Wed, 31 Dec 2014 07:22:50 +0100
Correct previous dual key logic pending first delivery installment.
michael@0 | 1 | /* |
michael@0 | 2 | *************************************************************************** |
michael@0 | 3 | * Copyright (C) 2008-2013, International Business Machines Corporation |
michael@0 | 4 | * and others. All Rights Reserved. |
michael@0 | 5 | *************************************************************************** |
michael@0 | 6 | * |
michael@0 | 7 | * uspoof_impl.h |
michael@0 | 8 | * |
michael@0 | 9 | * Implemenation header for spoof detection |
michael@0 | 10 | * |
michael@0 | 11 | */ |
michael@0 | 12 | |
michael@0 | 13 | #ifndef USPOOFIM_H |
michael@0 | 14 | #define USPOOFIM_H |
michael@0 | 15 | |
michael@0 | 16 | #include "unicode/utypes.h" |
michael@0 | 17 | #include "unicode/uspoof.h" |
michael@0 | 18 | #include "unicode/uscript.h" |
michael@0 | 19 | #include "unicode/udata.h" |
michael@0 | 20 | |
michael@0 | 21 | #include "utrie2.h" |
michael@0 | 22 | |
michael@0 | 23 | #if !UCONFIG_NO_NORMALIZATION |
michael@0 | 24 | |
michael@0 | 25 | #ifdef __cplusplus |
michael@0 | 26 | |
michael@0 | 27 | U_NAMESPACE_BEGIN |
michael@0 | 28 | |
michael@0 | 29 | // The maximium length (in UTF-16 UChars) of the skeleton replacement string resulting from |
michael@0 | 30 | // a single input code point. This is function of the unicode.org data. |
michael@0 | 31 | #define USPOOF_MAX_SKELETON_EXPANSION 20 |
michael@0 | 32 | |
michael@0 | 33 | // The default stack buffer size for copies or conversions or normalizations |
michael@0 | 34 | // of input strings being checked. (Used in multiple places.) |
michael@0 | 35 | #define USPOOF_STACK_BUFFER_SIZE 100 |
michael@0 | 36 | |
michael@0 | 37 | // Magic number for sanity checking spoof data. |
michael@0 | 38 | #define USPOOF_MAGIC 0x3845fdef |
michael@0 | 39 | |
michael@0 | 40 | class IdentifierInfo; |
michael@0 | 41 | class ScriptSet; |
michael@0 | 42 | class SpoofData; |
michael@0 | 43 | struct SpoofDataHeader; |
michael@0 | 44 | struct SpoofStringLengthsElement; |
michael@0 | 45 | |
michael@0 | 46 | /** |
michael@0 | 47 | * Class SpoofImpl corresponds directly to the plain C API opaque type |
michael@0 | 48 | * USpoofChecker. One can be cast to the other. |
michael@0 | 49 | */ |
michael@0 | 50 | class SpoofImpl : public UObject { |
michael@0 | 51 | public: |
michael@0 | 52 | SpoofImpl(SpoofData *data, UErrorCode &status); |
michael@0 | 53 | SpoofImpl(); |
michael@0 | 54 | virtual ~SpoofImpl(); |
michael@0 | 55 | |
michael@0 | 56 | /** Copy constructor, used by the user level uspoof_clone() function. |
michael@0 | 57 | */ |
michael@0 | 58 | SpoofImpl(const SpoofImpl &src, UErrorCode &status); |
michael@0 | 59 | |
michael@0 | 60 | static SpoofImpl *validateThis(USpoofChecker *sc, UErrorCode &status); |
michael@0 | 61 | static const SpoofImpl *validateThis(const USpoofChecker *sc, UErrorCode &status); |
michael@0 | 62 | |
michael@0 | 63 | /** Get the confusable skeleton transform for a single code point. |
michael@0 | 64 | * The result is a string with a length between 1 and 18. |
michael@0 | 65 | * @param tableMask bit flag specifying which confusable table to use. |
michael@0 | 66 | * One of USPOOF_SL_TABLE_FLAG, USPOOF_MA_TABLE_FLAG, etc. |
michael@0 | 67 | * @return The length in UTF-16 code units of the substition string. |
michael@0 | 68 | */ |
michael@0 | 69 | int32_t confusableLookup(UChar32 inChar, int32_t tableMask, UnicodeString &destBuf) const; |
michael@0 | 70 | |
michael@0 | 71 | /** Set and Get AllowedLocales, implementations of the corresponding API */ |
michael@0 | 72 | void setAllowedLocales(const char *localesList, UErrorCode &status); |
michael@0 | 73 | const char * getAllowedLocales(UErrorCode &status); |
michael@0 | 74 | |
