1 // Scintilla source code edit control
2 /** @file LexPython.cxx
3 ** Lexer for Python.
4 **/
5 // Copyright 1998-2002 by Neil Hodgson <neilh@scintilla.org>
6 // The License.txt file describes the conditions under which this software may be distributed.
7
8 #include <stdlib.h>
9 #include <string.h>
10 #include <ctype.h>
11 #include <stdio.h>
12 #include <stdarg.h>
13
14 #include "Platform.h"
15
16 #include "PropSet.h"
17 #include "Accessor.h"
18 #include "StyleContext.h"
19 #include "KeyWords.h"
20 #include "Scintilla.h"
21 #include "SciLexer.h"
22
23 enum kwType { kwOther, kwClass, kwDef, kwImport };
24
IsPyComment(Accessor & styler,int pos,int len)25 static bool IsPyComment(Accessor &styler, int pos, int len) {
26 return len > 0 && styler[pos] == '#';
27 }
28
IsPyStringStart(int ch,int chNext,int chNext2)29 static bool IsPyStringStart(int ch, int chNext, int chNext2) {
30 if (ch == '\'' || ch == '"')
31 return true;
32 if (ch == 'u' || ch == 'U') {
33 if (chNext == '"' || chNext == '\'')
34 return true;
35 if ((chNext == 'r' || chNext == 'R') && (chNext2 == '"' || chNext2 == '\''))
36 return true;
37 }
38 if ((ch == 'r' || ch == 'R') && (chNext == '"' || chNext == '\''))
39 return true;
40
41 return false;
42 }
43
44 /* Return the state to use for the string starting at i; *nextIndex will be set to the first index following the quote(s) */
GetPyStringState(Accessor & styler,int i,unsigned int * nextIndex)45 static int GetPyStringState(Accessor &styler, int i, unsigned int *nextIndex) {
46 char ch = styler.SafeGetCharAt(i);
47 char chNext = styler.SafeGetCharAt(i + 1);
48
49 // Advance beyond r, u, or ur prefix, but bail if there are any unexpected chars
50 if (ch == 'r' || ch == 'R') {
51 i++;
52 ch = styler.SafeGetCharAt(i);
53 chNext = styler.SafeGetCharAt(i + 1);
54 } else if (ch == 'u' || ch == 'U') {
55 if (chNext == 'r' || chNext == 'R')
56 i += 2;
57 else
58 i += 1;
59 ch = styler.SafeGetCharAt(i);
60 chNext = styler.SafeGetCharAt(i + 1);
61 }
62
63 if (ch != '"' && ch != '\'') {
64 *nextIndex = i + 1;
65 return SCE_P_DEFAULT;
66 }
67
68 if (ch == chNext && ch == styler.SafeGetCharAt(i + 2)) {
69 *nextIndex = i + 3;
70
71 if (ch == '"')
72 return SCE_P_TRIPLEDOUBLE;
73 else
74 return SCE_P_TRIPLE;
75 } else {
76 *nextIndex = i + 1;
77
78 if (ch == '"')
79 return SCE_P_STRING;
80 else
81 return SCE_P_CHARACTER;
82 }
83 }
84
IsAWordChar(int ch)85 static inline bool IsAWordChar(int ch) {
86 return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');
87 }
88
IsAWordStart(int ch)89 static inline bool IsAWordStart(int ch) {
90 return (ch < 0x80) && (isalnum(ch) || ch == '_');
91 }
92
ColourisePyDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler)93 static void ColourisePyDoc(unsigned int startPos, int length, int initStyle,
94 WordList *keywordlists[], Accessor &styler) {
95
96 int endPos = startPos + length;
97
98 // Backtrack to previous line in case need to fix its tab whinging
99 int lineCurrent = styler.GetLine(startPos);
100 if (startPos > 0) {
101 if (lineCurrent > 0) {
102 lineCurrent--;
103 startPos = styler.LineStart(lineCurrent);
104 if (startPos == 0)
105 initStyle = SCE_P_DEFAULT;
106 else
107 initStyle = styler.StyleAt(startPos - 1);
108 }
109 }
110
111 WordList &keywords = *keywordlists[0];
112 WordList &keywords2 = *keywordlists[1];
113
114 const int whingeLevel = styler.GetPropertyInt("tab.timmy.whinge.level");
115
116 initStyle = initStyle & 31;
117 if (initStyle == SCE_P_STRINGEOL) {
118 initStyle = SCE_P_DEFAULT;
119 }
120
121 kwType kwLast = kwOther;
122 int spaceFlags = 0;
123 styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
124 bool hexadecimal = false;
125
126 // Python uses a different mask because bad indentation is marked by oring with 32
127 StyleContext sc(startPos, endPos - startPos, initStyle, styler, 0x7f);
