1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/src/share/jaxws_classes/com/sun/xml/internal/org/jvnet/mimepull/MIMEParser.java Wed Apr 27 01:27:09 2016 +0800
1.3 @@ -0,0 +1,502 @@
1.4 +/*
1.5 + * Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
1.6 + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
1.7 + *
1.8 + * This code is free software; you can redistribute it and/or modify it
1.9 + * under the terms of the GNU General Public License version 2 only, as
1.10 + * published by the Free Software Foundation. Oracle designates this
1.11 + * particular file as subject to the "Classpath" exception as provided
1.12 + * by Oracle in the LICENSE file that accompanied this code.
1.13 + *
1.14 + * This code is distributed in the hope that it will be useful, but WITHOUT
1.15 + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
1.16 + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
1.17 + * version 2 for more details (a copy is included in the LICENSE file that
1.18 + * accompanied this code).
1.19 + *
1.20 + * You should have received a copy of the GNU General Public License version
1.21 + * 2 along with this work; if not, write to the Free Software Foundation,
1.22 + * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
1.23 + *
1.24 + * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
1.25 + * or visit www.oracle.com if you need additional information or have any
1.26 + * questions.
1.27 + */
1.28 +
1.29 +package com.sun.xml.internal.org.jvnet.mimepull;
1.30 +
1.31 +import java.io.InputStream;
1.32 +import java.io.IOException;
1.33 +import java.util.*;
1.34 +import java.util.logging.Logger;
1.35 +import java.nio.ByteBuffer;
1.36 +import java.util.logging.Level;
1.37 +
1.38 +/**
1.39 + * Pull parser for the MIME messages. Applications can use pull API to continue
1.40 + * the parsing MIME messages lazily.
1.41 + *
1.42 + * <pre>
1.43 + * for e.g.:
1.44 + * <p>
1.45 + *
1.46 + * MIMEParser parser = ...
1.47 + * Iterator<MIMEEvent> it = parser.iterator();
1.48 + * while(it.hasNext()) {
1.49 + * MIMEEvent event = it.next();
1.50 + * ...
1.51 + * }
1.52 + * </pre>
1.53 + *
1.54 + * @author Jitendra Kotamraju
1.55 + */
1.56 +class MIMEParser implements Iterable<MIMEEvent> {
1.57 +
1.58 + private static final Logger LOGGER = Logger.getLogger(MIMEParser.class.getName());
1.59 +
1.60 + private static final String HEADER_ENCODING = "ISO8859-1";
1.61 +
1.62 + // Actually, the grammar doesn't support whitespace characters
1.63 + // after boundary. But the mail implementation checks for it.
1.64 + // We will only check for these many whitespace characters after boundary
1.65 + private static final int NO_LWSP = 1000;
1.66 + private enum STATE {START_MESSAGE, SKIP_PREAMBLE, START_PART, HEADERS, BODY, END_PART, END_MESSAGE}
1.67 + private STATE state = STATE.START_MESSAGE;
1.68 +
1.69 + private final InputStream in;
1.70 + private final byte[] bndbytes;
1.71 + private final int bl;
1.72 + private final MIMEConfig config;
1.73 + private final int[] bcs = new int[128]; // BnM algo: Bad Character Shift table
1.74 + private final int[] gss; // BnM algo : Good Suffix Shift table
1.75 +
1.76 + /**
1.77 + * Have we parsed the data from our InputStream yet?
1.78 + */
1.79 + private boolean parsed;
1.80 +
1.81 + /*
1.82 + * Read and process body partsList until we see the
1.83 + * terminating boundary line (or EOF).
1.84 + */
1.85 + private boolean done = false;
1.86 +
1.87 + private boolean eof;
1.88 + private final int capacity;
1.89 + private byte[] buf;
1.90 + private int len;
1.91 + private boolean bol; // beginning of the line
1.92 +
1.93 + /*
1.94 + * Parses the MIME content. At the EOF, it also closes input stream
1.95 + */
1.96 + MIMEParser(InputStream in, String boundary, MIMEConfig config) {
1.97 + this.in = in;
1.98 + this.bndbytes = getBytes("--"+boundary);
1.99 + bl = bndbytes.length;
1.100 + this.config = config;
1.101 + gss = new int[bl];
1.102 + compileBoundaryPattern();
1.103 +
1.104 + // \r\n + boundary + "--\r\n" + lots of LWSP
1.105 + capacity = config.chunkSize+2+bl+4+NO_LWSP;
1.106 + createBuf(capacity);
1.107 + }
1.108 +
1.109 + /**
1.110 + * Returns iterator for the parsing events. Use the iterator to advance
1.111 + * the parsing.