michael@0 | 75 | // Add (union) to the UnicodeSet all of the characters for the scripts used for |
michael@0 | 76 | // the specified locale. Part of the implementation of setAllowedLocales. |
michael@0 | 77 | void addScriptChars(const char *locale, UnicodeSet *allowedChars, UErrorCode &status); |
michael@0 | 78 | |
michael@0 | 79 | |
michael@0 | 80 | /** parse a hex number. Untility used by the builders. */ |
michael@0 | 81 | static UChar32 ScanHex(const UChar *s, int32_t start, int32_t limit, UErrorCode &status); |
michael@0 | 82 | |
michael@0 | 83 | // Implementation for Whole Script tests. |
michael@0 | 84 | // Return the test bit flag to be ORed into the eventual user return value |
michael@0 | 85 | // if a Spoof opportunity is detected. |
michael@0 | 86 | void wholeScriptCheck( |
michael@0 | 87 | const UnicodeString &text, ScriptSet *result, UErrorCode &status) const; |
michael@0 | 88 | |
michael@0 | 89 | static UClassID U_EXPORT2 getStaticClassID(void); |
michael@0 | 90 | virtual UClassID getDynamicClassID(void) const; |
michael@0 | 91 | |
michael@0 | 92 | // IdentifierInfo Cache. IdentifierInfo objects are somewhat expensive to create. |
michael@0 | 93 | // Maintain a one-element cache, which is sufficient to avoid repeatedly |
michael@0 | 94 | // creating new ones unless we get multi-thread concurrency in spoof |
michael@0 | 95 | // check operations, which should be statistically uncommon. |
michael@0 | 96 | IdentifierInfo *getIdentifierInfo(UErrorCode &status) const; |
michael@0 | 97 | void releaseIdentifierInfo(IdentifierInfo *idInfo) const; |
michael@0 | 98 | |
michael@0 | 99 | // |
michael@0 | 100 | // Data Members |
michael@0 | 101 | // |
michael@0 | 102 | |
michael@0 | 103 | int32_t fMagic; // Internal sanity check. |
michael@0 | 104 | int32_t fChecks; // Bit vector of checks to perform. |
michael@0 | 105 | |
michael@0 | 106 | SpoofData *fSpoofData; |
michael@0 | 107 | |
michael@0 | 108 | const UnicodeSet *fAllowedCharsSet; // The UnicodeSet of allowed characters. |
michael@0 | 109 | // for this Spoof Checker. Defaults to all chars. |
michael@0 | 110 | |
michael@0 | 111 | const char *fAllowedLocales; // The list of allowed locales. |
michael@0 | 112 | URestrictionLevel fRestrictionLevel; // The maximum restriction level for an acceptable identifier. |
michael@0 | 113 | |
michael@0 | 114 | IdentifierInfo *fCachedIdentifierInfo; // Do not use directly. See getIdentifierInfo().:w |
michael@0 | 115 | }; |
michael@0 | 116 | |
michael@0 | 117 | |
michael@0 | 118 | |
michael@0 | 119 | // |
michael@0 | 120 | // Confusable Mappings Data Structures |
michael@0 | 121 | // |
michael@0 | 122 | // For the confusable data, we are essentially implementing a map, |
michael@0 | 123 | // key: a code point |
michael@0 | 124 | // value: a string. Most commonly one char in length, but can be more. |
michael@0 | 125 | // |
michael@0 | 126 | // The keys are stored as a sorted array of 32 bit ints. |
michael@0 | 127 | // bits 0-23 a code point value |
michael@0 | 128 | // bits 24-31 flags |
michael@0 | 129 | // 24: 1 if entry applies to SL table |
michael@0 | 130 | // 25: 1 if entry applies to SA table |
michael@0 | 131 | // 26: 1 if entry applies to ML table |
michael@0 | 132 | // 27: 1 if entry applies to MA table |
michael@0 | 133 | // 28: 1 if there are multiple entries for this code point. |
michael@0 | 134 | // 29-30: length of value string, in UChars. |
michael@0 | 135 | // values are (1, 2, 3, other) |
michael@0 | 136 | // The key table is sorted in ascending code point order. (not on the |
michael@0 | 137 | // 32 bit int value, the flag bits do not participate in the sorting.) |
michael@0 | 138 | // |
michael@0 | 139 | // Lookup is done by means of a binary search in the key table. |
michael@0 | 140 | // |
michael@0 | 141 | // The corresponding values are kept in a parallel array of 16 bit ints. |
