1 // Scintilla source code edit control
2 /** @file LexVB.cxx
3  ** Lexer for Visual Basic and VBScript.
4  **/
5 // Copyright 1998-2005 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 // Internal state, highlighted as number
24 #define SCE_B_FILENUMBER SCE_B_DEFAULT+100
25 
26 
IsVBComment(Accessor & styler,int pos,int len)27 static bool IsVBComment(Accessor &styler, int pos, int len) {
28 	return len > 0 && styler[pos] == '\'';
29 }
30 
IsTypeCharacter(int ch)31 static inline bool IsTypeCharacter(int ch) {
32 	return ch == '%' || ch == '&' || ch == '@' || ch == '!' || ch == '#' || ch == '$';
33 }
34 
35 // Extended to accept accented characters
IsAWordChar(int ch)36 static inline bool IsAWordChar(int ch) {
37 	return ch >= 0x80 ||
38 	       (isalnum(ch) || ch == '.' || ch == '_');
39 }
40 
IsAWordStart(int ch)41 static inline bool IsAWordStart(int ch) {
42 	return ch >= 0x80 ||
43 	       (isalpha(ch) || ch == '_');
44 }
45 
IsANumberChar(int ch)46 static inline bool IsANumberChar(int ch) {
47 	// Not exactly following number definition (several dots are seen as OK, etc.)
48 	// but probably enough in most cases.
49 	return (ch < 0x80) &&
50 	        (isdigit(ch) || toupper(ch) == 'E' ||
51              ch == '.' || ch == '-' || ch == '+');
52 }
53 
ColouriseVBDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler,bool vbScriptSyntax)54 static void ColouriseVBDoc(unsigned int startPos, int length, int initStyle,
55                            WordList *keywordlists[], Accessor &styler, bool vbScriptSyntax) {
56 
57 	WordList &keywords = *keywordlists[0];
58 	WordList &keywords2 = *keywordlists[1];
59 	WordList &keywords3 = *keywordlists[2];
60 	WordList &keywords4 = *keywordlists[3];
61 
62 	styler.StartAt(startPos);
63 
64 	int visibleChars = 0;
65 	int fileNbDigits = 0;
66 
67 	// Do not leak onto next line
68 	if (initStyle == SCE_B_STRINGEOL || initStyle == SCE_B_COMMENT || initStyle == SCE_B_PREPROCESSOR) {
69 		initStyle = SCE_B_DEFAULT;
70 	}
71 
72 	StyleContext sc(startPos, length, initStyle, styler);
73 
74 	for (; sc.More(); sc.Forward()) {
75 
76 		if (sc.state == SCE_B_OPERATOR) {
77 			sc.SetState(SCE_B_DEFAULT);
78 		} else if (sc.state == SCE_B_IDENTIFIER) {
79 			if (!IsAWordChar(sc.ch)) {
80 				// In Basic (except VBScript), a variable name or a function name
81 				// can end with a special character indicating the type of the value
82 				// held or returned.
83 				bool skipType = false;
84 				if (!vbScriptSyntax && IsTypeCharacter(sc.ch)) {
85 					sc.Forward();	// Skip it
86 					skipType = true;
87 				}
88 				if (sc.ch == ']') {
89 					sc.Forward();
90 				}
91 				char s[100];
92 				sc.GetCurrentLowered(s, sizeof(s));
93 				if (skipType) {
94 					s[strlen(s) - 1] = '\0';
95 				}
96 				if (strcmp(s, "rem") == 0) {
97 					sc.ChangeState(SCE_B_COMMENT);
98 				} else {
99 					if (keywords.InList(s)) {
100 						sc.ChangeState(SCE_B_KEYWORD);
101 					} else if (keywords2.InList(s)) {
102 						sc.ChangeState(SCE_B_KEYWORD2);
103 					} else if (keywords3.InList(s)) {
104 						sc.ChangeState(SCE_B_KEYWORD3);
105 					} else if (keywords4.InList(s)) {
106 						sc.ChangeState(SCE_B_KEYWORD4);
107 					}	// Else, it is really an identifier...
