changeset 112:d488c7638b87

Prototype Filter support with keyword filter as a playground
author Artem Tikhomirov <tikhomirov.artem@gmail.com>
date Tue, 01 Feb 2011 00:21:08 +0100 (2011-01-31)
parents 32e794c599d7
children 67ae317408c9
files TODO src/org/tmatesoft/hg/internal/Filter.java src/org/tmatesoft/hg/internal/KeywordFilter.java
diffstat 3 files changed, 305 insertions(+), 0 deletions(-) [+]
line wrap: on
line diff
--- a/TODO	Tue Feb 01 00:19:31 2011 +0100
+++ b/TODO	Tue Feb 01 00:21:08 2011 +0100
@@ -34,6 +34,7 @@
   * Tags are read and can be queried (cmdline Log does)
   
 - keywords
+  - filter with context. filters shall have weight (to allow certain filter come first)
 
 - newlines
 
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/src/org/tmatesoft/hg/internal/Filter.java	Tue Feb 01 00:21:08 2011 +0100
@@ -0,0 +1,36 @@
+/*
+ * Copyright (c) 2011 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@svnkit.com
+ */
+package org.tmatesoft.hg.internal;
+
+import java.nio.ByteBuffer;
+
+import org.tmatesoft.hg.core.Path;
+import org.tmatesoft.hg.repo.HgRepository;
+
+/**
+ *
+ * @author Artem Tikhomirov
+ * @author TMate Software Ltd.
+ */
+public interface Filter {
+
+	ByteBuffer filter(ByteBuffer src);
+
+	interface Factory {
+		Filter create(HgRepository hgRepo, Path path);
+	}
+}
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/src/org/tmatesoft/hg/internal/KeywordFilter.java	Tue Feb 01 00:21:08 2011 +0100
@@ -0,0 +1,268 @@
+/*
+ * Copyright (c) 2011 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@svnkit.com
+ */
+package org.tmatesoft.hg.internal;
+
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.FileOutputStream;
+import java.nio.ByteBuffer;
+import java.util.TreeMap;
+
+import org.tmatesoft.hg.core.Path;
+import org.tmatesoft.hg.repo.HgRepository;
+
+/**
+ *
+ * @author Artem Tikhomirov
+ * @author TMate Software Ltd.
+ */
+public class KeywordFilter implements Filter {
+	private final boolean isExpanding;
+	private final TreeMap<String,String> keywords;
+	private final int minBufferLen;
+
+	/**
+	 * 
+	 * @param expand <code>true</code> to expand keywords, <code>false</code> to shrink
+	 */
+	private KeywordFilter(boolean expand) {
+		isExpanding = expand;
+		keywords = new TreeMap<String,String>();
+		keywords.put("Id", "Id");
+		keywords.put("Revision", "Revision");
+		keywords.put("Author", "Author");
+		int l = 0;
+		for (String s : keywords.keySet()) {
+			if (s.length() > l) {
+				l = s.length();
+			}
+		}
+		// FIXME later may implement #filter() not to read full kw value (just "$kw:"). However, limit of maxLen + 2 would keep valid.
+		// for buffers less then minBufferLen, there are chances #filter() implementation would never end
+		// (i.e. for input "$LongestKey"$
+		minBufferLen = l + 2 + (isExpanding ? 0 : 120 /*any reasonable constant for max possible kw value length*/);
+	}
+
+	/**
+	 * @param src buffer ready to be read
+	 * @return buffer ready to be read and original buffer's position modified to reflect consumed bytes. IOW, if source buffer
+	 * on return has remaining bytes, they are assumed not-read (not processed) and next chunk passed to filter is supposed to 
+	 * start with them  
+	 */
+	public ByteBuffer filter(ByteBuffer src) {
+		if (src.capacity() < minBufferLen) {
+			throw new IllegalStateException(String.format("Need buffer of at least %d bytes to ensure filter won't hang", minBufferLen));
+		}
+		ByteBuffer rv = null;
+		int keywordStart = -1;
+		int x = src.position();
+		while (x < src.limit()) {
+			if (keywordStart == -1) {
+				int i = indexOf(src, '$', x, false);
+				if (i == -1) {
+					if (rv == null) {
+						return src;
+					} else {
+						copySlice(src, x, src.limit(), rv);
+						rv.flip();
+						src.position(src.limit());
+						return rv;
+					}
+				}
+				keywordStart = i;
+				// fall-through
+			}
+			if (keywordStart >= 0) {
+				int i = indexOf(src, '$', keywordStart+1, true);
+				if (i == -1) {
+					// end of buffer reached
+					if (rv == null) {
+						if (keywordStart == x) {
+							// FIXME in fact, x might be equal to keywordStart and to src.position() here ('$' is first character in the buffer, 
+							// and there are no other '$' not eols till the end of the buffer). This would lead to deadlock (filter won't consume any
+							// bytes). To prevent this, either shall copy bytes [keywordStart..buffer.limit()) to local buffer and use it on the next invocation,
+							// or add lookup of the keywords right after first '$' is found (do not wait for closing '$'). For now, large enough src buffer would be sufficient
+							// not to run into such situation
+							throw new IllegalStateException("Try src buffer of a greater size");
+						}
+						rv = ByteBuffer.allocateDirect(keywordStart - x);
+					}
