7178763: javadoc OutOfMemory error results in several jdk8 tl nightly failures
Reviewed-by: ksrini
/*
* Copyright (c) 2004, 2012, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
package com.sun.tools.javac.parser;
import com.sun.tools.javac.parser.Tokens.Comment;
import com.sun.tools.javac.parser.Tokens.Comment.CommentStyle;
import com.sun.tools.javac.util.*;
import java.nio.*;
import static com.sun.tools.javac.util.LayoutCharacters.*;
/** An extension to the base lexical analyzer that captures
* and processes the contents of doc comments. It does so by
* translating Unicode escape sequences and by stripping the
* leading whitespace and starts from each line of the comment.
*
* <p><b>This is NOT part of any supported API.
* If you write code that depends on this, you do so at your own risk.
* This code and its internal interfaces are subject to change or
* deletion without notice.</b>
*/
public class JavadocTokenizer extends JavaTokenizer {
/** Create a scanner from the input buffer. buffer must implement
* array() and compact(), and remaining() must be less than limit().
*/
protected JavadocTokenizer(ScannerFactory fac, CharBuffer buffer) {
super(fac, buffer);
}
/** Create a scanner from the input array. The array must have at
* least a single character of extra space.
*/
protected JavadocTokenizer(ScannerFactory fac, char[] input, int inputLength) {
super(fac, input, inputLength);
}
@Override
protected Comment processComment(int pos, int endPos, CommentStyle style) {
char[] buf = reader.getRawCharacters(pos, endPos);
return new JavadocComment(new DocReader(fac, buf, buf.length, pos), style);
}
/**
* This is a specialized version of UnicodeReader that keeps track of the
* column position within a given character stream (used for Javadoc processing),
* and which builds a table for mapping positions in the comment string to
* positions in the source file.
*/
static class DocReader extends UnicodeReader {
int col;
int startPos;
/**
* A buffer for building a table for mapping positions in {@link #sbuf}
* to positions in the source buffer.
*
* The array is organized as a series of pairs of integers: the first
* number in each pair specifies a position in the comment text,
* the second number in each pair specifies the corresponding position
* in the source buffer. The pairs are sorted in ascending order.
*
* Since the mapping function is generally continuous, with successive
* positions in the string corresponding to successive positions in the
* source buffer, the table only needs to record discontinuities in
* the mapping. The values of intermediate positions can be inferred.
*
* Discontinuities may occur in a number of places: when a newline
* is followed by whitespace and asterisks (which are ignored),
* when a tab is expanded into spaces, and when unicode escapes
* are used in the source buffer.
*
* Thus, to find the source position of any position, p, in the comment
* string, find the index, i, of the pair whose string offset
* ({@code pbuf[i] }) is closest to but not greater than p. Then,
* {@code sourcePos(p) = pbuf[i+1] + (p - pbuf[i]) }.
*/
int[] pbuf = new int[128];
/**
* The index of the next empty slot in the pbuf buffer.
*/
int pp = 0;
DocReader(ScannerFactory fac, char[] input, int inputLength, int startPos) {
super(fac, input, inputLength);
this.startPos = startPos;
}
@Override
protected void convertUnicode() {
if (ch == '\\' && unicodeConversionBp != bp) {
bp++; ch = buf[bp]; col++;
if (ch == 'u') {
do {
bp++; ch = buf[bp]; col++;
} while (ch == 'u');
int limit = bp + 3;
if (limit < buflen) {
int d = digit(bp, 16);
int code = d;
while (bp < limit && d >= 0) {
bp++; ch = buf[bp]; col++;
d = digit(bp, 16);
code = (code << 4) + d;
}
if (d >= 0) {
ch = (char)code;
unicodeConversionBp = bp;
return;
}
}
// "illegal.Unicode.esc", reported by base scanner
} else {
bp--;
ch = '\\';
col--;
}
}
}
@Override
protected void scanCommentChar() {
scanChar();
if (ch == '\\') {
if (peekChar() == '\\' && !isUnicode()) {
putChar(ch, false);
bp++; col++;
} else {
convertUnicode();
}
}
}
@Override
protected void scanChar() {
bp++;
ch = buf[bp];
switch (ch) {
case '\r': // return
col = 0;
break;
case '\n': // newline
if (bp == 0 || buf[bp-1] != '\r') {
col = 0;
}
break;
case '\t': // tab
col = (col / TabInc * TabInc) + TabInc;
break;
case '\\': // possible Unicode
col++;
convertUnicode();
break;
default:
col++;
break;
}
}
@Override
public void putChar(char ch, boolean scan) {
// At this point, bp is the position of the current character in buf,
// and sp is the position in sbuf where this character will be put.
