From 29bc63423aeba9327b2e60959f8dbb2c8e718a29 Mon Sep 17 00:00:00 2001
From: XmiliaH <45106915+XmiliaH@users.noreply.github.com>
Date: Tue, 6 Feb 2024 12:03:17 +0100
Subject: Cleanup fuzzer and add java docs (#61)
---
.../java/com/seaofnodes/simple/Fuzzer08Test.java | 146 +------
.../test/java/com/seaofnodes/simple/Reducer.java | 119 ------
.../com/seaofnodes/simple/ScriptGenerator.java | 305 --------------
.../java/com/seaofnodes/simple/fuzzer/Fuzzer.java | 128 ++++++
.../com/seaofnodes/simple/fuzzer/FuzzerUtils.java | 58 +++
.../java/com/seaofnodes/simple/fuzzer/Reducer.java | 144 +++++++
.../seaofnodes/simple/fuzzer/ScriptGenerator.java | 462 +++++++++++++++++++++
7 files changed, 801 insertions(+), 561 deletions(-)
delete mode 100644 chapter08/src/test/java/com/seaofnodes/simple/Reducer.java
delete mode 100644 chapter08/src/test/java/com/seaofnodes/simple/ScriptGenerator.java
create mode 100644 chapter08/src/test/java/com/seaofnodes/simple/fuzzer/Fuzzer.java
create mode 100644 chapter08/src/test/java/com/seaofnodes/simple/fuzzer/FuzzerUtils.java
create mode 100644 chapter08/src/test/java/com/seaofnodes/simple/fuzzer/Reducer.java
create mode 100644 chapter08/src/test/java/com/seaofnodes/simple/fuzzer/ScriptGenerator.java
(limited to 'chapter08/src/test/java/com/seaofnodes')
diff --git a/chapter08/src/test/java/com/seaofnodes/simple/Fuzzer08Test.java b/chapter08/src/test/java/com/seaofnodes/simple/Fuzzer08Test.java
index d8f10eef..b8c0a8e5 100644
--- a/chapter08/src/test/java/com/seaofnodes/simple/Fuzzer08Test.java
+++ b/chapter08/src/test/java/com/seaofnodes/simple/Fuzzer08Test.java
@@ -1,154 +1,26 @@
package com.seaofnodes.simple;
-import com.seaofnodes.simple.node.Node;
-import com.seaofnodes.simple.node.StopNode;
+import com.seaofnodes.simple.fuzzer.Fuzzer;
import org.junit.Ignore;
import org.junit.Test;
-import java.util.ArrayList;
-import java.util.Random;
import static org.junit.Assert.assertTrue;
+/**
+ * To use the fuzzer run the fuzzPeeps methods.
+ * This can be done from IntelliJ Gui or remove the `@Ignore` annotation and start the test case from the command line
+ * via mvn clean test -Dtest=com.seaofnodes.simple.Fuzzer08Test#fuzzPeeps
+ */
public class Fuzzer08Test {
- private final ArrayList exceptions = new ArrayList<>();
-
- public static StopNode parse(String script, boolean disablePeeps) {
- var parser = new Parser(script);
- Node._disablePeephole = disablePeeps;
- return parser.parse();
- }
-
- private static boolean sameException(Throwable e1, Throwable e2) {
- if (e1 == null || e2 == null) return e1 == e2;
- if (e1.getClass() != e2.getClass()) return false;
- if (e1.getMessage() != null && e2.getMessage() != null && !e1.getMessage().equals(e2.getMessage())) return false;
- var s1 = e1.getStackTrace();
- var s2 = e2.getStackTrace();
- if (s1 != null && s1.length > 0 && s2 != null && s2.length > 0 && !s1[0].equals(s2[0])) return false;
- if (!sameException(e1.getCause(), e2.getCause())) return false;
- var o1 = e1.getSuppressed();
- var o2 = e2.getSuppressed();
- if (o1 == null || o2 == null || o1.length != o2.length) return o1 == o2;
- for (int i=0; i=(])--"), ""),
- new Replace(Pattern.compile("\\b--"), "+"),
- new Replace(Pattern.compile("\\b\\+-"), "-"),
- new Replace(Pattern.compile("\\bfalse\\b"), "0"),
- new Replace(Pattern.compile("\\btrue\\b"), "1"),
- new Replace(Pattern.compile("-"), ""),
- new Replace(Pattern.compile("(?=])(?:-|\\*|/|>=|<=|>|<|!=|==)(?![+\\-*/<>=])"), "+"),
- new Replace(Pattern.compile("\\{(?:[^{}]|\\{})+}"), "{}"),
- new Replace(Pattern.compile("\\{((?:[^{}]|\\{})+)}"), "$1"),
- new Replace(Pattern.compile("\\(([^()]+)\\)"), "$1"),
- new Replace(Pattern.compile("\\b\\w+\\b"), "1"),
- new Replace(Pattern.compile("\\b\\w+\\b"), "0"),
- new Replace(Pattern.compile("\\d+\\+"), ""),
- new Replace(Pattern.compile("\\+\\d+"), ""),
- new Replace(Pattern.compile("(?<=\\n|^)[^\\n]*(?=\\n|$)"), "\n"),
- new Replace(Pattern.compile("\\n+"), "\n"),
- new Replace(Pattern.compile("^\\n+|\\n+$"), ""),
- new Replace(Pattern.compile("else "), ""),
- new Replace(Pattern.compile("\\bif\\([^()]+\\) *"), ""),
- new Replace(Pattern.compile("\\bwhile\\([^()]+\\) *"), ""),
- };
-
- private static boolean sameException(Throwable e1, Throwable e2) {
