1 // Scintilla source code edit control
2 /** @file LexMMIXAL.cxx
3 ** Lexer for MMIX Assembler Language.
4 ** Written by Christoph H�sler <christoph.hoesler@student.uni-tuebingen.de>
5 ** For information about MMIX visit http://www-cs-faculty.stanford.edu/~knuth/mmix.html
6 **/
7 // Copyright 1998-2003 by Neil Hodgson <neilh@scintilla.org>
8 // The License.txt file describes the conditions under which this software may be distributed.
9
10 #include <stdlib.h>
11 #include <string.h>
12 #include <stdio.h>
13 #include <stdarg.h>
14 #include <assert.h>
15 #include <ctype.h>
16
17 #include "ILexer.h"
18 #include "Scintilla.h"
19 #include "SciLexer.h"
20
21 #include "WordList.h"
22 #include "LexAccessor.h"
23 #include "Accessor.h"
24 #include "StyleContext.h"
25 #include "CharacterSet.h"
26 #include "LexerModule.h"
27
28 #ifdef SCI_NAMESPACE
29 using namespace Scintilla;
30 #endif
31
32
IsAWordChar(const int ch)33 static inline bool IsAWordChar(const int ch) {
34 return (ch < 0x80) && (isalnum(ch) || ch == ':' || ch == '_');
35 }
36
isMMIXALOperator(char ch)37 inline bool isMMIXALOperator(char ch) {
38 if (IsASCII(ch) && isalnum(ch))
39 return false;
40 if (ch == '+' || ch == '-' || ch == '|' || ch == '^' ||
41 ch == '*' || ch == '/' ||
42 ch == '%' || ch == '<' || ch == '>' || ch == '&' ||
43 ch == '~' || ch == '$' ||
44 ch == ',' || ch == '(' || ch == ')' ||
45 ch == '[' || ch == ']')
46 return true;
47 return false;
48 }
49
ColouriseMMIXALDoc(Sci_PositionU startPos,Sci_Position length,int initStyle,WordList * keywordlists[],Accessor & styler)50 static void ColouriseMMIXALDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[],
51 Accessor &styler) {
52
53 WordList &opcodes = *keywordlists[0];
54 WordList &special_register = *keywordlists[1];
55 WordList &predef_symbols = *keywordlists[2];
56
57 StyleContext sc(startPos, length, initStyle, styler);
58
59 for (; sc.More(); sc.Forward())
60 {
61 // No EOL continuation
62 if (sc.atLineStart) {
63 if (sc.ch == '@' && sc.chNext == 'i') {
64 sc.SetState(SCE_MMIXAL_INCLUDE);
65 } else {
66 sc.SetState(SCE_MMIXAL_LEADWS);
67 }
68 }
69
70 // Check if first non whitespace character in line is alphanumeric
71 if (sc.state == SCE_MMIXAL_LEADWS && !isspace(sc.ch)) { // LEADWS
72 if(!IsAWordChar(sc.ch)) {
73 sc.SetState(SCE_MMIXAL_COMMENT);
74 } else {
75 if(sc.atLineStart) {
76 sc.SetState(SCE_MMIXAL_LABEL);
77 } else {
78 sc.SetState(SCE_MMIXAL_OPCODE_PRE);
79 }
80 }
81 }
82
83 // Determine if the current state should terminate.
