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