- if (e1 == null || e2 == null) return e1 == e2;
- if (e1.getClass() != e2.getClass()) return false;
- if (e1.getMessage() != null && e2.getMessage() != null && !e1.getMessage().equals(e2.getMessage())) return false;
- var s1 = e1.getStackTrace();
- var s2 = e2.getStackTrace();
- if (s1 == null || s1.length == 0 || s2 == null || s2.length == 0) return true;
- return s1[0].equals(s2[0]);
- }
-
- private final boolean disablePeeps;
-
- private Reducer(boolean disablePeeps) {
- this.disablePeeps = disablePeeps;
- }
-
- private Throwable parse(String script) {
- try {
- var parser = new Parser(script);
- Node._disablePeephole = disablePeeps;
- parser.parse();
- return null;
- } catch (Throwable e) {
- return e;
- }
- }
-
- private String doReplacement(String script, Throwable ex, Pattern pattern, String with) {
- var matcher = pattern.matcher(script);
- var sb = new StringBuilder();
- int last = 0;
- while (matcher.find()) {
- var pos = sb.length();
- matcher.appendReplacement(sb, with);
- var tail = sb.length();
- matcher.appendTail(sb);
- var ex2 = parse(sb.toString());
- if (sameException(ex, ex2)) {
- sb.setLength(tail);
- } else {
- sb.setLength(pos);
- sb.append(script, last, matcher.end());
- }
- last = matcher.end();
- }
- matcher.appendTail(sb);
- return sb.toString();
- }
-
- private String cleanVariables(String script, Throwable ex) {
- var matcher = Pattern.compile("\\bint ([0-9a-zA-Z_]+)").matcher(script);
- int num = 0;
- while (matcher.find()) {
- var var = matcher.group(1);
- if (var.matches("v\\d+")) continue;
- while (script.contains("v"+num)) num++;
- var n = script.replaceAll("\\b"+var+"\\b", "v" + num);
- var ex2 = parse(n);
- if (sameException(ex, ex2)) {
- script = n;
- }
- }
- return script;
- }
-
- private String doAllReplacements(String script, Throwable ex) {
- for (var replace: REPLACEMENTS) {
- script = doReplacement(script, ex, replace.p, replace.r);
- }
- return script;
- }
-
- public static String reduce(String script, boolean disablePeeps) {
- var reducer = new Reducer(disablePeeps);
- var ex = reducer.parse(script);
- if (ex == null) return "return 0;";
- String old;
- do {
- old = script;
- script = reducer.doAllReplacements(script, ex);
- script = reducer.cleanVariables(script, ex);
- } while (!script.equals(old));
- return script;
- }
-
-}
diff --git a/chapter08/src/test/java/com/seaofnodes/simple/ScriptGenerator.java b/chapter08/src/test/java/com/seaofnodes/simple/ScriptGenerator.java
deleted file mode 100644
index 8c51089c..00000000
--- a/chapter08/src/test/java/com/seaofnodes/simple/ScriptGenerator.java
+++ /dev/null
@@ -1,305 +0,0 @@
-package com.seaofnodes.simple;
-
-import com.seaofnodes.simple.node.ScopeNode;
-
-import java.util.ArrayList;
-import java.util.HashSet;
-import java.util.Random;
-
-public class ScriptGenerator {
-
- @SuppressWarnings("unchecked")
- private static HashSet getKeywords() {
- try {
- var field = Parser.class.getDeclaredField("KEYWORDS");
- field.setAccessible(true);
- return (HashSet) field.get(new Parser(""));
- } catch (Throwable e) {
- throw new ExceptionInInitializerError(e);
- }
- }
-
- private static final int INDENTATION = 4;
- private static final String VAR_CHARS = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_0123456789";
- private static final HashSet KEYWORDS = getKeywords();
-
- private static final int FLAG_STOP = 0x01;
- private static final int FLAG_IF_WITHOUT_ELSE = 0x02;
-
- private final Random random;
- private int indentation = 0;
- private final StringBuilder sb;
- private int loopDepth = 0;
- private int currScopeStart = 0;
- private int depth = 0;
- private int exprDepth = 3;
- private final ArrayList variables = new ArrayList<>();
- private final boolean generateValid;
- private boolean maybeInvalid = false;
-
- public ScriptGenerator(Random random, StringBuilder sb, boolean generateValid) {
- this.random = random;
- this.sb = sb;
- this.generateValid = generateValid;
- }
-
- private int randLog(int max) {
- max--;
- int n = 1 << max;
- var num = random.nextInt(n);
- for(int i=max-1; i>=0; i--) {
- if ((num & (1 << i))!=0) return max-i-1;
- }
- return max;
- }
-
- private StringBuilder printIndentation() {
- return sb.repeat(' ', indentation);
