intl/icu/source/i18n/uspoof_impl.h

changeset 0
6474c204b198
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/intl/icu/source/i18n/uspoof_impl.h	Wed Dec 31 06:09:35 2014 +0100
     1.3 @@ -0,0 +1,345 @@
     1.4 +/*
     1.5 +***************************************************************************
     1.6 +* Copyright (C) 2008-2013, International Business Machines Corporation
     1.7 +* and others. All Rights Reserved.
     1.8 +***************************************************************************
     1.9 +*
    1.10 +*  uspoof_impl.h
    1.11 +*
    1.12 +*    Implemenation header for spoof detection
    1.13 +*
    1.14 +*/
    1.15 +
    1.16 +#ifndef USPOOFIM_H
    1.17 +#define USPOOFIM_H
    1.18 +
    1.19 +#include "unicode/utypes.h"
    1.20 +#include "unicode/uspoof.h"
    1.21 +#include "unicode/uscript.h"
    1.22 +#include "unicode/udata.h"
    1.23 +
    1.24 +#include "utrie2.h"
    1.25 +
    1.26 +#if !UCONFIG_NO_NORMALIZATION
    1.27 +
    1.28 +#ifdef __cplusplus
    1.29 +
    1.30 +U_NAMESPACE_BEGIN
    1.31 +
    1.32 +// The maximium length (in UTF-16 UChars) of the skeleton replacement string resulting from
    1.33 +//   a single input code point.  This is function of the unicode.org data.
    1.34 +#define USPOOF_MAX_SKELETON_EXPANSION 20
    1.35 +
    1.36 +// The default stack buffer size for copies or conversions or normalizations
    1.37 +// of input strings being checked.  (Used in multiple places.)
    1.38 +#define USPOOF_STACK_BUFFER_SIZE 100
    1.39 +
    1.40 +// Magic number for sanity checking spoof data.
    1.41 +#define USPOOF_MAGIC 0x3845fdef
    1.42 +
    1.43 +class IdentifierInfo;
    1.44 +class ScriptSet;
    1.45 +class SpoofData;
    1.46 +struct SpoofDataHeader;
    1.47 +struct SpoofStringLengthsElement;
    1.48 +
    1.49 +/**
    1.50 +  *  Class SpoofImpl corresponds directly to the plain C API opaque type
    1.51 +  *  USpoofChecker.  One can be cast to the other.
    1.52 +  */
    1.53 +class SpoofImpl : public UObject  {
    1.54 +public:
    1.55 +	SpoofImpl(SpoofData *data, UErrorCode &status);
    1.56 +	SpoofImpl();
    1.57 +	virtual ~SpoofImpl();
    1.58 +
    1.59 +    /** Copy constructor, used by the user level uspoof_clone() function.
    1.60 +     */
    1.61 +    SpoofImpl(const SpoofImpl &src, UErrorCode &status);
    1.62 +    
    1.63 +    static SpoofImpl *validateThis(USpoofChecker *sc, UErrorCode &status);
    1.64 +    static const SpoofImpl *validateThis(const USpoofChecker *sc, UErrorCode &status);
    1.65 +
    1.66 +    /** Get the confusable skeleton transform for a single code point.
    1.67 +     *  The result is a string with a length between 1 and 18.
    1.68 +     *  @param    tableMask  bit flag specifying which confusable table to use.
    1.69 +     *                       One of USPOOF_SL_TABLE_FLAG, USPOOF_MA_TABLE_FLAG, etc.
    1.70 +     *  @return   The length in UTF-16 code units of the substition string.
    1.71 +     */  
    1.72 +    int32_t confusableLookup(UChar32 inChar, int32_t tableMask, UnicodeString &destBuf) const;
    1.73 +
    1.74 +    /** Set and Get AllowedLocales, implementations of the corresponding API */
    1.75 +    void setAllowedLocales(const char *localesList, UErrorCode &status);
    1.76 +    const char * getAllowedLocales(UErrorCode &status);
    1.77 +
    1.78 +    // Add (union) to the UnicodeSet all of the characters for the scripts used for
    1.79 +    // the specified locale.  Part of the implementation of setAllowedLocales.
    1.80 +    void addScriptChars(const char *locale, UnicodeSet *allowedChars, UErrorCode &status);
    1.81 +
    1.82 +
    1.83 +    /** parse a hex number.  Untility used by the builders.   */
    1.84 +    static UChar32 ScanHex(const UChar *s, int32_t start, int32_t limit, UErrorCode &status);
    1.85 +
    1.86 +    // Implementation for Whole Script tests.
    1.87 +    // Return the test bit flag to be ORed into the eventual user return value
    1.88 +    //    if a Spoof opportunity is detected.
