57 */ |
57 */ |
58 protected JavadocTokenizer(ScannerFactory fac, char[] input, int inputLength) { |
58 protected JavadocTokenizer(ScannerFactory fac, char[] input, int inputLength) { |
59 super(fac, input, inputLength); |
59 super(fac, input, inputLength); |
60 } |
60 } |
61 |
61 |
62 /** The comment input buffer, index of next chacter to be read, |
62 @Override |
63 * index of one past last character in buffer. |
63 protected Comment processComment(int pos, int endPos, CommentStyle style) { |
|
64 char[] buf = reader.getRawCharacters(pos, endPos); |
|
65 return new JavadocComment(new ColReader(fac, buf, buf.length), style); |
|
66 } |
|
67 |
|
68 /** |
|
69 * This is a specialized version of UnicodeReader that keeps track of the |
|
70 * column position within a given character stream (used for Javadoc processing). |
64 */ |
71 */ |
65 private char[] buf; |
72 static class ColReader extends UnicodeReader { |
66 private int bp; |
73 |
67 private int buflen; |
74 int col; |
68 |
75 |
69 /** The current character. |
76 ColReader(ScannerFactory fac, char[] input, int inputLength) { |
70 */ |
77 super(fac, input, inputLength); |
71 private char ch; |
78 } |
72 |
79 |
73 /** The column number position of the current character. |
80 @Override |
74 */ |
81 protected void convertUnicode() { |
75 private int col; |
82 if (ch == '\\' && unicodeConversionBp != bp) { |
76 |
83 bp++; ch = buf[bp]; col++; |
77 /** The buffer index of the last converted Unicode character |
84 if (ch == 'u') { |
78 */ |
85 do { |
79 private int unicodeConversionBp = 0; |
86 bp++; ch = buf[bp]; col++; |
80 |
87 } while (ch == 'u'); |
81 /** |
88 int limit = bp + 3; |
82 * Buffer for doc comment. |
89 if (limit < buflen) { |
83 */ |
90 int d = digit(bp, 16); |
84 private char[] docCommentBuffer = new char[1024]; |
91 int code = d; |
85 |
92 while (bp < limit && d >= 0) { |
86 /** |
93 bp++; ch = buf[bp]; col++; |
87 * Number of characters in doc comment buffer. |
94 d = digit(bp, 16); |
88 */ |
95 code = (code << 4) + d; |
89 private int docCommentCount; |
96 } |
90 |
97 if (d >= 0) { |
91 /** |
98 ch = (char)code; |
92 * Translated and stripped contents of doc comment |
99 unicodeConversionBp = bp; |
93 */ |
100 return; |
94 private String docComment = null; |
101 } |
95 |
102 } |
96 |
103 // "illegal.Unicode.esc", reported by base scanner |
97 /** Unconditionally expand the comment buffer. |
104 } else { |
98 */ |
105 bp--; |
99 private void expandCommentBuffer() { |
106 ch = '\\'; |
100 char[] newBuffer = new char[docCommentBuffer.length * 2]; |
107 col--; |
101 System.arraycopy(docCommentBuffer, 0, newBuffer, |
108 } |
102 0, docCommentBuffer.length); |
109 } |
103 docCommentBuffer = newBuffer; |
110 } |
104 } |
111 |
105 |
112 @Override |
106 /** Convert an ASCII digit from its base (8, 10, or 16) |
113 protected void scanCommentChar() { |
107 * to its value. |
114 scanChar(); |
108 */ |
115 if (ch == '\\') { |
109 private int digit(int base) { |
116 if (peekChar() == '\\' && !isUnicode()) { |
110 char c = ch; |
117 putChar(ch, false); |
111 int result = Character.digit(c, base); |
118 bp++; col++; |
112 if (result >= 0 && c > 0x7f) { |
119 } else { |
113 ch = "0123456789abcdef".charAt(result); |
120 convertUnicode(); |
|
121 } |
|
122 } |
|
123 } |
|
124 |
|
125 @Override |
|
126 protected void scanChar() { |
|
127 bp++; |
|
128 ch = buf[bp]; |
|
129 switch (ch) { |
|
130 case '\r': // return |
|
131 col = 0; |
|
132 break; |
|
133 case '\n': // newline |
|
134 if (bp == 0 || buf[bp-1] != '\r') { |
|
135 col = 0; |
|
136 } |
|
137 break; |
|
138 case '\t': // tab |
|
139 col = (col / TabInc * TabInc) + TabInc; |
|
140 break; |
|
141 case '\\': // possible Unicode |
|
142 col++; |
|