1.112 + *
1.113 + * @return iterator for parsing events
1.114 + */
1.115 + @Override
1.116 + public Iterator<MIMEEvent> iterator() {
1.117 + return new MIMEEventIterator();
1.118 + }
1.119 +
1.120 + class MIMEEventIterator implements Iterator<MIMEEvent> {
1.121 +
1.122 + @Override
1.123 + public boolean hasNext() {
1.124 + return !parsed;
1.125 + }
1.126 +
1.127 + @Override
1.128 + public MIMEEvent next() {
1.129 + switch(state) {
1.130 + case START_MESSAGE :
1.131 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.START_MESSAGE);}
1.132 + state = STATE.SKIP_PREAMBLE;
1.133 + return MIMEEvent.START_MESSAGE;
1.134 +
1.135 + case SKIP_PREAMBLE :
1.136 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.SKIP_PREAMBLE);}
1.137 + skipPreamble();
1.138 + // fall through
1.139 + case START_PART :
1.140 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.START_PART);}
1.141 + state = STATE.HEADERS;
1.142 + return MIMEEvent.START_PART;
1.143 +
1.144 + case HEADERS :
1.145 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.HEADERS);}
1.146 + InternetHeaders ih = readHeaders();
1.147 + state = STATE.BODY;
1.148 + bol = true;
1.149 + return new MIMEEvent.Headers(ih);
1.150 +
1.151 + case BODY :
1.152 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.BODY);}
1.153 + ByteBuffer buf = readBody();
1.154 + bol = false;
1.155 + return new MIMEEvent.Content(buf);
1.156 +
1.157 + case END_PART :
1.158 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.END_PART);}
1.159 + if (done) {
1.160 + state = STATE.END_MESSAGE;
1.161 + } else {
1.162 + state = STATE.START_PART;
1.163 + }
1.164 + return MIMEEvent.END_PART;
1.165 +
1.166 + case END_MESSAGE :
1.167 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "MIMEParser state={0}", STATE.END_MESSAGE);}
1.168 + parsed = true;
1.169 + return MIMEEvent.END_MESSAGE;
1.170 +
1.171 + default :
1.172 + throw new MIMEParsingException("Unknown Parser state = "+state);
1.173 + }
1.174 + }
1.175 +
1.176 + @Override
1.177 + public void remove() {
1.178 + throw new UnsupportedOperationException();
1.179 + }
1.180 + }
1.181 +
1.182 + /**
1.183 + * Collects the headers for the current part by parsing mesage stream.
1.184 + *
1.185 + * @return headers for the current part
1.186 + */
1.187 + private InternetHeaders readHeaders() {
1.188 + if (!eof) {
1.189 + fillBuf();
1.190 + }
1.191 + return new InternetHeaders(new LineInputStream());
1.192 + }
1.193 +
1.194 + /**
1.195 + * Reads and saves the part of the current attachment part's content.
1.196 + * At the end of this method, buf should have the remaining data
1.197 + * at index 0.
1.198 + *
1.199 + * @return a chunk of the part's content
1.200 + *
1.201 + */
1.202 + private ByteBuffer readBody() {
1.203 + if (!eof) {
1.204 + fillBuf();
1.205 + }
1.206 + int start = match(buf, 0, len); // matches boundary
1.207 + if (start == -1) {
1.208 + // No boundary is found
1.209 + assert eof || len >= config.chunkSize;
1.210 + int chunkSize = eof ? len : config.chunkSize;
1.211 + if (eof) {
1.212 + done = true;
1.213 + throw new MIMEParsingException("Reached EOF, but there is no closing MIME boundary.");
1.214 + }
1.215 + return adjustBuf(chunkSize, len-chunkSize);
1.216 + }
1.217 + // Found boundary.
