@Test
public void testParse() throws Exception {
// generate a sample JSON array
StringBuilder sample = new StringBuilder();
sample.append("[");
int i = 0;
final int N = 5000;
for (; i < N; i++) {
sample.append(i);
sample.append(",");
}
sample.append(i);
sample.append("]");
String json = sample.toString();
StopWatch timer = new StopWatch();
for (int n = 0; n < 500; n++) {
JSONPullParser parser = new JSONPullParser(json);
JSONEvent e;
while ((e = parser.next()) != JSONEvent.EndJSON) {}
}
_logger.info("time: " + timer.getElapsedTime());
}
@Test
public void testLexerPerformance() throws Exception {
// generate a sample JSON array
StringBuilder sample = new StringBuilder();
sample.append("[");
int i = 0;
final int N = 5000;
for (; i < N; i++) {
sample.append(i);
sample.append(",");
}
sample.append(i);
sample.append("]");
String json = sample.toString();
StopWatch timer = new StopWatch();
for (int n = 0; n < 500; n++) {
JSONLexer lexer = new JSONLexer(json);
Token t;
while ((t = lexer.nextToken()) != null) {}
}
_logger.info("time: " + timer.getElapsedTime());
}
@Test
public void performanceOfCSVSplit() throws Exception {
StopWatch s = new StopWatch();
String line = null;
final int N = 1;
Pattern p = Pattern.compile(",");
s.reset();
for (int i = 0; i < N; i++) {
BufferedReader br =
new BufferedReader(
new InputStreamReader(
FileResource.find(SilkWalkerTest.class, "scaffold1.silk").openStream()));
while ((line = br.readLine()) != null) {
ArrayList<String> csv = StringUtil.splitCSV(line);
}
}
_logger.info("OpenCSV split:" + s.getElapsedTime());
}
public void SAIS(LSeq SA) {
StopWatch timer = new StopWatch();
_logger.info("SAIS: N=" + SA.textSize());
// Determin T[N-1]'s LS-type
// T[i] is SType if T[i,_) < T[i+1,_)
// T[i] is LType if T[i,_) > T[i+1,_)
{
long i = 0;
for (; i < N; ++i) {
long x = T.lookup((N + i - 1) % N);
long y = T.lookup((N + i) % N);
if (x == y) continue;
if (x < y) {
LS.set(N - 1, SType);
break;
} else {
LS.set(N - 1, LType);
break;
}
}
if (i == N) {
// When T = AAAA... , etc.
LS.set(N - 1, LType);
}
}
// T[i] is SType if T[i] < T[i+1] or T[i] = T[i+1] and T[i+1] is S-type
// T[i] is LType if T[i] > T[i+1] or T[i] = T[i+1] and T[i+1] is L-type
// Set the LS type of each character
for (long i = N - 1; i > 0; --i) {
long x = T.lookup(i);
long y = T.lookup(i - 1);
if (x < y) LS.set(i - 1, LType);
else if (x > y) LS.set(i - 1, SType);
else LS.set(i - 1, LS.get(i));
}
// Initialize the buckets.
// A bucket is a container of the suffixes sharing the same first character
{
_logger.trace("Initialize the buckets");
Arrays.fill(bucketEnd, 0);
// Compute the size of each bucket
for (long i = 0; i < N; ++i) {
++bucketEnd[(int) T.lookup(i)];
}
// Accumulate the character counts. The bucketStart holds the pointers to beginning of the
// buckets in SA
for (int i = 1; i < bucketEnd.length; ++i) {
bucketEnd[i] += bucketEnd[i - 1];
}
_logger.trace("Done.");
}
// initialize the suffix array
for (long i = 0; i < N; ++i) SA.set(i, N);
// Step 1: reduce the problem by at least 1/2
// Sort all the S-substrings
// Find LMS characters
long[] cursorInBucket = Arrays.copyOf(bucketEnd, bucketEnd.length);
for (long i = 0; i < N; ++i) {
if (isLMS(i)) SA.set(--cursorInBucket[(int) T.lookup(i)], i);
}
// Induced sorting LMS prefixes
{
_logger.trace(String.format("[N=%,d] induceSA", SA.textSize()));
induceSA(SA);
_logger.trace("Done.");
}
int numLMS = 0;
// Compact all the sorted substrings into the first M items of SA
// 2*M must be not larger than N