143 convertUnicode(); |
|
144 break; |
|
145 default: |
|
146 col++; |
|
147 break; |
|
148 } |
|
149 } |
|
150 } |
|
151 |
|
152 protected class JavadocComment extends JavaTokenizer.BasicComment<ColReader> { |
|
153 |
|
154 /** |
|
155 * Translated and stripped contents of doc comment |
|
156 */ |
|
157 private String docComment = null; |
|
158 |
|
159 JavadocComment(ColReader comment_reader, CommentStyle cs) { |
|
160 super(comment_reader, cs); |
114 } |
161 } |
115 return result; |
162 |
116 } |
163 public String getText() { |
117 |
164 if (!scanned && cs == CommentStyle.JAVADOC) { |
118 /** Convert Unicode escape; bp points to initial '\' character |
165 scanDocComment(); |
119 * (Spec 3.3). |
166 } |
120 */ |
167 return docComment; |
121 private void convertUnicode() { |
168 } |
122 if (ch == '\\' && unicodeConversionBp != bp) { |
169 |
123 bp++; ch = buf[bp]; col++; |
170 @Override |
124 if (ch == 'u') { |
171 @SuppressWarnings("fallthrough") |
125 do { |
172 protected void scanDocComment() { |
126 bp++; ch = buf[bp]; col++; |
173 try { |
127 } while (ch == 'u'); |
174 boolean firstLine = true; |
128 int limit = bp + 3; |
175 |
129 if (limit < buflen) { |
176 // Skip over first slash |
130 int d = digit(16); |
177 comment_reader.scanCommentChar(); |
131 int code = d; |
178 // Skip over first star |
132 while (bp < limit && d >= 0) { |
179 comment_reader.scanCommentChar(); |
133 bp++; ch = buf[bp]; col++; |
180 |
134 d = digit(16); |
181 // consume any number of stars |
135 code = (code << 4) + d; |
182 while (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '*') { |
136 } |
183 comment_reader.scanCommentChar(); |
137 if (d >= 0) { |
184 } |
138 ch = (char)code; |
185 // is the comment in the form /**/, /***/, /****/, etc. ? |
139 unicodeConversionBp = bp; |
186 if (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '/') { |
140 return; |
187 docComment = ""; |
141 } |
188 return; |
|
189 } |
|
190 |
|
191 // skip a newline on the first line of the comment. |
|
192 if (comment_reader.bp < comment_reader.buflen) { |
|
193 if (comment_reader.ch == LF) { |
|
194 comment_reader.scanCommentChar(); |
|
195 firstLine = false; |
|
196 } else if (comment_reader.ch == CR) { |
|
197 comment_reader.scanCommentChar(); |
|
198 if (comment_reader.ch == LF) { |
|
199 comment_reader.scanCommentChar(); |
|
200 firstLine = false; |
|
201 } |
|
202 } |
|
203 } |
|
204 |
|
205 outerLoop: |
|
206 |
|
207 // The outerLoop processes the doc comment, looping once |
|
208 // for each line. For each line, it first strips off |
|
209 // whitespace, then it consumes any stars, then it |
|
210 // puts the rest of the line into our buffer. |
|
211 while (comment_reader.bp < comment_reader.buflen) { |
|
212 |
|
213 // The wsLoop consumes whitespace from the beginning |
|
214 // of each line. |
|
215 wsLoop: |
|
216 |
|
217 while (comment_reader.bp < comment_reader.buflen) { |
|
218 switch(comment_reader.ch) { |
|
219 case ' ': |
|
220 comment_reader.scanCommentChar(); |
|
221 break; |
|
222 case '\t': |
|
223 comment_reader.col = ((comment_reader.col - 1) / TabInc * TabInc) + TabInc; |
|
224 comment_reader.scanCommentChar(); |
|
225 break; |
|
226 case FF: |
|
227 comment_reader.col = 0; |
|
228 comment_reader.scanCommentChar(); |
|
229 break; |
|
230 // Treat newline at beginning of line (blank line, no star) |
|
231 // as comment text. Old Javadoc compatibility requires this. |
|
232 /*---------------------------------* |
|
233 case CR: // (Spec 3.4) |
|
234 doc_reader.scanCommentChar(); |
|
235 if (ch == LF) { |
|
236 col = 0; |
|
237 doc_reader.scanCommentChar(); |
|
238 } |
|
239 break; |
|
240 case LF: // (Spec 3.4) |
|
241 doc_reader.scanCommentChar(); |
|
242 break; |
