| 1 | // Scintilla source code edit control |
| 2 | // Encoding: UTF-8 |
| 3 | /** @file LexMMIXAL.cxx |
| 4 | ** Lexer for MMIX Assembler Language. |
| 5 | ** Written by Christoph Hösler <christoph.hoesler@student.uni-tuebingen.de> |
| 6 | ** For information about MMIX visit http://www-cs-faculty.stanford.edu/~knuth/mmix.html |
| 7 | **/ |
| 8 | // Copyright 1998-2003 by Neil Hodgson <neilh@scintilla.org> |
| 9 | // The License.txt file describes the conditions under which this software may be distributed. |
| 10 | |
| 11 | #include <stdlib.h> |
| 12 | #include <string.h> |
| 13 | #include <stdio.h> |
| 14 | #include <stdarg.h> |
| 15 | #include <assert.h> |
| 16 | #include <ctype.h> |
| 17 | |
| 18 | #include <string> |
| 19 | #include <string_view> |
| 20 | |
| 21 | #include "ILexer.h" |
| 22 | #include "Scintilla.h" |
| 23 | #include "SciLexer.h" |
| 24 | |
| 25 | #include "WordList.h" |
| 26 | #include "LexAccessor.h" |
| 27 | #include "Accessor.h" |
| 28 | #include "StyleContext.h" |
| 29 | #include "CharacterSet.h" |
| 30 | #include "LexerModule.h" |
| 31 | |
| 32 | using namespace Lexilla; |
| 33 | |
| 34 | |
| 35 | static inline bool IsAWordChar(const int ch) { |
| 36 | return (ch < 0x80) && (isalnum(ch) || ch == ':' || ch == '_'); |
| 37 | } |
| 38 | |
| 39 | static inline bool isMMIXALOperator(char ch) { |
| 40 | if (IsASCII(ch) && isalnum(ch)) |
| 41 | return false; |
| 42 | if (ch == '+' || ch == '-' || ch == '|' || ch == '^' || |
| 43 | ch == '*' || ch == '/' || |
| 44 | ch == '%' || ch == '<' || ch == '>' || ch == '&' || |
| 45 | ch == '~' || ch == '$' || |
| 46 | ch == ',' || ch == '(' || ch == ')' || |
| 47 | ch == '[' || ch == ']') |
| 48 | return true; |
| 49 | return false; |
| 50 | } |
| 51 | |
| 52 | static void ColouriseMMIXALDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[], |
| 53 | Accessor &styler) { |
| 54 | |
| 55 | WordList &opcodes = *keywordlists[0]; |
| 56 | WordList &special_register = *keywordlists[1]; |
| 57 | WordList &predef_symbols = *keywordlists[2]; |
| 58 | |
| 59 | StyleContext sc(startPos, length, initStyle, styler); |
| 60 | |
| 61 | for (; sc.More(); sc.Forward()) |
| 62 | { |
| 63 | // No EOL continuation |
| 64 | if (sc.atLineStart) { |
| 65 | if (sc.ch == '@' && sc.chNext == 'i') { |
| 66 | sc.SetState(SCE_MMIXAL_INCLUDE); |
| 67 | } else { |
| 68 | sc.SetState(SCE_MMIXAL_LEADWS); |
| 69 | } |
| 70 | } |
| 71 | |
| 72 | // Check if first non whitespace character in line is alphanumeric |
| 73 | if (sc.state == SCE_MMIXAL_LEADWS && !isspace(sc.ch)) { // LEADWS |
| 74 | if(!IsAWordChar(sc.ch)) { |
| 75 | sc.SetState(SCE_MMIXAL_COMMENT); |
| 76 | } else { |
| 77 | if(sc.atLineStart) { |
| 78 | sc.SetState(SCE_MMIXAL_LABEL); |
| 79 | } else { |
| 80 | sc.SetState(SCE_MMIXAL_OPCODE_PRE); |
| 81 | } |
| 82 | } |
| 83 | } |
| 84 | |
| 85 | // Determine if the current state should terminate. |
| 86 | if (sc.state == SCE_MMIXAL_OPERATOR) { // OPERATOR |
| 87 | sc.SetState(SCE_MMIXAL_OPERANDS); |
| 88 | } else if (sc.state == SCE_MMIXAL_NUMBER) { // NUMBER |
| 89 | if (!isdigit(sc.ch)) { |
| 90 | if (IsAWordChar(sc.ch)) { |
| 91 | sc.ChangeState(SCE_MMIXAL_REF); |
| 92 | sc.SetState(SCE_MMIXAL_REF); |
| 93 | } else { |
| 94 | sc.SetState(SCE_MMIXAL_OPERANDS); |
| 95 | } |
| 96 | } |
| 97 | } else if (sc.state == SCE_MMIXAL_LABEL) { // LABEL |
