CQL Heuristics Calculator by gonzalotguerrero · Pull Request #1620 · WebFuzzing/EvoMaster

gonzalotguerrero · 2026-07-01T21:37:09Z

Added CassandraHeuristicsCalculator/CassandraOperationEvaluator to compute a distance heuristic for CQL WHERE clauses

Distance calculation for each operator/data type follows the spec from this document

…alculator

…sion-replacement-v2

- Small refactorings to follow dev guidelines - Refactor parsing of Duration literals and reorganise packages - Unify distance calculation for strings - Refactor operation evaluator - Add javadoc for calculator and CassandraRow abstraction

- More refactorings in parser utils - Add missing docs in parser utils

…sion-replacement-v2

…ent-v2 More CqlSession Replacements

- Use pattern for standard ISO - Improve CqlDurationLiteralParser and create unit tests class - Replace reflection with use of TemporalAmount

jgaleotti · 2026-07-15T18:16:28Z

+        } else if (op instanceof ContainsKeyOperation<?>) {
+            return evaluateContainsKey((ContainsKeyOperation<?>) op, candidateRow);
+        } else {
+            return FALSE_TRUTHNESS;


throw an exception for not supported evaluations

jgaleotti · 2026-07-15T18:19:39Z

+    private Truthness evaluateContains(ContainsOperation<?> op, CassandraRow candidateRow) {
+        Object rawCol = candidateRow.getValue(op.getColumnName());
+        if (rawCol == null) {
+            return FALSE_TRUTHNESS;


this means that the column value for that column is null, or that the column is not defined?

It means the column is not defined for that specific row! I'll modify the code so that it's not ambiguous

what if a null value is stored in the candidateRow, how can you disambiguate?

jgaleotti · 2026-07-15T18:20:05Z

+        Object rawCol = candidateRow.getValue(op.getColumnName());
+        if (rawCol == null) {
+            return FALSE_TRUTHNESS;
+        } else if (!(rawCol instanceof Map<?, ?>)) {


could a contain key be evaluated on a column that is not a map?

According to the CQL documentation, that's not possible

I'll add an exception for this scenario!

jgaleotti · 2026-07-15T18:20:24Z

+
+    private Truthness evaluateContainsKey(ContainsKeyOperation<?> op, CassandraRow candidateRow) {
+        Object rawCol = candidateRow.getValue(op.getColumnName());
+        if (rawCol == null) {


again, is this a valid value for candidateRow.getValue(String) ?

If the candidate row does not have that column defined, it could be possible

jgaleotti · 2026-07-15T18:23:02Z

+        Object rowValue   = candidateRow.getValue(op.getColumnName());
+        Object queryValue = op.getValue();
+
+        if (rowValue == null && queryValue == null) {


Is NULL literal valid in CQL? I think it is not valid. Therefore, queryValue cannot have null as value, right?

You're right, it's not. I'll change the code so that only rowValue is considered!

jgaleotti · 2026-07-15T18:32:55Z

+            CqlDurationLiteral literalDurationValue = CqlDurationLiteralParser.parse((String) literalValue);
+
+            return TruthnessUtils.buildAndAggregationTruthness(
+                    TruthnessUtils.getEqualityTruthness(months, literalDurationValue.months),


define a variable for each getEqualityTruthness to improve readability

jgaleotti · 2026-07-15T18:33:20Z

+
+    private Truthness evaluateEquals(Object a, Object b) {
+        if (a == null && b == null) {
+            return FALSE_TRUTHNESS;


again, check semantics of NULL==NULL

jgaleotti · 2026-07-15T18:33:47Z

+        } else if (a == null || b == null) {
+            return FALSE_TRUTHNESS_BETTER;
+        } else {
+            Truthness raw = compareByType(a, b, ComparisonType.EQUALS);


what raw means here?

It means it's the truthness instance before being scaled! I'll change the name

jgaleotti · 2026-07-15T18:34:36Z

+            if (t.isEmpty()) {
+                throw new IllegalArgumentException("Empty duration literal");
+            } else {
+                boolean isNegative = t.startsWith("-");


replace "-" with constant

jgaleotti · 2026-07-15T18:35:22Z

+            String upper = unsigned.toUpperCase();
+            if (upper.endsWith(ISO_WEEK_SUFFIX)) {
+                return parseIso8601WeekPattern(upper);
+            } else if (upper.contains("-")) {


same as before

jgaleotti · 2026-07-21T18:21:21Z

+    private Truthness evaluateContains(ContainsOperation<?> op, CassandraRow candidateRow) {
+        Object rawCol = candidateRow.getValue(op.getColumnName());
+        if (rawCol == null) {
+            return FALSE_TRUTHNESS;


what if a null value is stored in the candidateRow, how can you disambiguate?

arcuri82 · 2026-07-23T20:11:58Z

+public class CassandraHeuristicsCalculator {
+
+    public static final double C        = DistanceHelper.H_NOT_NULL;
+    public static final double C_BETTER = 0.15;


either this C_BETTER should be computer from H_NOT_NULL (eg H_NOT_NULL + 0.05), or should be added directly to DistanceHelper with a more general name. otherwise, what if in the future we refactor to have H_NOT_NULL=0.2 for example?

