1 // Scintilla source code edit control
2 /** @file LexYAML.cxx
3  ** Lexer for YAML.
4  **/
5 // Copyright 2003- by Sean O'Dell <sean@celsoft.com>
6 // Release under the same license as Scintilla/SciTE.
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 static const char * const yamlWordListDesc[] = {
24 	"Keywords",
25 	0
26 };
27 
AtEOL(Accessor & styler,unsigned int i)28 static inline bool AtEOL(Accessor &styler, unsigned int i) {
29 	return (styler[i] == '\n') ||
30 		((styler[i] == '\r') && (styler.SafeGetCharAt(i + 1) != '\n'));
31 }
32 
SpaceCount(char * lineBuffer)33 static unsigned int SpaceCount(char* lineBuffer) {
34 	if (lineBuffer == NULL)
35 		return 0;
36 
37 	char* headBuffer = lineBuffer;
38 
39 	while (*headBuffer == ' ')
40 		headBuffer++;
41 
42 	return headBuffer - lineBuffer;
43 }
44 
45 #define YAML_STATE_BITSIZE 16
46 #define YAML_STATE_MASK			(0xFFFF0000)
47 #define YAML_STATE_DOCUMENT		(1 << YAML_STATE_BITSIZE)
48 #define YAML_STATE_VALUE			(2 << YAML_STATE_BITSIZE)
49 #define YAML_STATE_COMMENT		(3 << YAML_STATE_BITSIZE)
50 #define YAML_STATE_TEXT_PARENT	(4 << YAML_STATE_BITSIZE)
51 #define YAML_STATE_TEXT			(5 << YAML_STATE_BITSIZE)
52 
ColouriseYAMLLine(char * lineBuffer,unsigned int currentLine,unsigned int lengthLine,unsigned int startLine,unsigned int endPos,WordList & keywords,Accessor & styler)53 static void ColouriseYAMLLine(
54 	char *lineBuffer,
55 	unsigned int currentLine,
56 	unsigned int lengthLine,
57 	unsigned int startLine,
58 	unsigned int endPos,
59 	WordList &keywords,
60 	Accessor &styler) {
61 
62 	unsigned int i = 0;
63 	bool bInQuotes = false;
64 	unsigned int indentAmount = SpaceCount(lineBuffer);
65 
66 	if (currentLine > 0) {
67 		int parentLineState = styler.GetLineState(currentLine - 1);
68 
69 		if ((parentLineState&YAML_STATE_MASK) == YAML_STATE_TEXT || (parentLineState&YAML_STATE_MASK) == YAML_STATE_TEXT_PARENT) {
70 			unsigned int parentIndentAmount = parentLineState&(~YAML_STATE_MASK);
71 			if (indentAmount > parentIndentAmount) {
72 				styler.SetLineState(currentLine, YAML_STATE_TEXT | parentIndentAmount);
73 				styler.ColourTo(endPos, SCE_YAML_TEXT);
74 				return;
75 			}
76 		}
77 	}
78 	styler.SetLineState(currentLine, 0);
79 	if (strncmp(lineBuffer, "---", 3) == 0) {	// Document marker
80 		styler.SetLineState(currentLine, YAML_STATE_DOCUMENT);
81 		styler.ColourTo(endPos, SCE_YAML_DOCUMENT);
82 		return;
83 	}
84 	// Skip initial spaces
85 	while ((i < lengthLine) && lineBuffer[i] == ' ') { // YAML always uses space, never TABS or anything else
86 		i++;
87 	}
88 	if (lineBuffer[i] == '\t') { // if we skipped all spaces, and we are NOT inside a text block, this is wrong
89 		styler.ColourTo(endPos, SCE_YAML_ERROR);
90 		return;
91 	}
92 	if (lineBuffer[i] == '#') {	// Comment
93 		styler.SetLineState(currentLine, YAML_STATE_COMMENT);
94 		styler.ColourTo(endPos, SCE_YAML_COMMENT);
95 		return;
96 	}
97 	while (i < lengthLine) {
98 		if (lineBuffer[i] == '\'' || lineBuffer[i] == '\"') {
99 			bInQuotes = !bInQuotes;
100 		} else if (lineBuffer[i] == ':' && !bInQuotes) {
101 			styler.ColourTo(startLine + i, SCE_YAML_IDENTIFIER);
102 			// Non-folding scalar
103 			i++;
104 			while ((i < lengthLine) && isspacechar(lineBuffer[i]))
105 				i++;
106 			unsigned int endValue = lengthLine - 1;
107 			while ((endValue >= i) && isspacechar(lineBuffer[endValue]))
108 				endValue--;
109 			lineBuffer[endValue + 1] = '\0';
110 			if (lineBuffer[i] == '|' || lineBuffer[i] == '>') {
111 				i++;
112 				if (lineBuffer[i] == '+' || lineBuffer[i] == '-')
113 					i++;
114 				while ((i < lengthLine) && isspacechar(lineBuffer[i]))
115 					i++;
116 				if (lineBuffer[i] == '\0') {
117 					styler.SetLineState(currentLine, YAML_STATE_TEXT_PARENT | indentAmount);
118 					styler.ColourTo(endPos, SCE_YAML_DEFAULT);
119 					return;
120 				} else if (lineBuffer[i] == '#') {
121 					styler.SetLineState(currentLine, YAML_STATE_TEXT_PARENT | indentAmount);
122 					styler.ColourTo(startLine + i - 1, SCE_YAML_DEFAULT);
123 					styler.ColourTo(endPos, SCE_YAML_COMMENT);
