scintilla/lexlib/CharacterSet.h

   1 // Scintilla source code edit control
   2 /** @file CharacterSet.h
   3  ** Encapsulates a set of characters. Used to test if a character is within a set.
   4  **/
   5 // Copyright 2007 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #ifndef CHARACTERSET_H
   9 #define CHARACTERSET_H
  10
  11 #ifdef SCI_NAMESPACE
  12 namespace Scintilla {
  13 #endif
  14
  15 class CharacterSet {
  16         int size;
  17         bool valueAfter;
  18         bool *bset;
  19 public:
  20         enum setBase {
  21                 setNone=0,
  22                 setLower=1,
  23                 setUpper=2,
  24                 setDigits=4,
  25                 setAlpha=setLower|setUpper,
  26                 setAlphaNum=setAlpha|setDigits
  27         };
  28         CharacterSet(setBase base=setNone, const char *initialSet="", int size_=0x80, bool valueAfter_=false) {
  29                 size = size_;
  30                 valueAfter = valueAfter_;
  31                 bset = new bool[size];
  32                 for (int i=0; i < size; i++) {
  33                         bset[i] = false;
  34                 }
  35                 AddString(initialSet);
  36                 if (base & setLower)
  37                         AddString("abcdefghijklmnopqrstuvwxyz");
  38                 if (base & setUpper)
  39                         AddString("ABCDEFGHIJKLMNOPQRSTUVWXYZ");
  40                 if (base & setDigits)
  41                         AddString("0123456789");
  42         }
  43         CharacterSet(const CharacterSet &other) {
  44                 size = other.size;
  45                 valueAfter = other.valueAfter;
  46                 bset = new bool[size];
  47                 for (int i=0; i < size; i++) {
  48                         bset[i] = other.bset[i];
  49                 }
  50         }
  51         ~CharacterSet() {
  52                 delete []bset;
  53                 bset = 0;
  54                 size = 0;
  55         }
  56         CharacterSet &operator=(const CharacterSet &other) {
  57                 if (this != &other) {
  58                         bool *bsetNew = new bool[other.size];
  59                         for (int i=0; i < other.size; i++) {
  60                                 bsetNew[i] = other.bset[i];
  61                         }
  62                         delete []bset;
  63                         size = other.size;
  64                         valueAfter = other.valueAfter;
  65                         bset = bsetNew;
  66                 }
  67                 return *this;
  68         }
  69         void Add(int val) {
  70                 assert(val >= 0);
  71                 assert(val < size);
  72                 bset[val] = true;
  73         }
  74         void AddString(const char *setToAdd) {
  75                 for (const char *cp=setToAdd; *cp; cp++) {
  76                         int val = static_cast<unsigned char>(*cp);
  77                         assert(val >= 0);
  78                         assert(val < size);
  79                         bset[val] = true;
  80                 }
  81         }
  82         bool Contains(int val) const {
  83                 assert(val >= 0);
  84                 if (val < 0) return false;
  85                 return (val < size) ? bset[val] : valueAfter;
  86         }
  87 };
  88
  89 // Functions for classifying characters
  90
  91 inline bool IsASpace(int ch) {
  92     return (ch == ' ') || ((ch >= 0x09) && (ch <= 0x0d));
  93 }
  94
  95 inline bool IsASpaceOrTab(int ch) {
  96         return (ch == ' ') || (ch == '\t');
  97 }
  98
  99 inline bool IsADigit(int ch) {
 100         return (ch >= '0') && (ch <= '9');
 101 }
 102
 103 inline bool IsADigit(int ch, int base) {
 104         if (base <= 10) {
 105                 return (ch >= '0') && (ch < '0' + base);
 106         } else {
 107                 return ((ch >= '0') && (ch <= '9')) ||
 108                        ((ch >= 'A') && (ch < 'A' + base - 10)) ||
 109                        ((ch >= 'a') && (ch < 'a' + base - 10));
 110         }
 111 }
 112
 113 inline bool IsASCII(int ch) {
 114         return (ch >= 0) && (ch < 0x80);
 115 }
 116
 117 inline bool IsLowerCase(int ch) {
 118         return (ch >= 'a') && (ch <= 'z');
 119 }
 120
 121 inline bool IsUpperCase(int ch) {
 122         return (ch >= 'A') && (ch <= 'Z');
 123 }
 124
 125 inline bool IsAlphaNumeric(int ch) {
 126         return
 127                 ((ch >= '0') && (ch <= '9')) ||
 128                 ((ch >= 'a') && (ch <= 'z')) ||
 129                 ((ch >= 'A') && (ch <= 'Z'));
 130 }
 131
 132 /**
 133  * Check if a character is a space.
 134  * This is ASCII specific but is safe with chars >= 0x80.
 135  */
 136 inline bool isspacechar(int ch) {
 137     return (ch == ' ') || ((ch >= 0x09) && (ch <= 0x0d));
 138 }
 139
 140 inline bool iswordchar(int ch) {
 141         return IsAlphaNumeric(ch) || ch == '.' || ch == '_';
 142 }
 143
 144 inline bool iswordstart(int ch) {
 145         return IsAlphaNumeric(ch) || ch == '_';
 146 }
 147
 148 inline bool isoperator(int ch) {
 149         if (IsAlphaNumeric(ch))
 150                 return false;
 151         if (ch == '%' || ch == '^' || ch == '&' || ch == '*' ||
 152                 ch == '(' || ch == ')' || ch == '-' || ch == '+' ||
 153                 ch == '=' || ch == '|' || ch == '{' || ch == '}' ||
 154                 ch == '[' || ch == ']' || ch == ':' || ch == ';' ||
 155                 ch == '<' || ch == '>' || ch == ',' || ch == '/' ||
 156                 ch == '?' || ch == '!' || ch == '.' || ch == '~')
 157                 return true;
 158         return false;
 159 }
 160
 161 // Simple case functions for ASCII.
 162
 163 inline char MakeUpperCase(char ch) {
 164         if (ch < 'a' || ch > 'z')
 165                 return ch;
 166         else
 167                 return static_cast<char>(ch - 'a' + 'A');
 168 }
 169
 170 int CompareCaseInsensitive(const char *a, const char *b);
 171 int CompareNCaseInsensitive(const char *a, const char *b, size_t len);
 172
 173 #ifdef SCI_NAMESPACE
 174 }
 175 #endif
 176
 177 #endif