neo4j-driver-java-skill

When to Use

• Java/Kotlin code connecting to Neo4j (Aura or self-managed)
• Setting up driver, sessions, transactions in Maven/Gradle projects
• Debugging result handling, error recovery, connection pool issues
• Async (CompletableFuture) or reactive (Project Reactor / RxJava) Neo4j access

When NOT to Use

• Cypher query authoring/optimization → neo4j-cypher-skill
• Driver version upgrades → neo4j-migration-skill
• Spring Data Neo4j (@Node, @Relationship, Neo4jRepository) → neo4j-spring-data-skill

---

Dependency

Maven

<dependency>
    <groupId>org.neo4j.driver</groupId>
    <artifactId>neo4j-java-driver</artifactId>
    <version>6.0.5</version>
</dependency>

Gradle

implementation 'org.neo4j.driver:neo4j-java-driver:6.0.5'

Check latest: https://central.sonatype.com/artifact/org.neo4j.driver/neo4j-java-driver

---

Environment Variables

Standard pattern for connection config — never hardcode credentials:

String uri      = System.getenv().getOrDefault("NEO4J_URI",      "neo4j://localhost:7687");
String user     = System.getenv().getOrDefault("NEO4J_USERNAME",  "neo4j");
String password = System.getenv().getOrDefault("NEO4J_PASSWORD",  "");
String database = System.getenv().getOrDefault("NEO4J_DATABASE",  "neo4j");

Spring Boot: inject via @Value("${spring.neo4j.uri}") or application.properties:

spring.neo4j.uri=neo4j+s://xxx.databases.neo4j.io
spring.neo4j.authentication.username=neo4j
spring.neo4j.authentication.password=secret

---

Driver Lifecycle

One Driver per application — thread-safe, expensive to create. Implement AutoCloseable or use try-with-resources.

// Long-lived singleton
var driver = GraphDatabase.driver(
    "neo4j+s://xxx.databases.neo4j.io",          // Aura TLS+routing
    AuthTokens.basic(user, password));
driver.verifyConnectivity();                      // fail fast

// Short-lived (tests / CLI)
try (var driver = GraphDatabase.driver(uri, AuthTokens.basic(user, password))) {
    driver.verifyConnectivity();
    // ...
}

URI schemes:

URI	Use
neo4j://localhost	Unencrypted, cluster routing
neo4j+s://xxx.databases.neo4j.io	TLS + cluster routing (Aura)
bolt://localhost:7687	Unencrypted, single instance
bolt+s://localhost:7687	TLS, single instance

Auth options: AuthTokens.basic(u,p) · AuthTokens.bearer(token) · AuthTokens.kerberos(b64) · AuthTokens.none()

---

Choosing the Right API

API	When	Auto-retry	Streaming
driver.executableQuery()	Default for most queries	✅	❌ eager
session.executeRead/Write()	Large results, callback control	✅	✅
session.beginTransaction()	Multi-method, external coordination	❌	✅
session.run()	Self-managing queries (CALL IN TRANSACTIONS)	❌	✅
driver.asyncSession()	Non-blocking CompletableFuture	✅	✅
driver.rxSession()	Reactor/RxJava backpressure	✅	✅

CALL { … } IN TRANSACTIONS and USING PERIODIC COMMIT self-manage their transaction — use session.run() only. executableQuery and executeRead/Write will fail for these queries.

---

executableQuery — Default

// Read — route to replicas
var result = driver.executableQuery("""
        MATCH (p:Person {name: $name})-[:KNOWS]->(friend)
        RETURN friend.name AS name
        """)
    .withParameters(Map.of("name", "Alice"))
    .withConfig(QueryConfig.builder()
        .withDatabase("neo4j")            // always specify — avoids home-db round-trip
        .withRouting(RoutingControl.READ)
        .build())
    .execute();

result.records().forEach(r -> System.out.println(r.get("name").asString()));
long ms = result.summary().resultAvailableAfter(TimeUnit.MILLISECONDS);

// Write
driver.executableQuery("CREATE (p:Person {name: $name, age: $age})")
    .withParameters(Map.of("name", "Bob", "age", 30))
    .withConfig(QueryConfig.builder().withDatabase("neo4j").build())
    .execute();

Never string-interpolate Cypher. Always .withParameters(Map.of(...)).

