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 // 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 <stdio.h>
11 #include <stdarg.h>
12 #include <assert.h>
13 #include <ctype.h>
14 
15 #include "ILexer.h"
16 #include "Scintilla.h"
17 #include "SciLexer.h"
18 
19 #include "WordList.h"
20 #include "LexAccessor.h"
21 #include "Accessor.h"
22 #include "StyleContext.h"
23 #include "CharacterSet.h"
24 #include "LexerModule.h"
25 
26 #ifdef SCI_NAMESPACE
27 using namespace Scintilla;
28 #endif
29 
30 static const char * const yamlWordListDesc[] = {
31 	"Keywords",
32 	0
33 };
34 
AtEOL(Accessor & styler,unsigned int i)35 static inline bool AtEOL(Accessor &styler, unsigned int i) {
36 	return (styler[i] == '\n') ||
37 		((styler[i] == '\r') && (styler.SafeGetCharAt(i + 1) != '\n'));
38 }
39 
SpaceCount(char * lineBuffer)40 static unsigned int SpaceCount(char* lineBuffer) {
41 	if (lineBuffer == NULL)
42 		return 0;
43 
44 	char* headBuffer = lineBuffer;
45 
46 	while (*headBuffer == ' ')
47 		headBuffer++;
48 
49 	return headBuffer - lineBuffer;
50 }
51 
52 #define YAML_STATE_BITSIZE 16
53 #define YAML_STATE_MASK			(0xFFFF0000)
54 #define YAML_STATE_DOCUMENT		(1 << YAML_STATE_BITSIZE)
55 #define YAML_STATE_VALUE			(2 << YAML_STATE_BITSIZE)
56 #define YAML_STATE_COMMENT		(3 << YAML_STATE_BITSIZE)
57 #define YAML_STATE_TEXT_PARENT	(4 << YAML_STATE_BITSIZE)
58 #define YAML_STATE_TEXT			(5 << YAML_STATE_BITSIZE)
59 
ColouriseYAMLLine(char * lineBuffer,unsigned int currentLine,unsigned int lengthLine,unsigned int startLine,unsigned int endPos,WordList & keywords,Accessor & styler)60 static void ColouriseYAMLLine(
61 	char *lineBuffer,
62 	unsigned int currentLine,
63 	unsigned int lengthLine,
64 	unsigned int startLine,
65 	unsigned int endPos,
66 	WordList &keywords,
67 	Accessor &styler) {
68 
69 	unsigned int i = 0;
70 	bool bInQuotes = false;
71 	unsigned int indentAmount = SpaceCount(lineBuffer);
72 
73 	if (currentLine > 0) {
74 		int parentLineState = styler.GetLineState(currentLine - 1);
75 
76 		if ((parentLineState&YAML_STATE_MASK) == YAML_STATE_TEXT || (parentLineState&YAML_STATE_MASK) == YAML_STATE_TEXT_PARENT) {
77 			unsigned int parentIndentAmount = parentLineState&(~YAML_STATE_MASK);
78 			if (indentAmount > parentIndentAmount) {
79 				styler.SetLineState(currentLine, YAML_STATE_TEXT | parentIndentAmount);
80 				styler.ColourTo(endPos, SCE_YAML_TEXT);
81 				return;
82 			}
83 		}
84 	}
85 	styler.SetLineState(currentLine, 0);
86 	if (strncmp(lineBuffer, "---", 3) == 0) {	// Document marker
87 		styler.SetLineState(currentLine, YAML_STATE_DOCUMENT);
88 		styler.ColourTo(endPos, SCE_YAML_DOCUMENT);
89 		return;
90 	}
91 	// Skip initial spaces
92 	while ((i < lengthLine) && lineBuffer[i] == ' ') { // YAML always uses space, never TABS or anything else
93 		i++;
94 	}
95 	if (lineBuffer[i] == '\t') { // if we skipped all spaces, and we are NOT inside a text block, this is wrong
96 		styler.ColourTo(endPos, SCE_YAML_ERROR);
97 		return;
98 	}
99 	if (lineBuffer[i] == '#') {	// Comment
100 		styler.SetLineState(currentLine, YAML_STATE_COMMENT);
101 		styler.ColourTo(endPos, SCE_YAML_COMMENT);
102 		return;
103 	}
104 	while (i < lengthLine) {
105 		if (lineBuffer[i] == '\'' || lineBuffer[i] == '\"') {
106 			bInQuotes = !bInQuotes;
107 		} else if (lineBuffer[i] == ':' && !bInQuotes) {
108 			styler.ColourTo(startLine + i - 1, SCE_YAML_IDENTIFIER);
109 			styler.ColourTo(startLine + i, SCE_YAML_OPERATOR);
110 			// Non-folding scalar
111 			i++;
112 			while ((i < lengthLine) && isspacechar(lineBuffer[i]))
113 				i++;
114 			unsigned int endValue = lengthLine - 1;
115 			while ((endValue >= i) && isspacechar(lineBuffer[endValue]))
116 				endValue--;
117 			lineBuffer[endValue + 1] = '\0';
118 			if (lineBuffer[i] == '|' || lineBuffer[i] == '>') {
119 				i++;
120 				if (lineBuffer[i] == '+' || lineBuffer[i] == '-')
121 					i++;
122 				while ((i < lengthLine) && isspacechar(lineBuffer[i]))
123 					i++;
124 				if (lineBuffer[i] == '\0') {
125 					styler.SetLineState(currentLine, YAML_STATE_TEXT_PARENT | indentAmount);
126 					styler.ColourTo(endPos, SCE_YAML_DEFAULT);
127 					return;
128 				} else if (lineBuffer[i] == '#') {
129 					styler.SetLineState(currentLine, YAML_STATE_TEXT_PARENT | indentAmount);
130 					styler.ColourTo(startLine + i - 1, SCE_YAML_DEFAULT);