- }
-
- public boolean genProgram() {
- this.maybeInvalid = false;
- variables.add(ScopeNode.ARG0);
- depth = random.nextInt(9) + 1;
- currScopeStart = variables.size();
- if ((genStatements() & FLAG_STOP) != 0)
- genReturn();
- return !this.maybeInvalid;
- }
-
- public int genStatements() {
- var num = random.nextInt(10);
- for (int i=0; i genBlock();
- case 2 -> genIf();
- case 3 -> genWhile();
- case 4, 5 -> genDecl();
- case 6, 7, 8 -> genAssignment();
- default -> genExit();
- };
- }
-
- public int genStatementBlock() {
- indentation += INDENTATION;
- int res;
- if (depth == 0) {
- res = random.nextBoolean() ? genAssignment() : genExit();
- } else {
- depth--;
- res = switch (random.nextInt(10)) {
- case 1, 2, 3, 4, 5 -> {
- depth++;
- indentation -= INDENTATION;
- var ret = genBlock();
- indentation += INDENTATION;
- depth--;
- yield ret;
- }
- case 6 -> genIf();
- case 7 -> genWhile();
- case 8 -> genAssignment();
- default -> genExit();
- };
- depth++;
- }
- indentation -= INDENTATION;
- return res;
- }
-
- public int genExit() {
- if (loopDepth == 0 && generateValid) return genReturn();
- return switch (random.nextInt(7)) {
- case 0,1,2 -> {
- if (loopDepth == 0) maybeInvalid = true;
- sb.append("continue;");
- yield FLAG_STOP;
- }
- case 3,4,5 -> {
- if (loopDepth == 0) maybeInvalid = true;
- sb.append("break;");
- yield FLAG_STOP;
- }
- default -> genReturn();
- };
- }
-
- public int genBlock() {
- if (depth == 0) {
- sb.append("{}");
- return 0;
- }
- depth--;
- int oldCSS = currScopeStart;
- currScopeStart = variables.size();
- sb.append("{\n");
- indentation += INDENTATION;
- var stop = genStatements();
- indentation -= INDENTATION;
- printIndentation().append("}");
- while (variables.size() > currScopeStart)
- variables.removeLast();
- currScopeStart = oldCSS;
- depth++;
- return stop & ~FLAG_IF_WITHOUT_ELSE;
- }
-
- public int genIf() {
- sb.append("if(");
- genExpression();
- sb.append(") ");
- var stop = genStatementBlock();
- if ((stop & FLAG_IF_WITHOUT_ELSE) == 0 && random.nextInt(10) > 3) {
- sb.append("\n");
- printIndentation().append("else ");
- stop &= genStatementBlock();
- } else {
- stop = FLAG_IF_WITHOUT_ELSE;
- }
- return stop;
- }
-
- public int genWhile() {
- sb.append("while(");
- genExpression();
- sb.append(") ");
- loopDepth++;
- var stop = genStatementBlock();
- loopDepth--;
- return stop & FLAG_IF_WITHOUT_ELSE;
- }
-
- private StringBuilder getRandomName() {
- int len = random.nextInt(10) + 1;
- StringBuilder sb = new StringBuilder(len);
- sb.append(VAR_CHARS.charAt(random.nextInt(VAR_CHARS.length()-10)));
- for (int i=1; i 0 && random.nextInt(10) > 7) {
- var v = variables.get(random.nextInt(currScopeStart));
- if (variables.lastIndexOf(v) < currScopeStart) return v;
- if (!generateValid) {
- maybeInvalid = true;
- return v;
- }
- }
- StringBuilder sb = getRandomName();
- var v = sb.toString();
- if (!generateValid && (KEYWORDS.contains(v) || variables.lastIndexOf(v) >= currScopeStart)) {
- maybeInvalid = true;
- return v;
- }
- while (KEYWORDS.contains(v) || variables.lastIndexOf(v) >= currScopeStart) {
- sb.append(VAR_CHARS.charAt(random.nextInt(VAR_CHARS.length())));
- v = sb.toString();
- }
- return v;
- }
-
- public int genDecl() {
- String name = getVarName();
- sb.append("int ").append(name).append("=");
- genExpression();
- sb.append(";");
- variables.add(name);
- return 0;
- }
-
- public int genAssignment() {
- if (variables.isEmpty()) return genDecl();
- genVariable();
- sb.append("=");
- genExpression();
- sb.append(";");
- return 0;
- }
-
- public int genReturn() {
- sb.append("return ");
- genExpression();
- sb.append(";");
- return FLAG_STOP;
- }
-
- private static final String[] EXPRESSIONS = {"==", "!=", "<", "<=", ">", ">=", "+", "-", "*", "/"};
- private static final String[] UNARY = {"-"};
-
- public void genExpression() {
- var num = randLog(10);
- while(num-->0) {
- genUnary();
- sb.append(EXPRESSIONS[random.nextInt(EXPRESSIONS.length)]);
- }
- genUnary();
- }
-
- public void genUnary() {
- var num = randLog(6);
- while(num-->0) {
- sb.append(UNARY[random.nextInt(UNARY.length)]);
- }
- genPrimary();
- }
-
- public void genPrimary() {
- switch (random.nextInt(10)) {
- case 0:
- sb.append("true");