// Record a new entry in pbuf if pbuf is empty or if sp and its
// corresponding source position are not equidistant from the
// corresponding values in the latest entry in the pbuf array.
// (i.e. there is a discontinuity in the map function.)
if ((pp == 0)
|| (sp - pbuf[pp - 2] != (startPos + bp) - pbuf[pp - 1])) {
if (pp + 1 >= pbuf.length) {
int[] new_pbuf = new int[pbuf.length * 2];
System.arraycopy(pbuf, 0, new_pbuf, 0, pbuf.length);
pbuf = new_pbuf;
}
pbuf[pp] = sp;
pbuf[pp + 1] = startPos + bp;
pp += 2;
}
super.putChar(ch, scan);
}
}
protected class JavadocComment extends JavaTokenizer.BasicComment<DocReader> {
/**
* Translated and stripped contents of doc comment
*/
private String docComment = null;
private int[] docPosns = null;
JavadocComment(DocReader reader, CommentStyle cs) {
super(reader, cs);
}
@Override
public String getText() {
if (!scanned && cs == CommentStyle.JAVADOC) {
scanDocComment();
}
return docComment;
}
@Override
public int getSourcePos(int pos) {
// Binary search to find the entry for which the string index is
// less than pos. Since docPosns is a list of pairs of integers
// we must make sure the index is always even.
// If we find an exact match for pos, the other item in the pair
// gives the source pos; otherwise, compute the source position
// relative to the best match found in the array.
if (pos < 0 || pos >= docComment.length())
throw new StringIndexOutOfBoundsException();
if (docPosns == null)
return -1;
int start = 0;
int end = docPosns.length;
while (start < end - 2) {
// find an even index midway between start and end
int index = ((start + end) / 4) * 2;
if (docPosns[index] < pos)
start = index;
else if (docPosns[index] == pos)
return docPosns[index + 1];
else
end = index;
}
return docPosns[start + 1] + (pos - docPosns[start]);
}
@Override
@SuppressWarnings("fallthrough")
protected void scanDocComment() {
try {
boolean firstLine = true;
// Skip over first slash
comment_reader.scanCommentChar();
// Skip over first star
comment_reader.scanCommentChar();
// consume any number of stars
while (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '*') {
comment_reader.scanCommentChar();
}
// is the comment in the form /**/, /***/, /****/, etc. ?
if (comment_reader.bp < comment_reader.buflen && comment_reader.ch == '/') {
docComment = "";
return;
}
// skip a newline on the first line of the comment.
if (comment_reader.bp < comment_reader.buflen) {
if (comment_reader.ch == LF) {
comment_reader.scanCommentChar();
firstLine = false;
} else if (comment_reader.ch == CR) {
comment_reader.scanCommentChar();
if (comment_reader.ch == LF) {
comment_reader.scanCommentChar();
firstLine = false;
}
}
}
outerLoop:
// The outerLoop processes the doc comment, looping once
// for each line. For each line, it first strips off
// whitespace, then it consumes any stars, then it
// puts the rest of the line into our buffer.