    1.89 +    void wholeScriptCheck(
    1.90 +        const UnicodeString &text, ScriptSet *result, UErrorCode &status) const;
    1.91 +	    
    1.92 +    static UClassID U_EXPORT2 getStaticClassID(void);
    1.93 +    virtual UClassID getDynamicClassID(void) const;
    1.94 +
    1.95 +    // IdentifierInfo Cache. IdentifierInfo objects are somewhat expensive to create.
    1.96 +    //                       Maintain a one-element cache, which is sufficient to avoid repeatedly
    1.97 +    //                       creating new ones unless we get multi-thread concurrency in spoof
    1.98 +    //                       check operations, which should be statistically uncommon.
    1.99 +    IdentifierInfo *getIdentifierInfo(UErrorCode &status) const; 
   1.100 +    void releaseIdentifierInfo(IdentifierInfo *idInfo) const;
   1.101 +
   1.102 +    //
   1.103 +    // Data Members
   1.104 +    //
   1.105 +
   1.106 +    int32_t           fMagic;             // Internal sanity check.
   1.107 +    int32_t           fChecks;            // Bit vector of checks to perform.
   1.108 +
   1.109 +    SpoofData        *fSpoofData;
   1.110 +    
   1.111 +    const UnicodeSet *fAllowedCharsSet;   // The UnicodeSet of allowed characters.
   1.112 +                                          //   for this Spoof Checker.  Defaults to all chars. 
   1.113 +
   1.114 +    const char       *fAllowedLocales;    // The list of allowed locales.
   1.115 +    URestrictionLevel fRestrictionLevel;  // The maximum restriction level for an acceptable identifier.
   1.116 +
   1.117 +    IdentifierInfo    *fCachedIdentifierInfo;    // Do not use directly. See getIdentifierInfo().:w
   1.118 +};
   1.119 +
   1.120 +
   1.121 +
   1.122 +//
   1.123 +//  Confusable Mappings Data Structures
   1.124 +//
   1.125 +//    For the confusable data, we are essentially implementing a map,
   1.126 +//       key:    a code point
   1.127 +//       value:  a string.  Most commonly one char in length, but can be more.
   1.128 +//
   1.129 +//    The keys are stored as a sorted array of 32 bit ints.
   1.130 +//             bits 0-23    a code point value
   1.131 +//             bits 24-31   flags
   1.132 +//                24:  1 if entry applies to SL table
   1.133 +//                25:  1 if entry applies to SA table
   1.134 +//                26:  1 if entry applies to ML table
   1.135 +//                27:  1 if entry applies to MA table
   1.136 +//                28:  1 if there are multiple entries for this code point.
   1.137 +//                29-30:  length of value string, in UChars.
   1.138 +//                         values are (1, 2, 3, other)
   1.139 +//        The key table is sorted in ascending code point order.  (not on the
   1.140 +//        32 bit int value, the flag bits do not participate in the sorting.)
   1.141 +//
   1.142 +//        Lookup is done by means of a binary search in the key table.
   1.143 +//
   1.144 +//    The corresponding values are kept in a parallel array of 16 bit ints.
   1.145 +//        If the value string is of length 1, it is literally in the value array.
   1.146 +//        For longer strings, the value array contains an index into the strings table.
   1.147 +//
   1.148 +//    String Table:
   1.149 +//       The strings table contains all of the value strings (those of length two or greater)
   1.150 +//       concatentated together into one long UChar (UTF-16) array.
   1.151 +//
   1.152 +//       The array is arranged by length of the strings - all strings of the same length
   1.153 +//       are stored together.  The sections are ordered by length of the strings -
   1.154 +//       all two char strings first, followed by all of the three Char strings, etc.
   1.155 +//
   1.156 +//       There is no nul character or other mark between adjacent strings.
   1.157 +//
   1.158 +//    String Lengths table
   1.159 +//       The length of strings from 1 to 3 is flagged in the key table.
   1.160 +//       For strings of length 4 or longer, the string length table provides a
   1.161 +//       mapping between an index into the string table and the corresponding length.
   1.162 +//       Strings of these lengths are rare, so lookup time is not an issue.
