Skip to content

Commit 401fece

Browse files
MirkodddMirkoddd
committed
perf(core): memoize SiftPatterns lambdas and add caching tests
- Introduced a memoize decorator in SiftPatterns and updated SiftPattern.preventBacktracking() to return caching implementations instead of pure lambdas. - Ensured that shake() and sieve() are computed only once per instance in a thread-safe manner (using volatile), drastically reducing CPU overhead for reused patterns. - Added explicit unit tests using assertSame to strictly verify that subsequent calls return the exact same String and Pattern objects, guaranteeing 100% coverage. - Cleaned up unused imports and enforced camelCase naming conventions for local variables in SiftPattern.java.
1 parent 1a365c1 commit 401fece

File tree

4 files changed

+105
-17
lines changed

4 files changed

+105
-17
lines changed

‎sift-core/src/main/java/com/mirkoddd/sift/core/SiftPatterns.java‎

Lines changed: 43 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -18,6 +18,7 @@
1818
import com.mirkoddd.sift.core.dsl.SiftPattern;
1919

2020
import java.util.Objects;
21+
import java.util.regex.Pattern;
2122

2223
/**
2324
* <h2>SiftPatterns - Component Factory</h2>
@@ -54,7 +55,7 @@ public static SiftPattern anyOf(SiftPattern option1, SiftPattern option2, SiftPa
5455
Objects.requireNonNull(opt, "Additional option cannot be null");
5556
}
5657

57-
return () -> {
58+
return memoize(() -> {
5859
StringBuilder sb = new StringBuilder();
5960
sb.append(RegexSyntax.NON_CAPTURING_GROUP_OPEN);
6061

@@ -69,7 +70,7 @@ public static SiftPattern anyOf(SiftPattern option1, SiftPattern option2, SiftPa
6970

7071
sb.append(RegexSyntax.GROUP_CLOSE);
7172
return sb.toString();
72-
};
73+
});
7374
}
7475

7576
/**
@@ -83,7 +84,7 @@ public static SiftPattern anyOf(SiftPattern option1, SiftPattern option2, SiftPa
8384
*/
8485
public static SiftPattern capture(SiftPattern pattern) {
8586
Objects.requireNonNull(pattern, "Pattern to capture cannot be null");
86-
return () -> RegexSyntax.GROUP_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE;
87+
return memoize(() -> RegexSyntax.GROUP_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE);
8788
}
8889

8990
/**
@@ -106,7 +107,7 @@ public static SiftPattern capture(SiftPattern pattern) {
106107
*/
107108
public static SiftPattern positiveLookahead(SiftPattern pattern) {
108109
Objects.requireNonNull(pattern, "Lookahead pattern cannot be null");
109-
return () -> RegexSyntax.POSITIVE_LOOKAHEAD_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE;
110+
return memoize(() -> RegexSyntax.POSITIVE_LOOKAHEAD_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE);
110111
}
111112

112113
/**
@@ -129,7 +130,7 @@ public static SiftPattern positiveLookahead(SiftPattern pattern) {
129130
*/
130131
public static SiftPattern negativeLookahead(SiftPattern pattern) {
131132
Objects.requireNonNull(pattern, "Lookahead pattern cannot be null");
132-
return () -> RegexSyntax.NEGATIVE_LOOKAHEAD_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE;
133+
return memoize(() -> RegexSyntax.NEGATIVE_LOOKAHEAD_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE);
133134
}
134135

135136
/**
@@ -152,7 +153,7 @@ public static SiftPattern negativeLookahead(SiftPattern pattern) {
152153
*/
153154
public static SiftPattern positiveLookbehind(SiftPattern pattern) {
154155
Objects.requireNonNull(pattern, "Lookbehind pattern cannot be null");
155-
return () -> RegexSyntax.POSITIVE_LOOKBEHIND_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE;
156+
return memoize(() -> RegexSyntax.POSITIVE_LOOKBEHIND_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE);
156157
}
157158

158159
/**
@@ -175,7 +176,7 @@ public static SiftPattern positiveLookbehind(SiftPattern pattern) {
175176
*/
176177
public static SiftPattern negativeLookbehind(SiftPattern pattern) {
177178
Objects.requireNonNull(pattern, "Lookbehind pattern cannot be null");
178-
return () -> RegexSyntax.NEGATIVE_LOOKBEHIND_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE;
179+
return memoize(() -> RegexSyntax.NEGATIVE_LOOKBEHIND_OPEN + pattern.shake() + RegexSyntax.GROUP_CLOSE);
179180
}
180181

181182
/**
@@ -213,7 +214,7 @@ public static SiftPattern group(SiftPattern first, SiftPattern... then) {
213214
Objects.requireNonNull(opt, "Additional option cannot be null");
214215
}
215216

216-
return () -> {
217+
return memoize(() -> {
217218
StringBuilder sb = new StringBuilder();
218219
sb.append(RegexSyntax.NON_CAPTURING_GROUP_OPEN);
219220

@@ -225,7 +226,7 @@ public static SiftPattern group(SiftPattern first, SiftPattern... then) {
225226

226227
sb.append(RegexSyntax.GROUP_CLOSE);
227228
return sb.toString();
228-
};
229+
});
229230
}
230231

231232
/**
@@ -254,11 +255,11 @@ public static SiftPattern literal(String text) {
254255
throw new IllegalArgumentException("Literal text cannot be empty. Use zero-width assertions if intentional.");
255256
}
256257

257-
return () -> {
258+
return memoize(() -> {
258259
StringBuilder sb = new StringBuilder();
259260
RegexEscaper.escapeString(text, sb);
260261
return sb.toString();
261-
};
262+
});
262263
}
263264