while (comment_reader.bp < comment_reader.buflen) {
int begin_bp = comment_reader.bp;
char begin_ch = comment_reader.ch;
// The wsLoop consumes whitespace from the beginning
// of each line.
wsLoop:
while (comment_reader.bp < comment_reader.buflen) {
switch(comment_reader.ch) {
case ' ':
comment_reader.scanCommentChar();
break;
case '\t':
comment_reader.col = ((comment_reader.col - 1) / TabInc * TabInc) + TabInc;
comment_reader.scanCommentChar();
break;
case FF:
comment_reader.col = 0;
comment_reader.scanCommentChar();
break;
// Treat newline at beginning of line (blank line, no star)
// as comment text. Old Javadoc compatibility requires this.
/*---------------------------------*
case CR: // (Spec 3.4)
doc_reader.scanCommentChar();
if (ch == LF) {
col = 0;
doc_reader.scanCommentChar();
}
break;
case LF: // (Spec 3.4)
doc_reader.scanCommentChar();
break;
*---------------------------------*/
default:
// we've seen something that isn't whitespace;
// jump out.
break wsLoop;
}
}
// Are there stars here? If so, consume them all
// and check for the end of comment.
if (comment_reader.ch == '*') {
// skip all of the stars
do {
comment_reader.scanCommentChar();
} while (comment_reader.ch == '*');
// check for the closing slash.
if (comment_reader.ch == '/') {
// We're done with the doc comment
// scanChar() and breakout.
break outerLoop;
}
} else if (! firstLine) {
// The current line does not begin with a '*' so we will
// treat it as comment
comment_reader.bp = begin_bp;
comment_reader.ch = begin_ch;
}
// The textLoop processes the rest of the characters
// on the line, adding them to our buffer.
textLoop:
while (comment_reader.bp < comment_reader.buflen) {
switch (comment_reader.ch) {
case '*':
// Is this just a star? Or is this the
// end of a comment?
comment_reader.scanCommentChar();
if (comment_reader.ch == '/') {
// This is the end of the comment,
// set ch and return our buffer.
break outerLoop;
}
// This is just an ordinary star. Add it to
// the buffer.
comment_reader.putChar('*', false);
break;
case ' ':
case '\t':
comment_reader.putChar(comment_reader.ch, false);
comment_reader.scanCommentChar();
break;
case FF:
comment_reader.scanCommentChar();
break textLoop; // treat as end of line
case CR: // (Spec 3.4)
comment_reader.scanCommentChar();
if (comment_reader.ch != LF) {
// Canonicalize CR-only line terminator to LF
comment_reader.putChar((char)LF, false);
break textLoop;
}
/* fall through to LF case */
case LF: // (Spec 3.4)
// We've seen a newline. Add it to our
// buffer and break out of this loop,
// starting fresh on a new line.
comment_reader.putChar(comment_reader.ch, false);
comment_reader.scanCommentChar();
break textLoop;
default:
// Add the character to our buffer.
comment_reader.putChar(comment_reader.ch, false);
comment_reader.scanCommentChar();
}
} // end textLoop
firstLine = false;
} // end outerLoop
if (comment_reader.sp > 0) {
int i = comment_reader.sp - 1;
trailLoop:
while (i > -1) {
switch (comment_reader.sbuf[i]) {
case '*':
i--;
break;
default:
break trailLoop;
}
}
comment_reader.sp = i + 1;
// Store the text of the doc comment
docComment = comment_reader.chars();
docPosns = new int[comment_reader.pp];
System.arraycopy(comment_reader.pbuf, 0, docPosns, 0, docPosns.length);
} else {
docComment = "";
}
} finally {
scanned = true;
comment_reader = null;
if (docComment != null &&
docComment.matches("(?sm).*^\\s*@deprecated( |$).*")) {
deprecatedFlag = true;
}
}
}
}
@Override
public Position.LineMap getLineMap() {
char[] buf = reader.getRawCharacters();
return Position.makeLineMap(buf, buf.length, true);
}
}