1.218 + // Is it at the start of a line ?
1.219 + int chunkLen = start;
1.220 + if (bol && start == 0) {
1.221 + // nothing to do
1.222 + } else if (start > 0 && (buf[start-1] == '\n' || buf[start-1] =='\r')) {
1.223 + --chunkLen;
1.224 + if (buf[start-1] == '\n' && start >1 && buf[start-2] == '\r') {
1.225 + --chunkLen;
1.226 + }
1.227 + } else {
1.228 + return adjustBuf(start+1, len-start-1); // boundary is not at beginning of a line
1.229 + }
1.230 +
1.231 + if (start+bl+1 < len && buf[start+bl] == '-' && buf[start+bl+1] == '-') {
1.232 + state = STATE.END_PART;
1.233 + done = true;
1.234 + return adjustBuf(chunkLen, 0);
1.235 + }
1.236 +
1.237 + // Consider all the whitespace in boundary+whitespace+"\r\n"
1.238 + int lwsp = 0;
1.239 + for(int i=start+bl; i < len && (buf[i] == ' ' || buf[i] == '\t'); i++) {
1.240 + ++lwsp;
1.241 + }
1.242 +
1.243 + // Check for \n or \r\n in boundary+whitespace+"\n" or boundary+whitespace+"\r\n"
1.244 + if (start+bl+lwsp < len && buf[start+bl+lwsp] == '\n') {
1.245 + state = STATE.END_PART;
1.246 + return adjustBuf(chunkLen, len-start-bl-lwsp-1);
1.247 + } else if (start+bl+lwsp+1 < len && buf[start+bl+lwsp] == '\r' && buf[start+bl+lwsp+1] == '\n') {
1.248 + state = STATE.END_PART;
1.249 + return adjustBuf(chunkLen, len-start-bl-lwsp-2);
1.250 + } else if (start+bl+lwsp+1 < len) {
1.251 + return adjustBuf(chunkLen+1, len-chunkLen-1); // boundary string in a part data
1.252 + } else if (eof) {
1.253 + done = true;
1.254 + throw new MIMEParsingException("Reached EOF, but there is no closing MIME boundary.");
1.255 + }
1.256 +
1.257 + // Some more data needed to determine if it is indeed a proper boundary
1.258 + return adjustBuf(chunkLen, len-chunkLen);
1.259 + }
1.260 +
1.261 + /**
1.262 + * Returns a chunk from the original buffer. A new buffer is
1.263 + * created with the remaining bytes.
1.264 + *
1.265 + * @param chunkSize create a chunk with these many bytes
1.266 + * @param remaining bytes from the end of the buffer that need to be copied to
1.267 + * the beginning of the new buffer
1.268 + * @return chunk
1.269 + */
1.270 + private ByteBuffer adjustBuf(int chunkSize, int remaining) {
1.271 + assert buf != null;
1.272 + assert chunkSize >= 0;
1.273 + assert remaining >= 0;
1.274 +
1.275 + byte[] temp = buf;
1.276 + // create a new buf and adjust it without this chunk
1.277 + createBuf(remaining);
1.278 + System.arraycopy(temp, len-remaining, buf, 0, remaining);
1.279 + len = remaining;
1.280 +
1.281 + return ByteBuffer.wrap(temp, 0, chunkSize);
1.282 + }
1.283 +
1.284 + private void createBuf(int min) {
1.285 + buf = new byte[min < capacity ? capacity : min];
1.286 + }
1.287 +
1.288 + /**
1.289 + * Skips the preamble to find the first attachment part
1.290 + */
1.291 + private void skipPreamble() {
1.292 +
1.293 + while(true) {
1.294 + if (!eof) {
1.295 + fillBuf();
1.296 + }
1.297 + int start = match(buf, 0, len); // matches boundary
1.298 + if (start == -1) {
1.299 + // No boundary is found
1.300 + if (eof) {
1.301 + throw new MIMEParsingException("Missing start boundary");
