1.1 --- a/src/share/classes/com/sun/tools/javac/parser/JavadocTokenizer.java Tue Nov 01 15:49:45 2011 -0700
1.2 +++ b/src/share/classes/com/sun/tools/javac/parser/JavadocTokenizer.java Fri Nov 04 12:36:40 2011 +0000
1.3 @@ -25,8 +25,8 @@
1.4
1.5 package com.sun.tools.javac.parser;
1.6
1.7 -import com.sun.tools.javac.file.JavacFileManager;
1.8 -import com.sun.tools.javac.parser.Tokens.Token;
1.9 +import com.sun.tools.javac.parser.Tokens.Comment;
1.10 +import com.sun.tools.javac.parser.Tokens.Comment.CommentStyle;
1.11 import com.sun.tools.javac.util.*;
1.12
1.13 import java.nio.*;
1.14 @@ -59,352 +59,295 @@
1.15 super(fac, input, inputLength);
1.16 }
1.17
1.18 - /** The comment input buffer, index of next chacter to be read,
1.19 - * index of one past last character in buffer.
1.20 - */
1.21 - private char[] buf;
1.22 - private int bp;
1.23 - private int buflen;
1.24 -
1.25 - /** The current character.
1.26 - */
1.27 - private char ch;
1.28 -
1.29 - /** The column number position of the current character.
1.30 - */
1.31 - private int col;
1.32 -
1.33 - /** The buffer index of the last converted Unicode character
1.34 - */
1.35 - private int unicodeConversionBp = 0;
1.36 + @Override
1.37 + protected Comment processComment(int pos, int endPos, CommentStyle style) {
1.38 + char[] buf = reader.getRawCharacters(pos, endPos);
1.39 + return new JavadocComment(new ColReader(fac, buf, buf.length), style);
1.40 + }
1.41
1.42 /**
1.43 - * Buffer for doc comment.
1.44 + * This is a specialized version of UnicodeReader that keeps track of the
1.45 + * column position within a given character stream (used for Javadoc processing).
1.46 */
1.47 - private char[] docCommentBuffer = new char[1024];
1.48 + static class ColReader extends UnicodeReader {
1.49
1.50 - /**
1.51 - * Number of characters in doc comment buffer.
1.52 - */
1.53 - private int docCommentCount;
1.54 + int col;
1.55
1.56 - /**
1.57 - * Translated and stripped contents of doc comment
1.58 - */
1.59 - private String docComment = null;
1.60 + ColReader(ScannerFactory fac, char[] input, int inputLength) {
1.61 + super(fac, input, inputLength);
1.62 + }
1.63
1.64 + @Override
1.65 + protected void convertUnicode() {
1.66 + if (ch == '\\' && unicodeConversionBp != bp) {
1.67 + bp++; ch = buf[bp]; col++;
1.68 + if (ch == 'u') {
1.69 + do {
1.70 + bp++; ch = buf[bp]; col++;
1.71 + } while (ch == 'u');
1.72 + int limit = bp + 3;
1.73 + if (limit < buflen) {
1.74 + int d = digit(bp, 16);
1.75 + int code = d;
1.76 + while (bp < limit && d >= 0) {
1.77 + bp++; ch = buf[bp]; col++;
1.78 + d = digit(bp, 16);
1.79 + code = (code << 4) + d;
1.80 + }
1.81 + if (d >= 0) {
1.82 + ch = (char)code;
1.83 + unicodeConversionBp = bp;
1.84 + return;
1.85 + }
1.86 + }
1.87 + // "illegal.Unicode.esc", reported by base scanner
1.88 + } else {
1.89 + bp--;
1.90 + ch = '\\';
1.91 + col--;
1.92 + }
1.93 + }
1.94 + }
1.95
1.96 - /** Unconditionally expand the comment buffer.