michael@0 | 142 | // If the value string is of length 1, it is literally in the value array. |
michael@0 | 143 | // For longer strings, the value array contains an index into the strings table. |
michael@0 | 144 | // |
michael@0 | 145 | // String Table: |
michael@0 | 146 | // The strings table contains all of the value strings (those of length two or greater) |
michael@0 | 147 | // concatentated together into one long UChar (UTF-16) array. |
michael@0 | 148 | // |
michael@0 | 149 | // The array is arranged by length of the strings - all strings of the same length |
michael@0 | 150 | // are stored together. The sections are ordered by length of the strings - |
michael@0 | 151 | // all two char strings first, followed by all of the three Char strings, etc. |
michael@0 | 152 | // |
michael@0 | 153 | // There is no nul character or other mark between adjacent strings. |
michael@0 | 154 | // |
michael@0 | 155 | // String Lengths table |
michael@0 | 156 | // The length of strings from 1 to 3 is flagged in the key table. |
michael@0 | 157 | // For strings of length 4 or longer, the string length table provides a |
michael@0 | 158 | // mapping between an index into the string table and the corresponding length. |
michael@0 | 159 | // Strings of these lengths are rare, so lookup time is not an issue. |
michael@0 | 160 | // Each entry consists of |
michael@0 | 161 | // uint16_t index of the _last_ string with this length |
michael@0 | 162 | // uint16_t the length |
michael@0 | 163 | // |
michael@0 | 164 | |
michael@0 | 165 | // Flag bits in the Key entries |
michael@0 | 166 | #define USPOOF_SL_TABLE_FLAG (1<<24) |
michael@0 | 167 | #define USPOOF_SA_TABLE_FLAG (1<<25) |
michael@0 | 168 | #define USPOOF_ML_TABLE_FLAG (1<<26) |
michael@0 | 169 | #define USPOOF_MA_TABLE_FLAG (1<<27) |
michael@0 | 170 | #define USPOOF_KEY_MULTIPLE_VALUES (1<<28) |
michael@0 | 171 | #define USPOOF_KEY_LENGTH_SHIFT 29 |
michael@0 | 172 | #define USPOOF_KEY_LENGTH_FIELD(x) (((x)>>29) & 3) |
michael@0 | 173 | |
michael@0 | 174 | |
michael@0 | 175 | struct SpoofStringLengthsElement { |
michael@0 | 176 | uint16_t fLastString; // index in string table of last string with this length |
michael@0 | 177 | uint16_t fStrLength; // Length of strings |
michael@0 | 178 | }; |
michael@0 | 179 | |
michael@0 | 180 | |
michael@0 | 181 | |
michael@0 | 182 | //------------------------------------------------------------------------------------- |
michael@0 | 183 | // |
michael@0 | 184 | // SpoofData |
michael@0 | 185 | // |
michael@0 | 186 | // A small class that wraps the raw (usually memory mapped) spoof data. |
michael@0 | 187 | // Serves two primary functions: |
michael@0 | 188 | // 1. Convenience. Contains real pointers to the data, to avoid dealing with |
michael@0 | 189 | // the offsets in the raw data. |
michael@0 | 190 | // 2. Reference counting. When a spoof checker is cloned, the raw data is shared |
michael@0 | 191 | // and must be retained until all checkers using the data are closed. |
michael@0 | 192 | // Nothing in this struct includes state that is specific to any particular |
michael@0 | 193 | // USpoofDetector object. |
michael@0 | 194 | // |
michael@0 | 195 | //--------------------------------------------------------------------------------------- |
michael@0 | 196 | class SpoofData: public UMemory { |
michael@0 | 197 | public: |
michael@0 | 198 | static SpoofData *getDefault(UErrorCode &status); // Load standard ICU spoof data. |
michael@0 | 199 | SpoofData(UErrorCode &status); // Create new spoof data wrapper. |
michael@0 | 200 | // Only used when building new data from rules. |
michael@0 | 201 | |
michael@0 | 202 | // Constructor for use when creating from prebuilt default data. |
michael@0 | 203 | // A UDataMemory is what the ICU internal data loading functions provide. |
