Browse Source

debug-text-hashfunctions: Test suite for content hashes

This is the test suite I was using to help understand why we had
such a high collision rate with RawTextComparator, and to select
a replacement function.

Since its not something we will run very often, lets make it a
program in the debug package rather than a JUnit test.  This way
we can run it on demand against any corpus of files we choose,
but we aren't bottlenecking our daily builds running tests with
no assertions.

Adding a new hash function to this suite is simple, just define
a new instance member of type "Hash" with the logic applied to
the region passed in.

Change-Id: Iec0b176adb464cf95b06cda157932b79c0b59886
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
stable-0.10
Shawn O. Pearce 14 years ago
parent
commit
af3fbb13f6
  1. 1
      org.eclipse.jgit.pgm/META-INF/services/org.eclipse.jgit.pgm.TextBuiltin
  2. 507
      org.eclipse.jgit.pgm/src/org/eclipse/jgit/pgm/debug/TextHashFunctions.java

1
org.eclipse.jgit.pgm/META-INF/services/org.eclipse.jgit.pgm.TextBuiltin

@ -31,6 +31,7 @@ org.eclipse.jgit.pgm.debug.ShowCacheTree
org.eclipse.jgit.pgm.debug.ShowCommands
org.eclipse.jgit.pgm.debug.ShowDirCache
org.eclipse.jgit.pgm.debug.ShowPackDelta
org.eclipse.jgit.pgm.debug.TextHashFunctions
org.eclipse.jgit.pgm.debug.WriteDirCache
org.eclipse.jgit.pgm.eclipse.Iplog

