1 // Scintilla source code edit control
2 /** @file LexLua.cxx
3 ** Lexer for Lua language.
4 **
5 ** Written by Paul Winwood.
6 ** Folder by Alexey Yutkin.
7 ** Modified by Marcos E. Wurzius & Philippe Lhoste
8 **/
9
10 #include <stdlib.h>
11 #include <string.h>
12 #include <ctype.h>
13 #include <stdarg.h>
14 #include <stdio.h>
15
16 #include "Platform.h"
17
18 #include "PropSet.h"
19 #include "Accessor.h"
20 #include "StyleContext.h"
21 #include "KeyWords.h"
22 #include "Scintilla.h"
23 #include "SciLexer.h"
24
25 // Extended to accept accented characters
IsAWordChar(int ch)26 static inline bool IsAWordChar(int ch) {
27 return ch >= 0x80 ||
28 (isalnum(ch) || ch == '.' || ch == '_');
29 }
30
IsAWordStart(int ch)31 static inline bool IsAWordStart(int ch) {
32 return ch >= 0x80 ||
33 (isalpha(ch) || ch == '_');
34 }
35
IsANumberChar(int ch)36 static inline bool IsANumberChar(int ch) {
37 // Not exactly following number definition (several dots are seen as OK, etc.)
38 // but probably enough in most cases.
39 return (ch < 0x80) &&
40 (isdigit(ch) || toupper(ch) == 'E' ||
41 ch == '.' || ch == '-' || ch == '+' ||
42 (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F'));
43 }
44
IsLuaOperator(int ch)45 static inline bool IsLuaOperator(int ch) {
46 if (ch >= 0x80 || isalnum(ch)) {
47 return false;
48 }
49 // '.' left out as it is used to make up numbers
50 if (ch == '*' || ch == '/' || ch == '-' || ch == '+' ||
51 ch == '(' || ch == ')' || ch == '=' ||
52 ch == '{' || ch == '}' || ch == '~' ||
53 ch == '[' || ch == ']' || ch == ';' ||
54 ch == '<' || ch == '>' || ch == ',' ||
55 ch == '.' || ch == '^' || ch == '%' || ch == ':' ||
56 ch == '#') {
57 return true;
58 }
59 return false;
60 }
61
62 // Test for [=[ ... ]=] delimiters, returns 0 if it's only a [ or ],
63 // return 1 for [[ or ]], returns >=2 for [=[ or ]=] and so on.
64 // The maximum number of '=' characters allowed is 254.
LongDelimCheck(StyleContext & sc)65 static int LongDelimCheck(StyleContext &sc) {
66 int sep = 1;
67 while (sc.GetRelative(sep) == '=' && sep < 0xFF)
68 sep++;
69 if (sc.GetRelative(sep) == sc.ch)
70 return sep;
71 return 0;
72 }
73
ColouriseLuaDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler)74 static void ColouriseLuaDoc(
75 unsigned int startPos,
76 int length,
77 int initStyle,
78 WordList *keywordlists[],
79 Accessor &styler) {
80
81 WordList &keywords = *keywordlists[0];
82 WordList &keywords2 = *keywordlists[1];
83 WordList &keywords3 = *keywordlists[2];
84 WordList &keywords4 = *keywordlists[3];
85 WordList &keywords5 = *keywordlists[4];
86 WordList &keywords6 = *keywordlists[5];
87 WordList &keywords7 = *keywordlists[6];
88 WordList &keywords8 = *keywordlists[7];
89
90 int currentLine = styler.GetLine(startPos);
91 // Initialize long string [[ ... ]] or block comment --[[ ... ]] nesting level,
92 // if we are inside such a string. Block comment was introduced in Lua 5.0,
93 // blocks with separators [=[ ... ]=] in Lua 5.1.
94 int nestLevel = 0;
95 int sepCount = 0;
96 if (initStyle == SCE_LUA_LITERALSTRING || initStyle == SCE_LUA_COMMENT) {
97 int lineState = styler.GetLineState(currentLine - 1);
98 nestLevel = lineState >> 8;
99 sepCount = lineState & 0xFF;
100 }
101
102 // Do not leak onto next line
103 if (initStyle == SCE_LUA_STRINGEOL || initStyle == SCE_LUA_COMMENTLINE || initStyle == SCE_LUA_PREPROCESSOR) {
104 initStyle = SCE_LUA_DEFAULT;
105 }
106
107 StyleContext sc(startPos, length, initStyle, styler);
108 if (startPos == 0 && sc.ch == '#') {
109 // shbang line: # is a comment only if first char of the script
110 sc.SetState(SCE_LUA_COMMENTLINE);
111 }
112 for (; sc.More(); sc.Forward()) {
113 if (sc.atLineEnd) {
114 // Update the line state, so it can be seen by next line
115 currentLine = styler.GetLine(sc.currentPos);
116 switch (sc.state) {
117 case SCE_LUA_LITERALSTRING:
118 case SCE_LUA_COMMENT:
119 // Inside a literal string or block comment, we set the line state
120 styler.SetLineState(currentLine, (nestLevel << 8) | sepCount);
121 break;
122 default:
123 // Reset the line state
124 styler.SetLineState(currentLine, 0);
125 break;
126 }
127 }
128 if (sc.atLineStart && (sc.state == SCE_LUA_STRING)) {
129 // Prevent SCE_LUA_STRINGEOL from leaking back to previous line
130 sc.SetState(SCE_LUA_STRING);
131 }
132
133 // Handle string line continuation
134 if ((sc.state == SCE_LUA_STRING || sc.state == SCE_LUA_CHARACTER) &&
135 sc.ch == '\\') {
136 if (sc.chNext == '\n' || sc.chNext == '\r') {
137 sc.Forward();
138 if (sc.ch == '\r' && sc.chNext == '\n') {
139 sc.Forward();
140 }
141 continue;
142 }
143 }
144
145 // Determine if the current state should terminate.