84 if (sc.state == SCE_MMIXAL_OPERATOR) { // OPERATOR
85 sc.SetState(SCE_MMIXAL_OPERANDS);
86 } else if (sc.state == SCE_MMIXAL_NUMBER) { // NUMBER
87 if (!isdigit(sc.ch)) {
88 if (IsAWordChar(sc.ch)) {
89 char s[100];
90 sc.GetCurrent(s, sizeof(s));
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 s[100];
104 sc.GetCurrent(s, sizeof(s));
105 if (*s == ':') { // ignore base prefix for match
106 for (size_t i = 0; i != sizeof(s)-1; ++i) {
107 *(s+i) = *(s+i+1);
108 }
109 }
110 if (special_register.InList(s)) {
111 sc.ChangeState(SCE_MMIXAL_REGISTER);
112 } else if (predef_symbols.InList(s)) {
113 sc.ChangeState(SCE_MMIXAL_SYMBOL);
114 }
115 sc.SetState(SCE_MMIXAL_OPERANDS);
116 }
117 } else if (sc.state == SCE_MMIXAL_OPCODE_PRE) { // OPCODE_PRE
118 if (!isspace(sc.ch)) {
119 sc.SetState(SCE_MMIXAL_OPCODE);
120 }
121 } else if (sc.state == SCE_MMIXAL_OPCODE) { // OPCODE
122 if (!IsAWordChar(sc.ch) ) {
123 char s[100];
124 sc.GetCurrent(s, sizeof(s));
125 if (opcodes.InList(s)) {
126 sc.ChangeState(SCE_MMIXAL_OPCODE_VALID);
127 } else {
128 sc.ChangeState(SCE_MMIXAL_OPCODE_UNKNOWN);
129 }
130 sc.SetState(SCE_MMIXAL_OPCODE_POST);
131 }
132 } else if (sc.state == SCE_MMIXAL_STRING) { // STRING
133 if (sc.ch == '\"') {
134 sc.ForwardSetState(SCE_MMIXAL_OPERANDS);
135 } else if (sc.atLineEnd) {
136 sc.ForwardSetState(SCE_MMIXAL_OPERANDS);
137 }
138 } else if (sc.state == SCE_MMIXAL_CHAR) { // CHAR
139 if (sc.ch == '\'') {
140 sc.ForwardSetState(SCE_MMIXAL_OPERANDS);
141 } else if (sc.atLineEnd) {
142 sc.ForwardSetState(SCE_MMIXAL_OPERANDS);
143 }
144 } else if (sc.state == SCE_MMIXAL_REGISTER) { // REGISTER
145 if (!isdigit(sc.ch)) {
146 sc.SetState(SCE_MMIXAL_OPERANDS);
147 }
148 } else if (sc.state == SCE_MMIXAL_HEX) { // HEX
149 if (!isxdigit(sc.ch)) {
150 sc.SetState(SCE_MMIXAL_OPERANDS);
151 }
152 }
153
154 // Determine if a new state should be entered.
155 if (sc.state == SCE_MMIXAL_OPCODE_POST || // OPCODE_POST
156 sc.state == SCE_MMIXAL_OPERANDS) { // OPERANDS
157 if (sc.state == SCE_MMIXAL_OPERANDS && isspace(sc.ch)) {
158 if (!sc.atLineEnd) {
159 sc.SetState(SCE_MMIXAL_COMMENT);
160 }
161 } else if (isdigit(sc.ch)) {
162 sc.SetState(SCE_MMIXAL_NUMBER);
163 } else if (IsAWordChar(sc.ch) || sc.Match('@')) {
164 sc.SetState(SCE_MMIXAL_REF);
165 } else if (sc.Match('\"')) {
166 sc.SetState(SCE_MMIXAL_STRING);
167 } else if (sc.Match('\'')) {
168 sc.SetState(SCE_MMIXAL_CHAR);
169 } else if (sc.Match('$')) {
170 sc.SetState(SCE_MMIXAL_REGISTER);
171 } else if (sc.Match('#')) {
172 sc.SetState(SCE_MMIXAL_HEX);
173 } else if (isMMIXALOperator(static_cast<char>(sc.ch))) {
174 sc.SetState(SCE_MMIXAL_OPERATOR);
175 }
176 }
177 }
178 sc.Complete();
179 }
180
181 static const char * const MMIXALWordListDesc[] = {
182 "Operation Codes",
183 "Special Register",
184 "Predefined Symbols",
185 0
186 };
187
188 LexerModule lmMMIXAL(SCLEX_MMIXAL, ColouriseMMIXALDoc, "mmixal", 0, MMIXALWordListDesc);
189
190