Visit the official documentation: https://gestalt-config.github.io/gestalt/
All documentation is being migrated to the official documentation site.
Gestalt is a powerful Java configuration library designed to simplify the way you handle and manage configurations in your software projects. In the rapidly evolving landscape of software development, having a flexible and reliable configuration management system is essential for ensuring the smooth operation of your applications.
Gestalt offers a comprehensive solution to the challenges of configuration management. It allows you to source configuration data from multiple inputs, merge them intelligently, and present them in a structured, type-safe manner. Whether you're working with Java beans, lists, sets, or primitive data types, Gestalt's automatic decoding based on data types simplifies the process.
This documentation will guide you through the key features of Gestalt, demonstrate how to get started quickly, and provide detailed insights into its capabilities. Whether you're a seasoned Java developer or just beginning your journey, Gestalt will empower you to manage your application configurations effectively and efficiently.
Let's dive in and explore how Gestalt can streamline your configuration management workflow and help you build more robust and adaptable software.
Automatic decoding based on type: Supports decoding into bean classes, lists, sets, or primitive types. This simplifies configuration retrieval.
Java Records: Full support for Java Records, constructing records from configuration using the Records Canonical Constructor.
Supports Multiple Formats: Load configurations from various sources, including Environment Variables, Property files, an in-memory map, and more.
Read Sub-sections of Your Config: Easily navigate to specific sub-sections within configurations using dot notation.
Kotlin interface: Full support for Kotlin with an easy-to-use Kotlin-esque interface, ideal for Kotlin projects.
Merge Multiple Sources: Seamlessly merge configurations from different sources to create comprehensive settings.
String Substitution: Build a config value by injecting Environment Variables, System Properties or other nodes into your strings. Evaluate the substitution at either load time or run time.
node Substitution: Include whole config nodes loaded from files or other places in the config tree anywhere in your config tree.
A/B Testing Segregate results based on groups or random results. See A/B Testing in the use cases section.
Flexible and Configurable: The library offers well-defined interfaces, allowing customization and extension.
Easy-to-Use Builder: Get started quickly with a user-friendly builder, or customize specific aspects of the library.
Receive All Errors Up Front: In case of configuration errors, receive multiple errors in a user-friendly log for efficient debugging.
Modular Support for Features: Include only the required features and dependencies in your build, keeping your application lightweight.
Zero Dependencies: The core library has zero external dependencies; add features and dependencies as needed.
Java 11 Minimum: Requires a minimum of Java 11 for compatibility with modern Java versions.
Java Modules: Supports Java 9 modules with proper exports.
Well Tested: Our codebase boasts an impressive ~92% code coverage, validated by over 1975 meaningful tests.
Versions 0.1.0 through version 0.11.0 require Java 8. Versions 0.12.0 plus require Java 11.
repositories {
mavenCentral()
}
Import gestalt-core, and the specific modules you need to support your use cases.
Gradle example:
implementation 'com.github.gestalt-config:gestalt-core:${version}'
implementation 'com.github.gestalt-config:gestalt-kotlin:${version}'
Or with the kotlin DSL:
implementation("com.github.gestalt-config:gestalt-core:$version")
implementation("com.github.gestalt-config:gestalt-kotlin:$version")
Maven Example:
<dependency>
<groupId>com.github.gestalt-config</groupId>
<artifactId>gestalt-core</artifactId>
<version>${version}</version>
</dependency>
Setup your configuration files
Multiple types of configurations are supported from multiple sources.
Here is an example of the default.properties
:
db.hosts[0].user=credmond
db.hosts[0].url=jdbc:postgresql://localhost:5432/mydb1
db.hosts[1].user=credmond
db.hosts[1].url=jdbc:postgresql://localhost:5432/mydb2
db.hosts[2].user=credmond
db.hosts[2].url=jdbc:postgresql://localhost:5432/mydb3
db.connectionTimeout=6000
db.idleTimeout=600
db.maxLifetime=60000.0
http.pool.maxTotal=100
http.pool.maxPerRoute=10
http.pool.validateAfterInactivity=6000
http.pool.keepAliveTimeoutMs=60000
http.pool.idleTimeoutSec=25
Here is an example of the dev.properties
:
db.hosts[0].url=jdbc:postgresql://dev.host.name1:5432/mydb
db.hosts[1].url=jdbc:postgresql://dev.host.name2:5432/mydb
db.hosts[2].url=jdbc:postgresql://dev.host.name3:5432/mydb
db.connectionTimeout=600
http.pool.maxTotal=1000
http.pool.maxPerRoute=50
Construct Gestalt using the builder.
Use the builder to construct the Gestalt library. It is possible to do this manually, but the builder greatly simplifies the construction of the library. It uses the service loader to automatically load all the default dependencies.
// Create a map of configurations we wish to inject.
Map<String, String> configs = new HashMap<>();
configs.put("db.hosts[0].password", "1234");
configs.put("db.hosts[1].password", "5678");
configs.put("db.hosts[2].password", "9012");
configs.put("db.idleTimeout", "123");
// Load the default property files from resources.
URL devFileURL = GestaltSample.class.getClassLoader().getResource("dev.properties");
File devFile = new File(devFileURL.getFile());
// using the builder to layer on the configuration files.
// The later ones layer on and over write any values in the previous
Gestalt gestalt = new GestaltBuilder()
.addSource(ClassPathConfigSourceBuilder.builder().setResource("/default.properties").build()) // Load the default property files from resources.
.addSource(FileConfigSourceBuilder.builder().setFile(devFile).build())
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.build();
// Loads and parses the configurations, this will throw exceptions if there are any errors.
gestalt.loadConfigs();
Retrieve configurations from Gestalt
Using the Gestalt Interface you can load sub nodes with dot notation into a wide variety of classes.
For non-generic classes you can pass in the class with getConfig("db.port", Integer.class)
or for classes with generic types we need to use a special TypeCapture wrapper that captures the generic type at runtime. This allows us to construct generic classes with such as Listnew TypeCapture<List<String>>() {}
Short myShortWrapper = gestalt.getConfig("http.pool.maxTotal", Short.class);
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
List<HttpPool> httpPoolList = gestalt.getConfig("http.pools", new TypeCapture<>() { });
var httpPoolList = gestalt.getConfig("http.pools", new TypeCapture<List<HttpPool>>() { });
The API to retrieve configurations:
/**
* Get a config for a path and a given class.
* If the config is missing or there are any errors it will throw a GestaltException
*/
<T> T getConfig(String path, Class<T> klass) throws GestaltException;
/**
* Get a config for a path and a given class.
* If the config is missing, invalid or there was an exception it will return the default value.
*/
<T> T getConfig(String path, T defaultVal, Class<T> klass);
/**
* Get a config Optional for a path and a given class.
* If the config is missing, invalid or there was an exception it will return an Optional.empty()
*/
<T> Optional<T> getConfigOptional(String path, Class<T> klass);
Adding a ConfigSource to the builder is the minimum step needed to build the Gestalt Library. You can add several ConfigSources to the builder and Gestalt, and they will be loaded in the order they are added. Where each new source will be merged with the existing source and where applicable overwrite the values of the previous sources. Each Config Source can be a diffrent format such as json, properties or Snake Case Env Vars, then internally they are converted into a common config tree.
Gestalt gestalt = builder
.addSource(FileConfigSourceBuilder.builder().setFile(defaults).build())
.addSource(FileConfigSourceBuilder.builder().setFile(devFile).build())
.addSource(EnvironmentConfigSourceBuilder.builder().setPrefix("MY_APP_CONFIG").build())
.build();
In the above example we first load a file defaults, then load a file devFile and overwrite any defaults, then overwrite any values from the Environment Variables. The priority will be Env Vars > devFile > defaults.
The config files are loaded and merged into a config tree. While loading into the config tree all node names and paths are converted to lower case and for environment variables we convert screaming snake case into dot notation. However, we do not convert other cases such as camel case into dot notation. So if your configs use a mix of dot notation and camel case, the nodes will not be merged. You can configure this conversion by providing your own Sentence Lexer
in the GestaltBuilder
. The config tree has a structure (sort of like json) where the root has one or more nodes or leafs. A node can have one or more nodes or leafs. A leaf can have a value but no further nodes or leafs. As we traverse the tree each node or leaf has a name and in combination it is called a path. A path can not have two leafs or both a node and a leaf at the same place. If this is detected Gestalt will throw an exception on loading with details on the path.
Valid:
db.connectionTimeout=6000
db.idleTimeout=600
db.maxLifetime=60000.0
http.pool.maxTotal=1000
http.pool.maxPerRoute=50
Invalid:
db.connectionTimeout=6000
db.idleTimeout=600
db=userTable #invalid the path db is both a node and a leaf.
http.pool.maxTotal=1000
http.pool.maxPerRoute=50
HTTP.pool.maxPerRoute=75 #invalid duplicate nodes at the same path.
All paths are converted to lower case as different sources have different naming conventions, Env Vars are typically Screaming Snake Case, properties are dot notation, json is camelCase. By normalizing them to lowercase it is easier to merge. However, we do not convert other cases such as camel case into dot notation. It is best to use a consistent case for your configurations.
To retrieve a configuration from Gestalt we need the path to the configuration as well as what type of class.
Gestalt is not case sensitive. Since Gestalt interops between Environment Variables and other sources with various cases, all strings in Gestalt are normalized to a lower case.
By default, Gestalt uses dot notation and allows indexing into arrays using a '[0]'.
If you want to use a different path style you can provide your own custom lexer to Gestalt. The SentenceLexer is used to convert the path passed to the Gestalt getConfig interface into tokens that Gestalt can use to navigate to your sub node.
// load a whole class, this works best with pojo's
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
// or get a specific config value from a class
short maxTotal gestalt.getConfig("HTTP.pool.maxTotal", Short.class);
// get with a default if you want a fallback from code
long maxConnectionsPerRoute = gestalt.getConfig("http.Pool.maxPerRoute", 24, Long.class);
// get a list of Host objects, or an PlaceHolder collection if there is no hosts found.
LinkedList<Host> hosts = gestalt.getConfig("db.hosts", new LinkedList(), new TypeCapture<LinkedList<Host>>() {});
// Get a class at a specific list index.
Host host = gestalt.getConfig("db.hosts[2]", Host.class);
// get a value of a class from a specific list index.
String password = gestalt.getConfig("db.hosts[2].password", String.class);
When decoding a path we do use path mappers to try different cases, but that only applies to the sub-tree starting at the path asked for. So if you call gestalt.getConfig("http.pool", HttpPool.class)
it will try and map the cases for nodes under path http.pool
but not the nodes in the path http-pool
.
// given the record.
public record HttpPool(String poolSize, int timeout) {}
And the properties:
booking.service.pool.size = 10
booking-service.timeout = 10
When getting the config will fail, because booking.service
will not match booking-service
so it will not find the timeout.
HttpPool pool = gestalt.getConfig("booking.service", HttpPool.class);
But given the properties
booking.service.pool.size = 10
booking.service.timeout = 10
These calls will succeed. When decoding poolSize
it will first try a lowercase match of poolsize
, but will not find the node, so it will try pool.size
for a combined path of booking.service.pool.size
which it will find.