131 					styler.ColourTo(endPos, SCE_YAML_COMMENT);
132 					return;
133 				} else {
134 					styler.ColourTo(endPos, SCE_YAML_ERROR);
135 					return;
136 				}
137 			} else if (lineBuffer[i] == '#') {
138 				styler.ColourTo(startLine + i - 1, SCE_YAML_DEFAULT);
139 				styler.ColourTo(endPos, SCE_YAML_COMMENT);
140 				return;
141 			}
142 			styler.SetLineState(currentLine, YAML_STATE_VALUE);
143 			if (lineBuffer[i] == '&' || lineBuffer[i] == '*') {
144 				styler.ColourTo(endPos, SCE_YAML_REFERENCE);
145 				return;
146 			}
147 			if (keywords.InList(&lineBuffer[i])) { // Convertible value (true/false, etc.)
148 				styler.ColourTo(endPos, SCE_YAML_KEYWORD);
149 				return;
150 			} else {
151 				unsigned int i2 = i;
152 				while ((i < lengthLine) && lineBuffer[i]) {
153 					if (!(IsASCII(lineBuffer[i]) && isdigit(lineBuffer[i])) && lineBuffer[i] != '-' && lineBuffer[i] != '.' && lineBuffer[i] != ',') {
154 						styler.ColourTo(endPos, SCE_YAML_DEFAULT);
155 						return;
156 					}
157 					i++;
158 				}
159 				if (i > i2) {
160 					styler.ColourTo(endPos, SCE_YAML_NUMBER);
161 					return;
162 				}
163 			}
164 			break; // shouldn't get here, but just in case, the rest of the line is coloured the default
165 		}
166 		i++;
167 	}
168 	styler.ColourTo(endPos, SCE_YAML_DEFAULT);
169 }
170 
ColouriseYAMLDoc(unsigned int startPos,int length,int,WordList * keywordLists[],Accessor & styler)171 static void ColouriseYAMLDoc(unsigned int startPos, int length, int, WordList *keywordLists[], Accessor &styler) {
172 	char lineBuffer[1024] = "";
173 	styler.StartAt(startPos);
174 	styler.StartSegment(startPos);
175 	unsigned int linePos = 0;
176 	unsigned int startLine = startPos;
177 	unsigned int endPos = startPos + length;
178 	unsigned int maxPos = styler.Length();
179 	unsigned int lineCurrent = styler.GetLine(startPos);
180 
181 	for (unsigned int i = startPos; i < maxPos && i < endPos; i++) {
182 		lineBuffer[linePos++] = styler[i];
183 		if (AtEOL(styler, i) || (linePos >= sizeof(lineBuffer) - 1)) {
184 			// End of line (or of line buffer) met, colourise it
185 			lineBuffer[linePos] = '\0';
186 			ColouriseYAMLLine(lineBuffer, lineCurrent, linePos, startLine, i, *keywordLists[0], styler);
187 			linePos = 0;
188 			startLine = i + 1;
189 			lineCurrent++;
190 		}
191 	}
192 	if (linePos > 0) {	// Last line does not have ending characters
193 		ColouriseYAMLLine(lineBuffer, lineCurrent, linePos, startLine, startPos + length - 1, *keywordLists[0], styler);
194 	}
195 }
196 
IsCommentLine(int line,Accessor & styler)197 static bool IsCommentLine(int line, Accessor &styler) {
198 	int pos = styler.LineStart(line);
199 	if (styler[pos] == '#')
200 		return true;
201 	return false;
202 }
203 
FoldYAMLDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)204 static void FoldYAMLDoc(unsigned int startPos, int length, int /*initStyle - unused*/,
205                       WordList *[], Accessor &styler) {
206 	const int maxPos = startPos + length;
207 	const int maxLines = styler.GetLine(maxPos - 1);             // Requested last line
208 	const int docLines = styler.GetLine(styler.Length() - 1);  // Available last line
209 	const bool foldComment = styler.GetPropertyInt("fold.comment.yaml") != 0;
210 
211 	// Backtrack to previous non-blank line so we can determine indent level
212 	// for any white space lines
213 	// and so we can fix any preceding fold level (which is why we go back
214 	// at least one line in all cases)
215 	int spaceFlags = 0;
216 	int lineCurrent = styler.GetLine(startPos);
217 	int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
218 	while (lineCurrent > 0) {
219 		lineCurrent--;
220 		indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
221 		if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG) &&
222 		        (!IsCommentLine(lineCurrent, styler)))
223 			break;
224 	}
225 	int indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
226 
227 	// Set up initial loop state
228 	int prevComment = 0;
229 	if (lineCurrent >= 1)
230 		prevComment = foldComment && IsCommentLine(lineCurrent - 1, styler);
231 
232 	// Process all characters to end of requested range
233 	// or comment that hangs over the end of the range.  Cap processing in all cases
234 	// to end of document (in case of unclosed comment at end).
