in multiple places.) 3585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_STACK_BUFFER_SIZE 100 3685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 3785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Magic number for sanity checking spoof data. 3885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_MAGIC 0x3845fdef 3985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 408393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Corneliusclass IdentifierInfo; 418393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Corneliusclass ScriptSet; 4285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hoclass SpoofData; 4385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hostruct SpoofDataHeader; 4485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hostruct SpoofStringLengthsElement; 4585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 4685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho/** 4785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * Class SpoofImpl corresponds directly to the plain C API opaque type 4885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * USpoofChecker. One can be cast to the other. 4985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho */ 5085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hoclass SpoofImpl : public UObject { 5185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hopublic: 5285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofImpl(SpoofData *data, UErrorCode &status); 5385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofImpl(); 5485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho virtual ~SpoofImpl(); 5585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 5685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho /** Copy constructor, used by the user level uspoof_clone() function. 5785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho */ 5885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofImpl(const SpoofImpl &src, UErrorCode &status); 5985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 6085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static SpoofImpl *validateThis(USpoofChecker *sc, UErrorCode &status); 6185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static const SpoofImpl *validateThis(const USpoofChecker *sc, UErrorCode &status); 6285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 6385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho /** Get the confusable skeleton transform for a single code point. 6485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * The result is a string with a length between 1 and 18. 6585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * @param tableMask bit flag specifying which confusable table to use. 6685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * One of USPOOF_SL_TABLE_FLAG, USPOOF_MA_TABLE_FLAG, etc. 6785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * @return The length in UTF-16 code units of the substition string. 6885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho */ 698393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius int32_t confusableLookup(UChar32 inChar, int32_t tableMask, UnicodeString &destBuf) const; 7085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 7185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho /** Set and Get AllowedLocales, implementations of the corresponding API */ 7285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void setAllowedLocales(const char *localesList, UErrorCode &status); 7385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho const char * getAllowedLocales(UErrorCode &status); 7485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 7585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Add (union) to the UnicodeSet all of the characters for the scripts used for 7685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // the specified locale. Part of the implementation of setAllowedLocales. 7785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void addScriptChars(const char *locale, UnicodeSet *allowedChars, UErrorCode &status); 7885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 7985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 8085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho /** parse a hex number. Untility used by the builders. */ 8185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static UChar32 ScanHex(const UChar *s, int32_t start, int32_t limit, UErrorCode &status); 8285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 8385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Implementation for Whole Script tests. 8485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Return the test bit flag to be ORed into the eventual user return value 8585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // if a Spoof opportunity is detected. 8685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void wholeScriptCheck( 878393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius const UnicodeString &text, ScriptSet *result, UErrorCode &status) const; 8885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 8985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static UClassID U_EXPORT2 getStaticClassID(void); 9085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho virtual UClassID getDynamicClassID(void) const; 9185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 928393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius // IdentifierInfo Cache. IdentifierInfo objects are somewhat expensive to create. 938393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius // Maintain a one-element cache, which is sufficient to avoid repeatedly 948393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius // creating new ones unless we get multi-thread concurrency in spoof 958393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius // check operations, which should be statistically uncommon. 968393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius IdentifierInfo *getIdentifierInfo(UErrorCode &status) const; 978393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius void releaseIdentifierInfo(IdentifierInfo *idInfo) const; 988393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius 9985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // 10085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Data Members 10185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // 10285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 10385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fMagic; // Internal sanity check. 10485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fChecks; // Bit vector of checks to perform. 10585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 10685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofData *fSpoofData; 10785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 10885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho const UnicodeSet *fAllowedCharsSet; // The UnicodeSet of allowed characters. 10985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // for this Spoof Checker. Defaults to all chars. 11085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 11185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho const char *fAllowedLocales; // The list of allowed locales. 1128393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius URestrictionLevel fRestrictionLevel; // The maximum restriction level for an acceptable identifier. 1138393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius 1148393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius IdentifierInfo *fCachedIdentifierInfo; // Do not use directly. See getIdentifierInfo().:w 11585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho}; 11685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 11785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 11885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 11985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 12085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Confusable Mappings Data Structures 12185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 12285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// For the confusable data, we are essentially implementing a map, 12385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// key: a code point 12485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// value: a string. Most commonly one char in length, but can be more. 12585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 12685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The keys are stored as a sorted array of 32 bit ints. 12785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// bits 0-23 a code point value 12885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// bits 24-31 flags 12985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 24: 1 if entry applies to SL table 13085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 25: 1 if entry applies to SA table 13185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 26: 1 if entry applies to ML table 13285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 27: 1 if entry applies to MA table 13385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 28: 1 if there are multiple entries for this code point. 13485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 29-30: length of value string, in UChars. 13585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// values are (1, 2, 3, other) 13685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The key table is sorted in ascending code point order. (not on the 13785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 32 bit int value, the flag bits do not participate in the sorting.) 