- break;
- case 1:
- sb.append("false");
- break;
- case 2:
- if (exprDepth == 0) {
- if (random.nextBoolean()) {
- genNumber();
- } else {
- genVariable();
- }
- } else {
- sb.append("(");
- exprDepth--;
- genExpression();
- exprDepth++;
- sb.append(")");
- }
- break;
- case 3, 4, 5, 6:
- genNumber();
- break;
- default:
- genVariable();
- }
- }
-
- public void genNumber() {
- sb.append(random.nextInt(100));
- }
-
- public void genVariable() {
- if (!generateValid && random.nextInt(100)>97) {
- sb.append(getRandomName());
- maybeInvalid = true;
- return;
- }
- if (variables.isEmpty()) {
- genNumber();
- } else {
- sb.append(variables.get(random.nextInt(variables.size())));
- }
- }
-
-}
diff --git a/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/Fuzzer.java b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/Fuzzer.java
new file mode 100644
index 00000000..2af22f1d
--- /dev/null
+++ b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/Fuzzer.java
@@ -0,0 +1,128 @@
+package com.seaofnodes.simple.fuzzer;
+
+import com.seaofnodes.simple.GraphEvaluator;
+import com.seaofnodes.simple.node.StopNode;
+
+import java.util.ArrayList;
+import java.util.Random;
+import java.util.function.Consumer;
+
+/**
+ * Implementation of the fuzzer.
+ * Random testcases are generated using the ScriptGenerator.
+ * This will generate a script similar to how the parser parses it but instead of parsing it will
+ * randomly choose a variation the parser would parse and generate it.
+ * These scripts are the parsed by the parser and all exceptions are caught and filtered
+ * to only show one occurrence of one problem. This includes script which generate different
+ * results when the compiled graph with and without peeps is executed.
+ * To aid debugging scripts that cause errors are then reduced by applying rules and checking
+ * that the same issue persists.
+ */
+public class Fuzzer {
+
+ private static final int EVAL_TIMEOUT = 1000;
+
+ /**
+ * List of exceptions already encountered.
+ * This is used to filter new exceptions and discard them if they are already found.
+ */
+ private final ArrayList exceptions = new ArrayList<>();
+
+ /**
+ * Filter and record an exception in a script.
+ * @param e The exception caused by the script
+ * @param script The script causing the exception
+ * @param reproducer Reproducer which should be used to reproduce the test case.
+ */
+ private void recordException(Throwable e, String script, Consumer reproducer) {
+ for (var ex : exceptions) {
+ if (FuzzerUtils.isExceptionFromSameCause(ex, e)) return;
+ }
+ exceptions.add(e);
+ System.out.println("========== Stack ==========");
+ e.printStackTrace(System.out);
+ System.out.println("========== Code ===========");
+ System.out.println(script);
+ System.out.println("========= Reduced =========");
+ System.out.println(Reducer.reduce(script, e, reproducer));
+ System.out.println("===========================");
+ System.out.flush();
+ }
+
+ /**
+ * Check that two graphs result in the same output when supplied with a value
+ * @param stop1 Stop node of the first graph
+ * @param stop2 Stop node of the second graph
+ * @param in The input value to both graphs to test for an equal output
+ */
+ private static void checkGraphs(StopNode stop1, StopNode stop2, long in) {
+ var e1 = GraphEvaluator.evaluateWithResult(stop1, in, EVAL_TIMEOUT);
+ var e2 = GraphEvaluator.evaluateWithResult(stop2, in, EVAL_TIMEOUT);
+ if (e1.type() == GraphEvaluator.ResultType.TIMEOUT || e2.type() == GraphEvaluator.ResultType.TIMEOUT) return;
+ if (e1.type() != e2.type())
+ throw new RuntimeException("Different calculations types " + e1.type() + " vs " + e2.type());
+ if (e1.value() != e2.value())
+ throw new RuntimeException("Different calculations values " + e1.value() + " vs " + e2.value());
+ }
+
+ /**
+ * Run checks for script. Compile the script with peeps enabled and disabled.
+ * Check that exceptions raised in the parser by both methods are the same and only happens if the script may be invalid.
+ * If the script was successfully parsed check that both version behave the same.
+ * @param script The script to test
+ * @param valid If the script is definitely valid. If not some exceptions may be suppressed.