HttpPool pool = gestalt.getConfig("booking.service", HttpPool.class);
Getting primitive and boxed types involves calling Gestalt and providing the class of the type you are trying to retrieve.
Short myShortWrapper = gestalt.getConfig("http.pool.maxTotal", Short.class);
short myShort = gestalt.getConfig("http.pool.maxTotal", short.class);
String serviceMode = gestalt.getConfig("serviceMode", String.class);
Gestalt will automatically decode and provide the value in the type you requested.
To retrieve a complex object, you need to pass in the class for Gestalt to return. Gestalt will automatically use reflection to create the object, determine all the fields in the requested class, and then lookup the values in the configurations to inject into the object. It will attempt to use the setter fields first, then fallback to directly setting the fields.
There are two configuration options that allow you to control when errors are thrown when decoding complex objects.
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
treatMissingValuesAsErrors
Treat missing field values in an object, proxy, record, or data object as errors. This will cause the API to either throw errors or return an empty optional.
true
, any time a value that is not discretionary is missing, it will fail and throw an exception.false
, a missing value will be returned as null
or the default initialization. Null
for objects and 0
for primitives.treatMissingDiscretionaryValuesAsErrors
Treat missing discretionary values (optional, fields with defaults, fields with default annotations) in an object, proxy, record, or data object as errors.
false
, you will be able to get the configuration with default values or an empty Optional.true
, if a field is missing and would have had a default, it will fail and throw an exception.@Nullable
annotationsIf a field or method is annotated with a @Nullable
annotation, it will treat a missing value as a discretionary value. So as long as treatMissingDiscretionaryValuesAsErrors
is not enabled, @Nullable
fields will allow null values without throwing errors.
There are multiple @Nullable
annotations and for this to work the annotations must use @Retention(RetentionPolicy.RUNTIME)
so the annotation is available at runtime for Gestalt.
One good library to use is jakarta.annotation:jakarta.annotation-api
that has a @Nullable
with @Retention(RetentionPolicy.RUNTIME)
.
Here are some examples of required and discretionary fields and which setting can control if they are treated as errors or allowed.
public class DBInfo {
// discretionary value controlled by treatMissingValuesAsErrors
private Optional<Integer> port; // default value Optional.empty()
private String uri = "my.sql.db"; // default value "my.sql.db"
private @Config(defaultVal = "100") Integer connections; // default value 100
// required value controlled by treatMissingDiscretionaryValuesAsErrors
private String password; // default value null
}
public interface DBInfoInterface {
Optional<Integer> getPort(); // default value Optional.empty()
default String getUri() { // default value "my.sql.db"
return "my.sql.db";
}
@Config(defaultVal = "100")
Integer getConnections(); // default value 100
// required value controlled by treatMissingDiscretionaryValuesAsErrors
String getPassword(); // default value null
}
public record DBInfoRecord(
// discretionary value controlled by treatMissingDiscretionaryValuesAsErrors
@Config(defaultVal = "100") Integer connections, // default value 100
Optional<Integer> port, // default value Optional.empty()
// required value controlled by treatMissingDiscretionaryValuesAsErrors
String uri, // default value null
String password // default value null
) {}
data class DBInfoDataDefault(
// discretionary value controlled by treatMissingValuesAsErrors
var port: Int?, // default value null
var uri: String = "my.sql.db", // default value "my.sql.db"
@Config(defaultVal = "100") var connections: Integer, // default value 100
// required value cam not disable treatMissingDiscretionaryValuesAsErrors and allow nulls.
var password: String, // required, can not be null.
)
To get an interface you need to pass in the interface class for gestalt to return. Gestalt will use a proxy object when requesting an interface. When you call a method on the proxy it will look up the similarly named property, decode and return it.
iHttpPool pool = gestalt.getConfig("http.pool", iHttpPool.class);
To get an interface you need to pass in a TypeCapture with the Generic value of the class for gestalt to return. Gestalt supports getting Generic objects such as Lists, Maps or Sets. However, due to type erasure we need to capture the type using the TypeCapture class. The Generic can be any type Gestalt supports decoding such as a primitive wrapper or an Object.
List<HttpPool> httpPoolList = gestalt.getConfig("http.pool", new TypeCapture<>() { });
var httpPoolMap = gestalt.getConfig("http.pool", new TypeCapture<Map<String, HttpPool>>() { });
Gestalt supports multiple varieties of List such as AbstractList, CopyOnWriteArrayList, ArrayList, LinkedList, Stack, Vector, and SequencedCollection. If asked for a List it will default to an ArrayList. Gestalt supports multiple varieties of Maps such as HashMap, TreeMap, ArrayList, LinkedHashMap and SequencedMap. If asked for a Map it will default to an HashMap. Gestalt supports multiple varieties of Sets such as HashSet, TreeSet, LinkedHashSet, LinkedHashMap and SequencedSet. If asked for a Set it will default to an HashSet.
For non-generic classes you can use the interface that accepts a class HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
, for Generic classes you need to use the interface that accepts a TypeCapture List<HttpPool> pools = gestalt.getConfig("http.pools", Collections.emptyList(), new TypeCapture<List<HttpPool>>() {});
to capture the generic type. This allows you to decode Lists, Sets and Maps with a generic type.
There are multiple ways to get a configuration with either a default, an Optional or the straight value. With the default and Optional Gestalt will not throw an exception if there is an error, instead returning a default or an PlaceHolder Option and log any warnings or errors.
The API also supports tagged configuration, where providing a tag will retrieve configs that match the specific tags or fallback to the default of no tags. You can implement profiles or environments using tags.
<T> T getConfig(String path, T defaultVal, Class<T> klass, Tags tags);
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class, Tags.of("environment", "dev"));
Most configuration sources support tagging them. So you can easily add tags to all properties in a source for your profile or environment.
Gestalt gestalt = new GestaltBuilder()
.addSource(ClassPathConfigSourceBuilder.builder().setResource("/default.properties").build()) // Load the default property files from resources.
.addSource(FileConfigSourceBuilder.builder().setFile(devFile).setTags(Tags.of("environment", "dev")).build())
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.build();
There are utility methods for common tags such as profile and environment.
Tags.profile("test") == Tags.of("profile", "test")
Tags.environment("dev") == Tags.of("environment", "dev")
You can set a default tag in the gestalt builder. The default tags are applied to all calls to get a gestalt configuration when tags are not provided. If the caller provides tags they will be used and the default tags will be ignored.
Gestalt gestalt = new GestaltBuilder()
.addSource(ClassPathConfigSourceBuilder.builder().setResource("/default.properties").build()) // Load the default property files from resources.
.addSource(FileConfigSourceBuilder.builder().setFile(devFile).setTags(Tags.profile("dev").build()))
.addSource(FileConfigSourceBuilder.builder().setFile(testFile).setTags(Tags.profile("test").build()))
.setDefaultTags(Tags.profile("dev"))
.build();
// has implicit Tags of Tags.profile("dev") that is applied as the default tags, so it will use values from the devFile.
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
// will use the Tags.profile("test") and ignore the default tags of Tags.profile("dev"), so it will use values from the testFile.
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class, Tags.profile("test"));
By default, Gestalt expects tags to be an exact match to select the roots to search. This is configurable by setting a different ConfigNodeTagResolutionStrategy
in the gestalt builder.
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs2)
.addTag(Tag.profile("orange"))
.addTag(Tag.profile("flower"))
.build())
.setConfigNodeTagResolutionStrategy(new SubsetTagsWithDefaultTagResolutionStrategy())
.build();
You can implement the interface ConfigNodeTagResolutionStrategy
to define your own resolution strategy.
The available strategies are:
name | Set Theory | Description |
---|---|---|
EqualTagsWithDefaultTagResolutionStrategy | Equals | Will Search two config node roots, the one that is an equal match to the tags and the root with no tags. Then return the config node roots to be searched. Only return the roots if they exist. |
SubsetTagsWithDefaultTagResolutionStrategy | Subset | Will Search for any roots that are a subset of the tags provided with a fallback of the default root. In combination with default tags, this can be used to create a profile system similar to Spring Config. |
You can provide tags to gestalt in two ways, setting the defaults in the gestalt config and passing in tags when getting a configuration.
Gestalt gestalt = new GestaltBuilder()
.addSource(ClassPathConfigSourceBuilder.builder().setResource("/default.properties").build()) // Load the default property files from resources.
.addSource(FileConfigSourceBuilder.builder().setFile(devFile).setTags(Tags.profile("dev").build()))
.addSource(FileConfigSourceBuilder.builder().setFile(testFile).setTags(Tags.profile("test").build()))
.setDefaultTags(Tags.profile("dev"))
.build();
// will use the Tags.profile("test") and ignore the default tags of Tags.profile("dev"), so it will use values from the testFile.
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class, Tags.profile("test"));
The default behaviour is to use the provided tags with the getConfig
and if not provided, fall back to the defaults.
By passing in the TagMergingStrategy to the GestaltBuilder, you can set your own strategy.
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs2)
.addTag(Tag.profile("orange"))
.addTag(Tag.profile("flower"))
.build())
.setTagMergingStrategy(new TagMergingStrategyCombine())
.build();
The available strategies are:
name | Set Theory | Description |
---|---|---|
TagMergingStrategyFallback | exclusive or | Use the provided tags with getConfig , and if not provided use a default fallback. |
TagMergingStrategyCombine | union | Merge the provided tags with getConfig , and the defaults |
You can provide your own strategy by implementing TagMergingStrategy.
Example of how to create and load a configuration objects using Gestalt:
public static class HttpPool {
public short maxTotal;
public long maxPerRoute;
public int validateAfterInactivity;
public double keepAliveTimeoutMs = 6000; // has a default value if not found in configurations
public OptionalInt idleTimeoutSec = 10; // has a default value if not found in configurations
public float defaultWait = 33.0F; // has a default value if not found in configurations
public HttpPool() {
}
}
public static class Host {
private String user;
private String url;
private String password;
private Optional<Integer> port;
public Host() {
}
// getter and setters ...
}
...
// load a whole class, this works best with pojo's
HttpPool pool = gestalt.getConfig("http.pool", HttpPool.class);
// or get a spcific config value
short maxTotal = gestalt.getConfig("http.pool.maxTotal", Short.class);
// get with a default if you want a fallback from code
long maxConnectionsPerRoute = gestalt.getConfig("http.pool.maxPerRoute", 24, Long.class);
// get a list of objects, or an PlaceHolder collection if there is no hosts found.
List<Host> hosts = gestalt.getConfig("db.hosts", Collections.emptyList(),
new TypeCapture<List<Host>>() {});
In Kotlin you dont need to specify the types if you used the kotlin extension methods provided in gestalt-kotlin
. It uses inline reified methods that automatically capture the type for you based on return type. If no configuration is found and the type is nullable, it will return null otherwise it will throw an GestaltException.
data class HttpPool(
var maxTotal: Short = 0,
var maxPerRoute: Long = 0,
var validateAfterInactivity: Int? = 0,
var keepAliveTimeoutMs: Double = 6000.0,
var idleTimeoutSec: Short = 10,
var defaultWait: Float = 33.0f
)
// load a kotlin data class
val pool: HttpPool = gestalt.getConfig("http.pool")
// get a list of objects, or an PlaceHolder collection if there is no hosts found.
val hosts: List<Host> = gestalt.getConfig("db.hosts", emptyList())
When decoding a Java Bean style class, a record, an interface or a Kotlin Data Class you can provide a custom annotation to override the path for the field as well as provide a default.