264265
/**
@@ -279,7 +280,7 @@ public static SiftPattern anythingBut(String chars) {
279280
throw new IllegalArgumentException("Excluded characters string cannot be empty");
280281
}
281282

282-
return () -> {
283+
return memoize(() -> {
283284
StringBuilder sb = new StringBuilder(3 + (chars.length() * 2));
284285
sb.append(RegexSyntax.CLASS_OPEN);
285286
sb.append(RegexSyntax.NEGATION);
@@ -290,6 +291,36 @@ public static SiftPattern anythingBut(String chars) {
290291

291292
sb.append(RegexSyntax.CLASS_CLOSE);
292293
return sb.toString();
294+
});
295+
}
296+
297+
/**
298+
* Internal helper to memoize the result of a pattern.
299+
* Guarantees that shake() and sieve() are computed only once in a thread-safe manner.
300+
*
301+
* @param generator The underlying SiftPattern generation logic.
302+
* @return A thread-safe, caching SiftPattern instance.
303+
*/
304+
private static SiftPattern memoize(SiftPattern generator) {
305+
return new SiftPattern() {
306+
private volatile String cachedRegex = null;
307+
private volatile Pattern cachedPattern = null;
308+
309+
@Override
310+
public String shake() {
311+
if (cachedRegex == null) {
312+
cachedRegex = generator.shake();
313+
}
314+
return cachedRegex;
315+
}
316+
317+
@Override
318+
public Pattern sieve() {
319+
if (cachedPattern == null) {
320+
cachedPattern = Pattern.compile(shake());
321+
}
322+
return cachedPattern;
323+
}
293324
};
294325
}
295326
}

‎sift-core/src/main/java/com/mirkoddd/sift/core/dsl/SiftPattern.java‎

Lines changed: 22 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -15,8 +15,6 @@
1515
*/
1616
package com.mirkoddd.sift.core.dsl;
1717

18-
import java.util.Collections;
19-
import java.util.Set;
2018
import java.util.regex.Pattern;
2119

2220
/**
@@ -54,9 +52,28 @@ public interface SiftPattern {
5452
* @return A new SiftPattern wrapped in an atomic group {@code (?>...)}.
5553
*/
5654
default SiftPattern preventBacktracking() {
57-
String ATOMIC_OPEN = "(?>";
58-
String ATOMIC_CLOSE = ")";
59-
return () -> ATOMIC_OPEN + this.shake() + ATOMIC_CLOSE;
55+
final String atomicOpen = "(?>";
56+
final String atomicClose = ")";
57+
return new SiftPattern() {
58+
private volatile String cachedRegex = null;
59+
private volatile Pattern cachedPattern = null;
60+
61+
@Override
62+
public String shake() {
63+
if (cachedRegex == null) {
64+
cachedRegex = atomicOpen + SiftPattern.this.shake() + atomicClose;
65+
}
66+
return cachedRegex;
67+
}
68+
69+
@Override
70+
public Pattern sieve() {
71+
if (cachedPattern == null) {
72+
cachedPattern = Pattern.compile(shake());
73+
}
74+
return cachedPattern;
75+
}
76+
};
6077
}
6178

6279
/**

‎sift-core/src/test/java/com/mirkoddd/sift/core/SiftPatternsTest.java‎

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -246,4 +246,24 @@ void testDefaultSieveImplementation() {
246246
assertEquals("[a-z]+", compiled.pattern(), "The compiled pattern should match the shake() output");
247247
assertTrue(compiled.matcher("abc").matches(), "The Pattern should correctly match valid strings");
248248
}
249+
250+
@Test
251+
void shouldMemoizeShakeAndSieveResults() {
252+
// We use literal() as it's wrapped by the memoize() helper
253+
SiftPattern memoizedPattern = SiftPatterns.literal("cache-test");
254+
255+
// 1. Verify shake() caching
256+
String firstShake = memoizedPattern.shake();
257+
String secondShake = memoizedPattern.shake();
258+
259+
assertEquals("cache-test", firstShake);
260+
assertSame(firstShake, secondShake, "shake() should return the exact same String instance from cache.");
261+
262+
// 2. Verify sieve() caching
263+
Pattern firstSieve = memoizedPattern.sieve();
264+
Pattern secondSieve = memoizedPattern.sieve();
265+
266+
assertEquals("cache-test", firstSieve.pattern());
267+
assertSame(firstSieve, secondSieve, "sieve() should return the exact same Pattern instance from cache.");
268+
}
249269
}

‎sift-core/src/test/java/com/mirkoddd/sift/core/SiftTest.java‎

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -908,6 +908,26 @@ void testLazyModifierEdgeCasesAndProtections() {
908908
assertEquals("^.+?", step3DoubleLazy.shake(),
909909
"Should ignore a second lazy call on the main pattern");
910910
}
911+
912+
@Test
913+
void shouldMemoizeShakeAndSieveForPreventBacktracking() {
914+
SiftPattern basePattern = SiftPatterns.literal("atomic");
915+
SiftPattern atomicPattern = basePattern.preventBacktracking();
916+
917+
// 1. Verify shake() caching
918+
String firstShake = atomicPattern.shake();
919+
String secondShake = atomicPattern.shake();
920+
921+
assertEquals("(?>atomic)", firstShake);
922+
assertSame(firstShake, secondShake, "shake() on preventBacktracking() should return the same String instance.");
923+
924+
// 2. Verify sieve() caching
925+
Pattern firstSieve = atomicPattern.sieve();
926+
Pattern secondSieve = atomicPattern.sieve();
927+
928+
assertEquals("(?>atomic)", firstSieve.pattern());
929+
assertSame(firstSieve, secondSieve, "sieve() on preventBacktracking() should return the same Pattern instance.");
930+
}
911931
}
912932

913933
@Nested

0 commit comments

Comments
 (0)