507
org.eclipse.jgit.pgm/src/org/eclipse/jgit/pgm/debug/TextHashFunctions.java

@ -0,0 +1,507 @@
/*
* Copyright (C) 2010, Google Inc.
* and other copyright owners as documented in the project's IP log.
*
* This program and the accompanying materials are made available
* under the terms of the Eclipse Distribution License v1.0 which
* accompanies this distribution, is reproduced below, and is
* available at http://www.eclipse.org/org/documents/edl-v10.php
*
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials provided
* with the distribution.
*
* - Neither the name of the Eclipse Foundation, Inc. nor the
* names of its contributors may be used to endorse or promote
* products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
* CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
package org.eclipse.jgit.pgm.debug;
import java.io.File;
import java.lang.reflect.Field;
import java.security.MessageDigest;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashSet;
import java.util.List;
import org.eclipse.jgit.diff.RawText;
import org.eclipse.jgit.diff.RawTextComparator;
import org.eclipse.jgit.errors.LargeObjectException;
import org.eclipse.jgit.lib.Constants;
import org.eclipse.jgit.lib.FileMode;
import org.eclipse.jgit.lib.MutableObjectId;
import org.eclipse.jgit.lib.ObjectReader;
import org.eclipse.jgit.lib.Repository;
import org.eclipse.jgit.lib.RepositoryBuilder;
import org.eclipse.jgit.lib.RepositoryCache;
import org.eclipse.jgit.pgm.CLIText;
import org.eclipse.jgit.pgm.TextBuiltin;
import org.eclipse.jgit.revwalk.RevWalk;
import org.eclipse.jgit.treewalk.TreeWalk;
import org.eclipse.jgit.util.FS;
import org.eclipse.jgit.util.NB;
import org.kohsuke.args4j.Option;
/**
* Scan repository to compute maximum number of collisions for hash functions.
*
* This is a test suite to help benchmark the collision rate of hash functions
* when applied to file contents in a Git repository. The test scans all text
* files in the HEAD revision of the repository it is run within. For each file
* it finds the unique lines, and then inserts those lines into a hash table to
* determine collision rates under the selected hash functions.
*
* To add another hash function to the test suite, declare a new instance member
* field of type {@link Hash} and implement the hashRegion method. The test
* suite will automatically pick up the new function through reflection.
*
* To add another folding function (method of squashing a 32 bit hash code into
* the hash tables smaller array index space), declare a new instance field of
* type {@link Fold} and implement the logic. The test suite will automatically
* pick up the new function through reflection.
*/
class TextHashFunctions extends TextBuiltin {
/** Standard SHA-1 on the line, using the first 4 bytes as the hash code. */
final Hash sha1 = new Hash() {
private final MessageDigest md = Constants.newMessageDigest();
@Override
protected int hashRegion(byte[] raw, int ptr, int end) {
md.reset();
md.update(raw, ptr, end - ptr);
return NB.decodeInt32(md.digest(), 0);
}
};
/** Professor Daniel J. Bernstein's rather popular string hash. */
final Hash djb = new Hash() {
@Override
protected int hashRegion(byte[] raw, int ptr, int end) {
int hash = 5381;
for (; ptr < end; ptr++)
hash = ((hash << 5) + hash) + (raw[ptr] & 0xff);
return hash;
}
};
/** Hash function commonly used by java.lang.String. */
final Hash string_hash31 = new Hash() {
@Override
protected int hashRegion(byte[] raw, int ptr, int end) {
int hash = 0;
for (; ptr < end; ptr++)
hash = 31 * hash + (raw[ptr] & 0xff);
return hash;
}
};
/** The Rabin polynomial hash that is used by our own DeltaIndex. */
final Hash rabin_DeltaIndex = new Hash() {