michael@0 | 204 | // The udm is adopted by the SpoofData. |
michael@0 | 205 | SpoofData(UDataMemory *udm, UErrorCode &status); |
michael@0 | 206 | |
michael@0 | 207 | // Constructor for use when creating from serialized data. |
michael@0 | 208 | // |
michael@0 | 209 | SpoofData(const void *serializedData, int32_t length, UErrorCode &status); |
michael@0 | 210 | |
michael@0 | 211 | // Check raw Spoof Data Version compatibility. |
michael@0 | 212 | // Return TRUE it looks good. |
michael@0 | 213 | static UBool validateDataVersion(const SpoofDataHeader *rawData, UErrorCode &status); |
michael@0 | 214 | ~SpoofData(); // Destructor not normally used. |
michael@0 | 215 | // Use removeReference() instead. |
michael@0 | 216 | // Reference Counting functions. |
michael@0 | 217 | // Clone of a user-level spoof detector increments the ref count on the data. |
michael@0 | 218 | // Close of a user-level spoof detector decrements the ref count. |
michael@0 | 219 | // If the data is owned by us, it will be deleted when count goes to zero. |
michael@0 | 220 | SpoofData *addReference(); |
michael@0 | 221 | void removeReference(); |
michael@0 | 222 | |
michael@0 | 223 | // Reserve space in the raw data. For use by builder when putting together a |
michael@0 | 224 | // new set of data. Init the new storage to zero, to prevent inconsistent |
michael@0 | 225 | // results if it is not all otherwise set by the requester. |
michael@0 | 226 | // Return: |
michael@0 | 227 | // pointer to the new space that was added by this function. |
michael@0 | 228 | void *reserveSpace(int32_t numBytes, UErrorCode &status); |
michael@0 | 229 | |
michael@0 | 230 | // initialize the pointers from this object to the raw data. |
michael@0 | 231 | void initPtrs(UErrorCode &status); |
michael@0 | 232 | |
michael@0 | 233 | // Reset all fields to an initial state. |
michael@0 | 234 | // Called from the top of all constructors. |
michael@0 | 235 | void reset(); |
michael@0 | 236 | |
michael@0 | 237 | SpoofDataHeader *fRawData; // Ptr to the raw memory-mapped data |
michael@0 | 238 | UBool fDataOwned; // True if the raw data is owned, and needs |
michael@0 | 239 | // to be deleted when refcount goes to zero. |
michael@0 | 240 | UDataMemory *fUDM; // If not NULL, our data came from a |
michael@0 | 241 | // UDataMemory, which we must close when |
michael@0 | 242 | // we are done. |
michael@0 | 243 | |
michael@0 | 244 | uint32_t fMemLimit; // Limit of available raw data space |
michael@0 | 245 | u_atomic_int32_t fRefCount; |
michael@0 | 246 | |
michael@0 | 247 | // Confusable data |
michael@0 | 248 | int32_t *fCFUKeys; |
michael@0 | 249 | uint16_t *fCFUValues; |
michael@0 | 250 | SpoofStringLengthsElement *fCFUStringLengths; |
michael@0 | 251 | UChar *fCFUStrings; |
michael@0 | 252 | |
michael@0 | 253 | // Whole Script Confusable Data |
michael@0 | 254 | UTrie2 *fAnyCaseTrie; |
michael@0 | 255 | UTrie2 *fLowerCaseTrie; |
michael@0 | 256 | ScriptSet *fScriptSets; |
michael@0 | 257 | }; |
michael@0 | 258 | |
michael@0 | 259 | |
michael@0 | 260 | //--------------------------------------------------------------------------------------- |
michael@0 | 261 | // |
michael@0 | 262 | // Raw Binary Data Formats, as loaded from the ICU data file, |
michael@0 | 263 | // or as built by the builder. |
michael@0 | 264 | // |
michael@0 | 265 | //--------------------------------------------------------------------------------------- |
michael@0 | 266 | struct SpoofDataHeader { |
michael@0 | 267 | int32_t fMagic; // (0x3845fdef) |
michael@0 | 268 | uint8_t fFormatVersion[4]; // Data Format. Same as the value in struct UDataInfo |
michael@0 | 269 | // if there is one associated with this data. |
michael@0 | 270 | int32_t fLength; // Total lenght in bytes of this spoof data, |