1.302 + } else {
1.303 + adjustBuf(len-bl+1, bl-1);
1.304 + continue;
1.305 + }
1.306 + }
1.307 +
1.308 + if (start > config.chunkSize) {
1.309 + adjustBuf(start, len-start);
1.310 + continue;
1.311 + }
1.312 + // Consider all the whitespace boundary+whitespace+"\r\n"
1.313 + int lwsp = 0;
1.314 + for(int i=start+bl; i < len && (buf[i] == ' ' || buf[i] == '\t'); i++) {
1.315 + ++lwsp;
1.316 + }
1.317 + // Check for \n or \r\n
1.318 + if (start+bl+lwsp < len && (buf[start+bl+lwsp] == '\n' || buf[start+bl+lwsp] == '\r') ) {
1.319 + if (buf[start+bl+lwsp] == '\n') {
1.320 + adjustBuf(start+bl+lwsp+1, len-start-bl-lwsp-1);
1.321 + break;
1.322 + } else if (start+bl+lwsp+1 < len && buf[start+bl+lwsp+1] == '\n') {
1.323 + adjustBuf(start+bl+lwsp+2, len-start-bl-lwsp-2);
1.324 + break;
1.325 + }
1.326 + }
1.327 + adjustBuf(start+1, len-start-1);
1.328 + }
1.329 + if (LOGGER.isLoggable(Level.FINE)) {LOGGER.log(Level.FINE, "Skipped the preamble. buffer len={0}", len);}
1.330 + }
1.331 +
1.332 + private static byte[] getBytes(String s) {
1.333 + char [] chars= s.toCharArray();
1.334 + int size = chars.length;
1.335 + byte[] bytes = new byte[size];
1.336 +
1.337 + for (int i = 0; i < size;) {
1.338 + bytes[i] = (byte) chars[i++];
1.339 + }
1.340 + return bytes;
1.341 + }
1.342 +
1.343 + /**
1.344 + * Boyer-Moore search method. Copied from java.util.regex.Pattern.java
1.345 + *
1.346 + * Pre calculates arrays needed to generate the bad character
1.347 + * shift and the good suffix shift. Only the last seven bits
1.348 + * are used to see if chars match; This keeps the tables small
1.349 + * and covers the heavily used ASCII range, but occasionally
1.350 + * results in an aliased match for the bad character shift.
1.351 + */
1.352 + private void compileBoundaryPattern() {
1.353 + int i, j;
1.354 +
1.355 + // Precalculate part of the bad character shift
1.356 + // It is a table for where in the pattern each
1.357 + // lower 7-bit value occurs
1.358 + for (i = 0; i < bndbytes.length; i++) {
1.359 + bcs[bndbytes[i]&0x7F] = i + 1;
1.360 + }
1.361 +
1.362 + // Precalculate the good suffix shift
1.363 + // i is the shift amount being considered
1.364 +NEXT: for (i = bndbytes.length; i > 0; i--) {
1.365 + // j is the beginning index of suffix being considered
1.366 + for (j = bndbytes.length - 1; j >= i; j--) {
1.367 + // Testing for good suffix
1.368 + if (bndbytes[j] == bndbytes[j-i]) {
1.369 + // src[j..len] is a good suffix
1.370 + gss[j-1] = i;
1.371 + } else {
1.372 + // No match. The array has already been
1.373 + // filled up with correct values before.
1.374 + continue NEXT;
1.375 + }
1.376 + }
1.377 + // This fills up the remaining of optoSft
1.378 + // any suffix can not have larger shift amount
1.379 + // then its sub-suffix. Why???
1.380 + while (j > 0) {
1.381 + gss[--j] = i;
1.382 + }
1.383 + }
1.384 + // Set the guard value because of unicode compression
1.385 + gss[bndbytes.length -1] = 1;
1.386 + }
1.387 +
1.388 + /**
1.389 + * Finds the boundary in the given buffer using Boyer-Moore algo.