The field annotation @Config
takes priority if both the field and method are annotated.
The class annotation @ConfigPrefix
allows the user to define the prefix for the config object as part of the class instead of the getConfig()
call. If you provide both the resulting prefix is first the path in getConfig then the prefix in the @ConfigPrefix
annotation.
For example using @ConfigPrefix(prefix = "connection")
with DBInfo pool = gestalt.getConfig("db", DBInfo.class);
the resulting path would be db.connection
.
@ConfigPrefix(prefix = "db")
public class DBInfo {
@Config(path = "channel.port", defaultVal = "1234")
private int port;
public int getPort() {
return port;
}
}
DBInfo pool = gestalt.getConfig("", DBInfo.class);
public class DBInfo {
private int port;
@Config(path = "channel.port", defaultVal = "1234")
public int getPort() {
return port;
}
}
DBInfo pool = gestalt.getConfig("db.connection", DBInfo.class);
The path provided in the annotation is used to find the configuration from the base path provided in the call to Gestalt getConfig.
So if the base path from gestalt.getConfig is db.connection
and the annotation is channel.port
the path the configuration will look for is db.connection.channel.port
The default accepts a string type and will be decoded into the property type using the gestalt decoders. For example if the property is an Integer and the default is "100" the integer value will be 100.
Certain annotations can be applied to a configuration using @{annotation}
, this will covert the annotation to metadata that can be applied to the node. Then the metadata is used to apply the intended behaviour to the node.
For example, we can apply the temporary node feature on a node by using the annotation @{temp:1}
my.password=abcdef@{temp:1}
annotation | parameter | description |
---|---|---|
temp | (int) Number of times this temp node can be read | restrict the number of times a value can be read before it is released |
encrypt | (boolean) if we should apply to this node | Encrypts the node in memory. |
nocache | (boolean) if we should apply to this node | Will not cache the node. If a node is part of a object the whole object will not be cached. |
secret | (boolean) if we should apply to this node | Treats the node as a secret, so it will not print it out in errors or the debug print. |
By default, white spaces before and after the annotation are trimmed. You can disable this feature using the gestalt builder and setting setAnnotationTrimWhiteSpace(false)
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.build())
.setAnnotationTrimWhiteSpace(false)
.build();
When decoding a class, we need to know what configuration to lookup for each field. To generate the name of the configuration to lookup, we first find the path as defined in the call to gestalt.getConfig("book.service", HttpPool.class)
where the path is book.service
. We do not apply the path mappers to the path, only the config tree notes from the path. Once at the path we check class for any annotations. If there are no annotations, then we search for the fields by exact match. So we look for a config value with the same name as the field. If it is unable to find the exact match, it will attempt to map it to a path based on camel case. Where the camel case words will be separated and converted to Kebab case, Snake case and Dot Notation, then used to search for the configuration.
The order is descending based on the priority of the mapper.
Casing | Priority | Class Name |
---|---|---|
Camel Case (exact match) | 1000 | StandardPathMapper |
Kebab Case | 600 | KebabCasePathMapper |
Snake Case | 550 | SnakeCasePathMapper |
Dot Notation | 500 | DotNotationPathMapper |
Given the following class lets see how it is translated to the different casings:
// With a class of
public static class DBConnection {
@Config(path = "host")
private String uri;
private int dbPort;
private String dbPath;
}
Kebab Case: All objects in Java use the standard Camel case, however in the config files you can use Kebab case, and if an exact match isnt found it will search for a config variable converting Camel case into Kebab case. Kebab case or an exact match are preferred as using dot notation could potentially cause some issues as it is parsed to a config tree. Using dot notation you would need to ensure that none of values break the tree rules.
// Given a config of
"users.host" => "myHost"
"users.uri" => "myHost"
"users.dbPort" => "1234"
"users.db-path" => "usersTable"
// the uri will map to host
// the dbPort will map to dbPort using Camel case using exact match.
// the dbPath will automatically map to db.path using Kebab case.
DBConnection connection = gestalt.getConfig("users", TypeCapture.of(DBConnection.class));
Snake Case: All objects in Java use the standard Camel case, however in the config files you can use Snake case, and if an exact match isnt found it will search for a config variable converting Camel case into snake case.
// Given a config of
"users.host" => "myHost"
"users.uri" => "myHost"
"users.dbPort" => "1234"
"users.db_path" => "usersTable"
// the uri will map to host
// the dbPort will map to dbPort using Camel case using exact match.
// the dbPath will automatically map to db_path using snake case
DBConnection connection = gestalt.getConfig("users", TypeCapture.of(DBConnection.class));
Dot Notation: All objects in Java use the standard Camel case, however in the config files you can use Dot Notation, and if an exact match isnt found it will search for a config variable converting Camel case into Dot Notation. Kebab case or an exact match are preferred as using dot notation could potentially cause some issues as it is parsed to a config tree. Using dot notation you would need to ensure that none of values break the tree rules.
// Given a config of
"users.host" => "myHost"
"users.uri" => "myHost"
"users.dbPort" => "1234"
"users.db.path" => "usersTable"
// the uri will map to host
// the dbPort will map to dbPort using Camel case using exact match.
// the dbPath will automatically map to db.path using dot notation.
DBConnection connection = gestalt.getConfig("users", TypeCapture.of(DBConnection.class));
For Kotlin Gestalt includes several extension methods that allow easier use of Gestalt by way of reified functions to better capture the generic type information. Using the extension functions you don't need to specify the type if the return type has enough information to be inferred. If nothing is found it will throw a GestaltException unless the type is nullable, then it will return null.
val pool: HttpPool = gestalt.getConfig("http.pool")
val hosts: List<Host> = gestalt.getConfig("db.hosts", emptyList())
Gestalt Version | Kotlin Version |
---|---|
0.25.0 + | 1.9 |
0.17.0 + | 1.8 |
0.13.0 to 0.16.6 | 1.7 |
0.10.0 to 0.12.0 | 1.6 |
0.9.0 to 0.9.3 | 1.5 |
0.1.0 to 0.8.1 | 1.4 |
Using the $include
keyword as part of a config path, you can include the referenced config node tree into the path provided. By default, the node is merged into the provided node under the current node as defaults that will be overridden. You can control the order of the nodes, by including a number where < 0 is included below the current node and > 0 is included above the current node. The root node is always 0. Having two nodes share the same order is undefined. For example: $include:-1
for included under the current node, and $include:1
for included over the current node.
If you are included multiple nodes each node must have an order, or the results are undefined, and some includes may be lost.
You can include into the root or any sub node. It also supports nested include.
The include node must provide a source that is used to determine how to include the source. Each source accepts different parameters that can be provided in the form of a key value with a comma separated list. One of the key value pairs must be source
that is used to determine the source type.
For example a classPath source with the resource includes.properties
would look like:
$include=source=classPath,resource=includes.properties
Example of include a classPath Node into a sub path with properties file imports.properties
.
b=b changed
c=c
In the first example we include the loaded file node with default settings of order -1 $include:-1
, where the root node is always order 0. So the node will be loaded under the current root nodes so will provide defaults that will be overwritten.
Map<String, String> configs = new HashMap<>();
configs.put("a", "a");
configs.put("b", "b");
configs.put("$include", "source=classPath,resource=includes.properties");
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.build();
gestalt.loadConfigs();
Assertions.assertEquals("a", gestalt.getConfig("a", String.class));
Assertions.assertEquals("b", gestalt.getConfig("b", String.class));
Assertions.assertEquals("c", gestalt.getConfig("c", String.class));
That is why we don't see b=b changed
as it will be overwritten by b=b
, but we still see c=c
as it was in the included defaults and not overwritten.
In this second example we include the node with $include:1
. Since the root node is always order 0, the included nodes will override the root.
Map<String, String> configs = new HashMap<>();
configs.put("a", "a");
configs.put("b", "b");
configs.put("$include:1", "source=classPath,resource=includes.properties");
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.build();
gestalt.loadConfigs();
Assertions.assertEquals("a", gestalt.getConfig("a", String.class));
Assertions.assertEquals("b changed", gestalt.getConfig("b", String.class));
Assertions.assertEquals("c", gestalt.getConfig("c", String.class));
That is why we see b=b changed
as it is overwritten the root b=b
.
In the final example, we include the loaded file node in the sub path sub
.
Map<String, String> configs = new HashMap<>();
configs.put("a", "a");
configs.put("b", "b");
configs.put("sub.a", "a");
configs.put("sub.$include:1", "source=classPath,resource=includes.properties");
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.build();
gestalt.loadConfigs();
Assertions.assertEquals("a", gestalt.getConfig("a", String.class));
Assertions.assertEquals("b", gestalt.getConfig("b", String.class));
Assertions.assertEquals("a", gestalt.getConfig("sub.a", String.class));
Assertions.assertEquals("b changed", gestalt.getConfig("sub.b", String.class));
Assertions.assertEquals("c", gestalt.getConfig("sub.c", String.class));
As you can see the nodes from the file includes.properties
were included in the sub path sub
. As can bee seen with sub.b = b changed
and sub.c = c
.
Supported substitution sources:
Source Type | Module | Parameter | Description |
---|---|---|---|
classPath | gestalt-core | resource | The name of the classpath resource to load. |
node | gestalt-core | path | Load an node at the given path into the current node. |
env | gestalt-core | failOnErrors | If we should fail on errors. Since Env Vars may not always conform to Gestalt expectations we can disable the errors and make it more lenient while loading Env Vars. |
prefix | Only include Env Vars that match the prefix. | ||
ignoreCaseOnPrefix | When matching the prefix should it ignore case. | ||
removePrefix | If we should remove the prefix after matching. | ||
file | gestalt-core | file | Load a file at a given location to the current node. |
path | Load a file as a path at a given location to the current node. | ||
k8Secret | gestalt-core | path | The directory to scan for kubernetes secrets. |
file | The file directory to scan for kubernetes secrets. | ||
system | gestalt-core | failOnErrors | If we should fail on errors. Since System Variables may not always conform to Gestalt expectations we can disable the errors and make it more lenient while loading System Vars. |
s3 | gestalt-aws | Module Config | To use S3 with the include node feature you must register an S3Client via the AWSModuleConfig: Gestalt gestalt = builder.addModuleConfig(AWSBuilder.builder().setRegion("us-east-1").setS3Client(s3Client).build()).build(); |
bucket | The S3 bucket to search in. | ||
key | The Key of the config file to load. | ||
blob | gestalt-azure | Module Config | To use Azure Blob with the include node feature you must register an BlobClient or a StorageSharedKeyCredential via the AzureModuleBuilder : Gestalt gestalt = builder.addModuleConfig(AzureModuleBuilder.builder().setBlobClient(blobClient).build())).build(); |
endpoint | Azure endpoint to access the blob storage. | ||
container | Azure Container containing the blob. | ||
blob | The blob with the file. | ||
git | gestalt-git | Module Config | When accessing private repos you must register the GitModuleConfig with Gestalt. Gestalt gestalt = new GestaltBuilder().addModuleConfig(GitModuleConfigBuilder.builder().setCredentials(new UsernamePasswordCredentialsProvider(userName, password)).build()).build(); |
repoURI | Where to locate the repo | ||
branch | What branch to find the config files. | ||
configFilePath | The subpath in the repo URI to find the config file. | ||
localRepoDirectory | Where to save the git files Gestalt Syncs. | ||
gcs | gestalt-google-cloud | Module Config | To use GCS with the include node feature you can register a Storage client via the GoogleModuleConfig : Gestalt gestalt = builder.addModuleConfig(GoogleModuleConfigBuilder.builder().setStorage(storage).build())).build(); . Otherwise it will fallback to the default storage client. |
bucketName | What bucket to find the config files. | ||
objectName | The specific config file to include. |
Gestalt supports string substitutions using ${}
at load time on configuration properties to dynamically modify configurations.