private final byte[] buf16 = new byte[16];
@Override
protected int hashRegion(byte[] raw, int ptr, int end) {
if (end - ptr < 16) {
Arrays.fill(buf16, (byte) 0);
System.arraycopy(raw, ptr, buf16, 0, end - ptr);
return rabin(buf16, 0);
} else {
return rabin(raw, ptr);
}
}
private int rabin(byte[] raw, int ptr) {
int hash;
// The first 4 steps collapse out into a 4 byte big-endian decode,
// with a larger right shift as we combined shift lefts together.
//
hash = ((raw[ptr] & 0xff) << 24) //
| ((raw[ptr + 1] & 0xff) << 16) //
| ((raw[ptr + 2] & 0xff) << 8) //
| (raw[ptr + 3] & 0xff);
hash ^= T[hash >>> 31];
hash = ((hash << 8) | (raw[ptr + 4] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 5] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 6] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 7] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 8] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 9] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 10] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 11] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 12] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 13] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 14] & 0xff)) ^ T[hash >>> 23];
hash = ((hash << 8) | (raw[ptr + 15] & 0xff)) ^ T[hash >>> 23];
return hash;
}
private final int[] T = { 0x00000000, 0xd4c6b32d, 0x7d4bd577,
0xa98d665a, 0x2e5119c3, 0xfa97aaee, 0x531accb4, 0x87dc7f99,
0x5ca23386, 0x886480ab, 0x21e9e6f1, 0xf52f55dc, 0x72f32a45,
0xa6359968, 0x0fb8ff32, 0xdb7e4c1f, 0x6d82d421, 0xb944670c,
0x10c90156, 0xc40fb27b, 0x43d3cde2, 0x97157ecf, 0x3e981895,
0xea5eabb8, 0x3120e7a7, 0xe5e6548a, 0x4c6b32d0, 0x98ad81fd,
0x1f71fe64, 0xcbb74d49, 0x623a2b13, 0xb6fc983e, 0x0fc31b6f,
0xdb05a842, 0x7288ce18, 0xa64e7d35, 0x219202ac, 0xf554b181,
0x5cd9d7db, 0x881f64f6, 0x536128e9, 0x87a79bc4, 0x2e2afd9e,
0xfaec4eb3, 0x7d30312a, 0xa9f68207, 0x007be45d, 0xd4bd5770,
0x6241cf4e, 0xb6877c63, 0x1f0a1a39, 0xcbcca914, 0x4c10d68d,
0x98d665a0, 0x315b03fa, 0xe59db0d7, 0x3ee3fcc8, 0xea254fe5,
0x43a829bf, 0x976e9a92, 0x10b2e50b, 0xc4745626, 0x6df9307c,
0xb93f8351, 0x1f8636de, 0xcb4085f3, 0x62cde3a9, 0xb60b5084,
0x31d72f1d, 0xe5119c30, 0x4c9cfa6a, 0x985a4947, 0x43240558,
0x97e2b675, 0x3e6fd02f, 0xeaa96302, 0x6d751c9b, 0xb9b3afb6,
0x103ec9ec, 0xc4f87ac1, 0x7204e2ff, 0xa6c251d2, 0x0f4f3788,
0xdb8984a5, 0x5c55fb3c, 0x88934811, 0x211e2e4b, 0xf5d89d66,
0x2ea6d179, 0xfa606254, 0x53ed040e, 0x872bb723, 0x00f7c8ba,
0xd4317b97, 0x7dbc1dcd, 0xa97aaee0, 0x10452db1, 0xc4839e9c,
0x6d0ef8c6, 0xb9c84beb, 0x3e143472, 0xead2875f, 0x435fe105,
0x97995228, 0x4ce71e37, 0x9821ad1a, 0x31accb40, 0xe56a786d,
0x62b607f4, 0xb670b4d9, 0x1ffdd283, 0xcb3b61ae, 0x7dc7f990,
0xa9014abd, 0x008c2ce7, 0xd44a9fca, 0x5396e053, 0x8750537e,
0x2edd3524, 0xfa1b8609, 0x2165ca16, 0xf5a3793b, 0x5c2e1f61,
0x88e8ac4c, 0x0f34d3d5, 0xdbf260f8, 0x727f06a2, 0xa6b9b58f,
0x3f0c6dbc, 0xebcade91, 0x4247b8cb, 0x96810be6, 0x115d747f,
0xc59bc752, 0x6c16a108, 0xb8d01225, 0x63ae5e3a, 0xb768ed17,
0x1ee58b4d, 0xca233860, 0x4dff47f9, 0x9939f4d4, 0x30b4928e,
0xe47221a3, 0x528eb99d, 0x86480ab0, 0x2fc56cea, 0xfb03dfc7,
0x7cdfa05e, 0xa8191373, 0x01947529, 0xd552c604, 0x0e2c8a1b,
0xdaea3936, 0x73675f6c, 0xa7a1ec41, 0x207d93d8, 0xf4bb20f5,
0x5d3646af, 0x89f0f582, 0x30cf76d3, 0xe409c5fe, 0x4d84a3a4,
0x99421089, 0x1e9e6f10, 0xca58dc3d, 0x63d5ba67, 0xb713094a,
0x6c6d4555, 0xb8abf678, 0x11269022, 0xc5e0230f, 0x423c5c96,
0x96faefbb, 0x3f7789e1, 0xebb13acc, 0x5d4da2f2, 0x898b11df,
0x20067785, 0xf4c0c4a8, 0x731cbb31, 0xa7da081c, 0x0e576e46,
0xda91dd6b, 0x01ef9174, 0xd5292259, 0x7ca44403, 0xa862f72e,
0x2fbe88b7, 0xfb783b9a, 0x52f55dc0, 0x8633eeed, 0x208a5b62,
0xf44ce84f, 0x5dc18e15, 0x89073d38, 0x0edb42a1, 0xda1df18c,
