scintilla/LexAsm.cxx

   1 // Scintilla source code edit control
   2 /** @file LexAsm.cxx
   3  ** Lexer for Assembler, just for the MASM syntax
   4  ** Written by The Black Horus
   5  ** Enhancements and NASM stuff by Kein-Hong Man, 2003-10
   6  ** SCE_ASM_COMMENTBLOCK and SCE_ASM_CHARACTER are for future GNU as colouring
   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 <ctype.h>
  14 #include <stdio.h>
  15 #include <stdarg.h>
  16
  17 #include "Platform.h"
  18
  19 #include "PropSet.h"
  20 #include "Accessor.h"
  21 #include "StyleContext.h"
  22 #include "KeyWords.h"
  23 #include "Scintilla.h"
  24 #include "SciLexer.h"
  25
  26 #ifdef SCI_NAMESPACE
  27 using namespace Scintilla;
  28 #endif
  29
  30 static inline bool IsAWordChar(const int ch) {
  31         return (ch < 0x80) && (isalnum(ch) || ch == '.' ||
  32                 ch == '_' || ch == '?');
  33 }
  34
  35 static inline bool IsAWordStart(const int ch) {
  36         return (ch < 0x80) && (isalnum(ch) || ch == '_' || ch == '.' ||
  37                 ch == '%' || ch == '@' || ch == '$' || ch == '?');
  38 }
  39
  40 static inline bool IsAsmOperator(const int ch) {
  41         if ((ch < 0x80) && (isalnum(ch)))
  42                 return false;
  43         // '.' left out as it is used to make up numbers
  44         if (ch == '*' || ch == '/' || ch == '-' || ch == '+' ||
  45                 ch == '(' || ch == ')' || ch == '=' || ch == '^' ||
  46                 ch == '[' || ch == ']' || ch == '<' || ch == '&' ||
  47                 ch == '>' || ch == ',' || ch == '|' || ch == '~' ||
  48                 ch == '%' || ch == ':')
  49                 return true;
  50         return false;
  51 }
  52
  53 static void ColouriseAsmDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
  54                             Accessor &styler) {
  55
  56         WordList &cpuInstruction = *keywordlists[0];
  57         WordList &mathInstruction = *keywordlists[1];
  58         WordList &registers = *keywordlists[2];
  59         WordList &directive = *keywordlists[3];
  60         WordList &directiveOperand = *keywordlists[4];
  61         WordList &extInstruction = *keywordlists[5];
  62
  63         // Do not leak onto next line
  64         if (initStyle == SCE_ASM_STRINGEOL)
  65                 initStyle = SCE_ASM_DEFAULT;
  66
  67         StyleContext sc(startPos, length, initStyle, styler);
  68
  69         for (; sc.More(); sc.Forward())
  70         {
  71
  72                 // Prevent SCE_ASM_STRINGEOL from leaking back to previous line
  73                 if (sc.atLineStart && (sc.state == SCE_ASM_STRING)) {
  74                         sc.SetState(SCE_ASM_STRING);
  75                 } else if (sc.atLineStart && (sc.state == SCE_ASM_CHARACTER)) {
  76                         sc.SetState(SCE_ASM_CHARACTER);
  77                 }
  78
  79                 // Handle line continuation generically.
  80                 if (sc.ch == '\\') {
  81                         if (sc.chNext == '\n' || sc.chNext == '\r') {
  82                                 sc.Forward();
  83                                 if (sc.ch == '\r' && sc.chNext == '\n') {
  84                                         sc.Forward();
  85                                 }
  86                                 continue;
  87                         }
  88                 }
  89
  90                 // Determine if the current state should terminate.