---

Managed Transactions (executeRead / executeWrite)

Sessions are NOT thread-safe — one per request/thread, always close.

try (var session = driver.session(SessionConfig.builder()
        .withDatabase("neo4j").build())) {

    // Read → replica routing
    var names = session.executeRead(tx -> {
        var result = tx.run(
            "MATCH (p:Person) WHERE p.name STARTS WITH $prefix RETURN p.name AS name",
            Map.of("prefix", "Al"));
        return result.stream().map(r -> r.get("name").asString()).toList(); // collect INSIDE
    });

    // Write → leader routing
    session.executeWriteWithoutResult(tx ->
        tx.run("CREATE (p:Person {name: $name})", Map.of("name", "Carol"))
    );
}

Result must be consumed INSIDE the callback

Result is a lazy cursor tied to the open transaction. Transaction closes when callback returns — any read after that throws ResultConsumedException.

// ❌ Returns Result — already closed by the time caller uses it
var result = session.executeRead(tx ->
    tx.run("MATCH (p:Person) RETURN p.name AS name"));
result.stream().forEach(...); // throws ResultConsumedException

// ✅ Collect to List inside callback
var names = session.executeRead(tx ->
    tx.run("MATCH (p:Person) RETURN p.name AS name")
      .stream().map(r -> r.get("name").asString()).toList());

Callback rules

• Consume each Result before next tx.run() — multiple open cursors = undefined behaviour.
• No side effects (HTTP, email, metric increments) — callback may be retried on transient errors.
• Use MERGE (idempotent), not CREATE, for retry-safe writes.
• executeRead → replica; executeWrite → leader.

TransactionConfig — timeouts & metadata

var config = TransactionConfig.builder()
    .withTimeout(Duration.ofSeconds(5))
    .withMetadata(Map.of("app", "myService", "user", userId))  // visible in SHOW TRANSACTIONS
    .build();
session.executeRead(tx -> { /* ... */ }, config);

---

Explicit Transactions

Use when work spans multiple methods or requires external coordination. Not auto-retried.

try (var session = driver.session(SessionConfig.builder().withDatabase("neo4j").build())) {
    var tx = session.beginTransaction();
    try {
        doPartA(tx);
        doPartB(tx);
        tx.commit();
    } catch (Exception e) {
        try { tx.rollback(); } catch (Exception rb) { e.addSuppressed(rb); }
        throw e;
    }
}

tx.rollback() is a network call — wrap in its own try/catch and use addSuppressed so the original exception is not lost.

Commit uncertainty: if tx.commit() throws ServiceUnavailableException, the commit may or may not have succeeded. Design writes as idempotent (MERGE + unique constraints) so retrying is safe.

Choose explicit vs managed:

• Auto-retry needed → executeRead / executeWrite
• Work spans multiple methods → explicit (pass tx as parameter)
• Coordinating with external I/O → explicit (commit only after I/O succeeds)

---

Error Handling

try {
    driver.executableQuery("...").execute();
} catch (ServiceUnavailableException e) {
    // No servers — check connection
} catch (SessionExpiredException e) {
    // Server closed session — open new one
} catch (TransientException e) {
    // Managed txns retry automatically; explicit txns need manual retry
} catch (Neo4jException e) {
    // Cypher/constraint error — e.code() gives GQL status code
}

Managed transactions auto-retry TransientException — no catch needed.

---

Data Types & Value Extraction

Cypher type	Java accessor
Integer	value.asLong() / value.asInt()
Float	value.asDouble()
String	value.asString()
Boolean	value.asBoolean()
List	value.asList()
Map	value.asMap()
Node	value.asNode()
Relationship	value.asRelationship()
Date	value.asLocalDate()
DateTime	value.asZonedDateTime()

var record = result.records().get(0);
String name = record.get("name").asString();
long age    = record.get("age").asLong();

var node = record.get("p").asNode();
String label = node.labels().iterator().next();
Map<String,Object> props = node.asMap();

Null safety — two distinct cases

Situation	record.get(key)	.asString()
Key present, value non-null	the value	returns string
Key present, value is graph null	Value where .isNull() = true	throws Uncoercible
Key absent (typo / not projected)	Value.NULL sentinel	throws NoSuchElementException

// Graph null — use default overload (safe only if key is always projected):
String city = record.get("city").asString("Unknown");

// Absent key — check containsKey first:
if (record.containsKey("city") && !record.get("city").isNull()) {
    String city = record.get("city").asString();
}

---

Object Mapping

Map query results to Java records/classes directly — eliminates manual accessor calls.