128
129 for (; sc.More(); sc.Forward()) {
130
131 if (sc.atLineStart) {
132 const char chBad = static_cast<char>(64);
133 const char chGood = static_cast<char>(0);
134 char chFlags = chGood;
135 if (whingeLevel == 1) {
136 chFlags = (spaceFlags & wsInconsistent) ? chBad : chGood;
137 } else if (whingeLevel == 2) {
138 chFlags = (spaceFlags & wsSpaceTab) ? chBad : chGood;
139 } else if (whingeLevel == 3) {
140 chFlags = (spaceFlags & wsSpace) ? chBad : chGood;
141 } else if (whingeLevel == 4) {
142 chFlags = (spaceFlags & wsTab) ? chBad : chGood;
143 }
144 sc.SetState(sc.state);
145 styler.SetFlags(chFlags, static_cast<char>(sc.state));
146 }
147
148 if (sc.atLineEnd) {
149 if ((sc.state == SCE_P_DEFAULT) ||
150 (sc.state == SCE_P_TRIPLE) ||
151 (sc.state == SCE_P_TRIPLEDOUBLE)) {
152 // Perform colourisation of white space and triple quoted strings at end of each line to allow
153 // tab marking to work inside white space and triple quoted strings
154 sc.SetState(sc.state);
155 }
156 lineCurrent++;
157 styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
158 if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
159 sc.ChangeState(SCE_P_STRINGEOL);
160 sc.ForwardSetState(SCE_P_DEFAULT);
161 }
162 if (!sc.More())
163 break;
164 }
165
166 bool needEOLCheck = false;
167
168 // Check for a state end
169 if (sc.state == SCE_P_OPERATOR) {
170 kwLast = kwOther;
171 sc.SetState(SCE_P_DEFAULT);
172 } else if (sc.state == SCE_P_NUMBER) {
173 if (!IsAWordChar(sc.ch) &&
174 !(!hexadecimal && ((sc.ch == '+' || sc.ch == '-') && (sc.chPrev == 'e' || sc.chPrev == 'E')))) {
175 sc.SetState(SCE_P_DEFAULT);
176 }
177 } else if (sc.state == SCE_P_IDENTIFIER) {
178 if ((sc.ch == '.') || (!IsAWordChar(sc.ch))) {
179 char s[100];
180 sc.GetCurrent(s, sizeof(s));
181 int style = SCE_P_IDENTIFIER;
182 if ((kwLast == kwImport) && (strcmp(s, "as") == 0)) {
183 style = SCE_P_WORD;
184 } else if (keywords.InList(s)) {
185 style = SCE_P_WORD;
186 } else if (kwLast == kwClass) {
187 style = SCE_P_CLASSNAME;
188 } else if (kwLast == kwDef) {
189 style = SCE_P_DEFNAME;
190 } else if (keywords2.InList(s)) {
191 style = SCE_P_WORD2;
192 }
193 sc.ChangeState(style);
194 sc.SetState(SCE_P_DEFAULT);
195 if (style == SCE_P_WORD) {
196 if (0 == strcmp(s, "class"))
197 kwLast = kwClass;
198 else if (0 == strcmp(s, "def"))
199 kwLast = kwDef;
200 else if (0 == strcmp(s, "import"))
201 kwLast = kwImport;
202 else
203 kwLast = kwOther;
204 } else {
205 kwLast = kwOther;
206 }
207 }
208 } else if ((sc.state == SCE_P_COMMENTLINE) || (sc.state == SCE_P_COMMENTBLOCK)) {
209 if (sc.ch == '\r' || sc.ch == '\n') {
210 sc.SetState(SCE_P_DEFAULT);
211 }
212 } else if (sc.state == SCE_P_DECORATOR) {
213 if (sc.ch == '\r' || sc.ch == '\n') {
214 sc.SetState(SCE_P_DEFAULT);
215 } else if (sc.ch == '#') {
216 sc.SetState((sc.chNext == '#') ? SCE_P_COMMENTBLOCK : SCE_P_COMMENTLINE);
217 }
218 } else if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
219 if (sc.ch == '\\') {
220 if ((sc.chNext == '\r') && (sc.GetRelative(2) == '\n')) {
221 sc.Forward();
222 }
223 sc.Forward();
224 } else if ((sc.state == SCE_P_STRING) && (sc.ch == '\"')) {
225 sc.ForwardSetState(SCE_P_DEFAULT);
226 needEOLCheck = true;
227 } else if ((sc.state == SCE_P_CHARACTER) && (sc.ch == '\'')) {
228 sc.ForwardSetState(SCE_P_DEFAULT);
229 needEOLCheck = true;
230 }
231 } else if (sc.state == SCE_P_TRIPLE) {
232 if (sc.ch == '\\') {
233 sc.Forward();
234 } else if (sc.Match("\'\'\'")) {
235 sc.Forward();
236 sc.Forward();
237 sc.ForwardSetState(SCE_P_DEFAULT);
238 needEOLCheck = true;
239 }
240 } else if (sc.state == SCE_P_TRIPLEDOUBLE) {
241 if (sc.ch == '\\') {
242 sc.Forward();
243 } else if (sc.Match("\"\"\"")) {
244 sc.Forward();
245 sc.Forward();
246 sc.ForwardSetState(SCE_P_DEFAULT);
247 needEOLCheck = true;