13885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 13985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Lookup is done by means of a binary search in the key table. 14085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 14185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The corresponding values are kept in a parallel array of 16 bit ints. 14285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// If the value string is of length 1, it is literally in the value array. 14385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// For longer strings, the value array contains an index into the strings table. 14485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 14585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// String Table: 14685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The strings table contains all of the value strings (those of length two or greater) 14785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// concatentated together into one long UChar (UTF-16) array. 14885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 14985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The array is arranged by length of the strings - all strings of the same length 15085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// are stored together. The sections are ordered by length of the strings - 15185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// all two char strings first, followed by all of the three Char strings, etc. 15285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 15385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// There is no nul character or other mark between adjacent strings. 15485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 15585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// String Lengths table 15685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The length of strings from 1 to 3 is flagged in the key table. 15785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// For strings of length 4 or longer, the string length table provides a 15885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// mapping between an index into the string table and the corresponding length. 15985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Strings of these lengths are rare, so lookup time is not an issue. 16085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Each entry consists of 16185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// uint16_t index of the _last_ string with this length 16285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// uint16_t the length 16385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 16485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 16585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Flag bits in the Key entries 16685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_SL_TABLE_FLAG (1<<24) 16785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_SA_TABLE_FLAG (1<<25) 16885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_ML_TABLE_FLAG (1<<26) 16985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_MA_TABLE_FLAG (1<<27) 17085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_KEY_MULTIPLE_VALUES (1<<28) 17185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_KEY_LENGTH_SHIFT 29 17285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#define USPOOF_KEY_LENGTH_FIELD(x) (((x)>>29) & 3) 17385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 17485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 17585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hostruct SpoofStringLengthsElement { 17685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho uint16_t fLastString; // index in string table of last string with this length 17785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho uint16_t fStrLength; // Length of strings 17885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho}; 17985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 18085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 18185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 18285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho//------------------------------------------------------------------------------------- 18385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 18485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// SpoofData 18585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 18685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// A small class that wraps the raw (usually memory mapped) spoof data. 18785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Serves two primary functions: 18885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 1. Convenience. Contains real pointers to the data, to avoid dealing with 18985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// the offsets in the raw data. 19085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 2. Reference counting. When a spoof checker is cloned, the raw data is shared 19185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// and must be retained until all checkers using the data are closed. 19285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Nothing in this struct includes state that is specific to any particular 19385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// USpoofDetector object. 19485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 19585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho//--------------------------------------------------------------------------------------- 19685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hoclass SpoofData: public UMemory { 19785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho public: 19885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static SpoofData *getDefault(UErrorCode &status); // Load standard ICU spoof data. 19985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofData(UErrorCode &status); // Create new spoof data wrapper. 20085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Only used when building new data from rules. 20185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 20285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Constructor for use when creating from prebuilt default data. 20385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // A UDataMemory is what the ICU internal data loading functions provide. 20485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // The udm is adopted by the SpoofData. 20585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofData(UDataMemory *udm, UErrorCode &status); 20685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 20785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Constructor for use when creating from serialized data. 20885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // 20985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofData(const void *serializedData, int32_t length, UErrorCode &status); 21085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 21185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Check raw Spoof Data Version compatibility. 21285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Return TRUE it looks good. 21385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho static UBool validateDataVersion(const SpoofDataHeader *rawData, UErrorCode &status); 21485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho ~SpoofData(); // Destructor not normally used. 21585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Use removeReference() instead. 21685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Reference Counting functions. 21785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Clone of a user-level spoof detector increments the ref count on the data. 21885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Close of a user-level spoof detector decrements the ref count. 21985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // If the data is owned by us, it will be deleted when count goes to zero. 22085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofData *addReference(); 22185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void removeReference(); 22285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 22385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Reserve space in the raw data. For use by builder when putting together a 22485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // new set of data. Init the new storage to zero, to prevent inconsistent 22585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // results if it is not all otherwise set by the requester. 22685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Return: 22785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // pointer to the new space that was added by this function. 22885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void *reserveSpace(int32_t numBytes, UErrorCode &status); 22985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 23085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // initialize the pointers from this object to the raw data. 23185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void initPtrs(UErrorCode &status); 23285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 23385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Reset all fields to an initial state. 23485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Called from the top of all constructors. 