// Domain record — field names match RETURN aliases (case-sensitive)
public record Person(String name, long age) {}

// Map single record
var person = driver.executableQuery("MATCH (p:Person {name: $name}) RETURN p.name AS name, p.age AS age")
    .withParameters(Map.of("name", "Alice"))
    .withConfig(QueryConfig.builder().withDatabase("neo4j").build())
    .execute()
    .records()
    .stream()
    .map(r -> r.get("name").asString())   // or: r.as(Person.class) — see note
    .findFirst()
    .orElseThrow();

// Using .as(Person.class) — maps RETURN keys to record fields by name
var person2 = driver.executableQuery("""
        MATCH (p:Person {name: $name})
        RETURN p.name AS name, p.age AS age
        """)
    .withParameters(Map.of("name", "Tom Hanks"))
    .withConfig(QueryConfig.builder().withDatabase("neo4j").build())
    .execute()
    .records()
    .stream()
    .map(record -> record.get("p").as(Person.class))
    .findFirst()
    .orElseThrow(() -> new RuntimeException("Person not found"));

Nested mapping — return a map projection and include COLLECT {} for lists:

public record Movie(String title, List<Person> actors) {}

var movieCypher = """
    MATCH (movie:Movie)
    LIMIT 1
    RETURN movie {
        .title,
        actors: COLLECT {
            MATCH (actor:Person)-[:ACTED_IN]->(movie)
            RETURN actor
        }
    }
    """;

var movie = driver.executableQuery(movieCypher)
    .withConfig(QueryConfig.builder().withDatabase("neo4j").build())
    .execute()
    .records()
    .stream()
    .map(r -> r.get("movie").as(Movie.class))
    .findFirst()
    .orElseThrow();

Only mapped properties defined in the record are populated — extra properties returned by Cypher are ignored.

---

Performance Patterns

Always specify database — omitting triggers home-db round-trip on every call.

Route reads to replicas — RoutingControl.READ in QueryConfig or use executeRead.

Batch writes with UNWIND — pass List<Map<String,Object>> (plain maps only; custom objects fail):

List<Map<String, Object>> rows = people.stream()
    .map(p -> Map.<String, Object>of("name", p.name(), "age", p.age()))
    .toList();

driver.executableQuery("UNWIND $items AS item MERGE (p:Person {name: item.name}) SET p.age = item.age")
    .withParameters(Map.of("items", rows))
    .withConfig(QueryConfig.builder().withDatabase("neo4j").build())
    .execute();

Allowed leaf types in parameter maps: String, Long/Integer/Short/Byte, Double/Float, Boolean, List<?>, Map<String,?>, null. Custom objects and LocalDate must be converted first.

Group writes in one transaction — one executeWrite with a loop, not one executeWrite per iteration.

Connection pool — default 100 connections. Tune if exhausted:

Config.builder()
    .withMaxConnectionPoolSize(50)
    .withConnectionAcquisitionTimeout(30, TimeUnit.SECONDS)
    .build()

---

Common Errors

Mistake	Fix
String-interpolate Cypher params	.withParameters(Map.of(...)) always
Omit database name	Set in QueryConfig / SessionConfig every time
New Driver per request	Create once at startup; share everywhere
Share Session across threads	One session per request/thread
Return Result from tx callback	Collect to List/Map inside callback
Leave Result open before next tx.run()	Consume before next call
Side effects in managed tx callback	Move outside — callback may retry
Pass custom objects to UNWIND params	Convert to List<Map<String,Object>>
asString() on graph null	.asString("default") or check .isNull()
asString() on absent key	containsKey() before optional access
Naked tx.rollback() in catch	Wrap in try/catch; use addSuppressed
Assume commit() failure = no commit	Commit uncertainty — design writes idempotent
Block inside async callback (.join())	Chain with thenCompose
Skip session close in async error path	exceptionallyCompose to close then re-throw
One transaction per write in loop	Batch with UNWIND or group in one callback
executeWrite for a read	Use executeRead — routes to replica

---

References

Load on demand:

• references/async-reactive.md — full async CompletableFuture patterns, reactive RxSession with Flux.usingWhen, deadlock avoidance
• references/advanced-config.md — full Config.builder() options, TLS, notification filtering, session-level auth, user impersonation, cross-session bookmarks, spatial types (Values.point/WGS-84/Cartesian)

Docs:

• Java Driver manual: https://neo4j.com/docs/java-manual/current/
• API reference: https://neo4j.com/docs/api/java-driver/current/

---

Checklist

• One Driver instance created at startup; closed on shutdown
• verifyConnectivity() called after driver creation
• Database name specified in every QueryConfig / SessionConfig
• Parameters used (never string-interpolated Cypher)
• Result consumed inside managed transaction callback
• No side effects inside executeRead/Write callbacks
• Sessions closed via try-with-resources
• Async sessions closed in both success and error paths (exceptionallyCompose)
• ServiceUnavailableException on commit handled as commit-uncertain
• UNWIND params are List<Map<String,Object>> (no custom objects)
• containsKey() checked before accessing optional result columns