Home | History | Annotate | Download | only in i18n 
      1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
      2 // License & terms of use: http://www.unicode.org/copyright.html
      3 /*
      4 ******************************************************************************
      5 *
      6 *   Copyright (C) 2008-2016, International Business Machines
      7 *   Corporation and others.  All Rights Reserved.
      8 *
      9 ******************************************************************************
     10 *   file name:  uspoof_conf.h
     11 *   encoding:   US-ASCII
     12 *   tab size:   8 (not used)
     13 *   indentation:4
     14 *
     15 *   created on: 2009Jan05
     16 *   created by: Andy Heninger
     17 *
     18 *   Internal classes for compiling confusable data into its binary (runtime) form.
     19 */
     20 
     21 #ifndef __USPOOF_BUILDCONF_H__
     22 #define __USPOOF_BUILDCONF_H__
     23 
     24 #if !UCONFIG_NO_NORMALIZATION
     25 
     26 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
     27 
     28 #include "unicode/uregex.h"
     29 #include "uhash.h"
     30 #include "uspoof_impl.h"
     31 
     32 U_NAMESPACE_BEGIN
     33 
     34 // SPUString
     35 //              Holds a string that is the result of one of the mappings defined
     36 //              by the confusable mapping data (confusables.txt from Unicode.org)
     37 //              Instances of SPUString exist during the compilation process only.
     38 
     39 struct SPUString : public UMemory {
     40     UnicodeString  *fStr;             // The actual string.
     41     int32_t         fCharOrStrTableIndex;   // Index into the final runtime data for this
     42                                       // string (or, for length 1, the single string char
     43                                       // itself, there being no string table entry for it.)
     44     SPUString(UnicodeString *s);
     45     ~SPUString();
     46 };
     47 
     48 
     49 //  String Pool   A utility class for holding the strings that are the result of
     50 //                the spoof mappings.  These strings will utimately end up in the
     51 //                run-time String Table.
     52 //                This is sort of like a sorted set of strings, except that ICU's anemic
     53 //                built-in collections don't support those, so it is implemented with a
     54 //                combination of a uhash and a UVector.
     55 
     56 
     57 class SPUStringPool : public UMemory {
     58   public:
     59     SPUStringPool(UErrorCode &status);
     60     ~SPUStringPool();
     61 
     62     // Add a string. Return the string from the table.
     63     // If the input parameter string is already in the table, delete the
     64     //  input parameter and return the existing string.
     65     SPUString *addString(UnicodeString *src, UErrorCode &status);
     66 
     67 
     68     // Get the n-th string in the collection.
     69     SPUString *getByIndex(int32_t i);
     70 
     71     // Sort the contents; affects the ordering of getByIndex().
     72     void sort(UErrorCode &status);
     73 
     74     int32_t size();
     75 
     76   private:
     77     UVector     *fVec;    // Elements are SPUString *
     78     UHashtable  *fHash;   // Key: UnicodeString  Value: SPUString
     79 };
     80 
     81 
     82 // class ConfusabledataBuilder
     83 //     An instance of this class exists while the confusable data is being built from source.
     84 //     It encapsulates the intermediate data structures that are used for building.
     85 //     It exports one static function, to do a confusable data build.
     86 
     87 class ConfusabledataBuilder : public UMemory {
     88   private:
     89     SpoofImpl  *fSpoofImpl;
     90     UChar      *fInput;
     91     UHashtable *fTable;
     92     UnicodeSet *fKeySet;     // A set of all keys (UChar32s) that go into the four mapping tables.
     93 
     94     // The binary data is first assembled into the following four collections, then
     95     //   copied to its final raw-memory destination.
     96     UVector            *fKeyVec;
     97     UVector            *fValueVec;
     98     UnicodeString      *fStringTable;
     99 
    100     SPUStringPool      *stringPool;
    101     URegularExpression *fParseLine;
    102     URegularExpression *fParseHexNum;
    103     int32_t             fLineNum;
    104 
    105     ConfusabledataBuilder(SpoofImpl *spImpl, UErrorCode &status);
    106     ~ConfusabledataBuilder();
    107     void build(const char * confusables, int32_t confusablesLen, UErrorCode &status);
    108 
    109     // Add an entry to the key and value tables being built
    110     //   input:  data from SLTable, MATable, etc.
    111     //   outut:  entry added to fKeyVec and fValueVec
    112     void addKeyEntry(UChar32     keyChar,     // The key character
    113                      UHashtable *table,       // The table, one of SATable, MATable, etc.
    114                      int32_t     tableFlag,   // One of USPOOF_SA_TABLE_FLAG, etc.
    115                      UErrorCode &status);
    116 
    117     // From an index into fKeyVec & fValueVec
    118     //   get a UnicodeString with the corresponding mapping.
    119     UnicodeString getMapping(int32_t index);
    120 
    121     // Populate the final binary output data array with the compiled data.
    122     void outputData(UErrorCode &status);
    123 
    124   public:
    125     static void buildConfusableData(SpoofImpl *spImpl, const char * confusables,
    126         int32_t confusablesLen, int32_t *errorType, UParseError *pe, UErrorCode &status);
    127 };
    128 U_NAMESPACE_END
    129 
    130 #endif
    131 #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS
    132 #endif  // __USPOOF_BUILDCONF_H__
    133