For example if we have a properties file with a Database connection you don't want to save your usernames and passwords in the properties files. Instead, you want to inject the username and passwords as Environment Variables.
db.user=${DB_USER}
db.password=${DB_PASSWORD}
You can use multiple string replacements within a single string to build a configuration property.
db.uri=jdbc:mysql://${DB_HOST}:${DB_PORT}/${environment}
Load time ${}
substitutions are evaluated when we load the configurations and build the config tree. This is done once on gestalt.load()
then all results are cached in the config tree and returned.
Run time #{}
substitutions are evaluated at runtime when you call gestalt.getConfig(...);
, the results are not cached and each time you call gestalt.getConfig(...);
you will re-evaluate the value.
It is recommended to use Load time ${}
substitutions in the vast majority of cases as it is more performant. The main use case for run time #{}
substitutions is for values you expect to change from one run to the next, such as wanting a different random number each time you call gestalt.getConfig(...);
.
Aside from evaluated time, the syntax and use of both ${}
and #{}
are otherwise identical, you can mix and match them as needed.
You can specify the substitution in the format ${transform:key} or ${key}. If you provide a transform name it will only check that one transform. Otherwise, it will check all the Transformer annotated with a @ConfigPriority
in descending order and will return the first matching value.
Unlike the rest of Gestalt, this is case-sensitive, and it does not tokenize the string (except the node transform). The key expects an exact match, so if the Environment Variable name is DB_USER you need to use the key DB_USER. Using db.user or db_user will not match.
db.uri=jdbc:mysql://${DB_HOST}:${map:DB_PORT}/${sys:environment}
You can provide a default for the substitution in the format ${transform:key:=default}
or ${key:=default}
. If you provide a default it will use the default value in the event that the key provided cant be found
db.uri=jdbc:mysql://${DB_HOST}:${map:DB_PORT:=3306}/${environment:=dev}
Using nested substitution, you can have a chain of defaults. Where you can fall back from one source to another.
test.duration=${sys:duration:=${env:TEST_DURATION:=120}}
In this example, it will first try the system variable duration
, then the Environment Variable TEST_DURATION
and finally if none of those are found, it will use the default 120
You can escape the value with '\' like \${my text}
to prevent the substitution. In Java you need to write \\
to escape the character in a normal string but not in a Text block
In nested substitutions you should escape both the opening token \${
and the closing token \}
to be clear what is escaped, otherwise you may get undetermined results.
user.block.message=You are blocked because \\${reason\\}
Gestalt supports nested and recursive substitutions. Where a substitution can happen inside another substitution and the results could trigger another substitution. Please use nested substitution sparingly, it can get very complex and confusing quickly. Using these variables:
Environment Variables:
DB_HOST=cloudHost
environment=dev
System Variables:
DB_HOST=localHost
environment=test
Map Variable:
DB_TRANSFORM=sys
DB_PORT=13306
config source:
db.uri=jdbc:mysql://${${DB_TRANSFORM}:DB_HOST}:${map:DB_PORT}/${sys:environment}
This will resolve ${DB_TRANSFORM} => sys
then resolve ${sys:DB_HOST} => localHost
For a configuration value of db.uri=jdbc:mysql://localHost:13306/test
Nested substitution resolving to a nested substitution. Given properties:
this.path = greeting
your.path = ${this.path}
my.path.greeting = good day
And a string to Substitute:
"${my.path.${your.path}}"
the result is good day
${your.path}
resolves to ${this.path}
${this.path}
is then resolved to greeting
And finally the path my.path.greeting
is resolved to good day
keyword | priority | source |
---|---|---|
env | 100 | Environment Variables |
envVar | 100 | Deprecated Environment Variables |
sys | 200 | Java System Properties |
map | 400 | A custom map provided to the constructor |
node | 300 | map to another leaf node in the configuration tree |
random | n/a | provides a random value |
base64Decode | n/a | decode a base 64 encoded string |
base64Encode | n/a | encode a base 64 encoded string |
classpath | n/a | load the contents of a file on the classpath into a string substitution. |
dist100 | n/a | Use a comma-separated list, where each element is a colon-separated pair of a threshold and its corresponding value. If an element has no threshold, it is treated as the default. For example, the format 10:red,30:green,blue defines ranges and outcomes for random distributions: numbers from 1 to 10 correspond to red , 11 to 30 correspond to green , and all numbers above 30 default to blue . This is best used with runtime evaluation using #{dist100:10:red,30:green,blue} |
file | n/a | load the contents of a file into a string substitution |
urlDecode | n/a | URL decode a string |
urlEncode | n/a | URL encode a string |
awsSecret | n/a | An AWS Secret is injected for the secret name and key. Configure the AWS Secret by registering a AWSModuleConfig using the AWSBuilder. Gestalt gestalt = builder.addModuleConfig(AWSBuilder.builder().setRegion("us-east-1").build()).build(); |
azureSecret | n/a | An Azure Secret is injected for the secret name and key. Configure the Azure Secret by registering a AzureModuleConfig using the AzureModuleBuilder. Gestalt gestalt = builder.addModuleConfig(AzureModuleBuilder.builder().setKeyVaultUri("test").setCredential(tokenCredential)).build(); |
gcpSecret | n/a | A Google Cloud Secret given the key provided. Optionally configure the GCP Secret by registering an GoogleModuleConfig using the GoogleBuilder, or let google use the defaults. Gestalt gestalt = builder.addModuleConfig(GoogleBuilder.builder().setProjectId("myProject").build()).build() |
vault | n/a | A vault Secret given the key provided. Configure the Vault Secret by registering an VaultModuleConfig using the VaultBuilder. Gestalt gestalt = builder.addModuleConfig(VaultBuilder.builder().setVault(vault).build()).build() . Uses the io.github.jopenlibs:vault-java-driver project to communicate with vault |
To inject a random variable during config node processing you can use the format ${random:type(origin, bound)} The random value is generated while loading the config, so you will always get the same random value when asking gestalt.
db.userId=dbUser-${random:int(5, 25)}
app.uuid=${random:uuid}
data type | format | notes |
---|---|---|
byte | byte | a random byte of data base 64 encoded |
byte | byte(length) | random bytes of provided length base 64 encoded |
int | int | a random int of all possible int values |
int | int(max) | a random int from 0 to the max value provided |
int | int(origin, bound) | a random int between origin and bound |
long | long | a random long of all possible long values |
long | long(max) | a random long from 0 to the max value provided |
long | long(origin, bound) | a random long between origin and bound |
float | float | a random float between 0 and 1 |
float | float(max) | a random float from 0 to the max value provided |
float | float(origin, bound) | a random float between origin and bound |
double | double | a random double of all possible long values |
double | double(max) | a random double from 0 to the max value provided |
double | double(origin, bound) | a random double between origin and bound |
boolean | boolean | a random boolean |
string | string | a random string of characters a-z of length 1 |
string | string(length) | a random string of characters a-z of length provided |
char | char | a random char of characters a-z |
uuid | uuid | a random uuid |
When adding a config source you are able to apply zero or more Tags to the source. Those tags are then applied to all configuration within that source. Tags are optional and can be omitted.
When retrieving the config it will first search for an exact match to the tags, if provided, then search for the configs with no tags. It will then merge the results.
If you provide 2 tags in the source, when retrieving the configuration you must provide those two exact tags.
// head.shot.multiplier = 1.3
// max.online.players = 32
ConfigSourcePackage pveConfig = ClassPathConfigSourceBuilder.builder().setResource("/test-pve.properties").setTags(Tags.of("mode", "pve")).build();
// head.shot.multiplier = 1.5
ConfigSourcePackage pvpConfig = ClassPathConfigSourceBuilder.builder().setResource("/test-pvp.properties").setTags(Tags.of("mode", "pvp")).build();
// head.shot.multiplier = 1.0
// gut.shot.multiplier = 1.0
ConfigSourcePackage defaultConfig = ClassPathConfigSourceBuilder.builder().setResource("/test.properties").setTags(Tags.of()).build(); // Tags.of() can be omitted
Gestalt gestalt = builder
.addSource(pveConfig)
.addSource(pvpConfig)
.addSource(defaultConfig)
.build();
// retrieving "head.shot.multiplier" values change depending on the tag.
float pvpHeadShot = gestalt.getConfig("head.shot.multiplier", Float.class, Tags.of("mode", "pve")); // 1.3
float pveHeadShot = gestalt.getConfig("head.shot.multiplier", Float.class, Tags.of("mode", "pvp")); // 1.5
float coopHeadShot = gestalt.getConfig("head.shot.multiplier", Float.class, Tags.of("mode", "coop")); // 1.0 fall back to default
float defaultHeadShot = gestalt.getConfig("head.shot.multiplier", Float.class); // 1.0
// Gut shot is only defined in the default, so it will always return the default.
float pvpGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "pve")); // 1.0
float pveGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "pvp")); // 1.0
float coopGutSoot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "coop")); // 1.0
float defaultGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class); // 1.0
// Max online players is only defined in the pvp, so it will only return with the pvp tags.
float pvpGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "pve")); // 32
float pveGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "pvp")); // not found
float coopGutSoot = gestalt.getConfig("gut.shot.multiplier", Float.class, Tags.of("mode", "coop")); // not found
float defaultGutShot = gestalt.getConfig("gut.shot.multiplier", Float.class); // not found
Config Source | module | Details |
---|---|---|
BlobConfigSource | gestalt-azure |
Loads a config source from Azure Blob, Must include package com.github.gestalt-config:gestalt-azure:version. |
ClassPathConfigSource | gestalt-core | Load a file from the java class path. Uses getResourceAsStream to find and load the InputStream. |
EnvironmentConfigSource | gestalt-core | Loads all Environment Variables in the system. It expects Env Vars to be in screaming snake case, and will parse the "_" as a path delimiter. will convert them to a list of key values from the Env Map for the config loader. You can provide a prefix to only load Environment Variables with the prefix. Then you can choose to keep the prefix or remove it. |
FileConfigSource | gestalt-core | Loads a file from the local file system. The format for the source will depend on the file extension of the file. For example if it is dev.properties, the format will be properties. Returns a InpuStream for the config loader. |
InputStreamConfigSource | gestalt-core | Load a configuration from a InputStream. The format for the source will depend on the file extension of the file. For example if it is dev.properties, the format will be properties. Returns a InpuStream for the config loader. |
KubernetesSecretConfigSource | gestalt-core | Specify a path to search for kubernetes secrets files. The directory is scanned and each file is added to the configuration. The name of the file is treated as the key for configuration and the content of the file is the value for the configuration. |
GCSConfigSource | gestalt-google |
Load a config from Google Cloud Storage. Requires a bucketName and a objectName. A google Storage object is optional, otherwise it defaults to the default instance. |
GitConfigSource | gestalt-git |
Syncs a remote repo locally then uses the files to build a configuration. This uses jgit and supports several forms of authentication. See GitConfigSourceTest.java for examples of use. |
MapConfigSource | gestalt-core | Allows you to pass in your own map, it will convert the map into a list of path and value for the config loader. |
StringConfigSource | gestalt-core | Takes any string and converts it into a InputStream. You must also provide the format type so we can match it to a loader. |
SystemPropertiesConfigSource | gestalt-core | Loads the Java System Properties and convert them to a list of key values or the config loader. |
S3ConfigSource | gestalt-aws |
Loads a config source from AWS S3, Must include package com.github.gestalt-config:gestalt-aws:version. |
URLConfigSource | gestalt-core | Loads a config source from a URL. |
Each config loader understands how to load a specific type of config. Often this is associated with a specific ConfigSource. For example the EnvironmentVarsLoader only loads the EnvironmentConfigSource. However, some loaders expect a format of the config, but accept it from multiple sources. For example the PropertyLoader expects the typical java property file, but it can come from any source as long as it is an input stream. It may be the system properties, local file, github, or S3.
