kitaev@213: /*
kitaev@213:  * Copyright (c) 2011 TMate Software Ltd
kitaev@213:  *  
kitaev@213:  * This program is free software; you can redistribute it and/or modify
kitaev@213:  * it under the terms of the GNU General Public License as published by
kitaev@213:  * the Free Software Foundation; version 2 of the License.
kitaev@213:  *
kitaev@213:  * This program is distributed in the hope that it will be useful,
kitaev@213:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
kitaev@213:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
kitaev@213:  * GNU General Public License for more details.
kitaev@213:  *
kitaev@213:  * For information on how to redistribute this software under
kitaev@213:  * the terms of a license other than GNU General Public License
kitaev@213:  * contact TMate Software at support@hg4j.com
kitaev@213:  */
kitaev@213: package org.tmatesoft.hg.internal;
kitaev@213: 
kitaev@213: import java.util.Arrays;
kitaev@213: import java.util.TreeSet;
kitaev@213: 
kitaev@213: import org.tmatesoft.hg.util.PathRewrite;
kitaev@213: 
kitaev@213: /**
kitaev@213:  * @see http://mercurial.selenic.com/wiki/CaseFoldingPlan
kitaev@213:  * @see http://mercurial.selenic.com/wiki/fncacheRepoFormat
kitaev@213:  * 
kitaev@213:  * @author Artem Tikhomirov
kitaev@213:  * @author TMate Software Ltd.
kitaev@213:  */
kitaev@213: class StoragePathHelper implements PathRewrite {
kitaev@213: 	
kitaev@213: 	private final boolean store;
kitaev@213: 	private final boolean fncache;
kitaev@213: 	private final boolean dotencode;
kitaev@213: 
kitaev@213: 	public StoragePathHelper(boolean isStore, boolean isFncache, boolean isDotencode) {
kitaev@213: 		store = isStore;
kitaev@213: 		fncache = isFncache;
kitaev@213: 		dotencode = isDotencode;
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	// FIXME document what path argument is, whether it includes .i or .d, and whether it's 'normalized' (slashes) or not.
kitaev@213: 	// since .hg/store keeps both .i files and files without extension (e.g. fncache), guees, for data == false 
kitaev@213: 	// we shall assume path has extension
kitaev@213: 	public String rewrite(String path) {
kitaev@213: 		final String STR_STORE = "store/";
kitaev@213: 		final String STR_DATA = "data/";
kitaev@213: 		final String STR_DH = "dh/";
kitaev@213: 		final String reservedChars = "\\:*?\"<>|";
kitaev@213: 		char[] hexByte = new char[2];
kitaev@213: 		
kitaev@213: 		path = path.replace(".hg/", ".hg.hg/").replace(".i/", ".i.hg/").replace(".d/", ".d.hg/");
kitaev@213: 		StringBuilder sb = new StringBuilder(path.length() << 1);
kitaev@213: 		if (store || fncache) {
kitaev@213: 			// encodefilename
kitaev@213: 			for (int i = 0; i < path.length(); i++) {
kitaev@213: 				final char ch = path.charAt(i);
kitaev@213: 				if (ch >= 'a' && ch <= 'z') {
kitaev@213: 					sb.append(ch); // POIRAE
kitaev@213: 				} else if (ch >= 'A' && ch <= 'Z') {
kitaev@213: 					sb.append('_');
kitaev@213: 					sb.append(Character.toLowerCase(ch)); // Perhaps, (char) (((int) ch) + 32)? Even better, |= 0x20? 