+ */
+ private static void runCheck(String script, boolean valid) {
+ StopNode stop1;
+ try {
+ stop1 = FuzzerUtils.parse(script, false);
+ } catch (RuntimeException e1) {
+ try {
+ FuzzerUtils.parse(script, true);
+ } catch (RuntimeException e2) {
+ if (FuzzerUtils.isExceptionFromSameCause(e1, e2)) {
+ if (!valid || e1.getClass() == RuntimeException.class) return;
+ } else {
+ e1.addSuppressed(e2);
+ }
+ }
+ throw e1;
+ }
+ var stop2 = FuzzerUtils.parse(script, true);
+ checkGraphs(stop1, stop2, 0);
+ checkGraphs(stop1, stop2, 1);
+ checkGraphs(stop1, stop2, 10);
+ }
+
+ /**
+ * Check that the script does not result in exceptions.
+ * @param script The script to test.
+ * @param valid If the script is valid or if it might contain syntax errors.
+ */
+ private void check(String script, boolean valid) {
+ try {
+ runCheck(script, valid);
+ } catch (Throwable e) {
+ recordException(e, script, s->runCheck(s, valid));
+ }
+ }
+
+ /**
+ * Run one test with the given seed.
+ * @param seed The seed to use for generating this test case
+ */
+ public void fuzzPeeps(long seed) {
+ var rand = new Random(seed);
+ var sb = new StringBuilder();
+ var valid = new ScriptGenerator(rand, sb, false).genProgram();
+ check(sb.toString(), valid);
+ }
+
+ /**
+ * Check that no exceptions happened.
+ * @return true if no exception happened
+ */
+ public boolean noExceptions() {
+ return exceptions.isEmpty();
+ }
+}
diff --git a/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/FuzzerUtils.java b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/FuzzerUtils.java
new file mode 100644
index 00000000..ad499836
--- /dev/null
+++ b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/FuzzerUtils.java
@@ -0,0 +1,58 @@
+package com.seaofnodes.simple.fuzzer;
+
+import com.seaofnodes.simple.Parser;
+import com.seaofnodes.simple.node.Node;
+import com.seaofnodes.simple.node.StopNode;
+
+/**
+ * Some utilities for the fuzzer.
+ */
+class FuzzerUtils {
+
+ /**
+ * Check that eiter one of the objects is null or that both are equal.
+ */
+ private static boolean equalOrNull(Object a, Object b) {
+ if (a == null || b == null) return true;
+ return a.equals(b);
+ }
+
+ /**
+ * Check if both stack traces have the same first entry. Empty stack traces can happen with NPE optimization
+ * and allow the other to be any stack trace.
+ */
+ private static boolean sameStackTraceOrigin(StackTraceElement[] a, StackTraceElement[] b) {
+ if (a == null || a.length == 0 || b == null || b.length == 0) return true; // One or both stack traces are missing.
+ return a[0].equals(b[0]);
+ }
+
+ /**
+ * Checks if two exceptions are caused by the same problem.
+ */
+ public static boolean isExceptionFromSameCause(Throwable a, Throwable b) {
+ if (a == null || b == null) return a == b; // Handle the null case
+ if (a.getClass() != b.getClass()) return false; // Different classes
+ if (!equalOrNull(a.getMessage(), b.getMessage())) return false; // Different messages. Null is allowed as wildcard
+ var s1 = a.getStackTrace();
+ var s2 = b.getStackTrace();
+ if (!sameStackTraceOrigin(s1, s2)) return false; // Different stack trace origins
+ if (!isExceptionFromSameCause(a.getCause(), b.getCause())) return false; // Different causes
+ var o1 = a.getSuppressed();
+ var o2 = b.getSuppressed();
+ if (o1 == null || o2 == null || o1.length != o2.length) return o1 == o2; // Different suppression
+ for (int i=0; i=(])--"), ""),
+ new Replace(Pattern.compile("\\b--"), "+"),
+ new Replace(Pattern.compile("\\b\\+-"), "-"),
+ new Replace(Pattern.compile("\\bfalse\\b"), "0"),
+ new Replace(Pattern.compile("\\btrue\\b"), "1"),
+ new Replace(Pattern.compile("-"), ""),
+ new Replace(Pattern.compile("(?=])(?:-|\\*|/|>=|<=|>|<|!=|==)(?![+\\-*/<>=])"), "+"),
+ new Replace(Pattern.compile("\\{(?:[^{}]|\\{})+}"), "{}"),
+ new Replace(Pattern.compile("\\{((?:[^{}]|\\{})+)}"), "$1"),
+ new Replace(Pattern.compile("\\(([^()]+)\\)"), "$1"),
+ new Replace(Pattern.compile("\\b\\w+\\b"), "1"),
+ new Replace(Pattern.compile("\\b\\w+\\b"), "0"),
+ new Replace(Pattern.compile("\\d+\\+"), ""),
+ new Replace(Pattern.compile("\\+\\d+"), ""),
+ new Replace(Pattern.compile("(?<=\\n|^)[^\\n]*(?=\\n|$)"), "\n"),
+ new Replace(Pattern.compile("\\n+"), "\n"),
+ new Replace(Pattern.compile("^\\n+|\\n+$"), ""),
+ new Replace(Pattern.compile("else *"), ""),
+ new Replace(Pattern.compile("\\bif\\([^()]+\\) *"), ""),
+ new Replace(Pattern.compile("\\bwhile\\([^()]+\\) *"), ""),
+ };
+
+ /**
+ * Predicate to test if the reduced script generates the same error.
