Mercurial > jhg
view src/org/tmatesoft/hg/internal/StoragePathHelper.java @ 576:3c4db86e8c1f
Issue 43: poor performance with InflaterDataAccess. Phase 2: inflate into buffer, effective skip and readByte/readBytes()
author | Artem Tikhomirov <tikhomirov.artem@gmail.com> |
---|---|
date | Tue, 16 Apr 2013 19:31:57 +0200 |
parents | 48f993aa2f41 |
children | 5e0313485eef |
line wrap: on
line source
/* * Copyright (c) 2011-2012 TMate Software Ltd * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; version 2 of the License. * * This program 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 for more details. * * For information on how to redistribute this software under * the terms of a license other than GNU General Public License * contact TMate Software at support@hg4j.com */ package org.tmatesoft.hg.internal; import java.nio.ByteBuffer; import java.nio.CharBuffer; import java.nio.charset.Charset; import java.nio.charset.CharsetEncoder; import java.util.Arrays; import java.util.TreeSet; import java.util.regex.Matcher; import java.util.regex.Pattern; import org.tmatesoft.hg.util.PathRewrite; /** * @see http://mercurial.selenic.com/wiki/CaseFoldingPlan * @see http://mercurial.selenic.com/wiki/fncacheRepoFormat * @see http://mercurial.selenic.com/wiki/EncodingStrategy * * @author Artem Tikhomirov * @author TMate Software Ltd. */ class StoragePathHelper implements PathRewrite { private final boolean store; private final boolean fncache; private final boolean dotencode; private final Pattern suffix2replace; private final CharsetEncoder csEncoder; private final char[] hexEncodedByte = new char[] {'~', '0', '0'}; private final ByteBuffer byteEncodingBuf; private final CharBuffer charEncodingBuf; public StoragePathHelper(boolean isStore, boolean isFncache, boolean isDotencode) { this(isStore, isFncache, isDotencode, Charset.defaultCharset()); } public StoragePathHelper(boolean isStore, boolean isFncache, boolean isDotencode, Charset fsEncoding) { assert fsEncoding != null; store = isStore; fncache = isFncache; dotencode = isDotencode; suffix2replace = Pattern.compile("\\.([id]|hg)/"); csEncoder = fsEncoding.newEncoder(); byteEncodingBuf = ByteBuffer.allocate(Math.round(csEncoder.maxBytesPerChar()) + 1/*in fact, need ceil, hence +1*/); charEncodingBuf = CharBuffer.allocate(1); } /** * path argument is repository-relative name of the user's file. * It has to be normalized (slashes) and shall not include extension .i or .d. */ public CharSequence rewrite(CharSequence p) { final String STR_STORE = "store/"; final String STR_DATA = "data/"; final String STR_DH = "dh/"; final String reservedChars = "\\:*?\"<>|"; Matcher suffixMatcher = suffix2replace.matcher(p); CharSequence path; // Matcher.replaceAll, but without extra toString boolean found = suffixMatcher.find(); if (found) { StringBuffer sb = new StringBuffer(p.length() + 20); do { suffixMatcher.appendReplacement(sb, ".$1.hg/"); } while (found = suffixMatcher.find()); suffixMatcher.appendTail(sb); path = sb; } else { path = p; } StringBuilder sb = new StringBuilder(path.length() << 1); if (store || fncache) { for (int i = 0; i < path.length(); i++) { final char ch = path.charAt(i); if (ch >= 'a' && ch <= 'z') { sb.append(ch); // POIRAE } else if (ch >= 'A' && ch <= 'Z') { sb.append('_'); sb.append(Character.toLowerCase(ch)); // Perhaps, (char) (((int) ch) + 32)? Even better, |= 0x20? } else if (reservedChars.indexOf(ch) != -1) { sb.append(toHexByte(ch)); } else if ((ch >= '~' /*126*/ && ch <= 255) || ch < ' ' /*32*/) { sb.append(toHexByte(ch)); } else if (ch == '_') { sb.append('_'); sb.append('_'); } else { // either ASCII char that doesn't require special handling, or an Unicode character to get encoded // according to filesystem/native encoding, see http://mercurial.selenic.com/wiki/EncodingStrategy // despite of what the page says, use of native encoding seems worst solution to me (repositories // can't be easily shared between OS'es with different encodings then, e.g. Win1251 and Linux UTF8). // If the ease of sharing was not the point, what's the reason to mangle with names at all then ( // lowercase and exclude reserved device names). if (ch < '~' /*126*/ || !csEncoder.canEncode(ch)) { sb.append(ch); } else { appendEncoded(sb, ch); } } } // auxencode if (fncache) { encodeWindowsDeviceNames(sb); } } final int MAX_PATH_LEN = 120; if (fncache && (sb.length() + STR_DATA.length() + ".i".length() > MAX_PATH_LEN)) { // TODO [post-1.0] Mercurial uses system encoding for paths, hence we need to pass bytes to DigestHelper // to ensure our sha1 value (default encoding of unicode string if one looks into DH impl) match that // produced by Mercurial (based on native string). String digest = new DigestHelper().sha1(STR_DATA, path, ".i").asHexString(); final int DIR_PREFIX_LEN = 