1.97 - */
1.98 - private void expandCommentBuffer() {
1.99 - char[] newBuffer = new char[docCommentBuffer.length * 2];
1.100 - System.arraycopy(docCommentBuffer, 0, newBuffer,
1.101 - 0, docCommentBuffer.length);
1.102 - docCommentBuffer = newBuffer;
1.103 - }
1.104 + @Override
1.105 + protected void scanCommentChar() {
1.106 + scanChar();
1.107 + if (ch == '\\') {
1.108 + if (peekChar() == '\\' && !isUnicode()) {
1.109 + putChar(ch, false);
1.110 + bp++; col++;
1.111 + } else {
1.112 + convertUnicode();
1.113 + }
1.114 + }
1.115 + }
1.116
1.117 - /** Convert an ASCII digit from its base (8, 10, or 16)
1.118 - * to its value.
1.119 - */
1.120 - private int digit(int base) {
1.121 - char c = ch;
1.122 - int result = Character.digit(c, base);
1.123 - if (result >= 0 && c > 0x7f) {
1.124 - ch = "0123456789abcdef".charAt(result);
1.125 + @Override
1.126 + protected void scanChar() {
1.127 + bp++;
1.128 + ch = buf[bp];
1.129 + switch (ch) {
1.130 + case '\r': // return
1.131 + col = 0;
1.132 + break;
1.133 + case '\n': // newline
1.134 + if (bp == 0 || buf[bp-1] != '\r') {
1.135 + col = 0;
1.136 + }
1.137 + break;
1.138 + case '\t': // tab
1.139 + col = (col / TabInc * TabInc) + TabInc;
1.140 + break;
1.141 + case '\\': // possible Unicode
1.142 + col++;
1.143 + convertUnicode();
1.144 + break;
1.145 + default:
1.146 + col++;
1.147 + break;
1.148 + }
1.149 + }
1.150 + }
1.151 +
1.152 + protected class JavadocComment extends JavaTokenizer.BasicComment<ColReader> {
1.153 +
1.154 + /**
1.155 + * Translated and stripped contents of doc comment
1.156 + */
1.157 + private String docComment = null;
1.158 +
1.159 + JavadocComment(ColReader comment_reader, CommentStyle cs) {
1.160 + super(comment_reader, cs);
1.161 }
1.162 - return result;
1.163 - }
1.164
1.165 - /** Convert Unicode escape; bp points to initial '\' character
1.166 - * (Spec 3.3).
1.167 - */
1.168 - private void convertUnicode() {
1.169 - if (ch == '\\' && unicodeConversionBp != bp) {
1.170 - bp++; ch = buf[bp]; col++;
1.171 - if (ch == 'u') {
1.172 - do {
1.173 - bp++; ch = buf[bp]; col++;
1.174 - } while (ch == 'u');
1.175 - int limit = bp + 3;
1.176 - if (limit < buflen) {
1.177 - int d = digit(16);
1.178 - int code = d;
1.179 - while (bp < limit && d >= 0) {
1.180 - bp++; ch = buf[bp]; col++;
1.181 - d = digit(16);
1.182 - code = (code << 4) + d;
1.183 - }
1.184 - if (d >= 0) {
1.185 - ch = (char)code;
1.186 - unicodeConversionBp = bp;
1.187 - return;
1.188 - }
1.189 + public String getText() {
1.190 + if (!scanned && cs == CommentStyle.JAVADOC) {
1.191 + scanDocComment();
1.192 + }
1.193 + return docComment;
1.194 + }
1.195 +
1.196 + @Override
1.197 + @SuppressWarnings("fallthrough")
1.198 + protected void scanDocComment() {
1.199 + try {
1.200 + boolean firstLine = true;
1.201 +
1.202 + // Skip over first slash
1.203 + comment_reader.scanCommentChar();
1.204 + // Skip over first star
1.205 + comment_reader.scanCommentChar();
1.206 +
1.207 + // consume any number of stars
1.208 + while (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '*') {
1.209 + comment_reader.scanCommentChar();
1.210 + }
1.211 + // is the comment in the form /**/, /***/, /****/, etc. ?