|
243 *---------------------------------*/ |
|
244 default: |
|
245 // we've seen something that isn't whitespace; |
|
246 // jump out. |
|
247 break wsLoop; |
|
248 } |
|
249 } |
|
250 |
|
251 // Are there stars here? If so, consume them all |
|
252 // and check for the end of comment. |
|
253 if (comment_reader.ch == '*') { |
|
254 // skip all of the stars |
|
255 do { |
|
256 comment_reader.scanCommentChar(); |
|
257 } while (comment_reader.ch == '*'); |
|
258 |
|
259 // check for the closing slash. |
|
260 if (comment_reader.ch == '/') { |
|
261 // We're done with the doc comment |
|
262 // scanChar() and breakout. |
|
263 break outerLoop; |
|
264 } |
|
265 } else if (! firstLine) { |
|
266 //The current line does not begin with a '*' so we will indent it. |
|
267 for (int i = 1; i < comment_reader.col; i++) { |
|
268 comment_reader.putChar(' ', false); |
|
269 } |
|
270 } |
|
271 // The textLoop processes the rest of the characters |
|
272 // on the line, adding them to our buffer. |
|
273 textLoop: |
|
274 while (comment_reader.bp < comment_reader.buflen) { |
|
275 switch (comment_reader.ch) { |
|
276 case '*': |
|
277 // Is this just a star? Or is this the |
|
278 // end of a comment? |
|
279 comment_reader.scanCommentChar(); |
|
280 if (comment_reader.ch == '/') { |
|
281 // This is the end of the comment, |
|
282 // set ch and return our buffer. |
|
283 break outerLoop; |
|
284 } |
|
285 // This is just an ordinary star. Add it to |
|
286 // the buffer. |
|
287 comment_reader.putChar('*', false); |
|
288 break; |
|
289 case ' ': |
|
290 case '\t': |
|
291 comment_reader.putChar(comment_reader.ch, false); |
|
292 comment_reader.scanCommentChar(); |
|
293 break; |
|
294 case FF: |
|
295 comment_reader.scanCommentChar(); |
|
296 break textLoop; // treat as end of line |
|
297 case CR: // (Spec 3.4) |
|
298 comment_reader.scanCommentChar(); |
|
299 if (comment_reader.ch != LF) { |
|
300 // Canonicalize CR-only line terminator to LF |
|
301 comment_reader.putChar((char)LF, false); |
|
302 break textLoop; |
|
303 } |
|
304 /* fall through to LF case */ |
|
305 case LF: // (Spec 3.4) |
|
306 // We've seen a newline. Add it to our |
|
307 // buffer and break out of this loop, |
|
308 // starting fresh on a new line. |
|
309 comment_reader.putChar(comment_reader.ch, false); |
|
310 comment_reader.scanCommentChar(); |
|
311 break textLoop; |
|
312 default: |
|
313 // Add the character to our buffer. |
|
314 comment_reader.putChar(comment_reader.ch, false); |
|
315 comment_reader.scanCommentChar(); |
|
316 } |
|
317 } // end textLoop |
|
318 firstLine = false; |
|
319 } // end outerLoop |
|
320 |
|
321 if (comment_reader.sp > 0) { |
|
322 int i = comment_reader.sp - 1; |
|
323 trailLoop: |
|
324 while (i > -1) { |
|
325 switch (comment_reader.sbuf[i]) { |
|
326 case '*': |
|
327 i--; |
|
328 break; |
|
329 default: |
|
330 break trailLoop; |
|
331 } |
|
332 } |
|
333 comment_reader.sp = i + 1; |
|
334 |
|
335 // Store the text of the doc comment |
|
336 docComment = comment_reader.chars(); |
|
337 } else { |
|
338 docComment = ""; |
142 } |
339 } |
143 // "illegal.Unicode.esc", reported by base scanner |
340 } finally { |
144 } else { |
341 scanned = true; |
145 bp--; |
342 if (docComment != null && |
146 ch = '\\'; |
343 docComment.matches("(?sm).*^\\s*@deprecated( |$).*")) { |
147 col--; |
344 deprecatedFlag = true; |
148 } |
|
149 } |
|
150 } |
|
151 |
|
152 |
|
153 /** Read next character. |
|
154 */ |
|
155 private void scanChar() { |
|
156 bp++; |
|
157 ch = buf[bp]; |
|
158 switch (ch) { |
|
159 case '\r': // return |
|
160 col = 0; |
|
161 break; |
|
162 case '\n': // newline |
|
163 if (bp == 0 || buf[bp-1] != '\r') { |
|
164 col = 0; |
|
165 } |
|
