Home | History | Annotate | Download | only in i18n 
      1 //  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:   UTF-8
     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 #include "unicode/utypes.h"
     25 
     26 #if !UCONFIG_NO_NORMALIZATION
     27 
     28 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
     29 
     30 #include "unicode/uregex.h"
     31 #include "uhash.h"
     32 #include "uspoof_impl.h"
     33 
     34 U_NAMESPACE_BEGIN
     35 
     36 // SPUString
     37 //              Holds a string that is the result of one of the mappings defined
     38 //              by the confusable mapping data (confusables.txt from Unicode.org)
     39 //              Instances of SPUString exist during the compilation process only.
     40 
     41 struct SPUString : public UMemory {
     42     UnicodeString  *fStr;             // The actual string.
     43     int32_t         fCharOrStrTableIndex;   // Index into the final runtime data for this
     44                                       // string (or, for length 1, the single string char
     45                                       // itself, there being no string table entry for it.)
     46     SPUString(UnicodeString *s);
     47     ~SPUString();
     48 };
     49 
     50 
     51 //  String Pool   A utility class for holding the strings that are the result of
     52 //                the spoof mappings.  These strings will utimately end up in the
     53 //                run-time String Table.
     54 //                This is sort of like a sorted set of strings, except that ICU's anemic
     55 //                built-in collections don't support those, so it is implemented with a
     56 //                combination of a uhash and a UVector.
     57 
     58 
     59 class SPUStringPool : public UMemory {
     60   public:
     61     SPUStringPool(UErrorCode &status);
     62     ~SPUStringPool();
     63 
     64     // Add a string. Return the string from the table.
     65     // If the input parameter string is already in the table, delete the
     66     //  input parameter and return the existing string.
     67     SPUString *addString(UnicodeString *src, UErrorCode &status);
     68 
     69 
     70     // Get the n-th string in the collection.
     71     SPUString *getByIndex(int32_t i);
     72 
     73     // Sort the contents; affects the ordering of getByIndex().
     74     void sort(UErrorCode &status);
     75 
     76     int32_t size();
     77 
     78   private:
     79     UVector     *fVec;    // Elements are SPUString *
     80     UHashtable  *fHash;   // Key: UnicodeString  Value: SPUString
     81 };
     82 
     83 
     84 // class ConfusabledataBuilder
     85 //     An instance of this class exists while the confusable data is being built from source.
     86 //     It encapsulates the intermediate data structures that are used for building.
     87 //     It exports one static function, to do a confusable data build.
     88 
     89 class ConfusabledataBuilder : public UMemory {
     90   private:
     91     SpoofImpl  *fSpoofImpl;
     92     UChar      *fInput;
     93     UHashtable *fTable;
     94     UnicodeSet *fKeySet;     // A set of all keys (UChar32s) that go into the four mapping tables.
     95 
     96     // The binary data is first assembled into the following four collections, then
     97     //   copied to its final raw-memory destination.
     98     UVector            *fKeyVec;
     99     UVector            *fValueVec;
    100     UnicodeString      *fStringTable;
    101 
    102     SPUStringPool      *stringPool;
    103     URegularExpression *fParseLine;
    104     URegularExpression *fParseHexNum;
    105     int32_t             fLineNum;
    106 
    107     ConfusabledataBuilder(SpoofImpl *spImpl, UErrorCode &status);
    108     ~ConfusabledataBuilder();
    109     void build(const char * confusables, int32_t confusablesLen, UErrorCode &status);
    110 
    111     // Add an entry to the key and value tables being built
    112     //   input:  data from SLTable, MATable, etc.
    113     //   outut:  entry added to fKeyVec and fValueVec
    114     void addKeyEntry(UChar32     keyChar,     // The key character
    115                      UHashtable *table,       // The table, one of SATable, MATable, etc.
    116                      int32_t     tableFlag,   // One of USPOOF_SA_TABLE_FLAG, etc.
    117                      UErrorCode &status);
    118 
    119     // From an index into fKeyVec & fValueVec
    120     //   get a UnicodeString with the corresponding mapping.
    121     UnicodeString getMapping(int32_t index);
    122 
    123     // Populate the final binary output data array with the compiled data.
    124     void outputData(UErrorCode &status);
    125 
    126   public:
    127     static void buildConfusableData(SpoofImpl *spImpl, const char * confusables,
    128         int32_t confusablesLen, int32_t *errorType, UParseError *pe, UErrorCode &status);
    129 };
    130 U_NAMESPACE_END
    131 
    132 #endif
    133 #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS
    134 #endif  // __USPOOF_BUILDCONF_H__
    135