248 }
249 }
250
251 // State exit code may have moved on to end of line
252 if (needEOLCheck && sc.atLineEnd) {
253 lineCurrent++;
254 styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
255 if (!sc.More())
256 break;
257 }
258
259 // Check for a new state starting character
260 if (sc.state == SCE_P_DEFAULT) {
261 if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
262 if (sc.ch == '0' && (sc.chNext == 'x' || sc.chNext == 'X')) {
263 hexadecimal = true;
264 } else {
265 hexadecimal = false;
266 }
267 sc.SetState(SCE_P_NUMBER);
268 } else if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch)) || sc.ch == '`') {
269 sc.SetState(SCE_P_OPERATOR);
270 } else if (sc.ch == '#') {
271 sc.SetState(sc.chNext == '#' ? SCE_P_COMMENTBLOCK : SCE_P_COMMENTLINE);
272 } else if (sc.ch == '@') {
273 sc.SetState(SCE_P_DECORATOR);
274 } else if (IsPyStringStart(sc.ch, sc.chNext, sc.GetRelative(2))) {
275 unsigned int nextIndex = 0;
276 sc.SetState(GetPyStringState(styler, sc.currentPos, &nextIndex));
277 while (nextIndex > (sc.currentPos + 1) && sc.More()) {
278 sc.Forward();
279 }
280 } else if (IsAWordStart(sc.ch)) {
281 sc.SetState(SCE_P_IDENTIFIER);
282 }
283 }
284 }
285 sc.Complete();
286 }
287
IsCommentLine(int line,Accessor & styler)288 static bool IsCommentLine(int line, Accessor &styler) {
289 int pos = styler.LineStart(line);
290 int eol_pos = styler.LineStart(line + 1) - 1;
291 for (int i = pos; i < eol_pos; i++) {
292 char ch = styler[i];
293 if (ch == '#')
294 return true;
295 else if (ch != ' ' && ch != '\t')
296 return false;
297 }
298 return false;
299 }
300
IsQuoteLine(int line,Accessor & styler)301 static bool IsQuoteLine(int line, Accessor &styler) {
302 int style = styler.StyleAt(styler.LineStart(line)) & 31;
303 return ((style == SCE_P_TRIPLE) || (style == SCE_P_TRIPLEDOUBLE));
304 }
305
306
FoldPyDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)307 static void FoldPyDoc(unsigned int startPos, int length, int /*initStyle - unused*/,
308 WordList *[], Accessor &styler) {
309 const int maxPos = startPos + length;
310 const int maxLines = styler.GetLine(maxPos - 1); // Requested last line
311 const int docLines = styler.GetLine(styler.Length() - 1); // Available last line
312 const bool foldComment = styler.GetPropertyInt("fold.comment.python") != 0;
313 const bool foldQuotes = styler.GetPropertyInt("fold.quotes.python") != 0;
314
315 // Backtrack to previous non-blank line so we can determine indent level
316 // for any white space lines (needed esp. within triple quoted strings)
317 // and so we can fix any preceding fold level (which is why we go back
318 // at least one line in all cases)
319 int spaceFlags = 0;
320 int lineCurrent = styler.GetLine(startPos);
321 int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
322 while (lineCurrent > 0) {
323 lineCurrent--;
324 indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
325 if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG) &&
326 (!IsCommentLine(lineCurrent, styler)) &&
327 (!IsQuoteLine(lineCurrent, styler)))
328 break;
329 }
330 int indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
331
332 // Set up initial loop state
333 startPos = styler.LineStart(lineCurrent);
334 int prev_state = SCE_P_DEFAULT & 31;
335 if (lineCurrent >= 1)
336 prev_state = styler.StyleAt(startPos - 1) & 31;
337 int prevQuote = foldQuotes && ((prev_state == SCE_P_TRIPLE) || (prev_state == SCE_P_TRIPLEDOUBLE));
338 int prevComment = 0;
339 if (lineCurrent >= 1)
340 prevComment = foldComment && IsCommentLine(lineCurrent - 1, styler);
341
342 // Process all characters to end of requested range or end of any triple quote
343 // or comment that hangs over the end of the range. Cap processing in all cases
344 // to end of document (in case of unclosed quote or comment at end).