124 					return;
125 				} else {
126 					styler.ColourTo(endPos, SCE_YAML_ERROR);
127 					return;
128 				}
129 			}
130 			styler.SetLineState(currentLine, YAML_STATE_VALUE);
131 			if (lineBuffer[i] == '&' || lineBuffer[i] == '*') {
132 				styler.ColourTo(endPos, SCE_YAML_REFERENCE);
133 				return;
134 			}
135 			if (keywords.InList(&lineBuffer[i])) { // Convertible value (true/false, etc.)
136 				styler.ColourTo(endPos, SCE_YAML_KEYWORD);
137 				return;
138 			} else {
139 				unsigned int i2 = i;
140 				while ((i < lengthLine) && lineBuffer[i]) {
141 					if (!isdigit(lineBuffer[i]) && lineBuffer[i] != '-' && lineBuffer[i] != '.' && lineBuffer[i] != ',') {
142 						styler.ColourTo(endPos, SCE_YAML_DEFAULT);
143 						return;
144 					}
145 					i++;
146 				}
147 				if (i > i2) {
148 					styler.ColourTo(endPos, SCE_YAML_NUMBER);
149 					return;
150 				}
151 			}
152 			break; // shouldn't get here, but just in case, the rest of the line is coloured the default
153 		}
154 		i++;
155 	}
156 	styler.ColourTo(endPos, SCE_YAML_DEFAULT);
157 }
158 
ColouriseYAMLDoc(unsigned int startPos,int length,int,WordList * keywordLists[],Accessor & styler)159 static void ColouriseYAMLDoc(unsigned int startPos, int length, int, WordList *keywordLists[], Accessor &styler) {
160 	char lineBuffer[1024];
161 	styler.StartAt(startPos);
162 	styler.StartSegment(startPos);
163 	unsigned int linePos = 0;
164 	unsigned int startLine = startPos;
165 	unsigned int endPos = startPos + length;
166 	unsigned int maxPos = styler.Length();
167 	unsigned int lineCurrent = styler.GetLine(startPos);
168 
169 	for (unsigned int i = startPos; i < maxPos && i < endPos; i++) {
170 		lineBuffer[linePos++] = styler[i];
171 		if (AtEOL(styler, i) || (linePos >= sizeof(lineBuffer) - 1)) {
172 			// End of line (or of line buffer) met, colourise it
173 			lineBuffer[linePos] = '\0';
174 			ColouriseYAMLLine(lineBuffer, lineCurrent, linePos, startLine, i, *keywordLists[0], styler);
175 			linePos = 0;
176 			startLine = i + 1;
177 			lineCurrent++;
178 		}
179 	}
180 	if (linePos > 0) {	// Last line does not have ending characters
181 		ColouriseYAMLLine(lineBuffer, lineCurrent, linePos, startLine, startPos + length - 1, *keywordLists[0], styler);
182 	}
183 }
184 
IsCommentLine(int line,Accessor & styler)185 static bool IsCommentLine(int line, Accessor &styler) {
186 	int pos = styler.LineStart(line);
187 	if (styler[pos] == '#')
188 		return true;
189 	return false;
190 }
191 
FoldYAMLDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)192 static void FoldYAMLDoc(unsigned int startPos, int length, int /*initStyle - unused*/,
193                       WordList *[], Accessor &styler) {
194 	const int maxPos = startPos + length;
195 	const int maxLines = styler.GetLine(maxPos - 1);             // Requested last line
196 	const int docLines = styler.GetLine(styler.Length() - 1);  // Available last line
197 	const bool foldComment = styler.GetPropertyInt("fold.comment.yaml") != 0;
198 
199 	// Backtrack to previous non-blank line so we can determine indent level
200 	// for any white space lines
201 	// and so we can fix any preceding fold level (which is why we go back
202 	// at least one line in all cases)
203 	int spaceFlags = 0;
204 	int lineCurrent = styler.GetLine(startPos);
205 	int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
206 	while (lineCurrent > 0) {
207 		lineCurrent--;
208 		indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
209 		if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG) &&
210 		        (!IsCommentLine(lineCurrent, styler)))
211 			break;
212 	}
213 	int indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
214 
215 	// Set up initial loop state
216 	int prevComment = 0;
217 	if (lineCurrent >= 1)
218 		prevComment = foldComment && IsCommentLine(lineCurrent - 1, styler);
219 
220 	// Process all characters to end of requested range
221 	// or comment that hangs over the end of the range.  Cap processing in all cases
222 	// to end of document (in case of unclosed comment at end).