   1.163 +//       Each entry consists of
   1.164 +//            uint16_t      index of the _last_ string with this length
   1.165 +//            uint16_t      the length
   1.166 +//
   1.167 +
   1.168 +// Flag bits in the Key entries
   1.169 +#define USPOOF_SL_TABLE_FLAG (1<<24)
   1.170 +#define USPOOF_SA_TABLE_FLAG (1<<25)
   1.171 +#define USPOOF_ML_TABLE_FLAG (1<<26)
   1.172 +#define USPOOF_MA_TABLE_FLAG (1<<27)
   1.173 +#define USPOOF_KEY_MULTIPLE_VALUES (1<<28)
   1.174 +#define USPOOF_KEY_LENGTH_SHIFT 29
   1.175 +#define USPOOF_KEY_LENGTH_FIELD(x) (((x)>>29) & 3)
   1.176 +
   1.177 +
   1.178 +struct SpoofStringLengthsElement {
   1.179 +    uint16_t      fLastString;         // index in string table of last string with this length
   1.180 +    uint16_t      fStrLength;           // Length of strings
   1.181 +};
   1.182 +
   1.183 +
   1.184 +
   1.185 +//-------------------------------------------------------------------------------------
   1.186 +//
   1.187 +//  SpoofData
   1.188 +//
   1.189 +//    A small class that wraps the raw (usually memory mapped) spoof data.
   1.190 +//    Serves two primary functions:
   1.191 +//      1.  Convenience.  Contains real pointers to the data, to avoid dealing with
   1.192 +//          the offsets in the raw data.
   1.193 +//      2.  Reference counting.  When a spoof checker is cloned, the raw data is shared
   1.194 +//          and must be retained until all checkers using the data are closed.
   1.195 +//    Nothing in this struct includes state that is specific to any particular
   1.196 +//    USpoofDetector object.
   1.197 +//
   1.198 +//---------------------------------------------------------------------------------------
   1.199 +class SpoofData: public UMemory {
   1.200 +  public:
   1.201 +    static SpoofData *getDefault(UErrorCode &status);   // Load standard ICU spoof data.
   1.202 +    SpoofData(UErrorCode &status);   // Create new spoof data wrapper.
   1.203 +                                     // Only used when building new data from rules.
   1.204 +    
   1.205 +    // Constructor for use when creating from prebuilt default data.
   1.206 +    //   A UDataMemory is what the ICU internal data loading functions provide.
   1.207 +    //   The udm is adopted by the SpoofData.
   1.208 +    SpoofData(UDataMemory *udm, UErrorCode &status);
   1.209 +
   1.210 +    // Constructor for use when creating from serialized data.
   1.211 +    //
   1.212 +    SpoofData(const void *serializedData, int32_t length, UErrorCode &status);
   1.213 +
   1.214 +    //  Check raw Spoof Data Version compatibility.
   1.215 +    //  Return TRUE it looks good.
   1.216 +    static UBool validateDataVersion(const SpoofDataHeader *rawData, UErrorCode &status);
   1.217 +    ~SpoofData();                    // Destructor not normally used.
   1.218 +                                     // Use removeReference() instead.
   1.219 +    // Reference Counting functions.
   1.220 +    //    Clone of a user-level spoof detector increments the ref count on the data.
   1.221 +    //    Close of a user-level spoof detector decrements the ref count.
   1.222 +    //    If the data is owned by us, it will be deleted when count goes to zero.
   1.223 +    SpoofData *addReference(); 
   1.224 +    void removeReference();
   1.225 +
   1.226 +    // Reserve space in the raw data.  For use by builder when putting together a
   1.227 +    //   new set of data.  Init the new storage to zero, to prevent inconsistent
   1.228 +    //   results if it is not all otherwise set by the requester.
   1.229 +    //  Return:
   1.230 +    //    pointer to the new space that was added by this function.
   1.231 +    void *reserveSpace(int32_t numBytes, UErrorCode &status);
   1.232 +
   1.233 +    // initialize the pointers from this object to the raw data.
   1.234 +    void initPtrs(UErrorCode &status);
   1.235 +
   1.236 +    // Reset all fields to an initial state.
   1.237 +    // Called from the top of all constructors.
   1.238 +    void reset();
   1.239 +    
   1.240 +    SpoofDataHeader             *fRawData;          // Ptr to the raw memory-mapped data
   1.241 +    UBool                       fDataOwned;         // True if the raw data is owned, and needs
   1.242 +                                                    //  to be deleted when refcount goes to zero.
   1.243 +    UDataMemory                 *fUDM;              // If not NULL, our data came from a
   1.244 +                                                    //   UDataMemory, which we must close when
   1.245 +                                                    //   we are done.
   1.246 +
   1.247 +    uint32_t                    fMemLimit;          // Limit of available raw data space
   1.248 +    u_atomic_int32_t            fRefCount;
   1.249 +
   1.250 +    // Confusable data
   1.251 +    int32_t                     *fCFUKeys;
   1.252 +    uint16_t                    *fCFUValues;
   1.253 +    SpoofStringLengthsElement   *fCFUStringLengths;
   1.254 +    UChar                       *fCFUStrings;
   1.255 +
   1.256 +    // Whole Script Confusable Data
   1.257 +    UTrie2                      *fAnyCaseTrie;
   1.258 +    UTrie2                      *fLowerCaseTrie;
   1.259 +    ScriptSet                   *fScriptSets;
   1.260 +    };
   1.261 +    
   1.262 +
   1.263 +//---------------------------------------------------------------------------------------
   1.264 +//
   1.265 +//  Raw Binary Data Formats, as loaded from the ICU data file,
   1.266 +//    or as built by the builder.
