tikhomirov@74: /* tikhomirov@74: * Copyright (c) 2011 TMate Software Ltd tikhomirov@74: * tikhomirov@74: * This program is free software; you can redistribute it and/or modify tikhomirov@74: * it under the terms of the GNU General Public License as published by tikhomirov@74: * the Free Software Foundation; version 2 of the License. tikhomirov@74: * tikhomirov@74: * This program is distributed in the hope that it will be useful, tikhomirov@74: * but WITHOUT ANY WARRANTY; without even the implied warranty of tikhomirov@74: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the tikhomirov@74: * GNU General Public License for more details. tikhomirov@74: * tikhomirov@74: * For information on how to redistribute this software under tikhomirov@74: * the terms of a license other than GNU General Public License tikhomirov@74: * contact TMate Software at support@svnkit.com tikhomirov@74: */ tikhomirov@74: package org.tmatesoft.hg.internal; tikhomirov@74: tikhomirov@74: import java.util.Arrays; tikhomirov@74: import java.util.TreeSet; tikhomirov@74: tikhomirov@74: import org.tmatesoft.hg.repo.HgRepository; tikhomirov@74: import org.tmatesoft.hg.util.PathRewrite; tikhomirov@74: tikhomirov@74: /** tikhomirov@74: * @see http://mercurial.selenic.com/wiki/CaseFoldingPlan tikhomirov@80: * @see http://mercurial.selenic.com/wiki/fncacheRepoFormat tikhomirov@74: * tikhomirov@74: * @author Artem Tikhomirov tikhomirov@74: * @author TMate Software Ltd. tikhomirov@74: */ tikhomirov@74: class StoragePathHelper implements PathRewrite { tikhomirov@74: tikhomirov@74: private final boolean store; tikhomirov@74: private final boolean fncache; tikhomirov@74: private final boolean dotencode; tikhomirov@74: tikhomirov@74: public StoragePathHelper(boolean isStore, boolean isFncache, boolean isDotencode) { tikhomirov@74: store = isStore; tikhomirov@74: fncache = isFncache; tikhomirov@74: dotencode = isDotencode; tikhomirov@74: } tikhomirov@74: tikhomirov@74: // FIXME document what path argument is, whether it includes .i or .d, and whether it's 'normalized' (slashes) or not. tikhomirov@74: // since .hg/store keeps both .i files and files without extension (e.g. fncache), guees, for data == false tikhomirov@74: // we shall assume path has extension tikhomirov@74: // FIXME much more to be done, see store.py:_hybridencode tikhomirov@74: public String rewrite(String path) { tikhomirov@74: final String STR_STORE = "store/"; tikhomirov@74: final String STR_DATA = "data/"; tikhomirov@74: final String STR_DH = "dh/"; tikhomirov@74: tikhomirov@74: path = path.replace(".hg/", ".hg.hg/").replace(".i/", ".i.hg/").replace(".d/", ".d.hg/"); tikhomirov@74: StringBuilder sb = new StringBuilder(path.length() << 1); tikhomirov@74: if (store || fncache) { tikhomirov@74: // encodefilename tikhomirov@74: final String reservedChars = "\\:*?