michael@0 | 271 | // including all sections, not just the header. |
michael@0 | 272 | |
michael@0 | 273 | // The following four sections refer to data representing the confusable data |
michael@0 | 274 | // from the Unicode.org data from "confusables.txt" |
michael@0 | 275 | |
michael@0 | 276 | int32_t fCFUKeys; // byte offset to Keys table (from SpoofDataHeader *) |
michael@0 | 277 | int32_t fCFUKeysSize; // number of entries in keys table (32 bits each) |
michael@0 | 278 | |
michael@0 | 279 | // TODO: change name to fCFUValues, for consistency. |
michael@0 | 280 | int32_t fCFUStringIndex; // byte offset to String Indexes table |
michael@0 | 281 | int32_t fCFUStringIndexSize; // number of entries in String Indexes table (16 bits each) |
michael@0 | 282 | // (number of entries must be same as in Keys table |
michael@0 | 283 | |
michael@0 | 284 | int32_t fCFUStringTable; // byte offset of String table |
michael@0 | 285 | int32_t fCFUStringTableLen; // length of string table (in 16 bit UChars) |
michael@0 | 286 | |
michael@0 | 287 | int32_t fCFUStringLengths; // byte offset to String Lengths table |
michael@0 | 288 | int32_t fCFUStringLengthsSize; // number of entries in lengths table. (2 x 16 bits each) |
michael@0 | 289 | |
michael@0 | 290 | |
michael@0 | 291 | // The following sections are for data from confusablesWholeScript.txt |
michael@0 | 292 | |
michael@0 | 293 | int32_t fAnyCaseTrie; // byte offset to the serialized Any Case Trie |
michael@0 | 294 | int32_t fAnyCaseTrieLength; // Length (bytes) of the serialized Any Case Trie |
michael@0 | 295 | |
michael@0 | 296 | int32_t fLowerCaseTrie; // byte offset to the serialized Lower Case Trie |
michael@0 | 297 | int32_t fLowerCaseTrieLength; // Length (bytes) of the serialized Lower Case Trie |
michael@0 | 298 | |
michael@0 | 299 | int32_t fScriptSets; // byte offset to array of ScriptSets |
michael@0 | 300 | int32_t fScriptSetsLength; // Number of ScriptSets (24 bytes each) |
michael@0 | 301 | |
michael@0 | 302 | |
michael@0 | 303 | // The following sections are for data from xidmodifications.txt |
michael@0 | 304 | |
michael@0 | 305 | |
michael@0 | 306 | int32_t unused[15]; // Padding, Room for Expansion |
michael@0 | 307 | |
michael@0 | 308 | }; |
michael@0 | 309 | |
michael@0 | 310 | |
michael@0 | 311 | |
michael@0 | 312 | |
michael@0 | 313 | // |
michael@0 | 314 | // Structure for the Whole Script Confusable Data |
michael@0 | 315 | // See Unicode UAX-39, Unicode Security Mechanisms, for a description of the |
michael@0 | 316 | // Whole Script confusable data |
michael@0 | 317 | // |
michael@0 | 318 | // The data provides mappings from code points to a set of scripts |
michael@0 | 319 | // that contain characters that might be confused with the code point. |
michael@0 | 320 | // There are two mappings, one for lower case only, and one for characters |
michael@0 | 321 | // of any case. |
michael@0 | 322 | // |
michael@0 | 323 | // The actual data consists of a utrie2 to map from a code point to an offset, |
michael@0 | 324 | // and an array of UScriptSets (essentially bit maps) that is indexed |
michael@0 | 325 | // by the offsets obtained from the Trie. |
michael@0 | 326 | // |
michael@0 | 327 | // |
michael@0 | 328 | |
michael@0 | 329 | |
michael@0 | 330 | U_NAMESPACE_END |
michael@0 | 331 | #endif /* __cplusplus */ |
michael@0 | 332 | |
michael@0 | 333 | /** |
michael@0 | 334 | * Endianness swap function for binary spoof data. |
michael@0 | 335 | * @internal |
michael@0 | 336 | */ |
michael@0 | 337 | U_CAPI int32_t U_EXPORT2 |
michael@0 | 338 | uspoof_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, |
michael@0 | 339 | UErrorCode *status); |
michael@0 | 340 | |
michael@0 | 341 | |
michael@0 | 342 | #endif |
michael@0 | 343 | |
michael@0 | 344 | #endif /* USPOOFIM_H */ |
michael@0 | 345 |