kitaev@213: 				} else if (reservedChars.indexOf(ch) != -1) {
kitaev@213: 					sb.append('~');
kitaev@213: 					sb.append(toHexByte(ch, hexByte));
kitaev@213: 				} else if ((ch >= '~' /*126*/ && ch <= 255) || ch < ' ' /*32*/) {
kitaev@213: 					sb.append('~');
kitaev@213: 					sb.append(toHexByte(ch, hexByte));
kitaev@213: 				} else if (ch == '_') {
kitaev@213: 					sb.append('_');
kitaev@213: 					sb.append('_');
kitaev@213: 				} else {
kitaev@213: 					sb.append(ch);
kitaev@213: 				}
kitaev@213: 			}
kitaev@213: 			// auxencode
kitaev@213: 			if (fncache) {
kitaev@213: 				encodeWindowsDeviceNames(sb);
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 		final int MAX_PATH_LEN = 120;
kitaev@213: 		if (fncache && (sb.length() + STR_DATA.length() + ".i".length() > MAX_PATH_LEN)) {
kitaev@213: 			String digest = new DigestHelper().sha1(STR_DATA, path, ".i").asHexString();
kitaev@213: 			final int DIR_PREFIX_LEN = 8;
kitaev@213: 			 // not sure why (-4) is here. 120 - 40 = up to 80 for path with ext. dh/ + ext(.i) = 3+2
kitaev@213: 			final int MAX_DIR_PREFIX = 8 * (DIR_PREFIX_LEN + 1) - 4;
kitaev@213: 			sb = new StringBuilder(MAX_PATH_LEN);
kitaev@213: 			for (int i = 0; i < path.length(); i++) {
kitaev@213: 				final char ch = path.charAt(i);
kitaev@213: 				if (ch >= 'a' && ch <= 'z') {
kitaev@213: 					sb.append(ch);
kitaev@213: 				} else if (ch >= 'A' && ch <= 'Z') {
kitaev@213: 					sb.append((char) (ch | 0x20)); // lowercase 
kitaev@213: 				} else if (reservedChars.indexOf(ch) != -1) {
kitaev@213: 					sb.append('~');
kitaev@213: 					sb.append(toHexByte(ch, hexByte));
kitaev@213: 				} else if ((ch >= '~' /*126*/ && ch <= 255) || ch < ' ' /*32*/) {
kitaev@213: 					sb.append('~');
kitaev@213: 					sb.append(toHexByte(ch, hexByte));
kitaev@213: 				} else {
kitaev@213: 					sb.append(ch);
kitaev@213: 				}
kitaev@213: 			}
kitaev@213: 			encodeWindowsDeviceNames(sb);
kitaev@213: 			int fnameStart = sb.lastIndexOf("/"); // since we rewrite file names, it never ends with slash (for dirs, I'd pass length-2);
kitaev@213: 			StringBuilder completeHashName = new StringBuilder(MAX_PATH_LEN);
kitaev@213: 			completeHashName.append(STR_STORE);
kitaev@213: 			completeHashName.append(STR_DH);
kitaev@213: 			if (fnameStart == -1) {
kitaev@213: 				// no dirs, just long filename
kitaev@213: 				sb.setLength(MAX_PATH_LEN - 40 /*digest.length()*/ - STR_DH.length() - ".i".length());
kitaev@213: 				completeHashName.append(sb);
kitaev@213: 			} else {
kitaev@213: 				StringBuilder sb2 = new StringBuilder(MAX_PATH_LEN);
kitaev@213: 				int x = 0;
kitaev@213: 				do {
kitaev@213: 					int i = sb.indexOf("/", x);
kitaev@213: 					final int sb2Len = sb2.length(); 
kitaev@213: 					if (i-x <= DIR_PREFIX_LEN) { // a b c d e f g h /
kitaev@213: 						sb2.append(sb, x, i + 1); // with slash
kitaev@213: 					} else {
kitaev@213: 						sb2.append(sb, x, x + DIR_PREFIX_LEN);
kitaev@213: 						// may unexpectedly end with bad character
kitaev@213: 						final int last = sb2.length()-1;
kitaev@213: 						char lastChar = sb2.charAt(last); 
kitaev@213: 						assert lastChar == sb.charAt(x + DIR_PREFIX_LEN - 1);