23585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho void reset(); 23685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 23785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofDataHeader *fRawData; // Ptr to the raw memory-mapped data 23885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UBool fDataOwned; // True if the raw data is owned, and needs 23985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // to be deleted when refcount goes to zero. 24085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UDataMemory *fUDM; // If not NULL, our data came from a 24185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // UDataMemory, which we must close when 242b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho // we are done. 24385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 24485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho uint32_t fMemLimit; // Limit of available raw data space 24559d709d503bab6e2b61931737e662dd293b40578ccornelius u_atomic_int32_t fRefCount; 24685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 24785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Confusable data 24885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t *fCFUKeys; 24985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho uint16_t *fCFUValues; 25085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho SpoofStringLengthsElement *fCFUStringLengths; 25185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UChar *fCFUStrings; 25285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 25385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // Whole Script Confusable Data 25485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UTrie2 *fAnyCaseTrie; 25585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UTrie2 *fLowerCaseTrie; 25685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho ScriptSet *fScriptSets; 25785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho }; 25885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 25985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 26085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho//--------------------------------------------------------------------------------------- 26185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 26285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Raw Binary Data Formats, as loaded from the ICU data file, 26385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// or as built by the builder. 26485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 26585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho//--------------------------------------------------------------------------------------- 26685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Hostruct SpoofDataHeader { 26727f654740f2a26ad62a5c155af9199af9e69b889claireho int32_t fMagic; // (0x3845fdef) 26885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho uint8_t fFormatVersion[4]; // Data Format. Same as the value in struct UDataInfo 26985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // if there is one associated with this data. 27085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fLength; // Total lenght in bytes of this spoof data, 27185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // including all sections, not just the header. 27285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 27385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // The following four sections refer to data representing the confusable data 27485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // from the Unicode.org data from "confusables.txt" 27585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 27685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUKeys; // byte offset to Keys table (from SpoofDataHeader *) 27785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUKeysSize; // number of entries in keys table (32 bits each) 27885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 27985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // TODO: change name to fCFUValues, for consistency. 28085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringIndex; // byte offset to String Indexes table 28185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringIndexSize; // number of entries in String Indexes table (16 bits each) 28285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // (number of entries must be same as in Keys table 28385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 28485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringTable; // byte offset of String table 28585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringTableLen; // length of string table (in 16 bit UChars) 28685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 28785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringLengths; // byte offset to String Lengths table 28885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fCFUStringLengthsSize; // number of entries in lengths table. (2 x 16 bits each) 28985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 29085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 29185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // The following sections are for data from confusablesWholeScript.txt 29285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 29385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fAnyCaseTrie; // byte offset to the serialized Any Case Trie 29485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fAnyCaseTrieLength; // Length (bytes) of the serialized Any Case Trie 29585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 29685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fLowerCaseTrie; // byte offset to the serialized Lower Case Trie 29785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fLowerCaseTrieLength; // Length (bytes) of the serialized Lower Case Trie 29885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 29985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fScriptSets; // byte offset to array of ScriptSets 30085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t fScriptSetsLength; // Number of ScriptSets (24 bytes each) 30185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 30285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 30385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho // The following sections are for data from xidmodifications.txt 30485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 30585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 30685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho int32_t unused[15]; // Padding, Room for Expansion 30785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 30885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho }; 30985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 31085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 31185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 31285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 31385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 31485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Structure for the Whole Script Confusable Data 31585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// See Unicode UAX-39, Unicode Security Mechanisms, for a description of the 31685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// Whole Script confusable data 31785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 31885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The data provides mappings from code points to a set of scripts 31985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// that contain characters that might be confused with the code point. 32085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// There are two mappings, one for lower case only, and one for characters 32185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// of any case. 32285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 32385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// The actual data consists of a utrie2 to map from a code point to an offset, 32485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// and an array of UScriptSets (essentially bit maps) that is indexed 32585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// by the offsets obtained from the Trie. 32685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 32785bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho// 32885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 32985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 33085bf2e2fbc60a9f938064abc8127d61da7d19882Claire HoU_NAMESPACE_END 331103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius#endif /* __cplusplus */ 33285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 33385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho/** 33485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * Endianness swap function for binary spoof data. 33585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * @internal 33685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho */ 33785bf2e2fbc60a9f938064abc8127d61da7d19882Claire HoU_CAPI int32_t U_EXPORT2 33885bf2e2fbc60a9f938064abc8127d61da7d19882Claire Houspoof_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, 33985bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho UErrorCode *status); 34085bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 34185bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 34285bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#endif 34385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 34485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho#endif /* USPOOFIM_H */ 34585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho 346