| 1 | // Scintilla source code edit control |
| 2 | /** @file LexSML.cxx |
| 3 | ** Lexer for SML. |
| 4 | **/ |
| 5 | // Copyright 2009 by James Moffatt and Yuzhou Xin |
| 6 | // Modified from LexCaml.cxx by Robert Roessler <robertr@rftp.com> Copyright 2005 |
| 7 | // The License.txt file describes the conditions under which this software may be distributed. |
| 8 | |
| 9 | #include <stdlib.h> |
| 10 | #include <string.h> |
| 11 | #include <stdio.h> |
| 12 | #include <stdarg.h> |
| 13 | #include <assert.h> |
| 14 | #include <ctype.h> |
| 15 | |
| 16 | #include <string> |
| 17 | #include <string_view> |
| 18 | |
| 19 | #include "ILexer.h" |
| 20 | #include "Scintilla.h" |
| 21 | #include "SciLexer.h" |
| 22 | |
| 23 | #include "WordList.h" |
| 24 | #include "LexAccessor.h" |
| 25 | #include "Accessor.h" |
| 26 | #include "StyleContext.h" |
| 27 | #include "CharacterSet.h" |
| 28 | #include "LexerModule.h" |
| 29 | |
| 30 | #if defined(__clang__) |
| 31 | #pragma clang diagnostic ignored "-Wcomma" |
| 32 | #endif |
| 33 | |
| 34 | inline int issml(int c) {return isalnum(c) || c == '_';} |
| 35 | inline int issmlf(int c) {return isalpha(c) || c == '_';} |
| 36 | inline int issmld(int c) {return isdigit(c) || c == '_';} |
| 37 | |
| 38 | |
| 39 | using namespace Lexilla; |
| 40 | |
| 41 | static void ColouriseSMLDoc( |
| 42 | Sci_PositionU startPos, Sci_Position length, |
| 43 | int initStyle, |
| 44 | WordList *keywordlists[], |
| 45 | Accessor &styler) |
| 46 | { |
| 47 | StyleContext sc(startPos, length, initStyle, styler); |
| 48 | int nesting = 0; |
| 49 | if (sc.state < SCE_SML_STRING) |
| 50 | sc.state = SCE_SML_DEFAULT; |
| 51 | if (sc.state >= SCE_SML_COMMENT) |
| 52 | nesting = (sc.state & 0x0f) - SCE_SML_COMMENT; |
| 53 | |
| 54 | Sci_PositionU chToken = 0; |
| 55 | int chBase = 0, chLit = 0; |
| 56 | WordList& keywords = *keywordlists[0]; |
| 57 | WordList& keywords2 = *keywordlists[1]; |
| 58 | WordList& keywords3 = *keywordlists[2]; |
| 59 | const int useMagic = styler.GetPropertyInt("lexer.caml.magic" , 0); |
| 60 | |
| 61 | while (sc.More()) { |
| 62 | int state2 = -1; |
| 63 | Sci_Position chColor = sc.currentPos - 1; |
| 64 | bool advance = true; |
| 65 | |
| 66 | switch (sc.state & 0x0f) { |
| 67 | case SCE_SML_DEFAULT: |
| 68 | chToken = sc.currentPos; |
| 69 | if (issmlf(sc.ch)) |
| 70 | state2 = SCE_SML_IDENTIFIER; |
| 71 | else if (sc.Match('`') && issmlf(sc.chNext)) |
| 72 | state2 = SCE_SML_TAGNAME; |
| 73 | else if (sc.Match('#')&&isdigit(sc.chNext)) |
| 74 | state2 = SCE_SML_LINENUM; |
| 75 | else if (sc.Match('#','\"')){ |
| 76 | state2 = SCE_SML_CHAR,chLit = 0; |
| 77 | sc.Forward(); |
| 78 | |
| 79 | } |
| 80 | else if (isdigit(sc.ch)) { |
| 81 | state2 = SCE_SML_NUMBER, chBase = 10; |
| 82 | if (sc.Match('0') && strchr("xX" , sc.chNext)) |
| 83 | chBase = 16, sc.Forward();} |
| 84 | else if (sc.Match('\"')&&sc.chPrev!='#') |
| 85 | state2 = SCE_SML_STRING; |
| 86 | else if (sc.Match('(', '*')){ |
| 87 | state2 = SCE_SML_COMMENT, |
| 88 | sc.ch = ' ', |
| 89 | sc.Forward();} |
| 90 | else if (strchr("!~" |
| 91 | "=<>@^+-*/" |
| 92 | "()[];,:.#" , sc.ch)) |
| 93 | state2 = SCE_SML_OPERATOR; |
| 94 | break; |
| 95 | |
| 96 | case SCE_SML_IDENTIFIER: |
| 97 | if (!(issml(sc.ch) || sc.Match('\''))) { |
| 98 | const Sci_Position n = sc.currentPos - chToken; |
| 99 | if (n < 24) { |
| 100 | char t[24]; |
| 101 | for (Sci_Position i = -n; i < 0; i++) |
| 102 | t[n + i] = static_cast<char>(sc.GetRelative(i)); |
| 103 | t[n] = '\0'; |
| 104 | if ((n == 1 && sc.chPrev == '_') || keywords.InList(t)) |
| 105 | sc.ChangeState(SCE_SML_KEYWORD); |
| 106 | else if (keywords2.InList(t)) |
| 107 | sc.ChangeState(SCE_SML_KEYWORD2); |
| 108 | else if (keywords3.InList(t)) |
| 109 | sc.ChangeState(SCE_SML_KEYWORD3); |