  91                 if (sc.state == SCE_ASM_OPERATOR) {
  92                         if (!IsAsmOperator(sc.ch)) {
  93                             sc.SetState(SCE_ASM_DEFAULT);
  94                         }
  95                 }else if (sc.state == SCE_ASM_NUMBER) {
  96                         if (!IsAWordChar(sc.ch)) {
  97                                 sc.SetState(SCE_ASM_DEFAULT);
  98                         }
  99                 } else if (sc.state == SCE_ASM_IDENTIFIER) {
 100                         if (!IsAWordChar(sc.ch) ) {
 101                                 char s[100];
 102                                 sc.GetCurrentLowered(s, sizeof(s));
 103
 104                                 if (cpuInstruction.InList(s)) {
 105                                         sc.ChangeState(SCE_ASM_CPUINSTRUCTION);
 106                                 } else if (mathInstruction.InList(s)) {
 107                                         sc.ChangeState(SCE_ASM_MATHINSTRUCTION);
 108                                 } else if (registers.InList(s)) {
 109                                         sc.ChangeState(SCE_ASM_REGISTER);
 110                                 }  else if (directive.InList(s)) {
 111                                         sc.ChangeState(SCE_ASM_DIRECTIVE);
 112                                 } else if (directiveOperand.InList(s)) {
 113                                         sc.ChangeState(SCE_ASM_DIRECTIVEOPERAND);
 114                                 } else if (extInstruction.InList(s)) {
 115                                         sc.ChangeState(SCE_ASM_EXTINSTRUCTION);
 116                                 }
 117                                 sc.SetState(SCE_ASM_DEFAULT);
 118                         }
 119                 }
 120                 else if (sc.state == SCE_ASM_COMMENT ) {
 121                         if (sc.atLineEnd) {
 122                                 sc.SetState(SCE_ASM_DEFAULT);
 123                         }
 124                 } else if (sc.state == SCE_ASM_STRING) {
 125                         if (sc.ch == '\\') {
 126                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 127                                         sc.Forward();
 128                                 }
 129                         } else if (sc.ch == '\"') {
 130                                 sc.ForwardSetState(SCE_ASM_DEFAULT);
 131                         } else if (sc.atLineEnd) {
 132                                 sc.ChangeState(SCE_ASM_STRINGEOL);
 133                                 sc.ForwardSetState(SCE_ASM_DEFAULT);
 134                         }
 135                 } else if (sc.state == SCE_ASM_CHARACTER) {
 136                         if (sc.ch == '\\') {
 137                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 138                                         sc.Forward();
 139                                 }
 140                         } else if (sc.ch == '\'') {
 141                                 sc.ForwardSetState(SCE_ASM_DEFAULT);
 142                         } else if (sc.atLineEnd) {
 143                                 sc.ChangeState(SCE_ASM_STRINGEOL);
 144                                 sc.ForwardSetState(SCE_ASM_DEFAULT);
 145                         }
 146                 }
 147
 148                 // Determine if a new state should be entered.
 149                 if (sc.state == SCE_ASM_DEFAULT) {
 150                         if (sc.ch == ';'){
 151                                 sc.SetState(SCE_ASM_COMMENT);
 152                         } else if (isascii(sc.ch) && (isdigit(sc.ch) || (sc.ch == '.' && isascii(sc.chNext) && isdigit(sc.chNext)))) {
 153                                 sc.SetState(SCE_ASM_NUMBER);
 154                         } else if (IsAWordStart(sc.ch)) {
 155                                 sc.SetState(SCE_ASM_IDENTIFIER);
 156                         } else if (sc.ch == '\"') {
 157                                 sc.SetState(SCE_ASM_STRING);
 158                         } else if (sc.ch == '\'') {
 159                                 sc.SetState(SCE_ASM_CHARACTER);
 160                         } else if (IsAsmOperator(sc.ch)) {
 161                                 sc.SetState(SCE_ASM_OPERATOR);
 162                         }
 163                 }
 164
 165         }
 166         sc.Complete();
 167 }
 168
 169 static const char * const asmWordListDesc[] = {
 170         "CPU instructions",
 171         "FPU instructions",
 172         "Registers",
 173         "Directives",
 174         "Directive operands",
 175         "Extended instructions",
 176         0
 177 };
 178
 179 LexerModule lmAsm(SCLEX_ASM, ColouriseAsmDoc, "asm", 0, asmWordListDesc);
 180