I added it to DistanceHelper so that the SQL calculator can use it directly from there.

I also did this for TRUE_TRUTHNESS, FALSE_TRUTHNESS , and FALSE_TRUTHNESS_BETTER, and added them to TruthnessUtils.

arcuri82 · 2026-07-23T20:15:20Z

+    // In CQL, a stored NULL and an absent/never-set column are the same thing at read time
+    // (writing NULL creates a tombstone, identical to never having written the column), so
+    // none of the null checks below (here and in evaluateContains, evaluateContainsKey and
+    // evaluateComparison) can tell these two cases apart.


for functions and fields, use /** */ so you get JavaDoc out of it

arcuri82 · 2026-07-23T20:20:27Z

+     * optional, mirroring the driver's own grammar, so bare {@code P} or {@code PT} are valid
+     * (zero-length) durations.
+     */
+    private static final Pattern ISO8601_PATTERN = Pattern.compile(


a lot of the code here seems generic about date handling, and not specific to Cassandra, isn't it? if so, should refactor in some common re-usable utility functions not bound to Cassandra (as it could be re-used in other contexts, if needed)

I tried moving some of the logic to a new class IsoDurationParser, let me know what you think!

- Abstract parsing of ISO duration format - Turn comments into javadoc - Unify distance and truthness related constants

gonzalotguerrero added 15 commits June 19, 2026 16:49

Merge branch 'master' into feature/cql-heuristics-calculator

76bb1fb

First calculator draft

92aa301

Merge branch 'master' into feature/cql-heuristics-calculator

56c806a

Merge branch 'feature/cql-query-parser' into feature/cql-heuristics-c…

c80e69d

…alculator

Fix comparison of time-related values

6b6ad91

Improve test class coverage

059e9a5

Add support for integer constants in time-related types

8455eb6

Merge branch 'master' into feature/cql-heuristics-calculator

b6e870b

Created CassandraOperationEvaluator

a796494

Minor refactorings

bcf43e2

Add more replacements for overloaded execute method

6dd2267

Add separate fields for keyspace and table in ExecutedCqlCommand

ec7e7f3

Merge branch 'master' into feature/cql-heuristics-calculator

d28282c

Merge branch 'master' into feature/cql-session-replacement-v2

f8ccc59

Merge branch 'feature/cql-heuristics-calculator' into feature/cql-ses…

f61c578

…sion-replacement-v2

gonzalotguerrero requested a review from jgaleotti July 6, 2026 21:20

jgaleotti requested changes Jul 8, 2026

View reviewed changes

gonzalotguerrero added 3 commits July 8, 2026 20:09

PR review comments

288be95

- Small refactorings to follow dev guidelines - Refactor parsing of Duration literals and reorganise packages - Unify distance calculation for strings - Refactor operation evaluator - Add javadoc for calculator and CassandraRow abstraction

Improve CqlParserUtils

8a566ee

- More refactorings in parser utils - Add missing docs in parser utils

Merge branch 'master' into feature/cql-heuristics-calculator

c9028dd

gonzalotguerrero requested a review from jgaleotti July 9, 2026 03:23

gonzalotguerrero added 2 commits July 9, 2026 01:45

Divide regex to improve readability

d19881f

Merge branch 'feature/cql-heuristics-calculator' into feature/cql-ses…

5263ce5

…sion-replacement-v2

jgaleotti requested changes Jul 9, 2026

View reviewed changes

arcuri82 and others added 4 commits July 10, 2026 12:53

Merge pull request #1629 from WebFuzzing/feature/cql-session-replacem…

432b4c7

…ent-v2 More CqlSession Replacements

Evaluator and parsers refactorings

2cc704e

Improve parsing and comparison of Duration values:

d2179ac

- Use pattern for standard ISO - Improve CqlDurationLiteralParser and create unit tests class - Replace reflection with use of TemporalAmount

More refactorings in evaluator

bfbe32b

gonzalotguerrero requested a review from jgaleotti July 15, 2026 02:52

jgaleotti requested changes Jul 15, 2026

View reviewed changes

gonzalotguerrero added 3 commits July 20, 2026 20:16

Refactorings and fix handling of unsupported operations

37aa780

Fix handling of missing columns in rows

e8af58a

Fix tests

f30c556

gonzalotguerrero requested a review from jgaleotti July 21, 2026 02:18

jgaleotti requested changes Jul 21, 2026

View reviewed changes

Add comment regarding NULL

7671a3e

gonzalotguerrero requested a review from jgaleotti July 21, 2026 21:03

jgaleotti approved these changes Jul 21, 2026

View reviewed changes

jgaleotti requested a review from arcuri82 July 21, 2026 21:22

Merge branch 'master' into feature/cql-heuristics-calculator

7ea66bc

arcuri82 requested changes Jul 23, 2026

View reviewed changes

Review comments:

ac4b916

- Abstract parsing of ISO duration format - Turn comments into javadoc - Unify distance and truthness related constants

gonzalotguerrero requested a review from arcuri82 July 24, 2026 04:53

Uh oh!

Conversation

gonzalotguerrero commented Jul 1, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gonzalotguerrero Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

gonzalotguerrero Jul 24, 2026 •

edited

Loading