1.390 + * Copied from java.util.regex.Pattern.java
1.391 + *
1.392 + * @param mybuf boundary to be searched in this mybuf
1.393 + * @param off start index in mybuf
1.394 + * @param len number of bytes in mybuf
1.395 + *
1.396 + * @return -1 if there is no match or index where the match starts
1.397 + */
1.398 + private int match(byte[] mybuf, int off, int len) {
1.399 + int last = len - bndbytes.length;
1.400 +
1.401 + // Loop over all possible match positions in text
1.402 +NEXT: while (off <= last) {
1.403 + // Loop over pattern from right to left
1.404 + for (int j = bndbytes.length - 1; j >= 0; j--) {
1.405 + byte ch = mybuf[off+j];
1.406 + if (ch != bndbytes[j]) {
1.407 + // Shift search to the right by the maximum of the
1.408 + // bad character shift and the good suffix shift
1.409 + off += Math.max(j + 1 - bcs[ch&0x7F], gss[j]);
1.410 + continue NEXT;
1.411 + }
1.412 + }
1.413 + // Entire pattern matched starting at off
1.414 + return off;
1.415 + }
1.416 + return -1;
1.417 + }
1.418 +
1.419 + /**
1.420 + * Fills the remaining buf to the full capacity
1.421 + */
1.422 + private void fillBuf() {
1.423 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "Before fillBuf() buffer len={0}", len);}
1.424 + assert !eof;
1.425 + while(len < buf.length) {
1.426 + int read;
1.427 + try {
1.428 + read = in.read(buf, len, buf.length-len);
1.429 + } catch(IOException ioe) {
1.430 + throw new MIMEParsingException(ioe);
1.431 + }
1.432 + if (read == -1) {
1.433 + eof = true;
1.434 + try {
1.435 + if (LOGGER.isLoggable(Level.FINE)) {LOGGER.fine("Closing the input stream.");}
1.436 + in.close();
1.437 + } catch(IOException ioe) {
1.438 + throw new MIMEParsingException(ioe);
1.439 + }
1.440 + break;
1.441 + } else {
1.442 + len += read;
1.443 + }
1.444 + }
1.445 + if (LOGGER.isLoggable(Level.FINER)) {LOGGER.log(Level.FINER, "After fillBuf() buffer len={0}", len);}
1.446 + }
1.447 +
1.448 + private void doubleBuf() {
1.449 + byte[] temp = new byte[2*len];
1.450 + System.arraycopy(buf, 0, temp, 0, len);
1.451 + buf = temp;
1.452 + if (!eof) {
1.453 + fillBuf();
1.454 + }
1.455 + }
1.456 +
1.457 + class LineInputStream {
1.458 + private int offset;
1.459 +
1.460 + /*
1.461 + * Read a line containing only ASCII characters from the input
1.462 + * stream. A line is terminated by a CR or NL or CR-NL sequence.
1.463 + * A common error is a CR-CR-NL sequence, which will also terminate
1.464 + * a line.
1.465 + * The line terminator is not returned as part of the returned
1.466 + * String. Returns null if no data is available. <p>
1.467 + *
1.468 + * This class is similar to the deprecated
1.469 + * <code>DataInputStream.readLine()</code>
1.470 + */
1.471 + public String readLine() throws IOException {
1.472 +
1.473 + int hdrLen = 0;
1.474 + int lwsp = 0;
1.475 + while(offset+hdrLen < len) {
1.476 + if (buf[offset+hdrLen] == '\n') {
1.477 + lwsp = 1;
1.478 + break;
1.479 + }
1.480 + if (offset+hdrLen+1 == len) {
1.481 + doubleBuf();
1.482 + }
1.483 + if (offset+hdrLen+1 >= len) { // No more data in the stream
1.484 + assert eof;
1.485 + return null;
1.486 + }
1.487 + if (buf[offset+hdrLen] == '\r' && buf[offset+hdrLen+1] == '\n') {
1.488 + lwsp = 2;
1.489 + break;
1.490 + }
1.491 + ++hdrLen;
1.492 + }
1.493 + if (hdrLen == 0) {
1.494 + adjustBuf(offset+lwsp, len-offset-lwsp);
1.495 + return null;
1.496 + }
1.497 +
1.498 + String hdr = new String(buf, offset, hdrLen, HEADER_ENCODING);
1.499 + offset += hdrLen+lwsp;
1.500 + return hdr;
1.501 + }
1.502 +
1.503 + }
1.504 +
1.505 +}