345 while ((lineCurrent <= docLines) && ((lineCurrent <= maxLines) || prevQuote || prevComment)) {
346
347 // Gather info
348 int lev = indentCurrent;
349 int lineNext = lineCurrent + 1;
350 int indentNext = indentCurrent;
351 int quote = false;
352 if (lineNext <= docLines) {
353 // Information about next line is only available if not at end of document
354 indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
355 int style = styler.StyleAt(styler.LineStart(lineNext)) & 31;
356 quote = foldQuotes && ((style == SCE_P_TRIPLE) || (style == SCE_P_TRIPLEDOUBLE));
357 }
358 const int quote_start = (quote && !prevQuote);
359 const int quote_continue = (quote && prevQuote);
360 const int comment = foldComment && IsCommentLine(lineCurrent, styler);
361 const int comment_start = (comment && !prevComment && (lineNext <= docLines) &&
362 IsCommentLine(lineNext, styler) && (lev > SC_FOLDLEVELBASE));
363 const int comment_continue = (comment && prevComment);
364 if ((!quote || !prevQuote) && !comment)
365 indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
366 if (quote)
367 indentNext = indentCurrentLevel;
368 if (indentNext & SC_FOLDLEVELWHITEFLAG)
369 indentNext = SC_FOLDLEVELWHITEFLAG | indentCurrentLevel;
370
371 if (quote_start) {
372 // Place fold point at start of triple quoted string
373 lev |= SC_FOLDLEVELHEADERFLAG;
374 } else if (quote_continue || prevQuote) {
375 // Add level to rest of lines in the string
376 lev = lev + 1;
377 } else if (comment_start) {
378 // Place fold point at start of a block of comments
379 lev |= SC_FOLDLEVELHEADERFLAG;
380 } else if (comment_continue) {
381 // Add level to rest of lines in the block
382 lev = lev + 1;
383 }
384
385 // Skip past any blank lines for next indent level info; we skip also
386 // comments (all comments, not just those starting in column 0)
387 // which effectively folds them into surrounding code rather
388 // than screwing up folding.
389
390 while (!quote &&
391 (lineNext < docLines) &&
392 ((indentNext & SC_FOLDLEVELWHITEFLAG) ||
393 (lineNext <= docLines && IsCommentLine(lineNext, styler)))) {
394
395 lineNext++;
396 indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
397 }
398
399 const int levelAfterComments = indentNext & SC_FOLDLEVELNUMBERMASK;
400 const int levelBeforeComments = Platform::Maximum(indentCurrentLevel,levelAfterComments);
401
402 // Now set all the indent levels on the lines we skipped
403 // Do this from end to start. Once we encounter one line
404 // which is indented more than the line after the end of
405 // the comment-block, use the level of the block before
406
407 int skipLine = lineNext;
408 int skipLevel = levelAfterComments;
409
410 while (--skipLine > lineCurrent) {
411 int skipLineIndent = styler.IndentAmount(skipLine, &spaceFlags, NULL);
412
413 if ((skipLineIndent & SC_FOLDLEVELNUMBERMASK) > levelAfterComments)
414 skipLevel = levelBeforeComments;
415
416 int whiteFlag = skipLineIndent & SC_FOLDLEVELWHITEFLAG;
417
418 styler.SetLevel(skipLine, skipLevel | whiteFlag);
419 }
420
421 // Set fold header on non-quote/non-comment line
422 if (!quote && !comment && !(indentCurrent & SC_FOLDLEVELWHITEFLAG) ) {
423 if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK))
424 lev |= SC_FOLDLEVELHEADERFLAG;
425 }
426
427 // Keep track of triple quote and block comment state of previous line
428 prevQuote = quote;
429 prevComment = comment_start || comment_continue;
430
431 // Set fold level for this line and move to next line
432 styler.SetLevel(lineCurrent, lev);
433 indentCurrent = indentNext;
434 lineCurrent = lineNext;
435 }
436
437 // NOTE: Cannot set level of last line here because indentCurrent doesn't have
438 // header flag set; the loop above is crafted to take care of this case!
439 //styler.SetLevel(lineCurrent, indentCurrent);
440 }
441
442 static const char * const pythonWordListDesc[] = {
443 "Keywords",
444 "Highlighted identifiers",
445 0
446 };
447
448 LexerModule lmPython(SCLEX_PYTHON, ColourisePyDoc, "python", FoldPyDoc,
449 pythonWordListDesc);
450