Config Loader | Formats supported | details | module |
---|---|---|---|
EnvironmentVarsLoader | envVars | Loads Environment Variables from the EnvironmentConfigSource, it expects a list not a InputStream. By default, it splits the paths using a "_". You can also disable failOnErrors if you are receiving errors from the environment variables, as you can not always control what is present. By treating Errors as warnings it will not fail if it finds a configuration the parser doesn't understand. Instead it will ignore the specific config. | core |
MapConfigLoader | mapConfig | Loads a user provided Map from the MapConfigSource, it expects a list not a InputStream. By default, it splits the paths using a "." and tokenizes arrays with a numeric index as "[0]". | core |
PropertyLoader | properties, props, and systemProperties | Loads a standard property file from an InputStream. By default, it splits the paths using a "." and tokenizes arrays with a numeric index as "[0]". | core |
JsonLoader | json | Leverages Jackson to load json files and convert them into a ConfigNode tree. | gestalt-json |
TomlLoader | toml | Leverages Jackson to load toml files and convert them into a ConfigNode tree. | gestalt-toml |
YamlLoader | yml and yaml | Leverages Jackson to load yaml files and convert them into a ConfigNode tree. | gestalt-yaml |
HoconLoader | config | Leverages com.typesafe:config to load hocon files, supports substitutions. | gestalt-hocon |
If you didn't manually add any ConfigLoaders as part of the GestaltBuilder, it will add the defaults. The GestaltBuilder uses the service loader to create instances of the Config loaders. It will configure them by passing in the GestaltConfig to applyConfig. To register your own default ConfigLoaders add them to the builder, or add it to a file in META-INF\services\org.github.gestalt.config.loader.ConfigLoader and add the full path to your ConfigLoader
By default, Gestalt expects Environment Variables to be screaming snake case, but you can configure it to have a different case.
By registering a EnvironmentVarsLoaderModuleConfig
with the GestaltBuilder
you can customize the Environment Loader.
In this example it will expect double __
as delimiter.
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(EnvironmentConfigSourceBuilder.builder().build())
.addModuleConfig(EnvironmentVarsLoaderModuleConfigBuilder
.builder()
.setLexer(new PathLexer("__"))
.build())
.build();
gestalt.loadConfigs();
You can also customize many of the Loaders such as the YamlLoader
, TomlLoader
, JsonLoader
and HoconLoader
by registering the Module Configs with the builder.
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(ClassPathConfigSourceBuilder.builder().setResource("/default.yaml").build())
.addModuleConfig(YamlModuleConfigBuilder.builder()
.setObjectMapper(customObjectmapper)
.build())
.build();
gestalt.loadConfigs();
Type | details |
---|---|
Array | Java primitive array type with any generic class, Can decode simple types from a single comma separated value, or from an array node. You can escape the comma with a \, so the values are not split. |
BigDecimal | |
BigInteger | |
Boolean | Boolean and boolean |
Byte | Byte and byte |
Char | Char and char |
ConfigContainer | A container that caches your config value, and updates it when there is a configuration change. |
Date | takes a DateTimeFormatter as a parameter, by default it uses DateTimeFormatter.ISO_DATE_TIME |
Double | Double and double |
Duration | Decodes a duration from either a number or an ISO 8601 standardized string format like "PT42S". |
Enum | |
File | |
Float | Float and float |
Instant | |
Integer | Integer and int |
List | a Java list with any Generic class, Can decode simple types from a single comma separated value, or from an array node. You can escape the comma with a \, so the values are not split. Supports multiple varieties of List such as AbstractList, CopyOnWriteArrayList, ArrayList, LinkedList, Stack, Vector, and SequencedCollection. If asked for a List it will default to an ArrayList. |
LocalDate | Takes a DateTimeFormatter as a parameter, by default it uses DateTimeFormatter.ISO_LOCAL_DATE |
LocalDateTime | Takes a DateTimeFormatter as a parameter, by default it uses DateTimeFormatter.ISO_DATE_TIME |
Long | Long or long |
Map | A map, Assumes that the key is a simple class that can be decoded from a single string. ie a Boolean, String, Int. The value can be any type we can decode. Supports parsing a map from a string with the format key1=value1, key2=value2. You can escape the comma with a \, so the values are not split. Supports multiple varieties of Maps such as HashMap, TreeMap, ArrayList, LinkedHashMap and SequencedMap. If asked for a Map it will default to an HashMap. |
Object | Decodes a java Bean style class, although it will work with any java class. Will fail if the constructor is private. Will construct the class even if there are missing values, the values will be null or the default. Then it will return errors which you can disable using treatMissingValuesAsErrors = true. Decodes member classes and lists as well. |
Optional | Decodes an optional value, if no value is found it will return an Optional.empty() |
OptionalDouble | Decodes an optional Double, if no value is found it will return an OptionalDouble.empty() |
OptionalInt | Decodes an optional Integer, if no value is found it will return an OptionaInt.empty() |
OptionalLong | Decodes an optional Long, if no value is found it will return an OptionalLong.empty() |
Path | |
Pattern | |
Proxy (interface) | Will create a proxy for an interface that will return the config value based on the java bean method name. So a method "getCar()" would match a config named "car". If a config is missing it will call the default method if provided. Has 2 modes, Cached and pass-through, the default is Cached. Cached will receive a cache of all values on creation and return those from an internal cache. Pass-though will result the object on creation, but when calling to get the values it will call gestalt for each value. This allows you to always get the most recent values. To set the mode on the builder use Gestalt gestalt = builder.setProxyDecoderMode(ProxyDecoderMode.PASSTHROUGH) |
Record | Decodes a Java record. All members of the record must have a value or construction will fail.So unlike the Object decoder it will not have the option to default to null or provide defaults. Will construct the record even if there are extra values, it will ignore all extra values. |
Set | A Set with any Generic class, Can decode simple types from a single comma separated value, or from an array node. You can escape the comma with a \, so the values are not split. Provides an unordered HashSet. Supports multiple varieties of Sets such as HashSet, TreeSet, LinkedHashSet, LinkedHashMap and SequencedSet. If asked for a Set it will default to an HashSet. |
Short | Short or short |
String | |
StringConstructor | Will decode a class that has a constructor that accepts a single string. This will only match for leaf nodes. It will send the value of the leaf node to the String constructor. |
UUID |
For Kotlin, the kotlin specific decoders are only selected when calling from the Kotlin Gestalt extension functions, or when using KTypeCapture. Otherwise, it will match the Java decoder. Kotlin decoders: Boolean, Byte, Char, Data class, Double, Duration, Float, Integer, Long, Short, String
For kotlin data classes it builds a Kotlin Data class by creating a map of parameters. If there are any missing required parameters it will fail.
Required parameters are ones that don't have a default and are not nullable. An exception will be thrown in this case.
If all members are optional, and we have no parameters we will try and create the class with the default PlaceHolder constructor.
If you didn't manually add any Decoders as part of the GestaltBuilder, it will add the defaults. The GestaltBuilder uses the service loader to create instances of the Decoders. It will configure them by passing in the GestaltConfig to applyConfig. To register your own default Decoders add them to the builder, or add it to a file in META-INF\services\org.github.gestalt.config.decoder.Decoder and add the full path to your Decoder
Gestalt is idempotent, as in on calling loadConfigs()
a config tree is built and will not be updated, even if the underlying sources have changed.
By using Reload strategies you can tell Gestalt when the specific config source has changed to dynamically update configuration on the fly. Once the config tree has been rebuilt, Gestalt will trigger its own Gestalt Core Reload Listener. So you can get an update that the reload has happened.
When adding a ConfigSource to the builder, you can choose to a reload strategy. The reload strategy triggers from either a file change, a timer event or a manual call from your code. Each reload strategy is for a specific source, and will not cause all sources to be reloaded, only that source. Once Gestalt has reloaded the config it will send out its own Gestalt Core Reload event. you can add a listener to the builder to get a notification when a Gestalt Core Reload has completed. The Gestalt Cache uses this to clear the cache when a Config Source has changed.
Gestalt gestalt = builder
.addSource(FileConfigSourceBuilder.builder()
.setFile(devFile)
.addConfigReloadStrategy(new FileChangeReloadStrategy())
.build())
.addCoreReloadListener(reloadListener)
.build();
Reload Strategy | Details |
---|---|
FileChangeReload | Specify a FileConfigSource, and the FileChangeReload will listen for changes on that file. When the file changes it will tell Gestalt to reload the file. Also works with symlink and will reload if the symlink change. |
TimedConfigReloadStrategy | Provide a ConfigSource and a Duration then the Reload Strategy will reload every period defined by the Duration |
ManualConfigReloadStrategy | You can manually call reload to force a source to reload. |
For example if you want to use a Map Config Source, and have updated values reflected in calls to Gestalt, you can register a ManualConfigReloadStrategy
with a Map Config Source. Then after you can update the values in the map call reload()
on the ManualConfigReloadStrategy
to tell Gestalt you want to rebuild its internal Config Tree. Future calls to Gestalt should reflect the updated values.