8; // not sure why (-4) is here. 120 - 40 = up to 80 for path with ext. dh/ + ext(.i) = 3+2 final int MAX_DIR_PREFIX = 8 * (DIR_PREFIX_LEN + 1) - 4; sb = new StringBuilder(MAX_PATH_LEN); for (int i = 0; i < path.length(); i++) { final char ch = path.charAt(i); if (ch >= 'a' && ch <= 'z') { sb.append(ch); } else if (ch >= 'A' && ch <= 'Z') { sb.append((char) (ch | 0x20)); // lowercase } else if (reservedChars.indexOf(ch) != -1) { sb.append(toHexByte(ch)); } else if ((ch >= '~' /*126*/ && ch <= 255) || ch < ' ' /*32*/) { sb.append(toHexByte(ch)); } else { if (ch < '~' /*126*/ || !csEncoder.canEncode(ch)) { sb.append(ch); } else { appendEncoded(sb, ch); } } } encodeWindowsDeviceNames(sb); int fnameStart = sb.lastIndexOf("/"); // since we rewrite file names, it never ends with slash (for dirs, I'd pass length-2); StringBuilder completeHashName = new StringBuilder(MAX_PATH_LEN); completeHashName.append(STR_STORE); completeHashName.append(STR_DH); if (fnameStart == -1) { // no dirs, just long filename sb.setLength(MAX_PATH_LEN - 40 /*digest.length()*/ - STR_DH.length() - ".i".length()); completeHashName.append(sb); } else { StringBuilder sb2 = new StringBuilder(MAX_PATH_LEN); int x = 0; do { int i = sb.indexOf("/", x); final int sb2Len = sb2.length(); if (i-x <= DIR_PREFIX_LEN) { // a b c d e f g h / sb2.append(sb, x, i + 1); // with slash } else { sb2.append(sb, x, x + DIR_PREFIX_LEN); // may unexpectedly end with bad character final int last = sb2.length()-1; char lastChar = sb2.charAt(last); assert lastChar == sb.charAt(x + DIR_PREFIX_LEN - 1); if (lastChar == '.' || lastChar == ' ') { sb2.setCharAt(last, '_'); } sb2.append('/'); } if (sb2.length()-1 > MAX_DIR_PREFIX) { sb2.setLength(sb2Len); // strip off last segment, it's too much break; } x = i+1; } while (x < fnameStart); assert sb2.charAt(sb2.length() - 1) == '/'; int left = MAX_PATH_LEN - sb2.length() - 40 /*digest.length()*/ - STR_DH.length() - ".i".length(); assert left >= 0; fnameStart++; // move from / to actual name if (fnameStart + left > sb.length()) { // there left less chars in the mangled name that we can fit sb2.append(sb, fnameStart, sb.length()); int stillAvailable = (fnameStart+left) - sb.length(); // stillAvailable > 0; sb2.append(".i", 0, stillAvailable > 2 ? 2 : stillAvailable); } else { // add as much as we can sb2.append(sb, fnameStart, fnameStart+left); } completeHashName.append(sb2); } completeHashName.append(digest); sb = completeHashName; } else if (store) { sb.insert(0, STR_STORE + STR_DATA); } sb.append(".i"); return sb.toString(); } private void encodeWindowsDeviceNames(StringBuilder sb) { int x = 0; // last segment start final TreeSet<String> windowsReservedFilenames = new TreeSet<String>(); windowsReservedFilenames.addAll(Arrays.asList("con prn aux nul com1 com2 com3 com4 com5 com6 com7 com8 com9 lpt1 lpt2 lpt3 lpt4 lpt5 lpt6 lpt7 lpt8 lpt9".split(" "))); do { int i = sb.indexOf("/", x); if (i == -1) { i = sb.length(); } // windows reserved filenames are at least of length 3 if (i - x >= 3) { boolean found = false; if (i-x == 3 || i-x == 4) { found = windowsReservedFilenames.contains(sb.subSequence(x, i)); } else if (sb.charAt(x+3) == '.') { // implicit i-x > 3 found = windowsReservedFilenames.contains(sb.subSequence(x, x+3)); } else if (i-x > 4 && sb.charAt(x+4) == '.') { found = windowsReservedFilenames.contains(sb.subSequence(x, x+4)); } if (found) { // x+2 as we change the third letter in device name replace(sb, x+2, toHexByte(sb.charAt(x+2))); i += 2; } } if (dotencode && (sb.charAt(x) == '.' || sb.charAt(x) == ' ')) { char dotOrSpace = sb.charAt(x); // beware, replace() below changes charAt(x), rather get a copy // not to get ~7e for '.' instead of ~2e, if later refactoring changes the logic replace(sb, x, toHexByte(dotOrSpace)); i += 2; } x = i+1; } while (x < sb.length()); } // shall be synchronized in case of multithreaded use private void appendEncoded(StringBuilder sb, char ch) { charEncodingBuf.clear(); byteEncodingBuf.clear(); charEncodingBuf.put(ch).flip(); csEncoder.encode(charEncodingBuf, byteEncodingBuf, false); byteEncodingBuf.flip(); while (byteEncodingBuf.hasRemaining()) { sb.append(toHexByte(byteEncodingBuf.get())); } } /** * replace char at sb[index] with a sequence */ private static void replace(StringBuilder sb, int index, char[] with) { // there's StringBuilder.replace(int, int+1, String), but with char[] - I don't want to make a string out of hexEncodedByte sb.setCharAt(index, with[0]); sb.insert(index+1, with, 1, with.length - 1); } /** * put hex representation of byte ch into buf from specified offset */ private char[] toHexByte(int ch) { final String hexDigits = "0123456789abcdef"; hexEncodedByte[1] = hexDigits.charAt((ch & 0x00F0) >>> 4); hexEncodedByte[2] = hexDigits.charAt(ch & 0x0F); return hexEncodedByte; } }