| 98 | if (!IsAWordChar(sc.ch) ) { |
| 99 | sc.SetState(SCE_MMIXAL_OPCODE_PRE); |
| 100 | } |
| 101 | } else if (sc.state == SCE_MMIXAL_REF) { // REF |
| 102 | if (!IsAWordChar(sc.ch) ) { |
| 103 | char s0[100]; |
| 104 | sc.GetCurrent(s0, sizeof(s0)); |
| 105 | const char *s = s0; |
| 106 | if (*s == ':') { // ignore base prefix for match |
| 107 | ++s; |
| 108 | } |
| 109 | if (special_register.InList(s)) { |
| 110 | sc.ChangeState(SCE_MMIXAL_REGISTER); |
| 111 | } else if (predef_symbols.InList(s)) { |
| 112 | sc.ChangeState(SCE_MMIXAL_SYMBOL); |
| 113 | } |
| 114 | sc.SetState(SCE_MMIXAL_OPERANDS); |
| 115 | } |
| 116 | } else if (sc.state == SCE_MMIXAL_OPCODE_PRE) { // OPCODE_PRE |
| 117 | if (!isspace(sc.ch)) { |
| 118 | sc.SetState(SCE_MMIXAL_OPCODE); |
| 119 | } |
| 120 | } else if (sc.state == SCE_MMIXAL_OPCODE) { // OPCODE |
| 121 | if (!IsAWordChar(sc.ch) ) { |
| 122 | char s[100]; |
| 123 | sc.GetCurrent(s, sizeof(s)); |
| 124 | if (opcodes.InList(s)) { |
| 125 | sc.ChangeState(SCE_MMIXAL_OPCODE_VALID); |
| 126 | } else { |
| 127 | sc.ChangeState(SCE_MMIXAL_OPCODE_UNKNOWN); |
| 128 | } |
| 129 | sc.SetState(SCE_MMIXAL_OPCODE_POST); |
| 130 | } |
| 131 | } else if (sc.state == SCE_MMIXAL_STRING) { // STRING |
| 132 | if (sc.ch == '\"') { |
| 133 | sc.ForwardSetState(SCE_MMIXAL_OPERANDS); |
| 134 | } else if (sc.atLineEnd) { |
| 135 | sc.ForwardSetState(SCE_MMIXAL_OPERANDS); |
| 136 | } |
| 137 | } else if (sc.state == SCE_MMIXAL_CHAR) { // CHAR |
| 138 | if (sc.ch == '\'') { |
| 139 | sc.ForwardSetState(SCE_MMIXAL_OPERANDS); |
| 140 | } else if (sc.atLineEnd) { |
| 141 | sc.ForwardSetState(SCE_MMIXAL_OPERANDS); |
| 142 | } |
| 143 | } else if (sc.state == SCE_MMIXAL_REGISTER) { // REGISTER |
| 144 | if (!isdigit(sc.ch)) { |
| 145 | sc.SetState(SCE_MMIXAL_OPERANDS); |
| 146 | } |
| 147 | } else if (sc.state == SCE_MMIXAL_HEX) { // HEX |
| 148 | if (!isxdigit(sc.ch)) { |
| 149 | sc.SetState(SCE_MMIXAL_OPERANDS); |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | // Determine if a new state should be entered. |
| 154 | if (sc.state == SCE_MMIXAL_OPCODE_POST || // OPCODE_POST |
| 155 | sc.state == SCE_MMIXAL_OPERANDS) { // OPERANDS |
| 156 | if (sc.state == SCE_MMIXAL_OPERANDS && isspace(sc.ch)) { |
| 157 | sc.SetState(SCE_MMIXAL_COMMENT); |
| 158 | } else if (isdigit(sc.ch)) { |
| 159 | sc.SetState(SCE_MMIXAL_NUMBER); |
| 160 | } else if (IsAWordChar(sc.ch) || sc.Match('@')) { |
| 161 | sc.SetState(SCE_MMIXAL_REF); |
| 162 | } else if (sc.Match('\"')) { |
| 163 | sc.SetState(SCE_MMIXAL_STRING); |
| 164 | } else if (sc.Match('\'')) { |
| 165 | sc.SetState(SCE_MMIXAL_CHAR); |
| 166 | } else if (sc.Match('$')) { |
| 167 | sc.SetState(SCE_MMIXAL_REGISTER); |
| 168 | } else if (sc.Match('#')) { |
| 169 | sc.SetState(SCE_MMIXAL_HEX); |
| 170 | } else if (isMMIXALOperator(static_cast<char>(sc.ch))) { |
| 171 | sc.SetState(SCE_MMIXAL_OPERATOR); |
| 172 | } |
| 173 | } |
| 174 | } |
| 175 | sc.Complete(); |
| 176 | } |
| 177 | |
| 178 | static const char * const MMIXALWordListDesc[] = { |
| 179 | "Operation Codes" , |
| 180 | "Special Register" , |
| 181 | "Predefined Symbols" , |
| 182 | 0 |
| 183 | }; |
| 184 | |
| 185 | LexerModule lmMMIXAL(SCLEX_MMIXAL, ColouriseMMIXALDoc, "mmixal" , 0, MMIXALWordListDesc); |
| 186 | |
| 187 | |