Map<String, String> configs = new HashMap<>();
configs.put("some.value", "value1");
var manualReload = new ManualConfigReloadStrategy();
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.addConfigReloadStrategy(manualReload)
.build())
.build();
gestalt.loadConfigs();
Assertions.assertEquals("value1", gestalt.getConfig("some.value", String.class));
configs.put("some.value", "value2");
manualReload.reload();
Assertions.assertEquals("value2", gestalt.getConfig("some.value", String.class));
Configuration | default | Details |
---|---|---|
treatWarningsAsErrors | false | if we treat warnings as errors Gestalt will fail on any warnings. When set to true it overrides the behaviour in the below configs. |
treatMissingArrayIndexAsError | false | By default Gestalt will insert null values into an array or list that is missing an index. By enabling this you will get an exception instead |
treatMissingValuesAsErrors | false | By default Gestalt will not update values in classes not found in the config. Null values will be left null and values with defaults will keep their defaults. By enabling this you will get an exception if any value is missing. |
treatMissingDiscretionaryValuesAsErrors | true | Sets treat missing discretionary values (optional, fields with defaults, fields with default annotations) as an error. If this is false you will be able to get the configuration with default values or an empty Optional. If this is true, if a field is missing and would have had a default it will fail and throw an exception. |
dateDecoderFormat | null | Pattern for a DateTimeFormatter, if left blank will use the default for the decoder |
localDateTimeFormat | null | Pattern for a DateTimeFormatter, if left blank will use the default for the decoder |
localDateFormat | null | Pattern for a DateTimeFormatter, if left blank will use the default for the decoder |
substitutionOpeningToken | ${ | Customize what tokens gestalt looks for when starting replacing substrings |
substitutionClosingToken | } | Customize what tokens gestalt looks for when ending replacing substrings |
maxSubstitutionNestedDepth | 5 | Get the maximum string substitution nested depth. If you have nested or recursive substitutions that go deeper than this it will fail. |
nodeIncludeKeyword | $include | The token used to denote a included node. If this is found in a path it will attempt to load the node into the tree at this location. |
nodeNestedIncludeLimit | 5 | The maximum number of nested Node Includes Gestalt will attempt. If you have nested or recursive Includes that go deeper than this it will fail. |
observationsEnabled | false | if observations should be enabled. This needs to be used in conjunction with the gestalt-micrometer or other observations library. |
proxyDecoderMode | CACHE | Either CACHE or PASSTHROUGH, where cache means we serve results through a cache that is never updated or pass through where each call is forwarded to Gestalt to be looked up. |
Configurations often contain secret information. To protect this information we apply a layered approach.
One layer of security used by Gestalt is to restrict the number of times a value can be read before it is released, GC'ed and no longer accessible in memory.
The Temporary Value feature allows us to specify the secret using a regex and the number of times it is accessible. Once the leaf value has been read the accessCount times, it will release the secret value of the node by setting it to null. Eventually the secret node should be garbage collected. However, while waiting for GC it may still be found in memory. These values will not be cached in the Gestalt Cache and should not be cached by the caller. Since they are not cached there a performance cost since each request has to be looked up.
To protect values you can either annotate a configuration with @{temp:int}
or use the addTemporaryNodeAccessCount
methods in the GestaltBuilder
, register a TemporarySecretModule
by using the TemporarySecretModuleBuilder
.
Using the annotation @{temp:int}
:
Map<String, String> configs = new HashMap<>();
configs.put("my.password", "abcdef@{temp:1}");
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.build())
.build();
gestalt.loadConfigs();
// the first call will get the node and reduce the access cound for the node to 0.
Assertions.assertEquals("abcdef", gestalt.getConfig("my.password", String.class));
// The second time we get the node the value has been released and will have no result.
Assertions.assertTrue(gestalt.getConfigOptional("some.value", String.class).isEmpty());
Or using the method addTemporaryNodeAccessCount
in the gestalt builder:
Map<String, String> configs = new HashMap<>();
configs.put("my.password", "abcdef");
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.build())
.addTemporaryNodeAccessCount("password", 1)
.build();
gestalt.loadConfigs();
// the first call will get the node and reduce the access cound for the node to 0.
Assertions.assertEquals("abcdef", gestalt.getConfig("my.password", String.class));
// The second time we get the node the value has been released and will have no result.
Assertions.assertTrue(gestalt.getConfigOptional("some.value", String.class).isEmpty());
Or using the TemporarySecretModule
TemporarySecretModuleBuilder builder = TemporarySecretModuleBuilder.builder().addSecretWithCount("secret", 1);
GestaltBuilder builder = new GestaltBuilder();
builder.addModuleConfig(builder.build());
Another layer of security used by Gestalt is to encrypt your secrets at rest in memory.
The Encrypted Secrets feature allows us to specify the which secrets to encrypt using a regex. During initialization the secret is encrypted and the source value is released to be GC'ed. The source value may still be accessible until it has been GC'ed. The configuration will be read from its source, and an encryption cipher with iv will be generated for each secret. The cipher and iv is used to encrypt the secret and the decryption cipher is passed into the node for decryption when needed.
These values will not be cached in the Gestalt Cache and should not be cached by the caller. Since they are not cached there a performance cost since each request has to be decrypted. This makes it harder to access your secrets, but a persist attacker will be able to find the cipher, and iv in memory can decrypt the value.
To encrypt values you can either annotate a configuration with @{encrypt}
, use the addEncryptedSecret
methods in the GestaltBuilder
, register a EncryptedSecretModule
by using the EncryptedSecretModuleBuilder
.
Using the annotation @{encrypt}
:
Map<String, String> configs = new HashMap<>();
configs.put("my.password", "abcdef@{encrypt}");
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.build())
.build();
gestalt.loadConfigs();
// the call will get the encrypted node but will return the decrypted results.
Assertions.assertEquals("abcdef", gestalt.getConfig("my.password", String.class));
Or using the addEncryptedSecret
method on the builder:
Map<String, String> configs = new HashMap<>();
configs.put("my.password", "abcdef");
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.build())
.addEncryptedSecret("password")
.build();
gestalt.loadConfigs();
// the call will get the encrypted node but will return the decrypted results.
Assertions.assertEquals("abcdef", gestalt.getConfig("my.password", String.class));
Gestalt leverages System.logger, the jdk logging library to provide a logging facade. Many logging libraries provide backends for System Logger.
To use log4j2 as the logging backend for the system logger include these dependencies. This is supported in version 2.13.2 of log4j2.
In Maven:
<dependency>
<groupId>org.apache.logging.log4j</groupId>
<artifactId>log4j-jpl</artifactId>
<version>${version}</version>
<scope>runtime</scope>
</dependency>
Or in Gradle
implementation("org.apache.logging.log4j:log4j-jpl:${version}")
To use logback as the logging backend for the system logger include these dependencies. This is supported in version 2+ of Logback.
In Maven:
<dependency>
<groupId>org.slf4j</groupId>
<artifactId>slf4j-jdk-platform-logging</artifactId>
<version>${version}</version>
</dependency>
Or in Gradle
implementation("org.slf4j:slf4j-jdk-platform-logging:${version}")
Several places in the library we will print out the contents of a node if there is an error, or you call the debug print functionality. To ensure that no secrets are leaked we conceal the secrets based on searching the path for several keywords. If the keyword is found in the path the leaf value will be replaced with a configurable mask.
How to configure the masking rules and the mask.
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.addSecurityMaskingRule("port")
.setSecurityMask("&&&&&")
.build();
gestalt.loadConfigs();
String rootNode = gestalt.debugPrint(Tags.of());
Assertions.assertEquals("MapNode{db=MapNode{password=LeafNode{value='test'}, " +
"port=LeafNode{value='*****'}, uri=LeafNode{value='my.sql.com'}}}", rootNode);
By default, the builder has several rules predefined here.
Gestalt exposes several observations and provides a implementation for micrometer.
To import the micrometer implementation add gestalt-micrometer
to your build files.
In Maven:
<dependency>
<groupId>com.github.gestalt-config</groupId>
<artifactId>gestalt-micrometer</artifactId>
<version>${version}</version>
</dependency>
Or in Gradle
implementation("com.github.gestalt-config:gestalt-micrometer:${version}")
Then when building gestalt, you need to register the module config MicrometerModuleConfig
using the MicrometerModuleConfigBuilder
.
SimpleMeterRegistry registry = new SimpleMeterRegistry();
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.setObservationsEnabled(true)
.addModuleConfig(MicrometerModuleConfigBuilder.builder()
.setMeterRegistry(registry)
.setPrefix("myApp")
.build())
.build();
gestalt.loadConfigs();
There are several options to configure the micrometer module.
Option | Description | Default |
---|---|---|
meterRegistry | Provide the micrometer registry to submit observations. | SimpleMeterRegistry |
includePath | When getting a config include the path in the observations tags. This can be a high cardinality observation so is not recommended. | false |
includeClass | When getting a config include the class in the observations tags. This can be a high cardinality observation so is not recommended. | false |
includeOptional | When getting a config include if the configuration is optional or default as a true or false in the observation tags. | false |
includeTags | When getting a config include the tags in the request in the observations tags. | false |
prefix | Add a prefix to the observations to better group your observations. | gestalt |
The following observations are exposed
Observations | Description | Type | tags |
---|---|---|---|
config.get | Recorded when we request a configuration from gestalt that is not cached. | Timer | default:true if a default or optional value is returned. exception:exception class if there was an exception. |
reload | Recorded when a configuration is reloaded. | Timer | source:source name. exception:exception class if there was an exception. |
get.config.missing | Incremented for each missing configuration, if decoding a class this can be more than one. | Counter | optional: true or false depending if the optional value is optional or has a default. |
get.config.error | Incremented for each error while getting a configuration, if decoding a class this can be more than one. | Counter | |
get.config.warning | Incremented for warning error while getting a configuration, if decoding a class this can be more than one. | Counter | |
cache.hit | Incremented for each request served from the cache. A cache miss would be recorded in the observations config.get | Counter |
Gestalt allows a validator to hook into and validate calls to get a configuration object. Gestalt includes a Hibernate Bean Validator implementation.
If the object decoded fails to validate, a GestaltException
is thrown with the details of the failed validations.
For calls to getConfig
with a default value it will log the failed validations then return the default value.
For calls to getConfigOptional
it will log the failed validations then return an Optional.empty()
.
To import the Hibernate Validator implementation add gestalt-validator-hibernate
to your build files.
In Maven:
<dependency>
<groupId>com.github.gestalt-config</groupId>
<artifactId>gestalt-validator-hibernate</artifactId>
<version>${version}</version>
</dependency>
Or in Gradle
implementation("com.github.gestalt-config:gestalt-validator-hibernate:${version}")
Then when building gestalt, you need to register the module config HibernateModuleConfig
using the HibernateModuleBuilder
.
ValidatorFactory factory = Validation.buildDefaultValidatorFactory();
Validator validator = factory.getValidator();
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.setValidationEnabled(true)
.addModuleConfig(HibernateModuleBuilder.builder()
.setValidator(validator)
.build())
.build();
gestalt.loadConfigs();
For details on how to use the Hibernate Validator see their documentation.
Allow Gestalt to inject configuration directly into your classes using Guice using the @InjectConfig
annotation on any class fields. This does not support constructor injection (due to Guice limitation)
To enable add the new GestaltModule(gestalt)
to your Guice Modules, then pass in your instance of Gestalt.
See the unit tests for examples of use.
Injector injector = Guice.createInjector(new GestaltModule(gestalt));
MyService service = injector.getInstance(MyService.class);
// use the InjectConfig along with the path to inject configuration.
public static class MyService {
@InjectConfig(path = "db.user") DBConnection connection;
public DBConnection getConnection() {
return connection;
}
}
When you are using Kodein you can use it to inject your configurations directly into your objects.