1.212 + if (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '/') {
1.213 + docComment = "";
1.214 + return;
1.215 + }
1.216 +
1.217 + // skip a newline on the first line of the comment.
1.218 + if (comment_reader.bp < comment_reader.buflen) {
1.219 + if (comment_reader.ch == LF) {
1.220 + comment_reader.scanCommentChar();
1.221 + firstLine = false;
1.222 + } else if (comment_reader.ch == CR) {
1.223 + comment_reader.scanCommentChar();
1.224 + if (comment_reader.ch == LF) {
1.225 + comment_reader.scanCommentChar();
1.226 + firstLine = false;
1.227 + }
1.228 + }
1.229 + }
1.230 +
1.231 + outerLoop:
1.232 +
1.233 + // The outerLoop processes the doc comment, looping once
1.234 + // for each line. For each line, it first strips off
1.235 + // whitespace, then it consumes any stars, then it
1.236 + // puts the rest of the line into our buffer.
1.237 + while (comment_reader.bp < comment_reader.buflen) {
1.238 +
1.239 + // The wsLoop consumes whitespace from the beginning
1.240 + // of each line.
1.241 + wsLoop:
1.242 +
1.243 + while (comment_reader.bp < comment_reader.buflen) {
1.244 + switch(comment_reader.ch) {
1.245 + case ' ':
1.246 + comment_reader.scanCommentChar();
1.247 + break;
1.248 + case '\t':
1.249 + comment_reader.col = ((comment_reader.col - 1) / TabInc * TabInc) + TabInc;
1.250 + comment_reader.scanCommentChar();
1.251 + break;
1.252 + case FF:
1.253 + comment_reader.col = 0;
1.254 + comment_reader.scanCommentChar();
1.255 + break;
1.256 + // Treat newline at beginning of line (blank line, no star)
1.257 + // as comment text. Old Javadoc compatibility requires this.
1.258 + /*---------------------------------*
1.259 + case CR: // (Spec 3.4)
1.260 + doc_reader.scanCommentChar();
1.261 + if (ch == LF) {
1.262 + col = 0;
1.263 + doc_reader.scanCommentChar();
1.264 + }
1.265 + break;
1.266 + case LF: // (Spec 3.4)
1.267 + doc_reader.scanCommentChar();
1.268 + break;
1.269 + *---------------------------------*/
1.270 + default:
1.271 + // we've seen something that isn't whitespace;
1.272 + // jump out.
1.273 + break wsLoop;
1.274 + }
1.275 + }
1.276 +
1.277 + // Are there stars here? If so, consume them all
1.278 + // and check for the end of comment.
1.279 + if (comment_reader.ch == '*') {
1.280 + // skip all of the stars
1.281 + do {
1.282 + comment_reader.scanCommentChar();
1.283 + } while (comment_reader.ch == '*');
1.284 +
1.285 + // check for the closing slash.
1.286 + if (comment_reader.ch == '/') {
1.287 + // We're done with the doc comment
1.288 + // scanChar() and breakout.
1.289 + break outerLoop;
1.290 + }
1.291 + } else if (! firstLine) {
1.292 + //The current line does not begin with a '*' so we will indent it.
1.293 + for (int i = 1; i < comment_reader.col; i++) {
1.294 + comment_reader.putChar(' ', false);
1.295 + }
1.296 + }
1.297 + // The textLoop processes the rest of the characters
1.298 + // on the line, adding them to our buffer.
1.299 + textLoop:
1.300 + while (comment_reader.bp < comment_reader.buflen) {
1.301 + switch (comment_reader.ch) {
1.302 + case '*':
1.303 + // Is this just a star? Or is this the
1.304 + // end of a comment?
1.305 + comment_reader.scanCommentChar();
1.306 + if (comment_reader.ch == '/') {
1.307 + // This is the end of the comment,
1.308 + // set ch and return our buffer.
1.309 + break outerLoop;
1.310 + }
1.311 + // This is just an ordinary star. Add it to
1.312 + // the buffer.