108 					sc.SetState(SCE_B_DEFAULT);
109 				}
110 			}
111 		} else if (sc.state == SCE_B_NUMBER) {
112 			// We stop the number definition on non-numerical non-dot non-eE non-sign char
113 			// Also accepts A-F for hex. numbers
114 			if (!IsANumberChar(sc.ch) && !(tolower(sc.ch) >= 'a' && tolower(sc.ch) <= 'f')) {
115 				sc.SetState(SCE_B_DEFAULT);
116 			}
117 		} else if (sc.state == SCE_B_STRING) {
118 			// VB doubles quotes to preserve them, so just end this string
119 			// state now as a following quote will start again
120 			if (sc.ch == '\"') {
121 				if (tolower(sc.chNext) == 'c') {
122 					sc.Forward();
123 				}
124 				sc.ForwardSetState(SCE_B_DEFAULT);
125 			} else if (sc.atLineEnd) {
126 				sc.ChangeState(SCE_B_STRINGEOL);
127 				sc.ForwardSetState(SCE_B_DEFAULT);
128 			}
129 		} else if (sc.state == SCE_B_COMMENT) {
130 			if (sc.atLineEnd) {
131 				sc.ForwardSetState(SCE_B_DEFAULT);
132 			}
133 		} else if (sc.state == SCE_B_PREPROCESSOR) {
134 			if (sc.atLineEnd) {
135 				sc.ForwardSetState(SCE_B_DEFAULT);
136 			}
137 		} else if (sc.state == SCE_B_FILENUMBER) {
138 			if (IsADigit(sc.ch)) {
139 				fileNbDigits++;
140 				if (fileNbDigits > 3) {
141 					sc.ChangeState(SCE_B_DATE);
142 				}
143 			} else if (sc.ch == '\r' || sc.ch == '\n' || sc.ch == ',') {
144 				// Regular uses: Close #1; Put #1, ...; Get #1, ... etc.
145 				// Too bad if date is format #27, Oct, 2003# or something like that...
146 				// Use regular number state
147 				sc.ChangeState(SCE_B_NUMBER);
148 				sc.SetState(SCE_B_DEFAULT);
149 			} else if (sc.ch == '#') {
150 				sc.ChangeState(SCE_B_DATE);
151 				sc.ForwardSetState(SCE_B_DEFAULT);
152 			} else {
153 				sc.ChangeState(SCE_B_DATE);
154 			}
155 			if (sc.state != SCE_B_FILENUMBER) {
156 				fileNbDigits = 0;
157 			}
158 		} else if (sc.state == SCE_B_DATE) {
159 			if (sc.atLineEnd) {
160 				sc.ChangeState(SCE_B_STRINGEOL);
161 				sc.ForwardSetState(SCE_B_DEFAULT);
162 			} else if (sc.ch == '#') {
163 				sc.ForwardSetState(SCE_B_DEFAULT);
164 			}
165 		}
166 
167 		if (sc.state == SCE_B_DEFAULT) {
168 			if (sc.ch == '\'') {
169 				sc.SetState(SCE_B_COMMENT);
170 			} else if (sc.ch == '\"') {
171 				sc.SetState(SCE_B_STRING);
172 			} else if (sc.ch == '#' && visibleChars == 0) {
173 				// Preprocessor commands are alone on their line
174 				sc.SetState(SCE_B_PREPROCESSOR);
175 			} else if (sc.ch == '#') {
176 				// It can be a date literal, ending with #, or a file number, from 1 to 511
177 				// The date literal depends on the locale, so anything can go between #'s.
178 				// Can be #January 1, 1993# or #1 Jan 93# or #05/11/2003#, etc.