By using the extension method gestalt
within the scope of the Kodein DI DSL you can specify the path to your configurations, and it will automatically inject configurations into your object.
See the unit tests for examples of use.
val kodein = DI {
bindInstance { gestalt!! }
bindSingleton { DBService1(gestalt("db")) }
bindSingleton { DBService2(gestalt("db", DBInfoPOJO(port = 1000, password = "default"))) }
}
val dbService1 = kodein.direct.instance<DBService1>()
When you are using Koin you can use it to inject your configurations directly into your objects.
By using the extension method gestalt
within the scope of the koin module DSL you can specify the path to your configurations, and it will automatically inject configurations into your object.
See the unit tests for examples of use.
val koinModule = module {
single { gestalt!! }
single { DBService1(gestalt("db")) }
single { DBService2(gestalt("db", DBInfoPOJO(port = 1000, password = "default"))) }
}
val myApp = koinApplication {
modules(koinModule)
}
val dbService1: DBService1 = myApp.koin.get()
Often you may wish to override a configuration value with a value provided on the command line.
One way to do this is to add a SystemPropertiesConfigSource
as the last source in Gestalt. This way it will have the highest priority and override all previous sources.
Then when running the project you provide the command line parameter -D
In this example we provide a config source for default and dev, but allow for the overriding those with the system properties.
with the property values
# default
http.pool.maxTotal=100
# dev
http.pool.maxTotal=1000
However, we override with a command line parameter of: -Dhttp.pool.maxTotal=200
// for this to work you need to set the following command line Options
// -Dhttp.pool.maxTotal=200
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(ClassPathConfigSourceBuilder.builder().setResource("default.properties").build())
.addSource(ClassPathConfigSourceBuilder.builder().setResource("dev.properties").build())
.addSource(SystemPropertiesConfigSourceBuilder.builder().build())
.build();
// Load the configurations, this will throw exceptions if there are any errors.
gestalt.loadConfigs();
GestaltConfigTest.HttpPool pool = gestalt.getConfig("http.pool", GestaltConfigTest.HttpPool.class);
Assertions.assertEquals(200, pool.maxTotal);
In the end we should get the value 200 based on the overridden command line parameter.
In a similar vein as overriding with command line variables, you can override with an Environment Variable.
There is two ways of doing this. You can use string substitution but an alternative is to use the EnvironmentConfigSource
.
In this example we provide a config source for default that uses string substitution to load an Env Var. It expects the Env Var to be an exact match, it does not translate it in any way. You can also provide a default that will be used if the Env Var is not found.
with the property values
# default
http.pool.maxTotal=${HTTP_POOL_MAXTOTAL:=1000}
Using an Environment Variable of: HTTP_POOL_MAXTOTAL=200
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(ClassPathConfigSourceBuilder.builder().setResource("default.properties").build())
.build();
// Load the configurations, this will throw exceptions if there are any errors.
gestalt.loadConfigs();
GestaltConfigTest.HttpPool pool = gestalt.getConfig("http.pool", GestaltConfigTest.HttpPool.class);
Assertions.assertEquals(200, pool.maxTotal);
In the end we should get the value 200 based on the Env Var. If we didnt provide the Env Var, it would default to 1000.
If you wish to use Env Vars to directly override values in your config you can use the EnvironmentConfigSource
as the last source in Gestalt. This way it will have the highest priority and override all previous sources.
The Environment Variables are expected to be Screaming Snake Case, then the path is created from the key split up by the underscore "_".
So HTTP_POOL_MAXTOTAL
becomes an equivalent path of http.pool.maxtotal
In this example we provide a config source for default and dev, but allow for the overriding those with the Env Var.
with the property values
# default
http.pool.maxTotal=100
# dev
http.pool.maxTotal=1000
However, we override with an Env Var of: HTTP_POOL_MAXTOTAL=200
// for this to work you need to set the following command line Options
// -Dhttp.pool.maxTotal=200
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(ClassPathConfigSourceBuilder.builder().setResource("default.properties").build())
.addSource(ClassPathConfigSourceBuilder.builder().setResource("dev.properties").build())
.addSource(EnvironmentConfigSource.builder().build())
.build();
// Load the configurations, this will throw exceptions if there are any errors.
gestalt.loadConfigs();
GestaltConfigTest.HttpPool pool = gestalt.getConfig("http.pool", GestaltConfigTest.HttpPool.class);
Assertions.assertEquals(200, pool.maxTotal);
In the end we should get the value 200 based on the overridden Environment Variable.
If you wish to use a different case then Screaming Snake Case, you would need to provide your own EnvironmentVarsLoader with your specific SentenceLexer lexer.
There are several configuration options on the EnvironmentConfigSource
,
Configuration Name | Default | Description |
---|---|---|
failOnErrors | false | If we should fail on errors. By default the Environment Config Source pulls in all Environment variables, and several may not parse correctly |
prefix | "" | By provide a prefix only Env Vars that start with the prefix will be included. |
ignoreCaseOnPrefix | false | Define if we want to ignore the case when matching the prefix. |
removePrefix | false | If we should remove the prefix and the following "_" or"." from the imported configuration |
Typically, when you get a configuration from Gestalt, you maintain a reference to the value in your class. You typically dont want to call Gestalt each time you want to check the value of the configuration. Although Gestalt has a cache, there is overhead in calling Gestalt each time. However, when you cache locally if the configuration in Gestalt change via a reload, you will still have a reference to the old value.
So, instead of getting your specific configuration you could request a ConfigContainer, or a proxy decoder (by providing an interface).
var myConfigValue = gestalt.getConfig("some.value", new TypeCapture<ConfigContainer<String>>() {});
The ConfigContainer will hold your configuration value with several options to get it.
var myValue = configContainer.orElseThrow();
var myOptionalValue = configContainer.getOptional();
Then, when there is a reload, the ConfigContainer or proxy decoder will get and cache the new configuration. Ensuring you always have the most recent value.
The following example shows a simple use case for ConfigContainer.
Map<String, String> configs = new HashMap<>();
configs.put("some.value", "value1");
var manualReload = new ManualConfigReloadStrategy();
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder()
.setCustomConfig(configs)
.addConfigReloadStrategy(manualReload)
.build())
.build();
gestalt.loadConfigs();
var configContainer = gestalt.getConfig("some.value", new TypeCapture<ConfigContainer<String>>() {});
Assertions.assertEquals("value1", configContainer.orElseThrow());
// Change the values in the config map
configs.put("some.value", "value2");
// let gestalt know the values have changed so it can update the config tree.
manualReload.reload();
// The config container is automatically updated.
Assertions.assertEquals("value2", configContainer.orElseThrow());
By default, we expect all paths to be dot notation. So in a properties file dot notation would look like db.uri=my-sql.dev.myCompany.com
and produce a config tree with a map node db
that has a map node uri
with a value node my-sql.dev.myCompany.com
.
For A properties file with db.pool-size=10
the db
path would translate into a map node, and the pool-size
would also be a map node with a value node of 10
. pool-size
would not be translated into a path pool
and size
, but during decoding the path mappers will attempt to map the variable poolSize
to pool-size
.
However, this only works for the nodes after the path we are looking for. It does not map nodes earlier in the path.
So connection-pool.size=10
and connection.pool.size=50
are two separate paths and will not be merged.
By modifying the delimiter in the default lexer, you can support converting snake, kebab and dot notation into similar paths. Where the lexer will split the path into tokens based on any of the ([._-])|(?<=[a-z])(?=[A-Z])|(?<=[A-Z])(?=[A-Z][a-z])|(?<=[0-9])(?=[A-Z][a-z])|(?<=[a-zA-Z])(?=[0-9])
. Or split them based on CamelCase, Snake Case, Dot Notation, or Kebab Case.
Map<String, String> configs = Map.of("db.uri", "test");
String json = "{\"db_port\":3306}";
String toml = "db-password = \"abc123\"";
SentenceLexer relaxedLexer = PathLexerBuilder.builder()
.setDelimiter("([._-])|(?<=[a-z])(?=[A-Z])|(?<=[A-Z])(?=[A-Z][a-z])|(?<=[0-9])(?=[A-Z][a-z])|(?<=[a-zA-Z])(?=[0-9])")
.build();
Gestalt gestalt = new GestaltBuilder()
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(configs).build())
.addSource(StringConfigSourceBuilder.builder().setConfig(json).setFormat("json").build())
.addSource(StringConfigSourceBuilder.builder().setConfig(toml).setFormat("toml").build())
.useCacheDecorator(false)
// do not normalize the sentence return it as is.
.setSentenceLexer(relaxedLexer)
.build();
The Environment Variable Loader doesn't use the default lexer as it supports Screaming Snake Case by default, whereas the default lexer for the rest of Gestalt is dot notation.
To modify the Environment variable's lexer you need to register a EnvironmentVarsLoaderModuleConfig
with the new lexer.
Gestalt gestalt = new GestaltBuilder()
.addModuleConfig(EnvironmentVarsLoaderModuleConfigBuilder
.builder()
.setLexer(relaxedLexer)
.build())
Create an A/B that gives a different value for a feature flag or other value each time we retrieve the value, and overrides per group for testing or segmentation.
By leveraging runtime string replacement evaluation in the form #{}
along with the dist100
you can configure the distribution of certain results and ensure that each call it is re-evaluated. Then use Tags with your groups to control the results for specific groups.
Map<String, String> customMap = new HashMap<>();
// we want 10% of traffic to be true for the default case
customMap.put("newFeature.enabled", "#{dist100:10:true,false}");
// for a user in test group 1 we want the feature to be enabled.
Map<String, String> customTest1Map = new HashMap<>();
customTest1Map.put("newFeature.enabled", "true");
// for a user in test group 2 we want the feature to be disabled.
Map<String, String> customTest2Map = new HashMap<>();
customTest2Map.put("newFeature.enabled", "false");
GestaltBuilder builder = new GestaltBuilder();
Gestalt gestalt = builder
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(customMap).build())
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(customTest1Map).setTags(Tags.of("group", "test1")).build())
.addSource(MapConfigSourceBuilder.builder().setCustomConfig(customTest2Map).setTags(Tags.of("group", "test2")).build())
.build();
gestalt.loadConfigs();
int enabled = 0;
int disabled = 0;
int passes = 100;
while (passes-- > 0) {
Boolean featEnabled = gestalt.getConfig("newFeature.enabled", TypeCapture.of(Boolean.class));
if (featEnabled) {
enabled++;
} else {
disabled++;
}
Boolean enabledTest1 = gestalt.getConfig("newFeature.enabled", TypeCapture.of(Boolean.class),Tags.of("group", "test1"));
Assertions.assertTrue(enabledTest1);
Boolean enabledTest2 = gestalt.getConfig("newFeature.enabled", TypeCapture.of(Boolean.class),Tags.of("group", "test2"));
Assertions.assertFalse(enabledTest2);
}
// we expect that we should get less than 15 % enabled
Assertions.assertTrue(enabled < 15);
// we expect that we should get more than 85 % enabled
Assertions.assertTrue(disabled > 85);
}
For more examples of how to use gestalt see the gestalt-sample or for Java 17 + samples gestalt-sample-java-latest
Gestalt tries its best to maintain backwards compatibility with external API's such as the builders, sources, decoders, and Gestalt itself. When changes are made, they are deprecated for several releases to give people a chance to migrate. For internal APIs backwards compatibility is not always guaranteed. Internal APIs are considered ConfigNodeProcessor, ResultProcessor, ConfigValidator, Transformer, PathMappers, ConfigLoaders and such that end users will rarely need to modify. Although they are exposed and meant to be used, it is rare that end users will need to modify them. For those advanced users, they may have extra burden of updating with releases.