+					// copy all from source till latest possible kw start 
+					copySlice(src, x, keywordStart, rv);
+					rv.flip();
+					// and tell caller we've consumed only to the potential kw start
+					src.position(keywordStart);
+					return rv;
+				} else if (src.get(i) == '$') {
+					// end of keyword, or start of a new one.
+					String keyword;
+					if ((keyword = matchKeyword(src, keywordStart, i)) != null) {
+						if (rv == null) {
+							rv = ByteBuffer.allocateDirect(isExpanding ? src.capacity() * 4 : src.capacity());
+						}
+						copySlice(src, x, keywordStart+1, rv);
+						rv.put(keyword.getBytes());
+						if (isExpanding) {
+							rv.put((byte) ':');
+							rv.put((byte) ' ');
+							expandKeywordValue(keyword, rv);
+							rv.put((byte) ' ');
+						}
+						rv.put((byte) '$');
+						keywordStart = -1;
+						x = i+1;
+						continue;
+					} else {
+						if (rv != null) {
+							// we've already did some substitution, thus need to copy bytes we've scanned. 
+							copySlice(src, x, i, rv);
+						} // no else in attempt to avoid rv creation if no real kw would be found  
+						keywordStart = i;
+						x = i; // '$' at i wasn't consumed, hence x points to i, not i+1. This is to avoid problems with case: "sdfsd $ asdfs $Id$ sdf"
+						continue;
+					}
+				} else {
+					assert src.get(i) == '\n' || src.get(i) == '\r';
+					// line break
+					if (rv != null) {
+						copySlice(src, x, i+1, rv);
+					}
+					x = i+1;
+					keywordStart = -1; // Wasn't keyword, really
+					continue; // try once again
+				}
+			}
+		}
+		if (keywordStart != -1) {
+			if (rv == null) {
+				// no expansion happened yet, and we have potential kw start
+				rv = ByteBuffer.allocateDirect(keywordStart - src.position());
+				copySlice(src, src.position(), keywordStart, rv);
+			}
+			src.position(keywordStart);
+		}
+		if (rv != null) {
+			rv.flip();
+			return rv;
+		}
+		return src;
+	}
+	
+	/**
+	 * @param keyword
+	 * @param rv
+	 */
+	private void expandKeywordValue(String keyword, ByteBuffer rv) {
+		if ("Id".equals(keyword)) {
+			rv.put(identityString().getBytes());
+		} else if ("Revision".equals(keyword)) {
+			rv.put(revision());
+		} else if ("Author".equals(keyword)) {
+			rv.put(username().getBytes());
+		}
+	}
+
+	private String matchKeyword(ByteBuffer src, int kwStart, int kwEnd) {
+		assert kwEnd - kwStart - 1 > 0;
+		assert src.get(kwStart) == src.get(kwEnd) && src.get(kwEnd) == '$';
+		char[] chars = new char[kwEnd - kwStart - 1];
+		int i;
+		for (i = 0; i < chars.length; i++) {
+			char c = (char) src.get(kwStart + 1 + i);
+			if (c == ':') {
+				break;
+			}
+			chars[i] = c;
+		}
+		String kw = new String(chars, 0, i);
+		return keywords.get(kw);
+	}
+	
+	// copies part of the src buffer, [from..to). doesn't modify src position
+	private static void copySlice(ByteBuffer src, int from, int to, ByteBuffer dst) {
+		if (to > src.limit()) {
+			throw new IllegalArgumentException("Bad right boundary");
+		}
+		if (dst.remaining() < to - from) {
+			throw new IllegalArgumentException("Not enough room in the destination buffer");
+		}
+		for (int i = from; i < to; i++) {
+			dst.put(src.get(i));
+		}
+	}
+
+	private static int indexOf(ByteBuffer b, char ch, int from, boolean newlineBreaks) {
+	for (int i = from; i < b.limit(); i++) {
+			byte c = b.get(i);
+			if (ch == c) {
+				return i;
+			}
+			if (newlineBreaks && (c == '\n' || c == '\r')) {
+				return i;
+			}
+		}
+		return -1;
+	}
+
+	private String identityString() {
+		return "sample/file.txt, asd";
+	}
+
+	private byte[] revision() {
+		return "1234567890ab".getBytes();
+	}
+	
+	private String username() {
+		/* ui.py: username()
+        Searched in this order: $HGUSER, [ui] section of hgrcs, $EMAIL
+        and stop searching if one of these is set.
+        If not found and ui.askusername is True, ask the user, else use
+        ($LOGNAME or $USER or $LNAME or $USERNAME) + "@full.hostname".
+        */
+		return "<Sample> sample@sample.org";
+	}
+
+	public static class Factory implements Filter.Factory {
+
+		public Filter create(HgRepository hgRepo, Path path) {
+			return new KeywordFilter(true);
+		}
+	}
+
+
+	public static void main(String[] args) throws Exception {
+		FileInputStream fis = new FileInputStream(new File("/temp/kwoutput.txt"));
+		FileOutputStream fos = new FileOutputStream(new File("/temp/kwoutput2.txt"));
+		ByteBuffer b = ByteBuffer.allocateDirect(256);
+		KeywordFilter kwFilter = new KeywordFilter(false);
+		while (fis.getChannel().read(b) != -1) {
+			b.flip(); // get ready to be read
+			ByteBuffer f = kwFilter.filter(b);
+			fos.getChannel().write(f);
+			if (b.hasRemaining()) {
+				ByteBuffer remaining = b.slice();
+				b.clear();
+				b.put(remaining);
+			} else {
+				b.clear();
+			}
+		}
+		fis.close();
+		fos.flush();
+		fos.close();
+	}
+}