235 	while ((lineCurrent <= docLines) && ((lineCurrent <= maxLines) || prevComment)) {
236 
237 		// Gather info
238 		int lev = indentCurrent;
239 		int lineNext = lineCurrent + 1;
240 		int indentNext = indentCurrent;
241 		if (lineNext <= docLines) {
242 			// Information about next line is only available if not at end of document
243 			indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
244 	}
245 		const int comment = foldComment && IsCommentLine(lineCurrent, styler);
246 		const int comment_start = (comment && !prevComment && (lineNext <= docLines) &&
247 		                           IsCommentLine(lineNext, styler) && (lev > SC_FOLDLEVELBASE));
248 		const int comment_continue = (comment && prevComment);
249 		if (!comment)
250 			indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
251 		if (indentNext & SC_FOLDLEVELWHITEFLAG)
252 			indentNext = SC_FOLDLEVELWHITEFLAG | indentCurrentLevel;
253 
254 		if (comment_start) {
255 			// Place fold point at start of a block of comments
256 			lev |= SC_FOLDLEVELHEADERFLAG;
257 		} else if (comment_continue) {
258 			// Add level to rest of lines in the block
259 			lev = lev + 1;
260 		}
261 
262 		// Skip past any blank lines for next indent level info; we skip also
263 		// comments (all comments, not just those starting in column 0)
264 		// which effectively folds them into surrounding code rather
265 		// than screwing up folding.
266 
267 		while ((lineNext < docLines) &&
268 		        ((indentNext & SC_FOLDLEVELWHITEFLAG) ||
269 		         (lineNext <= docLines && IsCommentLine(lineNext, styler)))) {
270 
271 			lineNext++;
272 			indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
273 		}
274 
275 		const int levelAfterComments = indentNext & SC_FOLDLEVELNUMBERMASK;
276 		const int levelBeforeComments = Maximum(indentCurrentLevel,levelAfterComments);
277 
278 		// Now set all the indent levels on the lines we skipped
279 		// Do this from end to start.  Once we encounter one line
280 		// which is indented more than the line after the end of
281 		// the comment-block, use the level of the block before
282 
283 		int skipLine = lineNext;
284 		int skipLevel = levelAfterComments;
285 
286 		while (--skipLine > lineCurrent) {
287 			int skipLineIndent = styler.IndentAmount(skipLine, &spaceFlags, NULL);
288 
289 			if ((skipLineIndent & SC_FOLDLEVELNUMBERMASK) > levelAfterComments)
290 				skipLevel = levelBeforeComments;
291 
292 			int whiteFlag = skipLineIndent & SC_FOLDLEVELWHITEFLAG;
293 
294 			styler.SetLevel(skipLine, skipLevel | whiteFlag);
295 		}
296 
297 		// Set fold header on non-comment line
298 		if (!comment && !(indentCurrent & SC_FOLDLEVELWHITEFLAG) ) {
299 			if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK))
300 				lev |= SC_FOLDLEVELHEADERFLAG;
301 		}
302 
303 		// Keep track of block comment state of previous line
304 		prevComment = comment_start || comment_continue;
305 
306 		// Set fold level for this line and move to next line
307 		styler.SetLevel(lineCurrent, lev);
308 		indentCurrent = indentNext;
309 		lineCurrent = lineNext;
310 	}
311 
312 	// NOTE: Cannot set level of last line here because indentCurrent doesn't have
313 	// header flag set; the loop above is crafted to take care of this case!
314 	//styler.SetLevel(lineCurrent, indentCurrent);
315 }
316 
317 LexerModule lmYAML(SCLEX_YAML, ColouriseYAMLDoc, "yaml", FoldYAMLDoc, yamlWordListDesc);
318