   1.267 +//
   1.268 +//---------------------------------------------------------------------------------------
   1.269 +struct SpoofDataHeader {
   1.270 +    int32_t       fMagic;                // (0x3845fdef)
   1.271 +    uint8_t       fFormatVersion[4];     // Data Format. Same as the value in struct UDataInfo
   1.272 +                                         //   if there is one associated with this data.
   1.273 +    int32_t       fLength;               // Total lenght in bytes of this spoof data,
   1.274 +                                         //   including all sections, not just the header.
   1.275 +
   1.276 +    // The following four sections refer to data representing the confusable data
   1.277 +    //   from the Unicode.org data from "confusables.txt"
   1.278 +
   1.279 +    int32_t       fCFUKeys;               // byte offset to Keys table (from SpoofDataHeader *)
   1.280 +    int32_t       fCFUKeysSize;           // number of entries in keys table  (32 bits each)
   1.281 +
   1.282 +    // TODO: change name to fCFUValues, for consistency.
   1.283 +    int32_t       fCFUStringIndex;        // byte offset to String Indexes table
   1.284 +    int32_t       fCFUStringIndexSize;    // number of entries in String Indexes table (16 bits each)
   1.285 +                                          //     (number of entries must be same as in Keys table
   1.286 +
   1.287 +    int32_t       fCFUStringTable;        // byte offset of String table
   1.288 +    int32_t       fCFUStringTableLen;     // length of string table (in 16 bit UChars)
   1.289 +
   1.290 +    int32_t       fCFUStringLengths;      // byte offset to String Lengths table
   1.291 +    int32_t       fCFUStringLengthsSize;  // number of entries in lengths table. (2 x 16 bits each)
   1.292 +
   1.293 +
   1.294 +    // The following sections are for data from confusablesWholeScript.txt
   1.295 +    
   1.296 +    int32_t       fAnyCaseTrie;           // byte offset to the serialized Any Case Trie
   1.297 +    int32_t       fAnyCaseTrieLength;     // Length (bytes) of the serialized Any Case Trie
   1.298 +    
   1.299 +    int32_t       fLowerCaseTrie;         // byte offset to the serialized Lower Case Trie
   1.300 +    int32_t       fLowerCaseTrieLength;   // Length (bytes) of the serialized Lower Case Trie
   1.301 +
   1.302 +    int32_t       fScriptSets;            // byte offset to array of ScriptSets
   1.303 +    int32_t       fScriptSetsLength;      // Number of ScriptSets (24 bytes each)
   1.304 +    
   1.305 +
   1.306 +    // The following sections are for data from xidmodifications.txt
   1.307 +    
   1.308 +    
   1.309 +    int32_t       unused[15];              // Padding, Room for Expansion
   1.310 +    
   1.311 + }; 
   1.312 +
   1.313 +
   1.314 +
   1.315 +    
   1.316 +//
   1.317 +//  Structure for the Whole Script Confusable Data
   1.318 +//    See Unicode UAX-39, Unicode Security Mechanisms, for a description of the
   1.319 +//    Whole Script confusable data
   1.320 +//
   1.321 +//  The data provides mappings from code points to a set of scripts
   1.322 +//    that contain characters that might be confused with the code point.
   1.323 +//  There are two mappings, one for lower case only, and one for characters
   1.324 +//    of any case.
   1.325 +//
   1.326 +//  The actual data consists of a utrie2 to map from a code point to an offset,
   1.327 +//  and an array of UScriptSets (essentially bit maps) that is indexed
   1.328 +//  by the offsets obtained from the Trie.
   1.329 +//
   1.330 +//
   1.331 +
   1.332 +
   1.333 +U_NAMESPACE_END
   1.334 +#endif /* __cplusplus */
   1.335 +
   1.336 +/**
   1.337 +  * Endianness swap function for binary spoof data.
   1.338 +  * @internal
   1.339 +  */
   1.340 +U_CAPI int32_t U_EXPORT2
   1.341 +uspoof_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData,
   1.342 +            UErrorCode *status);
   1.343 +
   1.344 +
   1.345 +#endif
   1.346 +
   1.347 +#endif  /* USPOOFIM_H */
   1.348 +

mercurial