kitaev@213: 						if (lastChar == '.' || lastChar == ' ') {
kitaev@213: 							sb2.setCharAt(last, '_');
kitaev@213: 						}
kitaev@213: 						sb2.append('/');
kitaev@213: 					}
kitaev@213: 					if (sb2.length()-1 > MAX_DIR_PREFIX) {
kitaev@213: 						sb2.setLength(sb2Len); // strip off last segment, it's too much
kitaev@213: 						break;
kitaev@213: 					}
kitaev@213: 					x = i+1; 
kitaev@213: 				} while (x < fnameStart);
kitaev@213: 				assert sb2.charAt(sb2.length() - 1) == '/';
kitaev@213: 				int left = MAX_PATH_LEN - sb2.length() - 40 /*digest.length()*/ - STR_DH.length() - ".i".length();
kitaev@213: 				assert left >= 0;
kitaev@213: 				fnameStart++; // move from / to actual name
kitaev@213: 				sb2.append(sb, fnameStart, fnameStart + left > sb.length() ? sb.length() : fnameStart+left);
kitaev@213: 				completeHashName.append(sb2);
kitaev@213: 			}
kitaev@213: 			completeHashName.append(digest);
kitaev@213: 			sb = completeHashName;
kitaev@213: 		} else if (store) {
kitaev@213: 			sb.insert(0, STR_STORE + STR_DATA);
kitaev@213: 		}
kitaev@213: 		sb.append(".i");
kitaev@213: 		return sb.toString();
kitaev@213: 	}
kitaev@213: 	
kitaev@213: 	private void encodeWindowsDeviceNames(StringBuilder sb) {
kitaev@213: 		char[] hexByte = new char[2];
kitaev@213: 		int x = 0; // last segment start
kitaev@213: 		final TreeSet<String> windowsReservedFilenames = new TreeSet<String>();
kitaev@213: 		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(" "))); 
kitaev@213: 		do {
kitaev@213: 			int i = sb.indexOf("/", x);
kitaev@213: 			if (i == -1) {
kitaev@213: 				i = sb.length();
kitaev@213: 			}
kitaev@213: 			// windows reserved filenames are at least of length 3 
kitaev@213: 			if (i - x >= 3) {
kitaev@213: 				boolean found = false;
kitaev@213: 				if (i-x == 3 || i-x == 4) {
kitaev@213: 					found = windowsReservedFilenames.contains(sb.subSequence(x, i));
kitaev@213: 				} else if (sb.charAt(x+3) == '.') { // implicit i-x > 3
kitaev@213: 					found = windowsReservedFilenames.contains(sb.subSequence(x, x+3));
kitaev@213: 				} else if (i-x > 4 && sb.charAt(x+4) == '.') {
kitaev@213: 					found = windowsReservedFilenames.contains(sb.subSequence(x, x+4));
kitaev@213: 				}
kitaev@213: 				if (found) {
kitaev@213: 					sb.insert(x+3, toHexByte(sb.charAt(x+2), hexByte));
kitaev@213: 					sb.setCharAt(x+2, '~');
kitaev@213: 					i += 2;
kitaev@213: 				}
kitaev@213: 			}
kitaev@213: 			if (dotencode && (sb.charAt(x) == '.' || sb.charAt(x) == ' ')) {
kitaev@213: 				sb.insert(x+1, toHexByte(sb.charAt(x), hexByte));
kitaev@213: 				sb.setCharAt(x, '~'); // setChar *after* charAt/insert to get ~2e, not ~7e for '.'
kitaev@213: 				i += 2;
kitaev@213: 			}
kitaev@213: 			x = i+1;
kitaev@213: 		} while (x < sb.length());
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	private static char[] toHexByte(int ch, char[] buf) {
kitaev@213: 		assert buf.length > 1;
kitaev@213: 		final String hexDigits = "0123456789abcdef";
kitaev@213: 		buf[0] = hexDigits.charAt((ch & 0x00F0) >>> 4);
kitaev@213: 		buf[1] = hexDigits.charAt(ch & 0x0F);
kitaev@213: 		return buf;
kitaev@213: 	}
kitaev@213: }