for (long i = 0; i < N; ++i) {
long si = SA.lookup(i);
if (si != N && isLMS(si)) SA.set(numLMS++, si);
}
// Initialize the name array buffer
for (long i = numLMS; i < N; ++i) SA.set(i, N);
// Find the lexicographic names of the LMS substrings
_logger.trace("Sorting LMS substrings: N=" + SA.textSize());
int name = 1;
SA.set(numLMS + (SA.lookup(0) / 2), name++);
for (long i = 1; i < numLMS; ++i) {
final long prev = SA.lookup(i - 1);
final long current = SA.lookup(i);
if (!isEqualLMSSubstring(prev, current)) {
name++;
}
SA.set(numLMS + (current / 2), name - 1);
}
for (long i = N - 1, j = N - 1; i >= numLMS; --i) {
if (SA.lookup(i) != N) SA.set(j--, SA.lookup(i) - 1);
}
// Step 2: solve the reduced problem
// Create SA1, a view of SA[0, numLMS-1]
_logger.trace("Solving the reduced problem: N=" + SA.textSize());
LSeq SA1 = new ArrayWrap(SA, 0, numLMS);
LSeq T1 = new ArrayWrap(SA, N - numLMS, numLMS);
if (name - 1 != numLMS) {
new CyclicSAIS(T1, name - 1).SAIS(SA1);
} else {
// When all LMS substrings have unique names
for (long i = 0; i < numLMS; i++) SA1.set(T1.lookup(i), i);
}
// Step 3: Induce SA from SA1
// Construct P1 using T1 buffer
for (long i = 0, j = 0; i < N; ++i) {
if (isLMS(i)) T1.set(j++, i); //
}
// Translate short name into the original index at T
// SA1 now holds the LMS-substring indexes
for (long i = 0; i < numLMS; ++i) {
SA1.set(i, T1.lookup(SA1.lookup(i)));
}
// Step 3-1: Put all the items in SA1 into corresponding S-type buckets of SA
// Clear SA[N1 .. N-1]
for (long i = numLMS; i < N; ++i) {
SA.set(i, N);
}
// Put SA1 contents into S-type buckets of SA
System.arraycopy(bucketEnd, 0, cursorInBucket, 0, bucketEnd.length);
for (int i = numLMS - 1; i >= 0; --i) {
long si = SA1.lookup(i);
SA.set(i, N);
long index = --cursorInBucket[(int) T.lookup(si)];
SA.set(index, si);
}
// SA.set(0, T.textSize() - 1);
// Step 3-2, 3-3
_logger.trace("Inducing SA from SA1: N=" + SA.textSize());
induceSA(SA);
_logger.info(String.format("done. %.2f sec.", timer.getElapsedTime()));
}
@Test
public void performanceOfTabSplit() throws Exception {
StopWatch s = new StopWatch();
String line = null;
final int N = 1;
Pattern p = Pattern.compile("\t");
s.reset();
for (int i = 0; i < N; i++) {
BufferedReader br =
new BufferedReader(
new InputStreamReader(
FileResource.find(SilkWalkerTest.class, "scaffold1.silk").openStream()));
while ((line = br.readLine()) != null) {
String[] tab = p.split(line);
}
}
_logger.info("default tab split:" + s.getElapsedTime());
s.reset();
for (int i = 0; i < N; i++) {
BufferedReader br =
new BufferedReader(
new InputStreamReader(
FileResource.find(SilkWalkerTest.class, "scaffold1.silk").openStream()));
while ((line = br.readLine()) != null) {
ArrayList<String> tokens = StringUtil.splitAtTab(line);
}
}
_logger.info("my tab split:" + s.getElapsedTime());
s.reset();
for (int i = 0; i < N; i++) {
BufferedReader br =
new BufferedReader(
new InputStreamReader(
FileResource.find(SilkWalkerTest.class, "scaffold1.silk").openStream()));
while ((line = br.readLine()) != null) {
StringTokenizer t = new StringTokenizer(line, "\t");
ArrayList<String> tokens = new ArrayList<String>();
while (t.hasMoreTokens()) {
tokens.add(t.nextToken());
}
}
}
_logger.info("tokenizer tab split:" + s.getElapsedTime());
s.reset();
for (int i = 0; i < N; i++) {
FastBufferedReader fb =
new FastBufferedReader(
new InputStreamReader(
FileResource.find(SilkWalkerTest.class, "scaffold1.silk").openStream()));
while ((line = fb.readLine()) != null) {