166 break; |
|
167 case '\t': // tab |
|
168 col = (col / TabInc * TabInc) + TabInc; |
|
169 break; |
|
170 case '\\': // possible Unicode |
|
171 col++; |
|
172 convertUnicode(); |
|
173 break; |
|
174 default: |
|
175 col++; |
|
176 break; |
|
177 } |
|
178 } |
|
179 |
|
180 @Override |
|
181 public Token readToken() { |
|
182 docComment = null; |
|
183 Token tk = super.readToken(); |
|
184 tk.docComment = docComment; |
|
185 return tk; |
|
186 } |
|
187 |
|
188 /** |
|
189 * Read next character in doc comment, skipping over double '\' characters. |
|
190 * If a double '\' is skipped, put in the buffer and update buffer count. |
|
191 */ |
|
192 private void scanDocCommentChar() { |
|
193 scanChar(); |
|
194 if (ch == '\\') { |
|
195 if (buf[bp+1] == '\\' && unicodeConversionBp != bp) { |
|
196 if (docCommentCount == docCommentBuffer.length) |
|
197 expandCommentBuffer(); |
|
198 docCommentBuffer[docCommentCount++] = ch; |
|
199 bp++; col++; |
|
200 } else { |
|
201 convertUnicode(); |
|
202 } |
|
203 } |
|
204 } |
|
205 |
|
206 /** |
|
207 * Process a doc comment and make the string content available. |
|
208 * Strips leading whitespace and stars. |
|
209 */ |
|
210 @SuppressWarnings("fallthrough") |
|
211 protected void processComment(int pos, int endPos, CommentStyle style) { |
|
212 if (style != CommentStyle.JAVADOC) { |
|
213 return; |
|
214 } |
|
215 |
|
216 buf = reader.getRawCharacters(pos, endPos); |
|
217 buflen = buf.length; |
|
218 bp = 0; |
|
219 col = 0; |
|
220 |
|
221 docCommentCount = 0; |
|
222 |
|
223 boolean firstLine = true; |
|
224 |
|
225 // Skip over first slash |
|
226 scanDocCommentChar(); |
|
227 // Skip over first star |
|
228 scanDocCommentChar(); |
|
229 |
|
230 // consume any number of stars |
|
231 while (bp < buflen && ch == '*') { |
|
232 scanDocCommentChar(); |
|
233 } |
|
234 // is the comment in the form /**/, /***/, /****/, etc. ? |
|
235 if (bp < buflen && ch == '/') { |
|
236 docComment = ""; |
|
237 return; |
|
238 } |
|
239 |
|
240 // skip a newline on the first line of the comment. |
|
241 if (bp < buflen) { |
|
242 if (ch == LF) { |
|
243 scanDocCommentChar(); |
|
244 firstLine = false; |
|
245 } else if (ch == CR) { |
|
246 scanDocCommentChar(); |
|
247 if (ch == LF) { |
|
248 scanDocCommentChar(); |
|
249 firstLine = false; |
|
250 } |
345 } |
251 } |
346 } |
252 } |
347 } |
253 |
348 } |
254 outerLoop: |
349 |
255 |
350 @Override |
256 // The outerLoop processes the doc comment, looping once |
|
257 // for each line. For each line, it first strips off |
|
258 // whitespace, then it consumes any stars, then it |
|
259 // puts the rest of the line into our buffer. |
|
260 while (bp < buflen) { |
|
261 |
|
262 // The wsLoop consumes whitespace from the beginning |
|
263 // of each line. |
|
264 wsLoop: |
|
265 |
|
266 while (bp < buflen) { |
|
267 switch(ch) { |
|
268 case ' ': |
|
269 scanDocCommentChar(); |
|
270 break; |
|
271 case '\t': |
|
272 col = ((col - 1) / TabInc * TabInc) + TabInc; |
|
273 scanDocCommentChar(); |
|
274 break; |
|
275 case FF: |
|
276 col = 0; |
|
277 scanDocCommentChar(); |
|
278 break; |
|
279 // Treat newline at beginning of line (blank line, no star) |
|
280 // as comment text. Old Javadoc compatibility requires this. |
|
281 /*---------------------------------* |
|
282 case CR: // (Spec 3.4) |
|
283 scanDocCommentChar(); |
|
284 if (ch == LF) { |
|
285 col = 0; |
|
286 scanDocCommentChar(); |
|
287 } |
|
288 break; |
|
289 case LF: // (Spec 3.4) |
|
290 scanDocCommentChar(); |
|
291 break; |
|
292 *---------------------------------*/ |
|
293 default: |
|
294 // we've seen something that isn't whitespace; |
|
295 // jump out. |
|
296 break wsLoop; |
|
297 } |
|
298 } |
|
299 |