| 110 | } |
| 111 | state2 = SCE_SML_DEFAULT, advance = false; |
| 112 | } |
| 113 | break; |
| 114 | |
| 115 | case SCE_SML_TAGNAME: |
| 116 | if (!(issml(sc.ch) || sc.Match('\''))) |
| 117 | state2 = SCE_SML_DEFAULT, advance = false; |
| 118 | break; |
| 119 | |
| 120 | case SCE_SML_LINENUM: |
| 121 | if (!isdigit(sc.ch)) |
| 122 | state2 = SCE_SML_DEFAULT, advance = false; |
| 123 | break; |
| 124 | |
| 125 | case SCE_SML_OPERATOR: { |
| 126 | const char* o = 0; |
| 127 | if (issml(sc.ch) || isspace(sc.ch) |
| 128 | || (o = strchr(")]};,\'\"`#" , sc.ch),o) |
| 129 | || !strchr("!$%&*+-./:<=>?@^|~" , sc.ch)) { |
| 130 | if (o && strchr(")]};," , sc.ch)) { |
| 131 | if ((sc.Match(')') && sc.chPrev == '(') |
| 132 | || (sc.Match(']') && sc.chPrev == '[')) |
| 133 | sc.ChangeState(SCE_SML_KEYWORD); |
| 134 | chColor++; |
| 135 | } else |
| 136 | advance = false; |
| 137 | state2 = SCE_SML_DEFAULT; |
| 138 | } |
| 139 | break; |
| 140 | } |
| 141 | |
| 142 | case SCE_SML_NUMBER: |
| 143 | if (issmld(sc.ch) || IsADigit(sc.ch, chBase)) |
| 144 | break; |
| 145 | if ((sc.Match('l') || sc.Match('L') || sc.Match('n')) |
| 146 | && (issmld(sc.chPrev) || IsADigit(sc.chPrev, chBase))) |
| 147 | break; |
| 148 | if (chBase == 10) { |
| 149 | if (sc.Match('.') && issmld(sc.chPrev)) |
| 150 | break; |
| 151 | if ((sc.Match('e') || sc.Match('E')) |
| 152 | && (issmld(sc.chPrev) || sc.chPrev == '.')) |
| 153 | break; |
| 154 | if ((sc.Match('+') || sc.Match('-')) |
| 155 | && (sc.chPrev == 'e' || sc.chPrev == 'E')) |
| 156 | break; |
| 157 | } |
| 158 | state2 = SCE_SML_DEFAULT, advance = false; |
| 159 | break; |
| 160 | |
| 161 | case SCE_SML_CHAR: |
| 162 | if (sc.Match('\\')) { |
| 163 | chLit = 1; |
| 164 | if (sc.chPrev == '\\') |
| 165 | sc.ch = ' '; |
| 166 | } else if ((sc.Match('\"') && sc.chPrev != '\\') || sc.atLineEnd) { |
| 167 | state2 = SCE_SML_DEFAULT; |
| 168 | chLit = 1; |
| 169 | if (sc.Match('\"')) |
| 170 | chColor++; |
| 171 | else |
| 172 | sc.ChangeState(SCE_SML_IDENTIFIER); |
| 173 | } else if (chLit < 1 && sc.currentPos - chToken >= 3) |
| 174 | sc.ChangeState(SCE_SML_IDENTIFIER), advance = false; |
| 175 | break; |
| 176 | |
| 177 | case SCE_SML_STRING: |
| 178 | if (sc.Match('\\') && sc.chPrev == '\\') |
| 179 | sc.ch = ' '; |
| 180 | else if (sc.Match('\"') && sc.chPrev != '\\') |
| 181 | state2 = SCE_SML_DEFAULT, chColor++; |
| 182 | break; |
| 183 | |
| 184 | case SCE_SML_COMMENT: |
| 185 | case SCE_SML_COMMENT1: |
| 186 | case SCE_SML_COMMENT2: |
| 187 | case SCE_SML_COMMENT3: |
| 188 | if (sc.Match('(', '*')) |
| 189 | state2 = sc.state + 1, chToken = sc.currentPos, |
| 190 | sc.ch = ' ', |
| 191 | sc.Forward(), nesting++; |
| 192 | else if (sc.Match(')') && sc.chPrev == '*') { |
| 193 | if (nesting) |
| 194 | state2 = (sc.state & 0x0f) - 1, chToken = 0, nesting--; |
| 195 | else |
| 196 | state2 = SCE_SML_DEFAULT; |
| 197 | chColor++; |
| 198 | } else if (useMagic && sc.currentPos - chToken == 4 |
| 199 | && sc.Match('c') && sc.chPrev == 'r' && sc.GetRelative(-2) == '@') |
| 200 | sc.state |= 0x10; |
| 201 | break; |
| 202 | } |
| 203 | |
| 204 | if (state2 >= 0) |
| 205 | styler.ColourTo(chColor, sc.state), sc.ChangeState(state2); |
| 206 | if (advance) |
| 207 | sc.Forward(); |
| 208 | } |
| 209 | |
| 210 | sc.Complete(); |
| 211 | } |
| 212 | |
| 213 | static void FoldSMLDoc( |
| 214 | Sci_PositionU, Sci_Position, |
| 215 | int, |
| 216 | WordList *[], |
| 217 | Accessor &) |
| 218 | { |
| 219 | } |
| 220 | |
| 221 | static const char * const SMLWordListDesc[] = { |
| 222 | "Keywords" , |
| 223 | "Keywords2" , |
| 224 | "Keywords3" , |
| 225 | 0 |
| 226 | }; |
| 227 | |
| 228 | LexerModule lmSML(SCLEX_SML, ColouriseSMLDoc, "SML" , FoldSMLDoc, SMLWordListDesc); |
| 229 | |
| 230 | |