ArrayList<String> tab = StringUtil.splitAtTab(line);
}
}
_logger.info("fast reader:" + s.getElapsedTime());
}
@Test
public void perfTest() throws Exception {
StopWatch s = new StopWatch();
File in = FileResource.copyToTemp(SilkWalkerTest.class, "scaffold1.silk", new File("target"));
final int N = 2;
{
s.reset();
int lineCount = 0;
char[] buf = new char[8192];
for (int i = 0; i < N; i++) {
BufferedReader fb = new BufferedReader(new FileReader(in));
for (int readSize = 0; (readSize = fb.read(buf)) != -1; ) {
for (int c = 0; c < readSize; c++) {
if (buf[c] == '\r') {
lineCount++;
if (c + 1 < readSize && buf[c + 1] == '\n') {
c++;
}
} else if (buf[c] == '\n') {
lineCount++;
}
}
}
}
_logger.info(String.format("BufferedReader: %.2f", s.getElapsedTime()));
}
{
s.reset();
int lineCount = 0;
for (int i = 0; i < N; i++) {
BufferedScanner fb = new BufferedScanner(new FileInputStream(in));
for (CharSequence line; (line = fb.nextLine()) != null; lineCount++) {}
}
_logger.info(
String.format("BufferedScanner nextLine: %.2f, line:%d", s.getElapsedTime(), lineCount));
}
{
s.reset();
int lineCount = 0;
for (int i = 0; i < N; i++) {
BufferedReader fb = new BufferedReader(new FileReader(in));
for (String line; (line = fb.readLine()) != null; lineCount++) {}
}
_logger.info(
String.format("BufferedReader readLine: %.2f, line:%d", s.getElapsedTime(), lineCount));
}
{
s.reset();
int lineCount = 0;
for (int i = 0; i < N; i++) {
BufferedScanner fb = new BufferedScanner(new FileInputStream(in));
for (CharSequence line; (line = fb.nextLine()) != null; lineCount++) {
String st = line.toString();
}
}
_logger.info(
String.format(
"BufferedScanner nextLine (with String conversion): %.2f, line:%d",
s.getElapsedTime(), lineCount));
}
{
s.reset();
int lineCount = 0;
for (int i = 0; i < N; i++) {
BufferedScanner fb = new BufferedScanner(new FileReader(in));
for (CharSequence line; (line = fb.nextLine()) != null; lineCount++) {}
}
_logger.info(
String.format(
"BufferedScanner nextLine with Reader input: %.2f, line:%d",
s.getElapsedTime(), lineCount));
}
{
s.reset();
int lineCount = 0;
char[] buf = new char[8192];
for (int i = 0; i < N; i++) {
FastBufferedReader fb = new FastBufferedReader(new FileReader(in));
for (int readSize = 0; (readSize = fb.read(buf)) != -1; ) {
for (int c = 0; c < readSize; c++) {
if (buf[c] == '\r') {
lineCount++;
if (c + 1 < readSize && buf[c + 1] == '\n') {
c++;
}
} else if (buf[c] == '\n') {
lineCount++;
}
}
}
}
_logger.info(String.format("FastBufferedReader: %.2f", s.getElapsedTime()));
}
{
s.reset();
int lineCount = 0;
byte[] buf = new byte[8192];
for (int i = 0; i < N; i++) {
FastBufferedInputStream fb = new FastBufferedInputStream(new FileInputStream(in));
for (int readSize = 0; (readSize = fb.read(buf)) != -1; ) {
for (int c = 0; c < readSize; c++) {
if (buf[c] == '\r') {
lineCount++;
if (c + 1 < readSize && buf[c + 1] == '\n') {
c++;
}
} else if (buf[c] == '\n') {
lineCount++;
}
}
}
}
_logger.info(String.format("FastBufferedInputStream: %.2f", s.getElapsedTime()));
}
{
s.reset();
int lineCount = 0;
byte[] buf = new byte[8192];
for (int i = 0; i < N; i++) {
BufferedInputStream fb = new BufferedInputStream(new FileInputStream(in));
for (int readSize = 0; (readSize = fb.read(buf)) != -1; ) {
for (int c = 0; c < readSize; c++) {
if (buf[c] == '\r') {
lineCount++;
if (c + 1 < readSize && buf[c + 1] == '\n') {
c++;
}
} else if (buf[c] == '\n') {
lineCount++;
}
}
}
}
_logger.info(String.format("BufferedInputStream: %.2f", s.getElapsedTime()));
}
}