This section is more for those wishing to know more about how Gestalt works, or how to add their own functionality. If you only wish to get configuration from Gestalt As Is, then feel free to skip it.
A config source provides an interface for providing a configuration that a ConfigLoader can understand and parse into a config node. Each ConfigSource has a format that a specific ConfigLoader will understand. So a ConfigLoader that loads "property" files can load them from multiple sources. A ConfigSource can provide either a InputStream or list of pairs of paths and values. You can write your own ConfigSource by implementing the interface and passing though the format that represents your source. For example, you could add a new URL ConfigSource that loads from a URL, depending on the file extension, has a different format. Each source must have a unique ID, that Gestalt uses to keep track of the source, the config node tree built from the source and when reloading the id of the source.
/**
* The format of the config source, for example this can be envVars, the extension of a file (properties, json, ect).
*
* @return The format of the config source
*/
String format();
/**
* If this config source has a stream, this will return the stream of data.
* Or if not supported it will throw an exception.
*
* @return input stream of data
* @throws GestaltException if there are any IO or if this is an unsupported operation
*/
InputStream loadStream() throws GestaltException;
/**
* provides a list of config values.
*
* @return provides a list of config values
* @throws GestaltException if there are any IO or if this is an unsupported operation
*/
List<Pair<String, String>> loadList() throws GestaltException;
A ConfigLoader accepts a specific source format. It reads in the config source as either a list or input stream. It is then responsible for converting the sources into a GResultOf with either a config node tree or validation errors. You can write your own ConfigLoader by implementing the interface and accepting a specific format. Then read in the provided ConfigSource InputStream or list and parse the values. For example you can add a json loader that takes an InputStream and uses Jackson to load and build a config tree.
/**
* True if the config loader accepts the format.
*
* @param format config format.
* @return True if the config loader accepts the format.
*/
boolean accepts(String format);
/**
* Load a ConfigSource then build the validated config node.
*
* @param source source we want to load with this config loader.
* @return the validated config node.
* @throws GestaltException any exceptions
*/
GResultOf<ConfigNode> loadSource(ConfigSource source) throws GestaltException;
Gestalt uses a SentenceLexer's in several places, to convert a string path into tokens that can be followed and to in the ConfigParser to turn the configuration paths into tokens then into config nodes. You can customize the SentenceLexer to use your own format of path. For example in Gestalt Environment Variables use a '_' to delimitate the tokens whereas property files use '.'. If you wanted to use camel case you could build a sentence lexer for that.
Decoders allow Gestalt to decode a config node into a specific value, class or collection. A Decoder can either work on a leaf and decode a single value, or it can work on a Map or Array node and decode a class or collection. You can create your own decoder by implementing the Decoder interface. By returning true for the matches Gestalt will ask your decoder to decode the current node by calling your Decoders decode method. Gestalt will pass in the current path, the current node to decode and the DecoderService so we can decode any subnodes.
/**
* true if this decoder matches the type capture.
*
* @param path the current path
* @param tags the tags for the current request
* @param node the current node we are decoding.
* @param type the type of object we are decoding.
* @return true if this decoder matches the type capture
*/
boolean canDecode(path: String, tags: Tags, configNode:ConfigNode, TypeCapture<?> klass);
/**
* Decode the current node. If the current node is a class or list we may need to decode sub nodes.
*
* @param path the current path
* @param node the current node we are decoding.
* @param type the type of object we are decoding.
* @param decoderService decoder Service used to decode members if needed. Such as class fields.
* @return GResultOf the current node with details of either success or failures.
*/
GResultOf<T> decode(String path, ConfigNode node, TypeCapture<?> type, DecoderService decoderService);
You are able to reload a single source and rebuild the config tree by implementing your own ConfigReloadStrategy.
The ConfigNodeService is the central storage for the merged config node tree along with holding the original config nodes stored in a ConfigNodeContainer with the original source id. This is so when we reload a config source, we can link the source being reloaded with the config tree it produces. Gestalt uses the ConfigNodeService to save, merge, result the config tree, navigate and find the node Gestalt is looking for.
The external facing portion of Java Config Library, it is the keystone of the system and is responsible for bringing together all the pieces of project. Since Gestalt relies on multiple services, the Builder makes it simple to construct a functional and default version of Gestalt.
The built Gestalt is used to load the config sources by adding them to the builder and then passed through to the Gestalt constructor. Gestalt will use the ConfigLoaderService to find a ConfigLoader that will load the source by a format. It will add the config node tree loaded to the ConfigNodeService to be added with the rest of the config trees. The new config tree will be merged and where applicable overwrite any of the existing config nodes.
When a source needs to be reloaded, it will be passed into the reload function. The sources will then be converted into a Config node as in the loading. Then Gestalt will use the ConfigNodeService to reload the source. Since the ConfigNodeService holds onto the source ID with the ConfigNodeContainer we are able to determine with config node to reload then take all the config nodes and re-merge them in the same order to rebuild the config tree with the newly loaded node.
To implement your own Config Node Processor you need to inherit from ConfigNodeProcessor.
/**
* Interface for the Config Node Processing. This will be run against every node in the tree.
*
* @author <a href="https://github.com/gestalt-config/gestalt/blob/main/mailto:colin.redmond@outlook.com"> Colin Redmond </a> (c) 2024.
*/
public interface ConfigNodeProcessor {
/**
* run the config node process the current node. You need to return a node, so if your config node processor does nothing to the node
* return the original node.
*
* @param path current path
* @param currentNode current node to process.
* @return the node after running through the processor.
*/
GResultOf<ConfigNode> process(String path, ConfigNode currentNode);
/**
* Apply the ConfigNodeProcessorConfig to the config node Processor. Needed when building via the ServiceLoader
* It is a default method as most Config Node Processor don't need to apply configs.
*
* @param config GestaltConfig to update the Processor
*/
default void applyConfig(ConfigNodeProcessorConfig config) {
}
}
When you write your own applyConfig method, each node of the config tree will be passed into the process method. You can either modify the current node or return it as is. The return value will be used to replace the tree, so if you return nothing your tree will be lost.
You can re-write any intermediate node or only modify the leaf nodes as TransformerConfigNodeProcessor
does.
To register your own default ConfigNodeProcessor
, add it to a file in META-INF\services\org.github.gestalt.config.processor.config.ConfigNodeProcessor
and add the full path to your ConfigNodeProcessor
.
The TransformerConfigNodeProcessor
is a specific type of ConfigNodeProcessor
that allows you to replace strings in a leaf node that match ${transformer:key}
into a config value. where the transformer is the name of a Transformer registered with the TransformerConfigNodeProcessor, such as in the above ConfigNodeProcessor section with envMap, sys, and map. The key is a string lookup into the transformer.
To implement your own Transformer you need to implement the Transformer class.
/**
* Allows you to add your own custom source for the TransformerConfigNodeProcessor.
* Whenever the TransformerConfigNodeProcessor sees a value ${name:key} the transform is selected that matches the same name
*/
public interface Transformer {
/**
* the name that will match the ${name:key} the transform is selected that matches the same name
* @return
*/
String name();
/**
* When a match is found for ${name:key} the key and the path are passed into the process method.
* The returned value replaces the whole ${name:key}
* @param path the current path
* @param key the key to lookup int this transform.
* @return the value to replace the ${name:key}
*/
Optional<String> process(String path, String key);
}
To register your own default Transformer, add it to a file in META-INF\services\org.github.gestalt.config.processor.config.transform.Transformer
and add the full path to your Transformer.
the annotation @ConfigPriority(100)
, specifies the descending priority order to check your transformer when a substitution has been made without specifying the source ${key}
Result Processors are used to modify the result of getting a configuration and decoding it.
Each processor has an annotation @ConfigPriority
so we run them in order passing the output of one Result Processor as the input to the next.
Gestalt has two core result processors ErrorResultProcessor
and DefaultResultProcessor
.
The ErrorResultProcessor
throws a GestaltException
if there is an unrecoverable error.
The DefaultResultProcessor
will convert the result into a default value if there is no result.
To implement your own Result Processors you need to inherit from ResultProcessor.
To automatically register your own default ResultProcessor
, add it to a file in META-INF\services\org.github.gestalt.config.processor.result.ResultProcessor
and add the full package of classpath your ResultProcessor
.
Alternatively, you can implement the interface and register it with the gestalt builder addResultProcessors(List<ResultProcessor> resultProcessorSet)
.
public interface ResultProcessor {
/**
* If your Result Processor needs access to the Gestalt Config.
*
* @param config Gestalt configuration
*/
default void applyConfig(GestaltConfig config) {}
/**
* Returns the {@link GResultOf} with any processed results.
* You can modify the results, errors or any combination.
* If your post processor does nothing to the node, return the original node.
*
* @param results GResultOf to process.
* @param path path the object was located at
* @param isOptional if the result is optional (an Optional or has a default.
* @param defaultVal value to return in the event of failure.
* @param klass the type of object.
* @param tags any tags used to retrieve te object
* @return The validation results with either errors or a successful obj.
* @param <T> Class of the object.
* @throws GestaltException for any exceptions while processing the results, such as if there are errors in the result.
*/
<T> GResultOf<T> processResults(GResultOf<T> results, String path, boolean isOptional,
T defaultVal, TypeCapture<T> klass, Tags tags)
throws GestaltException;
}
Validations are implemented as a Result Processor as well in ValidationResultProcessor
.
To ensure a simple API for validations it does not use the ResultProcessor interface but a ConfigValidator
interface.
To automatically register your own default ConfigValidator
, add it to a file in META-INF\services\org.github.gestalt.config.processor.result.validation.ConfigValidator
and add the full package of classpath ConfigValidator
. This is how gestalt-validator-hibernate
automatically is discovered.
Alternatively, you can implement the interface and register it with the gestalt builder addValidators(List<ConfigValidator> validatorsSet)
.
/**
* Interface for validating objects.
*
* @author <a href="https://github.com/gestalt-config/gestalt/blob/main/mailto:colin.redmond@outlook.com"> Colin Redmond </a> (c) 2024.
*/
public interface ConfigValidator {
/**
* If your Validator needs access to the Gestalt Config.
*
* @param config Gestalt configuration
*/
default void applyConfig(GestaltConfig config) {}
/**
* Returns the {@link GResultOf} with the validation results. If the object is ok it will return the result with no errors.
* If there are validation errors they will be returned.
*
* @param obj object to validate.
* @param path path the object was located at
* @param klass the type of object.
* @param tags any tags used to retrieve te object
* @return The validation results with either errors or a successful obj.
* @param <T> Class of the object.
*/
<T> GResultOf<T> validator(T obj, String path, TypeCapture<T> klass, Tags tags);
}