1.313 + comment_reader.putChar('*', false);
1.314 + break;
1.315 + case ' ':
1.316 + case '\t':
1.317 + comment_reader.putChar(comment_reader.ch, false);
1.318 + comment_reader.scanCommentChar();
1.319 + break;
1.320 + case FF:
1.321 + comment_reader.scanCommentChar();
1.322 + break textLoop; // treat as end of line
1.323 + case CR: // (Spec 3.4)
1.324 + comment_reader.scanCommentChar();
1.325 + if (comment_reader.ch != LF) {
1.326 + // Canonicalize CR-only line terminator to LF
1.327 + comment_reader.putChar((char)LF, false);
1.328 + break textLoop;
1.329 + }
1.330 + /* fall through to LF case */
1.331 + case LF: // (Spec 3.4)
1.332 + // We've seen a newline. Add it to our
1.333 + // buffer and break out of this loop,
1.334 + // starting fresh on a new line.
1.335 + comment_reader.putChar(comment_reader.ch, false);
1.336 + comment_reader.scanCommentChar();
1.337 + break textLoop;
1.338 + default:
1.339 + // Add the character to our buffer.
1.340 + comment_reader.putChar(comment_reader.ch, false);
1.341 + comment_reader.scanCommentChar();
1.342 + }
1.343 + } // end textLoop
1.344 + firstLine = false;
1.345 + } // end outerLoop
1.346 +
1.347 + if (comment_reader.sp > 0) {
1.348 + int i = comment_reader.sp - 1;
1.349 + trailLoop:
1.350 + while (i > -1) {
1.351 + switch (comment_reader.sbuf[i]) {
1.352 + case '*':
1.353 + i--;
1.354 + break;
1.355 + default:
1.356 + break trailLoop;
1.357 + }
1.358 + }
1.359 + comment_reader.sp = i + 1;
1.360 +
1.361 + // Store the text of the doc comment
1.362 + docComment = comment_reader.chars();
1.363 + } else {
1.364 + docComment = "";
1.365 }
1.366 - // "illegal.Unicode.esc", reported by base scanner
1.367 - } else {
1.368 - bp--;
1.369 - ch = '\\';
1.370 - col--;
1.371 + } finally {
1.372 + scanned = true;
1.373 + if (docComment != null &&
1.374 + docComment.matches("(?sm).*^\\s*@deprecated( |$).*")) {
1.375 + deprecatedFlag = true;
1.376 + }
1.377 }
1.378 }
1.379 }
1.380
1.381 -
1.382 - /** Read next character.
1.383 - */
1.384 - private void scanChar() {
1.385 - bp++;
1.386 - ch = buf[bp];
1.387 - switch (ch) {
1.388 - case '\r': // return
1.389 - col = 0;
1.390 - break;
1.391 - case '\n': // newline
1.392 - if (bp == 0 || buf[bp-1] != '\r') {
1.393 - col = 0;
1.394 - }
1.395 - break;
1.396 - case '\t': // tab
1.397 - col = (col / TabInc * TabInc) + TabInc;
1.398 - break;
1.399 - case '\\': // possible Unicode
1.400 - col++;
1.401 - convertUnicode();
1.402 - break;
1.403 - default:
1.404 - col++;
1.405 - break;
1.406 - }
1.407 - }
1.408 -
1.409 @Override
1.410 - public Token readToken() {
1.411 - docComment = null;
1.412 - Token tk = super.readToken();
1.413 - tk.docComment = docComment;
1.414 - return tk;
1.415 - }
1.416 -
1.417 - /**
1.418 - * Read next character in doc comment, skipping over double '\' characters.
1.419 - * If a double '\' is skipped, put in the buffer and update buffer count.
