Home | History | Annotate | Download | only in i18n 
      1 /*
      2 ******************************************************************************
      3 *
      4 *   Copyright (C) 2008-2009, International Business Machines
      5 *   Corporation and others.  All Rights Reserved.
      6 *
      7 ******************************************************************************
      8 *   file name:  uspoof_conf.h
      9 *   encoding:   US-ASCII
     10 *   tab size:   8 (not used)
     11 *   indentation:4
     12 *
     13 *   created on: 2009Jan05
     14 *   created by: Andy Heninger
     15 *
     16 *   Internal classes for compiling confusable data into its binary (runtime) form.
     17 */
     18 
     19 #ifndef __USPOOF_BUILDCONF_H__
     20 #define __USPOOF_BUILDCONF_H__
     21 
     22 #if !UCONFIG_NO_NORMALIZATION
     23 
     24 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
     25 
     26 #include "uspoof_impl.h"
     27 
     28 U_NAMESPACE_BEGIN
     29 
     30 // SPUString
     31 //              Holds a string that is the result of one of the mappings defined
     32 //              by the confusable mapping data (confusables.txt from Unicode.org)
     33 //              Instances of SPUString exist during the compilation process only.
     34 
     35 struct SPUString : public UMemory {
     36     UnicodeString  *fStr;             // The actual string.
     37     int32_t         fStrTableIndex;   // Index into the final runtime data for this string.
     38                                       //  (or, for length 1, the single string char itself,
     39                                       //   there being no string table entry for it.)
     40     SPUString(UnicodeString *s);
     41     ~SPUString();
     42 };
     43 
     44 
     45 //  String Pool   A utility class for holding the strings that are the result of
     46 //                the spoof mappings.  These strings will utimately end up in the
     47 //                run-time String Table.
     48 //                This is sort of like a sorted set of strings, except that ICU's anemic
     49 //                built-in collections don't support those, so it is implemented with a
     50 //                combination of a uhash and a UVector.
     51 
     52 
     53 class SPUStringPool : public UMemory {
     54   public:
     55     SPUStringPool(UErrorCode &status);
     56     ~SPUStringPool();
     57 
     58     // Add a string. Return the string from the table.
     59     // If the input parameter string is already in the table, delete the
     60     //  input parameter and return the existing string.
     61     SPUString *addString(UnicodeString *src, UErrorCode &status);
     62 
     63 
     64     // Get the n-th string in the collection.
     65     SPUString *getByIndex(int32_t i);
     66 
     67     // Sort the contents; affects the ordering of getByIndex().
     68     void sort(UErrorCode &status);
     69 
     70     int32_t size();
     71 
     72   private:
     73     UVector     *fVec;    // Elements are SPUString *
     74     UHashtable  *fHash;   // Key: UnicodeString  Value: SPUString
     75 };
     76 
     77 
     78 // class ConfusabledataBuilder
     79 //     An instance of this class exists while the confusable data is being built from source.
     80 //     It encapsulates the intermediate data structures that are used for building.
     81 //     It exports one static function, to do a confusable data build.
     82 
     83 class ConfusabledataBuilder : public UMemory {
     84   private:
     85     SpoofImpl  *fSpoofImpl;
     86     UChar      *fInput;
     87     UHashtable *fSLTable;
     88     UHashtable *fSATable;
     89     UHashtable *fMLTable;
     90     UHashtable *fMATable;
     91     UnicodeSet *fKeySet;     // A set of all keys (UChar32s) that go into the four mapping tables.
     92 
     93     // The binary data is first assembled into the following four collections, then
     94     //   copied to its final raw-memory destination.
     95     UVector            *fKeyVec;
     96     UVector            *fValueVec;
     97     UnicodeString      *fStringTable;
     98     UVector            *fStringLengthsTable;
     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 key);
    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