179 				// So we set the FILENUMBER state, and switch to DATE if it isn't a file number
180 				sc.SetState(SCE_B_FILENUMBER);
181 			} else if (sc.ch == '&' && tolower(sc.chNext) == 'h') {
182 				// Hexadecimal number
183 				sc.SetState(SCE_B_NUMBER);
184 				sc.Forward();
185 			} else if (sc.ch == '&' && tolower(sc.chNext) == 'o') {
186 				// Octal number
187 				sc.SetState(SCE_B_NUMBER);
188 				sc.Forward();
189 			} else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
190 				sc.SetState(SCE_B_NUMBER);
191 			} else if (IsAWordStart(sc.ch) || (sc.ch == '[')) {
192 				sc.SetState(SCE_B_IDENTIFIER);
193 			} else if (isoperator(static_cast<char>(sc.ch)) || (sc.ch == '\\')) {	// Integer division
194 				sc.SetState(SCE_B_OPERATOR);
195 			}
196 		}
197 
198 		if (sc.atLineEnd) {
199 			visibleChars = 0;
200 		}
201 		if (!IsASpace(sc.ch)) {
202 			visibleChars++;
203 		}
204 	}
205 	sc.Complete();
206 }
207 
FoldVBDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)208 static void FoldVBDoc(unsigned int startPos, int length, int,
209 						   WordList *[], Accessor &styler) {
210 	int endPos = startPos + length;
211 
212 	// Backtrack to previous line in case need to fix its fold status
213 	int lineCurrent = styler.GetLine(startPos);
214 	if (startPos > 0) {
215 		if (lineCurrent > 0) {
216 			lineCurrent--;
217 			startPos = styler.LineStart(lineCurrent);
218 		}
219 	}
220 	int spaceFlags = 0;
221 	int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, IsVBComment);
222 	char chNext = styler[startPos];
223 	for (int i = startPos; i < endPos; i++) {
224 		char ch = chNext;
225 		chNext = styler.SafeGetCharAt(i + 1);
226 
227 		if ((ch == '\r' && chNext != '\n') || (ch == '\n') || (i == endPos)) {
228 			int lev = indentCurrent;
229 			int indentNext = styler.IndentAmount(lineCurrent + 1, &spaceFlags, IsVBComment);
230 			if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG)) {
231 				// Only non whitespace lines can be headers
232 				if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK)) {
233 					lev |= SC_FOLDLEVELHEADERFLAG;
234 				} else if (indentNext & SC_FOLDLEVELWHITEFLAG) {
235 					// Line after is blank so check the next - maybe should continue further?
236 					int spaceFlags2 = 0;
237 					int indentNext2 = styler.IndentAmount(lineCurrent + 2, &spaceFlags2, IsVBComment);
238 					if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext2 & SC_FOLDLEVELNUMBERMASK)) {
239 						lev |= SC_FOLDLEVELHEADERFLAG;
240 					}
241 				}
242 			}
243 			indentCurrent = indentNext;
244 			styler.SetLevel(lineCurrent, lev);
245 			lineCurrent++;
246 		}
247 	}
248 }
249 
ColouriseVBNetDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler)250 static void ColouriseVBNetDoc(unsigned int startPos, int length, int initStyle,
251                            WordList *keywordlists[], Accessor &styler) {
252 	ColouriseVBDoc(startPos, length, initStyle, keywordlists, styler, false);
253 }
254 
ColouriseVBScriptDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler)255 static void ColouriseVBScriptDoc(unsigned int startPos, int length, int initStyle,
256                            WordList *keywordlists[], Accessor &styler) {
257 	ColouriseVBDoc(startPos, length, initStyle, keywordlists, styler, true);
258 }
259 
260 static const char * const vbWordListDesc[] = {
261 	"Keywords",
262 	"user1",
263 	"user2",
264 	"user3",
265 	0
266 };
267 
268 LexerModule lmVB(SCLEX_VB, ColouriseVBNetDoc, "vb", FoldVBDoc, vbWordListDesc);
269 LexerModule lmVBScript(SCLEX_VBSCRIPT, ColouriseVBScriptDoc, "vbscript", FoldVBDoc, vbWordListDesc);
270 
271