|
300 // Are there stars here? If so, consume them all |
|
301 // and check for the end of comment. |
|
302 if (ch == '*') { |
|
303 // skip all of the stars |
|
304 do { |
|
305 scanDocCommentChar(); |
|
306 } while (ch == '*'); |
|
307 |
|
308 // check for the closing slash. |
|
309 if (ch == '/') { |
|
310 // We're done with the doc comment |
|
311 // scanChar() and breakout. |
|
312 break outerLoop; |
|
313 } |
|
314 } else if (! firstLine) { |
|
315 //The current line does not begin with a '*' so we will indent it. |
|
316 for (int i = 1; i < col; i++) { |
|
317 if (docCommentCount == docCommentBuffer.length) |
|
318 expandCommentBuffer(); |
|
319 docCommentBuffer[docCommentCount++] = ' '; |
|
320 } |
|
321 } |
|
322 |
|
323 // The textLoop processes the rest of the characters |
|
324 // on the line, adding them to our buffer. |
|
325 textLoop: |
|
326 while (bp < buflen) { |
|
327 switch (ch) { |
|
328 case '*': |
|
329 // Is this just a star? Or is this the |
|
330 // end of a comment? |
|
331 scanDocCommentChar(); |
|
332 if (ch == '/') { |
|
333 // This is the end of the comment, |
|
334 // set ch and return our buffer. |
|
335 break outerLoop; |
|
336 } |
|
337 // This is just an ordinary star. Add it to |
|
338 // the buffer. |
|
339 if (docCommentCount == docCommentBuffer.length) |
|
340 expandCommentBuffer(); |
|
341 docCommentBuffer[docCommentCount++] = '*'; |
|
342 break; |
|
343 case ' ': |
|
344 case '\t': |
|
345 if (docCommentCount == docCommentBuffer.length) |
|
346 expandCommentBuffer(); |
|
347 docCommentBuffer[docCommentCount++] = ch; |
|
348 scanDocCommentChar(); |
|
349 break; |
|
350 case FF: |
|
351 scanDocCommentChar(); |
|
352 break textLoop; // treat as end of line |
|
353 case CR: // (Spec 3.4) |
|
354 scanDocCommentChar(); |
|
355 if (ch != LF) { |
|
356 // Canonicalize CR-only line terminator to LF |
|
357 if (docCommentCount == docCommentBuffer.length) |
|
358 expandCommentBuffer(); |
|
359 docCommentBuffer[docCommentCount++] = (char)LF; |
|
360 break textLoop; |
|
361 } |
|
362 /* fall through to LF case */ |
|
363 case LF: // (Spec 3.4) |
|
364 // We've seen a newline. Add it to our |
|
365 // buffer and break out of this loop, |
|
366 // starting fresh on a new line. |
|
367 if (docCommentCount == docCommentBuffer.length) |
|
368 expandCommentBuffer(); |
|
369 docCommentBuffer[docCommentCount++] = ch; |
|
370 scanDocCommentChar(); |
|
371 break textLoop; |
|
372 default: |
|
373 // Add the character to our buffer. |
|
374 if (docCommentCount == docCommentBuffer.length) |
|
375 expandCommentBuffer(); |
|
376 docCommentBuffer[docCommentCount++] = ch; |
|
377 scanDocCommentChar(); |
|
378 } |
|
379 } // end textLoop |
|
380 firstLine = false; |
|
381 } // end outerLoop |
|
382 |
|
383 if (docCommentCount > 0) { |
|
384 int i = docCommentCount - 1; |
|
385 trailLoop: |
|
386 while (i > -1) { |
|
387 switch (docCommentBuffer[i]) { |
|
388 case '*': |
|
389 i--; |
|
390 break; |
|
391 default: |
|
392 break trailLoop; |
|
393 } |
|
394 } |
|
395 docCommentCount = i + 1; |
|
396 |
|
397 // Store the text of the doc comment |
|
398 docComment = new String(docCommentBuffer, 0 , docCommentCount); |
|
399 } else { |
|
400 docComment = ""; |
|
401 } |
|
402 } |
|
403 |
|
404 /** Build a map for translating between line numbers and |
|
405 * positions in the input. |
|
406 * |
|
407 * @return a LineMap */ |
|
408 public Position.LineMap getLineMap() { |
351 public Position.LineMap getLineMap() { |
409 char[] buf = reader.getRawCharacters(); |
352 char[] buf = reader.getRawCharacters(); |
410 return Position.makeLineMap(buf, buf.length, true); |
353 return Position.makeLineMap(buf, buf.length, true); |
411 } |
354 } |
412 } |
355 } |