146 if (sc.state == SCE_LUA_OPERATOR) {
147 sc.SetState(SCE_LUA_DEFAULT);
148 } else if (sc.state == SCE_LUA_NUMBER) {
149 // We stop the number definition on non-numerical non-dot non-eE non-sign non-hexdigit char
150 if (!IsANumberChar(sc.ch)) {
151 sc.SetState(SCE_LUA_DEFAULT);
152 }
153 } else if (sc.state == SCE_LUA_IDENTIFIER) {
154 if (!IsAWordChar(sc.ch) || sc.Match('.', '.')) {
155 char s[100];
156 sc.GetCurrent(s, sizeof(s));
157 if (keywords.InList(s)) {
158 sc.ChangeState(SCE_LUA_WORD);
159 } else if (keywords2.InList(s)) {
160 sc.ChangeState(SCE_LUA_WORD2);
161 } else if (keywords3.InList(s)) {
162 sc.ChangeState(SCE_LUA_WORD3);
163 } else if (keywords4.InList(s)) {
164 sc.ChangeState(SCE_LUA_WORD4);
165 } else if (keywords5.InList(s)) {
166 sc.ChangeState(SCE_LUA_WORD5);
167 } else if (keywords6.InList(s)) {
168 sc.ChangeState(SCE_LUA_WORD6);
169 } else if (keywords6.InList(s)) {
170 sc.ChangeState(SCE_LUA_WORD6);
171 } else if (keywords7.InList(s)) {
172 sc.ChangeState(SCE_LUA_WORD7);
173 } else if (keywords8.InList(s)) {
174 sc.ChangeState(SCE_LUA_WORD8);
175 }
176 sc.SetState(SCE_LUA_DEFAULT);
177 }
178 } else if (sc.state == SCE_LUA_COMMENTLINE || sc.state == SCE_LUA_PREPROCESSOR) {
179 if (sc.atLineEnd) {
180 sc.ForwardSetState(SCE_LUA_DEFAULT);
181 }
182 } else if (sc.state == SCE_LUA_STRING) {
183 if (sc.ch == '\\') {
184 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
185 sc.Forward();
186 }
187 } else if (sc.ch == '\"') {
188 sc.ForwardSetState(SCE_LUA_DEFAULT);
189 } else if (sc.atLineEnd) {
190 sc.ChangeState(SCE_LUA_STRINGEOL);
191 sc.ForwardSetState(SCE_LUA_DEFAULT);
192 }
193 } else if (sc.state == SCE_LUA_CHARACTER) {
194 if (sc.ch == '\\') {
195 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
196 sc.Forward();
197 }
198 } else if (sc.ch == '\'') {
199 sc.ForwardSetState(SCE_LUA_DEFAULT);
200 } else if (sc.atLineEnd) {
201 sc.ChangeState(SCE_LUA_STRINGEOL);
202 sc.ForwardSetState(SCE_LUA_DEFAULT);
203 }
204 } else if (sc.state == SCE_LUA_LITERALSTRING || sc.state == SCE_LUA_COMMENT) {
205 if (sc.ch == '[') {
206 int sep = LongDelimCheck(sc);
207 if (sep == 1 && sepCount == 1) { // [[-only allowed to nest
208 nestLevel++;
209 sc.Forward();
210 }
211 } else if (sc.ch == ']') {
212 int sep = LongDelimCheck(sc);
213 if (sep == 1 && sepCount == 1) { // un-nest with ]]-only
214 nestLevel--;
215 sc.Forward();
216 if (nestLevel == 0) {
217 sc.ForwardSetState(SCE_LUA_DEFAULT);
218 }
219 } else if (sep > 1 && sep == sepCount) { // ]=]-style delim
220 sc.Forward(sep);
221 sc.ForwardSetState(SCE_LUA_DEFAULT);
222 }
223 }
224 }
225
226 // Determine if a new state should be entered.
