1 | // Scintilla source code edit control |
2 | /** @file LexR.cxx |
3 | ** Lexer for R, S, SPlus Statistics Program (Heavily derived from CPP Lexer). |
4 | ** |
5 | **/ |
6 | // Copyright 1998-2002 by Neil Hodgson <neilh@scintilla.org> |
7 | // The License.txt file describes the conditions under which this software may be distributed. |
8 | |
9 | #include <stdlib.h> |
10 | #include <string.h> |
11 | #include <stdio.h> |
12 | #include <stdarg.h> |
13 | #include <assert.h> |
14 | #include <ctype.h> |
15 | |
16 | #include <string> |
17 | #include <string_view> |
18 | |
19 | #include "ILexer.h" |
20 | #include "Scintilla.h" |
21 | #include "SciLexer.h" |
22 | |
23 | #include "WordList.h" |
24 | #include "LexAccessor.h" |
25 | #include "Accessor.h" |
26 | #include "StyleContext.h" |
27 | #include "CharacterSet.h" |
28 | #include "LexerModule.h" |
29 | |
30 | using namespace Lexilla; |
31 | |
32 | static inline bool IsAWordChar(const int ch) { |
33 | return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_'); |
34 | } |
35 | |
36 | static inline bool IsAWordStart(const int ch) { |
37 | return (ch < 0x80) && (isalnum(ch) || ch == '_'); |
38 | } |
39 | |
40 | static inline bool IsAnOperator(const int ch) { |
41 | if (IsASCII(ch) && 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 == '}' || ch == '{' || ch == '[' || |
49 | ch == ']') |
50 | return true; |
51 | return false; |
52 | } |
53 | |
54 | static void ColouriseRDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[], |
55 | Accessor &styler) { |
56 | |
57 | WordList &keywords = *keywordlists[0]; |
58 | WordList &keywords2 = *keywordlists[1]; |
59 | WordList &keywords3 = *keywordlists[2]; |
60 | |
61 | |
62 | // Do not leak onto next line |
63 | if (initStyle == SCE_R_INFIXEOL) |
64 | initStyle = SCE_R_DEFAULT; |
65 | |
66 | |
67 | StyleContext sc(startPos, length, initStyle, styler); |
68 | |
69 | for (; sc.More(); sc.Forward()) { |
70 | |
71 | if (sc.atLineStart && (sc.state == SCE_R_STRING)) { |
72 | // Prevent SCE_R_STRINGEOL from leaking back to previous line |
73 | sc.SetState(SCE_R_STRING); |
74 | } |
75 | |
76 | // Determine if the current state should terminate. |
77 | if (sc.state == SCE_R_OPERATOR) { |
78 | sc.SetState(SCE_R_DEFAULT); |
79 | } else if (sc.state == SCE_R_NUMBER) { |
80 | if (!IsADigit(sc.ch) && !(sc.ch == '.' && IsADigit(sc.chNext))) { |
81 | sc.SetState(SCE_R_DEFAULT); |
82 | } |
83 | } else if (sc.state == SCE_R_IDENTIFIER) { |
84 | if (!IsAWordChar(sc.ch)) { |
85 | char s[100]; |
86 | sc.GetCurrent(s, sizeof(s)); |
87 | if (keywords.InList(s)) { |
88 | sc.ChangeState(SCE_R_KWORD); |
89 | } else if (keywords2.InList(s)) { |
90 | sc.ChangeState(SCE_R_BASEKWORD); |
91 | } else if (keywords3.InList(s)) { |
92 | sc.ChangeState(SCE_R_OTHERKWORD); |
93 | } |
94 | sc.SetState(SCE_R_DEFAULT); |
95 | } |
96 | } else if (sc.state == SCE_R_COMMENT) { |
97 | if (sc.ch == '\r' || sc.ch == '\n') { |
98 | sc.SetState(SCE_R_DEFAULT); |
99 | } |
100 | } else if (sc.state == SCE_R_STRING) { |
101 | if (sc.ch == '\\') { |
102 | if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') { |
103 | sc.Forward(); |
104 | } |
105 | } else if (sc.ch == '\"') { |
106 | sc.ForwardSetState(SCE_R_DEFAULT); |
107 | } |
108 | } else if (sc.state == SCE_R_INFIX) { |
109 | if (sc.ch == '%') { |
110 | sc.ForwardSetState(SCE_R_DEFAULT); |
111 | } else if (sc.atLineEnd) { |
112 | sc.ChangeState(SCE_R_INFIXEOL); |
113 | sc.ForwardSetState(SCE_R_DEFAULT); |
114 | } |
115 | }else if (sc.state == SCE_R_STRING2) { |