0x739097d6, 0xa75624fb, 0x7c2868e4, 0xa8eedbc9, 0x0163bd93,
0xd5a50ebe, 0x52797127, 0x86bfc20a, 0x2f32a450, 0xfbf4177d,
0x4d088f43, 0x99ce3c6e, 0x30435a34, 0xe485e919, 0x63599680,
0xb79f25ad, 0x1e1243f7, 0xcad4f0da, 0x11aabcc5, 0xc56c0fe8,
0x6ce169b2, 0xb827da9f, 0x3ffba506, 0xeb3d162b, 0x42b07071,
0x9676c35c, 0x2f49400d, 0xfb8ff320, 0x5202957a, 0x86c42657,
0x011859ce, 0xd5deeae3, 0x7c538cb9, 0xa8953f94, 0x73eb738b,
0xa72dc0a6, 0x0ea0a6fc, 0xda6615d1, 0x5dba6a48, 0x897cd965,
0x20f1bf3f, 0xf4370c12, 0x42cb942c, 0x960d2701, 0x3f80415b,
0xeb46f276, 0x6c9a8def, 0xb85c3ec2, 0x11d15898, 0xc517ebb5,
0x1e69a7aa, 0xcaaf1487, 0x632272dd, 0xb7e4c1f0, 0x3038be69,
0xe4fe0d44, 0x4d736b1e, 0x99b5d833 };
};
/** Bitwise-and to extract only the low bits. */
final Fold truncate = new Fold() {
@Override
public int fold(int hash, int bits) {
return hash & ((1 << bits) - 1);
}
};
/** Applies the golden ratio and takes the upper bits. */
final Fold golden_ratio = new Fold() {
@Override
public int fold(int hash, int bits) {
/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */
return (hash * 0x9e370001) >>> (32 - bits);
}
};
// -----------------------------------------------------------------------
//
// Implementation of the suite lives below this line.
//
//
@Option(name = "--hash", multiValued = true, metaVar = "NAME", usage = "Enable hash function(s)")
List<String> hashFunctions = new ArrayList<String>();
@Option(name = "--fold", multiValued = true, metaVar = "NAME", usage = "Enable fold function(s)")
List<String> foldFunctions = new ArrayList<String>();
@Option(name = "--text-limit", metaVar = "LIMIT", usage = "Maximum size in KiB to scan")
int textLimit = 15 * 1024; // 15 MiB as later we do * 1024.
@Option(name = "--repository", aliases = { "-r" }, multiValued = true, metaVar = "GIT_DIR", usage = "Repository to scan")
List<File> gitDirs = new ArrayList<File>();
@Override
protected boolean requiresRepository() {
return false;
}
@Override
protected void run() throws Exception {
if (gitDirs.isEmpty()) {
RepositoryBuilder rb = new RepositoryBuilder() //
.setGitDir(gitdir) //
.readEnvironment() //
.findGitDir();
if (rb.getGitDir() == null)
throw die(CLIText.get().cantFindGitDirectory);
gitDirs.add(rb.getGitDir());
}
for (File dir : gitDirs) {
RepositoryBuilder rb = new RepositoryBuilder();
if (RepositoryCache.FileKey.isGitRepository(dir, FS.DETECTED))
rb.setGitDir(dir);
else
rb.findGitDir(dir);
Repository db = rb.build();
try {
run(db);
} finally {
db.close();
}
}
}
private void run(Repository db) throws Exception {
List<Function> all = init();
long fileCnt = 0;
long lineCnt = 0;
ObjectReader or = db.newObjectReader();
try {
final MutableObjectId id = new MutableObjectId();
RevWalk rw = new RevWalk(or);
TreeWalk tw = new TreeWalk(or);
tw.reset(rw.parseTree(db.resolve(Constants.HEAD)));
tw.setRecursive(true);
while (tw.next()) {
FileMode fm = tw.getFileMode(0);
if (!FileMode.REGULAR_FILE.equals(fm)
&& !FileMode.EXECUTABLE_FILE.equals(fm))
continue;
byte[] raw;
try {
tw.getObjectId(id, 0);
raw = or.open(id).getCachedBytes(textLimit * 1024);
} catch (LargeObjectException tooBig) {
continue;
}
if (RawText.isBinary(raw))
continue;
RawText txt = new RawText(raw);
int[] lines = new int[txt.size()];
int cnt = 0;
HashSet<Line> u = new HashSet<Line>();
for (int i = 0; i < txt.size(); i++) {
if (u.add(new Line(txt, i)))
lines[cnt++] = i;
}
fileCnt++;
lineCnt += cnt;
for (Function fun : all)
testOne(fun, txt, lines, cnt);
}
} finally {
or.release();
}
if (db.getDirectory() != null) {
String name = db.getDirectory().getName();
File parent = db.getDirectory().getParentFile();
if (name.equals(Constants.DOT_GIT_EXT) && parent != null)