227 if (sc.state == SCE_LUA_DEFAULT) {
228 if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
229 sc.SetState(SCE_LUA_NUMBER);
230 if (sc.ch == '0' && toupper(sc.chNext) == 'X') {
231 sc.Forward(1);
232 }
233 } else if (IsAWordStart(sc.ch)) {
234 sc.SetState(SCE_LUA_IDENTIFIER);
235 } else if (sc.ch == '\"') {
236 sc.SetState(SCE_LUA_STRING);
237 } else if (sc.ch == '\'') {
238 sc.SetState(SCE_LUA_CHARACTER);
239 } else if (sc.ch == '[') {
240 sepCount = LongDelimCheck(sc);
241 if (sepCount == 0) {
242 sc.SetState(SCE_LUA_OPERATOR);
243 } else {
244 nestLevel = 1;
245 sc.SetState(SCE_LUA_LITERALSTRING);
246 sc.Forward(sepCount);
247 }
248 } else if (sc.Match('-', '-')) {
249 sc.SetState(SCE_LUA_COMMENTLINE);
250 if (sc.Match("--[")) {
251 sc.Forward(2);
252 sepCount = LongDelimCheck(sc);
253 if (sepCount > 0) {
254 nestLevel = 1;
255 sc.ChangeState(SCE_LUA_COMMENT);
256 sc.Forward(sepCount);
257 }
258 } else {
259 sc.Forward();
260 }
261 } else if (sc.atLineStart && sc.Match('$')) {
262 sc.SetState(SCE_LUA_PREPROCESSOR); // Obsolete since Lua 4.0, but still in old code
263 } else if (IsLuaOperator(static_cast<char>(sc.ch))) {
264 sc.SetState(SCE_LUA_OPERATOR);
265 }
266 }
267 }
268 sc.Complete();
269 }
270
FoldLuaDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)271 static void FoldLuaDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
272 Accessor &styler) {
273 unsigned int lengthDoc = startPos + length;
274 int visibleChars = 0;
275 int lineCurrent = styler.GetLine(startPos);
276 int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
277 int levelCurrent = levelPrev;
278 char chNext = styler[startPos];
279 bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
280 int styleNext = styler.StyleAt(startPos);
281 char s[10];
282
283 for (unsigned int i = startPos; i < lengthDoc; i++) {
284 char ch = chNext;
285 chNext = styler.SafeGetCharAt(i + 1);
286 int style = styleNext;
287 styleNext = styler.StyleAt(i + 1);
288 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
289 if (style == SCE_LUA_WORD) {
290 if (ch == 'i' || ch == 'd' || ch == 'f' || ch == 'e' || ch == 'r' || ch == 'u') {
291 for (unsigned int j = 0; j < 8; j++) {
292 if (!iswordchar(styler[i + j])) {
293 break;
294 }
295 s[j] = styler[i + j];
296 s[j + 1] = '\0';
297 }
298
299 if ((strcmp(s, "if") == 0) || (strcmp(s, "do") == 0) || (strcmp(s, "function") == 0) || (strcmp(s, "repeat") == 0)) {
300 levelCurrent++;
301 }
302 if ((strcmp(s, "end") == 0) || (strcmp(s, "elseif") == 0) || (strcmp(s, "until") == 0)) {
303 levelCurrent--;
304 }
305 }
306 } else if (style == SCE_LUA_OPERATOR) {
307 if (ch == '{' || ch == '(') {
308 levelCurrent++;
309 } else if (ch == '}' || ch == ')') {
310 levelCurrent--;
311 }
312 } else if (style == SCE_LUA_LITERALSTRING || style == SCE_LUA_COMMENT) {
313 if (ch == '[') {
314 levelCurrent++;
315 } else if (ch == ']') {
316 levelCurrent--;
317 }
318 }
319
320 if (atEOL) {
321 int lev = levelPrev;
322 if (visibleChars == 0 && foldCompact) {
323 lev |= SC_FOLDLEVELWHITEFLAG;
324 }
325 if ((levelCurrent > levelPrev) && (visibleChars > 0)) {
326 lev |= SC_FOLDLEVELHEADERFLAG;
327 }
328 if (lev != styler.LevelAt(lineCurrent)) {
329 styler.SetLevel(lineCurrent, lev);
330 }
331 lineCurrent++;
332 levelPrev = levelCurrent;
333 visibleChars = 0;
334 }
335 if (!isspacechar(ch)) {
336 visibleChars++;
337 }
338 }
339 // Fill in the real level of the next line, keeping the current flags as they will be filled in later
340
341 int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
342 styler.SetLevel(lineCurrent, levelPrev | flagsNext);
343 }
344
345 static const char * const luaWordListDesc[] = {
346 "Keywords",
347 "Basic functions",
348 "String, (table) & math functions",
349 "(coroutines), I/O & system facilities",
350 "user1",
351 "user2",
352 "user3",
353 "user4",
354 0
355 };
356
357 LexerModule lmLua(SCLEX_LUA, ColouriseLuaDoc, "lua", FoldLuaDoc, luaWordListDesc);
358