kitaev@213: /*
kitaev@213:  * Copyright (c) 2010-2011 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.repo;
kitaev@213: 
kitaev@213: import java.io.BufferedReader;
kitaev@213: import java.io.File;
kitaev@213: import java.io.FileReader;
kitaev@213: import java.io.IOException;
kitaev@213: import java.util.ArrayList;
kitaev@213: import java.util.Collections;
kitaev@213: import java.util.List;
kitaev@213: import java.util.regex.Pattern;
kitaev@213: 
kitaev@213: import org.tmatesoft.hg.util.Path;
kitaev@213: 
kitaev@213: /**
kitaev@213:  * Handling of ignored paths according to .hgignore configuration
kitaev@213:  * 
kitaev@213:  * @author Artem Tikhomirov
kitaev@213:  * @author TMate Software Ltd.
kitaev@213:  */
kitaev@213: public class HgIgnore {
kitaev@213: 
kitaev@213: 	private List<Pattern> entries;
kitaev@213: 
kitaev@213: 	HgIgnore() {
kitaev@213: 		entries = Collections.emptyList();
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	/* package-local */void read(File hgignoreFile) throws IOException {
kitaev@213: 		if (!hgignoreFile.exists()) {
kitaev@213: 			return;
kitaev@213: 		}
kitaev@213: 		ArrayList<Pattern> result = new ArrayList<Pattern>(entries); // start with existing
kitaev@213: 		String syntax = "regex"; // or "glob"
kitaev@213: 		BufferedReader fr = new BufferedReader(new FileReader(hgignoreFile));
kitaev@213: 		String line;
kitaev@213: 		while ((line = fr.readLine()) != null) {
kitaev@213: 			line = line.trim();
kitaev@213: 			if (line.startsWith("syntax:")) {
kitaev@213: 				syntax = line.substring("syntax:".length()).trim();
kitaev@213: 				if (!"regex".equals(syntax) && !"glob".equals(syntax)) {
kitaev@213: 					throw new IllegalStateException(line);
kitaev@213: 				}
kitaev@213: 			} else if (line.length() > 0) {
kitaev@213: 				// shall I account for local paths in the file (i.e.
kitaev@213: 				// back-slashed on windows)?
kitaev@213: 				int x;
kitaev@213: 				if ((x = line.indexOf('#')) >= 0) {
kitaev@213: 					line = line.substring(0, x).trim();
kitaev@213: 					if (line.length() == 0) {
kitaev@213: 						continue;
kitaev@213: 					}
kitaev@213: 				}
kitaev@213: 				if ("glob".equals(syntax)) {
kitaev@213: 					// hgignore(5)
kitaev@213: 					// (http://www.selenic.com/mercurial/hgignore.5.html) says slashes '\' are escape characters,
kitaev@213: 					// hence no special  treatment of Windows path
kitaev@213: 					// however, own attempts make me think '\' on Windows are not treated as escapes
kitaev@213: 					line = glob2regex(line);
kitaev@213: 				}
kitaev@213: 				result.add(Pattern.compile(line)); // case-sensitive
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 		result.trimToSize();
kitaev@213: 		entries = result;
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	// note, #isIgnored(), even if queried for directories and returned positive reply, may still get
kitaev@213: 	// a file from that ignored folder to get examined. Thus, patterns like "bin" shall match not only a folder,
kitaev@213: 	// but any file under that folder as well
kitaev@213: 	// Alternatively, file walker may memorize folder is ignored and uses this information for all nested files. However,
kitaev@213: 	// this approach would require walker (a) return directories (b) provide nesting information. This may become
kitaev@213: 	// troublesome when one walks not over io.File, but Eclipse's IResource or any other custom VFS.
kitaev@213: 	//
kitaev@213: 	//
kitaev@213: 	// might be interesting, although looks like of no direct use in my case 
kitaev@213: 	// @see http://stackoverflow.com/questions/1247772/is-there-an-equivalent-of-java-util-regex-for-glob-type-patterns
kitaev@213: 	private String glob2regex(String line) {
kitaev@213: 		assert line.length() > 0;
kitaev@213: 		StringBuilder sb = new StringBuilder(line.length() + 10);
kitaev@213: 		sb.append('^'); // help avoid matcher.find() to match 'bin' pattern in the middle of the filename
kitaev@213: 		int start = 0, end = line.length() - 1;
kitaev@213: 		// '*' at the beginning and end of a line are useless for Pattern
kitaev@213: 		// XXX although how about **.txt - such globs can be seen in a config, are they valid for HgIgnore?
kitaev@213: 		while (start <= end && line.charAt(start) == '*') start++;
kitaev@213: 		while (end > start && line.charAt(end) == '*') end--;
kitaev@213: 
kitaev@213: 		for (int i = start; i <= end; i++) {
kitaev@213: 			char ch = line.charAt(i);
kitaev@213: 			if (ch == '.' || ch == '\\') {
kitaev@213: 				sb.append('\\');
kitaev@213: 			} else if (ch == '?') {
kitaev@213: 				// simple '.' substitution might work out, however, more formally 
kitaev@213: 				// a char class seems more appropriate to avoid accidentally
kitaev@213: 				// matching a subdirectory with ? char (i.e. /a/b?d against /a/bad, /a/bed and /a/b/d)
kitaev@213: 				// @see http://pubs.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_13_03
kitaev@213: 				// quote: "The slash character in a pathname shall be explicitly matched by using one or more slashes in the pattern; 
kitaev@213: 				// it shall neither be matched by the asterisk or question-mark special characters nor by a bracket expression" 
kitaev@213: 				sb.append("[^/]");
kitaev@213: 				continue;
kitaev@213: 			} else if (ch == '*') {
kitaev@213: 				sb.append("[^/]*?");
kitaev@213: 				continue;
kitaev@213: 			}
kitaev@213: 			sb.append(ch);
kitaev@213: 		}
kitaev@213: 		return sb.toString();
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	// TODO use PathGlobMatcher
kitaev@213: 	public boolean isIgnored(Path path) {
kitaev@213: 		for (Pattern p : entries) {
kitaev@213: 			if (p.matcher(path).find()) {
kitaev@213: 				return true;
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 		return false;
kitaev@213: 	}
kitaev@213: }