1.420 - */
1.421 - private void scanDocCommentChar() {
1.422 - scanChar();
1.423 - if (ch == '\\') {
1.424 - if (buf[bp+1] == '\\' && unicodeConversionBp != bp) {
1.425 - if (docCommentCount == docCommentBuffer.length)
1.426 - expandCommentBuffer();
1.427 - docCommentBuffer[docCommentCount++] = ch;
1.428 - bp++; col++;
1.429 - } else {
1.430 - convertUnicode();
1.431 - }
1.432 - }
1.433 - }
1.434 -
1.435 - /**
1.436 - * Process a doc comment and make the string content available.
1.437 - * Strips leading whitespace and stars.
1.438 - */
1.439 - @SuppressWarnings("fallthrough")
1.440 - protected void processComment(int pos, int endPos, CommentStyle style) {
1.441 - if (style != CommentStyle.JAVADOC) {
1.442 - return;
1.443 - }
1.444 -
1.445 - buf = reader.getRawCharacters(pos, endPos);
1.446 - buflen = buf.length;
1.447 - bp = 0;
1.448 - col = 0;
1.449 -
1.450 - docCommentCount = 0;
1.451 -
1.452 - boolean firstLine = true;
1.453 -
1.454 - // Skip over first slash
1.455 - scanDocCommentChar();
1.456 - // Skip over first star
1.457 - scanDocCommentChar();
1.458 -
1.459 - // consume any number of stars
1.460 - while (bp < buflen && ch == '*') {
1.461 - scanDocCommentChar();
1.462 - }
1.463 - // is the comment in the form /**/, /***/, /****/, etc. ?
1.464 - if (bp < buflen && ch == '/') {
1.465 - docComment = "";
1.466 - return;
1.467 - }
1.468 -
1.469 - // skip a newline on the first line of the comment.
1.470 - if (bp < buflen) {
1.471 - if (ch == LF) {
1.472 - scanDocCommentChar();
1.473 - firstLine = false;
1.474 - } else if (ch == CR) {
1.475 - scanDocCommentChar();
1.476 - if (ch == LF) {
1.477 - scanDocCommentChar();
1.478 - firstLine = false;
1.479 - }
1.480 - }
1.481 - }
1.482 -
1.483 - outerLoop:
1.484 -
1.485 - // The outerLoop processes the doc comment, looping once
1.486 - // for each line. For each line, it first strips off
1.487 - // whitespace, then it consumes any stars, then it
1.488 - // puts the rest of the line into our buffer.
1.489 - while (bp < buflen) {
1.490 -
1.491 - // The wsLoop consumes whitespace from the beginning
1.492 - // of each line.
1.493 - wsLoop:
1.494 -
1.495 - while (bp < buflen) {
1.496 - switch(ch) {
1.497 - case ' ':
1.498 - scanDocCommentChar();
1.499 - break;
1.500 - case '\t':
1.501 - col = ((col - 1) / TabInc * TabInc) + TabInc;
1.502 - scanDocCommentChar();
1.503 - break;
1.504 - case FF:
1.505 - col = 0;
1.506 - scanDocCommentChar();
1.507 - break;
1.508 -// Treat newline at beginning of line (blank line, no star)
1.509 -// as comment text. Old Javadoc compatibility requires this.
1.510 -/*---------------------------------*
1.511 - case CR: // (Spec 3.4)
1.512 - scanDocCommentChar();
1.513 - if (ch == LF) {
1.514 - col = 0;
1.515 - scanDocCommentChar();
1.516 - }
1.517 - break;
1.518 - case LF: // (Spec 3.4)
1.519 - scanDocCommentChar();
1.520 - break;
1.521 -*---------------------------------*/
1.522 - default:
1.523 - // we've seen something that isn't whitespace;
1.524 - // jump out.
1.525 - break wsLoop;
1.526 - }
1.527 - }
1.528 -
1.529 - // Are there stars here? If so, consume them all
1.530 - // and check for the end of comment.
1.531 - if (ch == '*') {
1.532 - // skip all of the stars
1.533 - do {
1.534 - scanDocCommentChar();
1.535 - } while (ch == '*');
1.536 -
1.537 - // check for the closing slash.
1.538 - if (ch == '/') {
1.539 - // We're done with the doc comment
1.540 - // scanChar() and breakout.