+ */
+ private final Predicate tester;
+
+ /**
+ * Create a reducer with a test function.
+ * @param tester Test function which returns true when the same issue is found in the script.
+ */
+ private Reducer(Predicate tester) {
+ this.tester = tester;
+ }
+
+ /**
+ * Rewrites the script with one rewrite rule.
+ */
+ private String doReplacement(String script, Pattern pattern, String with) {
+ var matcher = pattern.matcher(script);
+ var sb = new StringBuilder();
+ int last = 0;
+ while (matcher.find()) {
+ var pos = sb.length();
+ matcher.appendReplacement(sb, with);
+ var tail = sb.length();
+ matcher.appendTail(sb);
+ if (tester.test(sb.toString())) {
+ sb.setLength(tail);
+ } else {
+ sb.setLength(pos);
+ sb.append(script, last, matcher.end());
+ }
+ last = matcher.end();
+ }
+ matcher.appendTail(sb);
+ return sb.toString();
+ }
+
+ /**
+ * Clean up variable names from the random ones generated by the fuzzer.
+ */
+ private String cleanVariables(String script) {
+ var matcher = Pattern.compile("\\bint ([0-9a-zA-Z_]+)").matcher(script);
+ int num = 0;
+ while (matcher.find()) {
+ var var = matcher.group(1);
+ if (var.matches("v\\d+")) continue;
+ while (script.contains("v"+num)) num++;
+ var n = script.replaceAll("\\b"+var+"\\b", "v" + num);
+ if (tester.test(n)) {
+ script = n;
+ }
+ }
+ return script;
+ }
+
+ /**
+ * Run all rewrite rules once.
+ */
+ private String doAllReplacements(String script) {
+ for (var replace: REPLACEMENTS) {
+ script = doReplacement(script, replace.p, replace.r);
+ }
+ return script;
+ }
+
+ /**
+ * Reduce the script to a minimal version which produces the same problem with a test function.
+ * @param tester Test function which returns true when the same issue is found in the script.
+ */
+ public static String reduce(String script, Predicate tester) {
+ var reducer = new Reducer(tester);
+ String old;
+ do {
+ old = script;
+ script = reducer.doAllReplacements(script);
+ script = reducer.cleanVariables(script);
+ } while (!script.equals(old));
+ return script;
+ }
+
+ /**
+ * Reduce the script to a minimal version which produces the same problem with a reproducer.
+ * @param ex The exception originally caused by the script
+ * @param reproducer Test function which throws an exception which will be checked against ex.
+ */
+ public static String reduce(String script, Throwable ex, Consumer reproducer) {
+ return reduce(script, s->{
+ Throwable nex = null;
+ try {
+ reproducer.accept(s);
+ } catch (Throwable e) {
+ nex = e;
+ }
+ return FuzzerUtils.isExceptionFromSameCause(nex, ex);
+ });
+ }
+
+}
diff --git a/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/ScriptGenerator.java b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/ScriptGenerator.java
new file mode 100644
index 00000000..9df3dfcc
--- /dev/null
+++ b/chapter08/src/test/java/com/seaofnodes/simple/fuzzer/ScriptGenerator.java
@@ -0,0 +1,462 @@
+package com.seaofnodes.simple.fuzzer;
+
+import com.seaofnodes.simple.Parser;
+import com.seaofnodes.simple.node.ScopeNode;
+
+import java.util.ArrayList;
+import java.util.HashSet;
+import java.util.Random;
+
+/**
+ * Generate a pseudo random script.
+ * This generator will generate the same script for two random number generators initialized with the same seed.
+ * To generate valid code it implements the parser but instead of parsing it emits the statements and expressions.
+ * It is guaranteed to terminate.
+ */
+public class ScriptGenerator {
+
+ /**
+ * Get the list of keywords from the parser.
+ */
+ @SuppressWarnings("unchecked")
+ private static HashSet getKeywords() {
+ try {
+ var field = Parser.class.getDeclaredField("KEYWORDS");
+ field.setAccessible(true);
+ return (HashSet) field.get(new Parser(""));
+ } catch (Throwable e) {
+ throw new ExceptionInInitializerError(e);
+ }
+ }
+
+ /**
+ * Number of spaces per indentation
+ */
+ private static final int INDENTATION = 4;
+ /**
+ * Valid characters for identifiers. The last 10 are not valid for the first character.
+ */
+ private static final String VAR_CHARS = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_0123456789";
+ /**
+ * List of keywords not valid for identifiers.
+ */
+ private static final HashSet KEYWORDS = getKeywords();
+
+ /**
+ * Flag for a statement that it does not pass on control flow to the next statement.
+ */
+ private static final int FLAG_STOP = 0x01;
+ /**
+ * Flag for a statement that it contains a final if statement without an else branch.