223 	while ((lineCurrent <= docLines) && ((lineCurrent <= maxLines) || prevComment)) {
224 
225 		// Gather info
226 		int lev = indentCurrent;
227 		int lineNext = lineCurrent + 1;
228 		int indentNext = indentCurrent;
229 		if (lineNext <= docLines) {
230 			// Information about next line is only available if not at end of document
231 			indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
232 	}
233 		const int comment = foldComment && IsCommentLine(lineCurrent, styler);
234 		const int comment_start = (comment && !prevComment && (lineNext <= docLines) &&
235 		                           IsCommentLine(lineNext, styler) && (lev > SC_FOLDLEVELBASE));
236 		const int comment_continue = (comment && prevComment);
237 		if (!comment)
238 			indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
239 		if (indentNext & SC_FOLDLEVELWHITEFLAG)
240 			indentNext = SC_FOLDLEVELWHITEFLAG | indentCurrentLevel;
241 
242 		if (comment_start) {
243 			// Place fold point at start of a block of comments
244 			lev |= SC_FOLDLEVELHEADERFLAG;
245 		} else if (comment_continue) {
246 			// Add level to rest of lines in the block
247 			lev = lev + 1;
248 		}
249 
250 		// Skip past any blank lines for next indent level info; we skip also
251 		// comments (all comments, not just those starting in column 0)
252 		// which effectively folds them into surrounding code rather
253 		// than screwing up folding.
254 
255 		while ((lineNext < docLines) &&
256 		        ((indentNext & SC_FOLDLEVELWHITEFLAG) ||
257 		         (lineNext <= docLines && IsCommentLine(lineNext, styler)))) {
258 
259 			lineNext++;
260 			indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
261 		}
262 
263 		const int levelAfterComments = indentNext & SC_FOLDLEVELNUMBERMASK;
264 		const int levelBeforeComments = Platform::Maximum(indentCurrentLevel,levelAfterComments);
265 
266 		// Now set all the indent levels on the lines we skipped
267 		// Do this from end to start.  Once we encounter one line
268 		// which is indented more than the line after the end of
269 		// the comment-block, use the level of the block before
270 
271 		int skipLine = lineNext;
272 		int skipLevel = levelAfterComments;
273 
274 		while (--skipLine > lineCurrent) {
275 			int skipLineIndent = styler.IndentAmount(skipLine, &spaceFlags, NULL);
276 
277 			if ((skipLineIndent & SC_FOLDLEVELNUMBERMASK) > levelAfterComments)
278 				skipLevel = levelBeforeComments;
279 
280 			int whiteFlag = skipLineIndent & SC_FOLDLEVELWHITEFLAG;
281 
282 			styler.SetLevel(skipLine, skipLevel | whiteFlag);
283 		}
284 
285 		// Set fold header on non-comment line
286 		if (!comment && !(indentCurrent & SC_FOLDLEVELWHITEFLAG) ) {
287 			if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK))
288 				lev |= SC_FOLDLEVELHEADERFLAG;
289 		}
290 
291 		// Keep track of block comment state of previous line
292 		prevComment = comment_start || comment_continue;
293 
294 		// Set fold level for this line and move to next line
295 		styler.SetLevel(lineCurrent, lev);
296 		indentCurrent = indentNext;
297 		lineCurrent = lineNext;
298 	}
299 
300 	// NOTE: Cannot set level of last line here because indentCurrent doesn't have
301 	// header flag set; the loop above is crafted to take care of this case!
302 	//styler.SetLevel(lineCurrent, indentCurrent);
303 }
304 
305 LexerModule lmYAML(SCLEX_YAML, ColouriseYAMLDoc, "yaml", FoldYAMLDoc, yamlWordListDesc);
306