name = parent.getName();
out.println(name + ":");
}
out.format(" %6d files; %5d avg. unique lines/file\n", //
fileCnt, //
lineCnt / fileCnt);
out.format("%-20s %-15s %9s\n", "Hash", "Fold", "Max Len");
out.println("-----------------------------------------------");
String lastHashName = null;
for (Function fun : all) {
String hashName = fun.hash.name;
if (hashName.equals(lastHashName))
hashName = "";
out.format("%-20s %-15s %9d\n", //
hashName, //
fun.fold.name, //
fun.maxChainLength);
lastHashName = fun.hash.name;
}
out.println();
out.flush();
}
private void testOne(Function fun, RawText txt, int[] elements, int cnt) {
final Hash cmp = fun.hash;
final Fold fold = fun.fold;
final int bits = tableBits(cnt);
final int[] buckets = new int[1 << bits];
for (int i = 0; i < cnt; i++)
buckets[fold.fold(cmp.hash(txt, elements[i]), bits)]++;
int maxChainLength = 0;
for (int i = 0; i < buckets.length; i++)
maxChainLength = Math.max(maxChainLength, buckets[i]);
fun.maxChainLength = Math.max(fun.maxChainLength, maxChainLength);
}
private List<Function> init() {
List<Hash> hashes = new ArrayList<Hash>();
List<Fold> folds = new ArrayList<Fold>();
try {
for (Field f : TextHashFunctions.class.getDeclaredFields()) {
if (f.getType() == Hash.class) {
f.setAccessible(true);
Hash cmp = (Hash) f.get(this);
cmp.name = f.getName();
hashes.add(cmp);
} else if (f.getType() == Fold.class) {
f.setAccessible(true);
Fold fold = (Fold) f.get(this);
fold.name = f.getName();
folds.add(fold);
}
}
} catch (IllegalArgumentException e) {
throw new RuntimeException("Cannot determine names", e);
} catch (IllegalAccessException e) {
throw new RuntimeException("Cannot determine names", e);
}
List<Function> all = new ArrayList<Function>();
for (Hash cmp : hashes) {
if (include(cmp.name, hashFunctions)) {
for (Fold f : folds) {
if (include(f.name, foldFunctions)) {
all.add(new Function(cmp, f));
}
}
}
}
return all;
}
private static boolean include(String name, List<String> want) {
if (want.isEmpty())
return true;
for (String s : want) {
if (s.equalsIgnoreCase(name))
return true;
}
return false;
}
private static class Function {
final Hash hash;
final Fold fold;
int maxChainLength;
Function(Hash cmp, Fold fold) {
this.hash = cmp;
this.fold = fold;
}
}
/** Base class for any hashCode function to be tested. */
private static abstract class Hash extends RawTextComparator {
String name;
@Override
public boolean equals(RawText a, int ai, RawText b, int bi) {
return RawTextComparator.DEFAULT.equals(a, ai, b, bi);
}
}
/** Base class for any hashCode folding function to be tested. */
private static abstract class Fold {
String name;
/**
* Fold the given 32-bit hash code into only {@code bits} of space.
*
* @param hash
* the 32 bit hash code to be folded into a smaller value.
* @param bits
* total number of bits that can appear in the output. The
* output value must be in the range {@code [0, 1 << bits)}.
* When bits = 2, valid outputs are 0, 1, 2, 3.
* @return the folded hash, squeezed into only {@code bits}.
*/
abstract int fold(int hash, int bits);
}
/** Utility to help us identify unique lines in a file. */
private class Line {
private final RawText txt;
private final int pos;
Line(RawText txt, int pos) {
this.txt = txt;
this.pos = pos;
}
@Override
public int hashCode() {
return RawTextComparator.DEFAULT.hash(txt, pos);
}
@Override
public boolean equals(Object obj) {
if (obj instanceof Line) {
Line e = (Line) obj;
return RawTextComparator.DEFAULT.equals(txt, pos, e.txt, e.pos);
}
return false;
}
}
private static int tableBits(final int sz) {
int bits = 31 - Integer.numberOfLeadingZeros(sz);
if (bits == 0)
bits = 1;
if (1 << bits < sz)
bits++;
return bits;
}
}
Loading…
Cancel
Save