How to Build Flexible Server‑Side Queries with Spring Data Querydsl

Overview: This blog introduces how to implement server‑side query parsing and execution using the Spring Querydsl framework.

Defining Query Requests

Backend services must parse query requests sent from the UI according to a shared specification. The article shows how Baidu and Google encode the search string as a q parameter in the URL.

Two typical query patterns are described:

Google‑like query

A single search box where the whole query string is placed in an HTTP parameter; the server decides which tables and fields to search.

Field‑specific SQL‑like query

Queries are expressed as field: value pairs, optionally combined with logical operators AND and OR. Examples:

name:bill

name:bill AND city:LA

name:bill OR city:LA

The server must translate these strings into a type‑safe predicate that can be executed against the database.

Using Spring Data Querydsl

Querydsl provides a fluent, type‑safe API for building SQL‑like queries. It was created to avoid string‑concatenated HQL and to keep queries in sync with domain model changes.

Official site: http://querydsl.com/

Spring Data integrates Querydsl through the QuerydslPredicateExecutor interface. By extending a repository with this interface, developers can write predicates such as:

public interface QuerydslPredicateExecutor<T> {
    // findById, findAll, count, exists …
}

Example of a predicate built with Querydsl:

Predicate predicate = user.firstname.equals("dave")
    .and(user.lastname.startsWith("mathews"));
userRepository.findAll(predicate);

This translates to the SQL fragment where firstname = 'dave' and lastname = 'mathews%'.

Implementing Dynamic Queries with Spring Querydsl

The article provides a concrete example using a Student model with fields id, gender, firstName, lastName, createdAt, and isGraduated. Queries can be composed freely, e.g.:

firstname:li AND lastname:hua

firstname:li OR lastname:hua AND gender:male

Parsing logic is handled by a QueryAnalysis class that splits the query string, extracts field, operator and value using a regular expression, and distributes AND and OR parts into separate containers.

Two builder classes are shown:

PredicateBuilder : collects SearchCriteria objects for AND and OR groups and creates BooleanExpression objects.

BooleanExpressionBuilder : converts a single SearchCriteria into a Querydsl BooleanExpression based on the field type (String, Date, Integer, Boolean) and the operator ( :, !:, >, <).

Key points include regular‑expression‑driven parsing, validation of query strings, separate handling of logical groups, and type‑specific predicate generation.

Challenges

AND queries involving parent and child tables

When a parent entity contains a collection (e.g., a customer with multiple markets), a query such as

customerNumber:5135116903 AND markets.active:false AND markets.marketId:A1

must ensure that the AND conditions apply to the same child record, not merely to any element of the collection. This requires careful predicate construction to avoid false positives.

Overall, the article demonstrates how to turn flexible, user‑defined query strings into type‑safe Querydsl predicates that can be executed efficiently on the backend.