1.541 - break outerLoop;
1.542 - }
1.543 - } else if (! firstLine) {
1.544 - //The current line does not begin with a '*' so we will indent it.
1.545 - for (int i = 1; i < col; i++) {
1.546 - if (docCommentCount == docCommentBuffer.length)
1.547 - expandCommentBuffer();
1.548 - docCommentBuffer[docCommentCount++] = ' ';
1.549 - }
1.550 - }
1.551 -
1.552 - // The textLoop processes the rest of the characters
1.553 - // on the line, adding them to our buffer.
1.554 - textLoop:
1.555 - while (bp < buflen) {
1.556 - switch (ch) {
1.557 - case '*':
1.558 - // Is this just a star? Or is this the
1.559 - // end of a comment?
1.560 - scanDocCommentChar();
1.561 - if (ch == '/') {
1.562 - // This is the end of the comment,
1.563 - // set ch and return our buffer.
1.564 - break outerLoop;
1.565 - }
1.566 - // This is just an ordinary star. Add it to
1.567 - // the buffer.
1.568 - if (docCommentCount == docCommentBuffer.length)
1.569 - expandCommentBuffer();
1.570 - docCommentBuffer[docCommentCount++] = '*';
1.571 - break;
1.572 - case ' ':
1.573 - case '\t':
1.574 - if (docCommentCount == docCommentBuffer.length)
1.575 - expandCommentBuffer();
1.576 - docCommentBuffer[docCommentCount++] = ch;
1.577 - scanDocCommentChar();
1.578 - break;
1.579 - case FF:
1.580 - scanDocCommentChar();
1.581 - break textLoop; // treat as end of line
1.582 - case CR: // (Spec 3.4)
1.583 - scanDocCommentChar();
1.584 - if (ch != LF) {
1.585 - // Canonicalize CR-only line terminator to LF
1.586 - if (docCommentCount == docCommentBuffer.length)
1.587 - expandCommentBuffer();
1.588 - docCommentBuffer[docCommentCount++] = (char)LF;
1.589 - break textLoop;
1.590 - }
1.591 - /* fall through to LF case */
1.592 - case LF: // (Spec 3.4)
1.593 - // We've seen a newline. Add it to our
1.594 - // buffer and break out of this loop,
1.595 - // starting fresh on a new line.
1.596 - if (docCommentCount == docCommentBuffer.length)
1.597 - expandCommentBuffer();
1.598 - docCommentBuffer[docCommentCount++] = ch;
1.599 - scanDocCommentChar();
1.600 - break textLoop;
1.601 - default:
1.602 - // Add the character to our buffer.
1.603 - if (docCommentCount == docCommentBuffer.length)
1.604 - expandCommentBuffer();
1.605 - docCommentBuffer[docCommentCount++] = ch;
1.606 - scanDocCommentChar();
1.607 - }
1.608 - } // end textLoop
1.609 - firstLine = false;
1.610 - } // end outerLoop
1.611 -
1.612 - if (docCommentCount > 0) {
1.613 - int i = docCommentCount - 1;
1.614 - trailLoop:
1.615 - while (i > -1) {
1.616 - switch (docCommentBuffer[i]) {
1.617 - case '*':
1.618 - i--;
1.619 - break;
1.620 - default:
1.621 - break trailLoop;
1.622 - }
1.623 - }
1.624 - docCommentCount = i + 1;
1.625 -
1.626 - // Store the text of the doc comment
1.627 - docComment = new String(docCommentBuffer, 0 , docCommentCount);
1.628 - } else {
1.629 - docComment = "";
1.630 - }
1.631 - }
1.632 -
1.633 - /** Build a map for translating between line numbers and
1.634 - * positions in the input.
1.635 - *
1.636 - * @return a LineMap */
1.637 public Position.LineMap getLineMap() {
1.638 char[] buf = reader.getRawCharacters();
1.639 return Position.makeLineMap(buf, buf.length, true);