+ */
+ private static final int FLAG_IF_WITHOUT_ELSE = 0x02;
+ /**
+ * Binary operators.
+ */
+ private static final String[] BINARY_OP = {"==", "!=", "<", "<=", ">", ">=", "+", "-", "*", "/"};
+ /**
+ * Unary operators.
+ */
+ private static final String[] UNARY_OP = {"-"};
+
+
+ /**
+ * The random number generator used for random decisions.
+ */
+ private final Random random;
+ /**
+ * Indentation depth.
+ */
+ private int indentation = 0;
+ /**
+ * Output string builder
+ */
+ private final StringBuilder sb;
+ /**
+ * Number of nested loops
+ */
+ private int loopDepth = 0;
+ /**
+ * Current scope start in the variables list.
+ */
+ private int currScopeStart = 0;
+ /**
+ * The depth of blocks allowed.
+ */
+ private int depth = 7;
+ /**
+ * The depth of expressions allowed.
+ */
+ private int exprDepth = 3;
+ /**
+ * Current variables in scope.
+ */
+ private final ArrayList variables = new ArrayList<>();
+ /**
+ * If this needs to generate valid scripts or is also allowed to generate invalid ones.
+ */
+ private final boolean generateValid;
+ /**
+ * If this script might be invalid.
+ */
+ private boolean maybeInvalid = false;
+
+ /**
+ *
+ * @param random the random number generator.
+ * @param sb the script will be written to this string builder.
+ * @param generateValid if the script should be guaranteed valid or if it is allowed to contain bugs.
+ */
+ public ScriptGenerator(Random random, StringBuilder sb, boolean generateValid) {
+ this.random = random;
+ this.sb = sb;
+ this.generateValid = generateValid;
+ }
+
+ /**
+ * Random number with log distribution
+ * @param max The maximum value
+ * @return A random number in the range [0, max) with higher numbers being more uncommon.
+ */
+ private int randLog(int max) {
+ max--;
+ int n = 1<=0; i--) {
+ if ((num & (1< 0 && random.nextInt(10) > 7) {
+ // Use a variable outside the current scope.
+ var v = variables.get(random.nextInt(currScopeStart));
+ if (variables.lastIndexOf(v) < currScopeStart) return v;
+ if (!generateValid) {
+ maybeInvalid = true;
+ return v;
+ }
+ }
+ // Generate a new random variable name
+ StringBuilder sb = getRandomName();
+ var v = sb.toString();
+ if (!generateValid && (KEYWORDS.contains(v) || variables.lastIndexOf(v) >= currScopeStart)) {
+ maybeInvalid = true;
+ return v;
+ }
+ while (KEYWORDS.contains(v) || variables.lastIndexOf(v) >= currScopeStart) {
+ sb.append(VAR_CHARS.charAt(random.nextInt(VAR_CHARS.length())));
+ v = sb.toString();
+ }
+ return v;
+ }
+
+
+
+ /**
+ * Generates a program and writes it into the string builder supplied in the constructor.
+ * @return If the program is allowed to be invalid returns if something potentially invalid was generated.
+ */
+ public boolean genProgram() {
+ this.maybeInvalid = false;
+ variables.add(ScopeNode.ARG0);
+ currScopeStart = variables.size();
+ if ((genStatements() & FLAG_STOP) == 0) {
+ if (generateValid || random.nextInt(10)<7) {
+ genReturn();
+ } else {
+ this.maybeInvalid = true;
+ }
+ }
+ return !this.maybeInvalid;
+ }
+
+ /**
+ * Generate a list of statements.
+ * @return flags FLAG_STOP and FLAG_IF_WITHOUT_ELSE for the last statement generated
+ */
+ public int genStatements() {
+ var num = random.nextInt(10);
+ for (int i=0; i genBlock();
+ case 2 -> genIf();
+ case 3 -> genWhile();
+ case 4, 5 -> genDecl();
+ case 6, 7, 8 -> genAssignment();
+ default -> genExit();
+ };
+ }
+
+ /**
+ * Generate a statement for if, else and while blocks.
+ * This is special since it disallows declarations and prefers to generate blocks.
+ * @return flags FLAG_STOP and FLAG_IF_WITHOUT_ELSE for the generated statement
+ */
+ public int genStatementBlock() {
+ indentation += INDENTATION;
+ int res;
+ if (depth == 0) {
+ res = random.nextBoolean() ? genAssignment() : genExit();
+ } else {
+ depth--;
+ res = switch (random.nextInt(10)) {
+ case 1, 2, 3, 4, 5 -> {
+ depth++;
+ indentation -= INDENTATION;
+ var ret = genBlock();
+ indentation += INDENTATION;
+ depth--;
+ yield ret;
+ }
+ case 6 -> genIf();
+ case 7 -> genWhile();
+ case 8 -> genAssignment();
+ default -> genExit();
+ };
+ depth++;
+ }
+ indentation -= INDENTATION;
+ return res;
+ }
+
+ /**
+ * Generate an exit. This can be return or break and continue when in a loop.