\"<>|"; tikhomirov@74: // in fact, \\ is unlikely to match, ever - we've replaced all of them already, above. Just regards to store.py tikhomirov@74: int x; tikhomirov@74: char[] hexByte = new char[2]; tikhomirov@74: for (int i = 0; i < path.length(); i++) { tikhomirov@74: final char ch = path.charAt(i); tikhomirov@74: if (ch >= 'a' && ch <= 'z') { tikhomirov@74: sb.append(ch); // POIRAE tikhomirov@74: } else if (ch >= 'A' && ch <= 'Z') { tikhomirov@74: sb.append('_'); tikhomirov@74: sb.append(Character.toLowerCase(ch)); // Perhaps, (char) (((int) ch) + 32)? Even better, |= 0x20? tikhomirov@74: } else if ( (x = reservedChars.indexOf(ch)) != -1) { tikhomirov@74: sb.append('~'); tikhomirov@74: sb.append(toHexByte(reservedChars.charAt(x), hexByte)); tikhomirov@74: } else if ((ch >= '~' /*126*/ && ch <= 255) || ch < ' ' /*32*/) { tikhomirov@74: sb.append('~'); tikhomirov@74: sb.append(toHexByte(ch, hexByte)); tikhomirov@74: } else if (ch == '_') { tikhomirov@74: // note, encoding from store.py:_buildencodefun and :_build_lower_encodefun tikhomirov@74: // differ in the way they process '_' (latter doesn't escape it) tikhomirov@74: sb.append('_'); tikhomirov@74: sb.append('_'); tikhomirov@74: } else { tikhomirov@74: sb.append(ch); tikhomirov@74: } tikhomirov@74: } tikhomirov@74: // auxencode tikhomirov@74: if (fncache) { tikhomirov@74: x = 0; // last segment start tikhomirov@74: final TreeSet windowsReservedFilenames = new TreeSet(); tikhomirov@74: 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(" "))); tikhomirov@74: do { tikhomirov@74: int i = sb.indexOf("/", x); tikhomirov@74: if (i == -1) { tikhomirov@74: i = sb.length(); tikhomirov@74: } tikhomirov@74: // windows reserved filenames are at least of length 3 tikhomirov@74: if (i - x >= 3) { tikhomirov@74: boolean found = false; tikhomirov@74: if (i-x == 3) { tikhomirov@74: found = windowsReservedFilenames.contains(sb.subSequence(x, i)); tikhomirov@74: } else if (sb.charAt(x+3) == '.') { // implicit i-x > 3 tikhomirov@74: found = windowsReservedFilenames.contains(sb.subSequence(x, x+3)); tikhomirov@74: } else if (i-x > 4 && sb.charAt(x+4) == '.') { tikhomirov@74: found = windowsReservedFilenames.contains(sb.subSequence(x, x+4)); tikhomirov@74: } tikhomirov@74: if (found) { tikhomirov@74: sb.setCharAt(x, '~'); tikhomirov@74: sb.insert(x+1, toHexByte(sb.charAt(x+2), hexByte)); tikhomirov@74: i += 2; tikhomirov@74: } tikhomirov@74: } tikhomirov@74: if (dotencode && (sb.charAt(x) == '.' || sb.charAt(x) == ' ')) { tikhomirov@74: sb.insert(x+1, toHexByte(sb.charAt(x), hexByte)); tikhomirov@74: sb.setCharAt(x, '~'); // setChar *after* charAt/insert to get ~2e, not ~7e for '.' tikhomirov@74: i += 2; tikhomirov@74: } tikhomirov@74: x = i+1; tikhomirov@74: } while (x < sb.length()); tikhomirov@74: } tikhomirov@74: } tikhomirov@74: final int MAX_PATH_LEN_IN_HGSTORE = 120; tikhomirov@74: if (fncache && (sb.length() + STR_DATA.length() > MAX_PATH_LEN_IN_HGSTORE)) { tikhomirov@74: throw HgRepository.notImplemented(); // FIXME digest and fncache use tikhomirov@74: } tikhomirov@74: if (store) { tikhomirov@74: sb.insert(0, STR_STORE + STR_DATA); tikhomirov@74: } tikhomirov@74: sb.append(".i"); tikhomirov@74: return sb.toString(); tikhomirov@74: } tikhomirov@74: tikhomirov@74: private static char[] toHexByte(int ch, char[] buf) { tikhomirov@74: assert buf.length > 1; tikhomirov@74: final String hexDigits = "0123456789abcdef"; tikhomirov@74: buf[0] = hexDigits.charAt((ch & 0x00F0) >>> 4); tikhomirov@74: buf[1] = hexDigits.charAt(ch & 0x0F); tikhomirov@74: return buf; tikhomirov@74: } tikhomirov@74: }