116 | if (sc.ch == '\\') { |
117 | if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') { |
118 | sc.Forward(); |
119 | } |
120 | } else if (sc.ch == '\'') { |
121 | sc.ForwardSetState(SCE_R_DEFAULT); |
122 | } |
123 | } |
124 | |
125 | // Determine if a new state should be entered. |
126 | if (sc.state == SCE_R_DEFAULT) { |
127 | if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) { |
128 | sc.SetState(SCE_R_NUMBER); |
129 | } else if (IsAWordStart(sc.ch) ) { |
130 | sc.SetState(SCE_R_IDENTIFIER); |
131 | } else if (sc.Match('#')) { |
132 | sc.SetState(SCE_R_COMMENT); |
133 | } else if (sc.ch == '\"') { |
134 | sc.SetState(SCE_R_STRING); |
135 | } else if (sc.ch == '%') { |
136 | sc.SetState(SCE_R_INFIX); |
137 | } else if (sc.ch == '\'') { |
138 | sc.SetState(SCE_R_STRING2); |
139 | } else if (IsAnOperator(sc.ch)) { |
140 | sc.SetState(SCE_R_OPERATOR); |
141 | } |
142 | } |
143 | } |
144 | sc.Complete(); |
145 | } |
146 | |
147 | // Store both the current line's fold level and the next lines in the |
148 | // level store to make it easy to pick up with each increment |
149 | // and to make it possible to fiddle the current level for "} else {". |
150 | static void FoldRDoc(Sci_PositionU startPos, Sci_Position length, int, WordList *[], |
151 | Accessor &styler) { |
152 | bool foldCompact = styler.GetPropertyInt("fold.compact" , 1) != 0; |
153 | bool foldAtElse = styler.GetPropertyInt("fold.at.else" , 0) != 0; |
154 | Sci_PositionU endPos = startPos + length; |
155 | int visibleChars = 0; |
156 | Sci_Position lineCurrent = styler.GetLine(startPos); |
157 | int levelCurrent = SC_FOLDLEVELBASE; |
158 | if (lineCurrent > 0) |
159 | levelCurrent = styler.LevelAt(lineCurrent-1) >> 16; |
160 | int levelMinCurrent = levelCurrent; |
161 | int levelNext = levelCurrent; |
162 | char chNext = styler[startPos]; |
163 | int styleNext = styler.StyleAt(startPos); |
164 | for (Sci_PositionU i = startPos; i < endPos; i++) { |
165 | char ch = chNext; |
166 | chNext = styler.SafeGetCharAt(i + 1); |
167 | int style = styleNext; |
168 | styleNext = styler.StyleAt(i + 1); |
169 | bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n'); |
170 | if (style == SCE_R_OPERATOR) { |
171 | if (ch == '{') { |
172 | // Measure the minimum before a '{' to allow |
173 | // folding on "} else {" |
174 | if (levelMinCurrent > levelNext) { |
175 | levelMinCurrent = levelNext; |
176 | } |
177 | levelNext++; |
178 | } else if (ch == '}') { |
179 | levelNext--; |
180 | } |
181 | } |
182 | if (atEOL) { |
183 | int levelUse = levelCurrent; |
184 | if (foldAtElse) { |
185 | levelUse = levelMinCurrent; |
186 | } |
187 | int lev = levelUse | levelNext << 16; |
188 | if (visibleChars == 0 && foldCompact) |
189 | lev |= SC_FOLDLEVELWHITEFLAG; |
190 | if (levelUse < levelNext) |
191 | lev |= SC_FOLDLEVELHEADERFLAG; |
192 | if (lev != styler.LevelAt(lineCurrent)) { |
193 | styler.SetLevel(lineCurrent, lev); |
194 | } |
195 | lineCurrent++; |
196 | levelCurrent = levelNext; |
197 | levelMinCurrent = levelCurrent; |
198 | visibleChars = 0; |
199 | } |
200 | if (!isspacechar(ch)) |
201 | visibleChars++; |
202 | } |
203 | } |
204 | |
205 | |
206 | static const char * const RWordLists[] = { |
207 | "Language Keywords" , |
208 | "Base / Default package function" , |
209 | "Other Package Functions" , |
210 | "Unused" , |
211 | "Unused" , |
212 | 0, |
213 | }; |
214 | |
215 | |
216 | |
217 | LexerModule lmR(SCLEX_R, ColouriseRDoc, "r" , FoldRDoc, RWordLists); |
218 | |