+ * @return FLAG_STOP
+ */
+ public int genExit() {
+ if (loopDepth == 0 && generateValid) return genReturn();
+ return switch (random.nextInt(7)) {
+ case 0,1,2 -> {
+ if (loopDepth == 0) maybeInvalid = true;
+ sb.append("continue;");
+ yield FLAG_STOP;
+ }
+ case 3,4,5 -> {
+ if (loopDepth == 0) maybeInvalid = true;
+ sb.append("break;");
+ yield FLAG_STOP;
+ }
+ default -> genReturn();
+ };
+ }
+
+ /**
+ * Generate a block statement. After a certain depth only empty blocks are generated to ensure termination of the generator.
+ * @return flag FLAG_STOP for the generated statement
+ */
+ public int genBlock() {
+ if (depth == 0) {
+ sb.append("{}");
+ return 0;
+ }
+ depth--;
+ int oldCSS = currScopeStart;
+ currScopeStart = variables.size();
+ sb.append("{\n");
+ indentation += INDENTATION;
+ var stop = genStatements();
+ indentation -= INDENTATION;
+ printIndentation().append("}");
+ while (variables.size() > currScopeStart)
+ variables.removeLast();
+ currScopeStart = oldCSS;
+ depth++;
+ return stop & ~FLAG_IF_WITHOUT_ELSE;
+ }
+
+ /**
+ * Generate an if statement with an optional else block.
+ * @return flags FLAG_STOP and FLAG_IF_WITHOUT_ELSE for the generated statement
+ */
+ public int genIf() {
+ sb.append("if(");
+ genExpression();
+ sb.append(") ");
+ var stop = genStatementBlock();
+ if ((stop & FLAG_IF_WITHOUT_ELSE) == 0 && random.nextInt(10) > 3) {
+ sb.append("\n");
+ printIndentation().append("else ");
+ stop &= genStatementBlock();
+ } else {
+ stop = FLAG_IF_WITHOUT_ELSE;
+ }
+ return stop;
+ }
+
+ /**
+ * Generate a while loop.
+ * @return 0
+ */
+ public int genWhile() {
+ sb.append("while(");
+ genExpression();
+ sb.append(") ");
+ loopDepth++;
+ genStatementBlock();
+ loopDepth--;
+ return 0;
+ }
+
+ /**
+ * Generate a declaration statement.
+ * @return 0
+ */
+ public int genDecl() {
+ String name = getVarName();
+ sb.append("int ").append(name).append("=");
+ genExpression();
+ sb.append(";");
+ variables.add(name);
+ return 0;
+ }
+
+ /**
+ * Generate an assignment statement.
+ * @return 0
+ */
+ public int genAssignment() {
+ if (variables.isEmpty()) return genDecl();
+ genVariable();
+ sb.append("=");
+ genExpression();
+ sb.append(";");
+ return 0;
+ }
+
+ /**
+ * Generate a return statement.
+ * @return FLAG_STOP
+ */
+ public int genReturn() {
+ sb.append("return ");
+ genExpression();
+ sb.append(";");
+ return FLAG_STOP;
+ }
+
+
+ /**
+ * Generate a binary expression.
+ * This method does not care about operator precedence.
+ */
+ public void genExpression() {
+ var num = randLog(10);
+ while(num-->0) {
+ genUnary();
+ sb.append(BINARY_OP[random.nextInt(BINARY_OP.length)]);
+ }
+ genUnary();
+ }
+
+ /**
+ * Generate a unary expression.
+ */
+ public void genUnary() {
+ var num = randLog(6);
+ while(num-->0) {
+ sb.append(UNARY_OP[random.nextInt(UNARY_OP.length)]);
+ }
+ genPrimary();
+ }
+
+ /**
+ * Generate a primary expression.
+ */
+ public void genPrimary() {
+ switch (random.nextInt(10)) {
+ case 0:
+ sb.append("true");
+ break;
+ case 1:
+ sb.append("false");
+ break;
+ case 2:
+ if (exprDepth == 0) {
+ // Ensure termination of the generator.
+ if (random.nextBoolean()) {
+ genNumber();
+ } else {
+ genVariable();
+ }
+ } else {
+ sb.append("(");
+ exprDepth--;
+ genExpression();
+ exprDepth++;
+ sb.append(")");
+ }
+ break;
+ case 3, 4, 5, 6:
+ genNumber();
+ break;
+ default:
+ genVariable();
+ }
+ }
+
+ /**
+ * Generate a number.
+ */
+ public void genNumber() {
+ sb.append(random.nextInt(100));
+ }
+
+ /**
+ * Generate a variable. If there are none generate a number instead.
+ * If the program is allowed to be invalid a random name can be generated in some cases.
+ */
+ public void genVariable() {
+ if (!generateValid && random.nextInt(100)>97) {
+ sb.append(getRandomName());
+ maybeInvalid = true;
+ return;
+ }
+ if (variables.isEmpty()) {
+ genNumber();
+ } else {
+ sb.append(variables.get(random.nextInt(variables.size())));
+ }
+ }
+
+}
--
cgit v1.2.3