GR8Conf 2009: Practical Groovy DSL by Guillaume Laforge
Post on 19-Jan-2015
2067 Views
Preview:
DESCRIPTION
Transcript
Practical Domain-Specific Languages with Groovy
All the techniques to create your own DSLs
Guillaume Laforge
Head of Groovy Development
• Groovy Project Manager
• JSR-241 Spec Lead
• Head of Groovy Developmentat SpringSource
• Initiator of the Grails framework
• Co-author of Groovy in Action
• Speaker: JavaOne, QCon, JavaZone, Sun TechDays, Devoxx, The Spring Experience, JAX, Dynamic Language World, IJTC, and more...
Guillaume Laforge
3
A few words about Groovy
• Groovy is a dynamic language for the JVM
–with a Meta Object Protocol
–compiles directly to bytecode, seamless Java interop
• Open Source ASL 2 project hosted at Codehaus
• Relaxed grammar derived from Java 5
–+ borrowed good ideas from Ruby, Python, Smalltalk
• Fast... for a dynlang on the JVM
• Closures, properties, optional typing, BigDecimal by default, nice wrapper APIs, and more...
4
• The context and the usual issues we face
• Some real-life examples of Domain-Specific Languages
• Groovy’s DSL capabilities
• Integrating a DSL in your application
• Considerations to remember when designing your own DSL
Agenda
5
The context
Subject Matter Experts,Business analysts...
HAI
CAN HAS STDIO?
I HAS A VAR
IM IN YR LOOP
UP VAR!!1
VISIBLE VAR
IZ VAR BIGGER THAN 10?
KTHXBYE
IM OUTTA YR LOOP
KTHXBYE
Developer producing LOLCODE
Lots of languages...
And in the end......nobody understands each other
11
Expressing requirements...
DSL: a potential solution?
•Use a more expressive language than a general purpose one
•Share a common metaphore of understanding between developers and subject matter experts
•Have domain experts help with the design of the business logic of an application
•Avoid cluttering business code with too much boilerplate technical code
•Cleanly separate business logic from application code
•Let business rules have their own lifecycle
12
Towards more readability (1)
13
Towards more readability (1)
13
Towards more readability (1)
20%
13
Towards more readability (2)
14
Towards more readability (2)
14
Towards more readability (2)
80%
14
• The context and the usual issues we face
• Some real-life examples of Domain-Specific Languages
• Groovy’s DSL capabilities
• Integrating a DSL in your application
• Considerations to remember when designing your own DSL
Agenda
15
• In our everyday life, we’re surrounded by DSLs
–Technical dialects
–Notations
–Business languages
16
A collection of DSLs
17
Technical dialects
SQL
19
^[\w-\.]+@([\w-]){2,4}$
20
Notations
1. e4 e52. Nf3 Nc63. Bb5 a6
L2 U F-1 B L2 F B -1 U L2
Visual!
24
Business languages
Real-life Groovy examples
• Anti-malaria drug resistance simulation
• Human Resources employee skills representation
• Insurance policies risk calculation engine
• Loan acceptance rules engine for a financial platform
• Mathematica-like lingua for nuclear safety simulations
• Market data feeds evolution scenarios
• and more...
25
• The context and the usual issues we face
• Some real-life examples of Domain-Specific Languages
• Groovy’s DSL capabilities
• Integrating a DSL in your application
• Considerations to remember when designing your own DSL
Agenda
26
A flexible & malleable syntax
• No need to write full-blown classes, use scripts
• Optional typing (def)
–in scripts, you can even omit the def keyword
• Native syntax constructs
• Parentheses & semi-colons are optional
• Named arguments
• BigDecimal by default for decimal numbers
• Closures for custom control structures
• Operator overloading
27
Scripts vs classes
• Hide all the boilerplate technical code
–an end-user doesn’t need to know about classes
–public class Rule { public static void main(String[] args) { System.out.println(“Hello”); }}
–println “Hello”
28
Optional typing
• No need to bother with types or even generics
–unless you want to!
• Imagine an interest rate lookup table method returning some generified type:
–Rate<LoanType, Duration, BigDecimal>[] lookupTable() { ... }def table = lookupTable()
• No need to repeat the horrible generics type info!
29
Native syntax constructs
• Lists
–[Monday, Tuesday, Wednesday]
• Maps
–[CA: ‘California’, TX: ‘Texas’]
• Ranges
–def bizDays = Monday..Friday
–def allowedAge = 18..65
–You can create your own custom ranges
30
Optional parens & semis
• Make statements and expressions look more like natural languages
–move(left);
–move left
31
Named arguments
• In Groovy you can mix named and unnamed arguments for method parameters
–named params are actually put in a map parameter
–plus optional parens & semis
• take 1.pill, of: Chloroquinine, after: 6.hours
• Corresponds to a method signature like:
–def take(Map m, MedicineQuantity mq)
32
BigDecimal by default
• Main reason why financial institutions often decide to use Groovy for their business rules!
–Although these days rounding issues are overrated!
• Java vs Groovy for a simple interpolation equation
• BigDecimal uMinusv = c.subtract(a); BigDecimal vMinusl = b.subtract(c); BigDecimal uMinusl = a.subtract(b); return e.multiply(uMinusv) .add(d.multiply(vMinusl)) .divide(uMinusl, 10, BigDecimal.ROUND_HALF_UP);
• (d * (b - c) + e * (c - a)) / (a - b)
33
• When closures are last, they can be put “out” of the parentheses surrounding parameters
•unless (account.balance > 100.euros, { account.debit 100.euros })
•unless (account.balance > 100.euros) { account.debit 100.euros}
• Signature def unless(boolean b, Closure c)
34
Custom control structuresThanks to closures
Operator overloading
•Currency amounts–15.euros + 10.dollars
•Distance handling–10.kilometers - 10.meters
•Workflow, concurrency
–taskA | taskB & taskC
•Credit an account–account << 10.dollarsaccount += 10.dollarsaccount.credit 10.dollars
a + b a.plus(b)
a - b a.minus(b)
a * b a.multiply(b)
a / b a.divide(b)
a % b a.modulo(b)
a ** b a.power(b)
a | b a.or(b)
a & b a.and(b)
a ^ b a.xor(b)
a[b] a.getAt(b)
a << b a.leftShift(b)
a >> b a.rightShift(b)
+a a.positive()
-a a.negative()
~a a.bitwiseNegate()
35
Groovy’s dynamic heart:
The MOP!MetaObject Protocol
Groovy’s MOP
• All the accesses to methods, properties, constructors, operators, etc. can be intercepted thanks to the MOP
• While Java’s behavior is hard-wired at compile-time in the class
• Groovy’s runtime behavior is adaptable at runtime through the metaclass.
• Different hooks for changing the runtime behavior
–GroovyObject, custom MetaClass implementation, categories, ExpandoMetaClass
37
GroovyObject
• All instances of classes created in Groovy implement the GroovyObject interface:
–getProperty(String name)
–setProperty(String name, Object value)
–invokeMethod(String name, Object[] params)
–getMetaClass()
–setMetaClass(MetaClass mc)
• A GO can have “pretended” methods and properties
38
MetaClass
• The core of Groovy’s MOP system
–invokeConstructor()
–invokeMethod() and invokeStaticMethod()
–invokeMissingMethod()
–getProperty() and setProperty()
–getAttribute() and setAttribute()
–respondsTo() and hasProperty()
• MetaClasses can change the behavior of existing third-party classes — even from the JDK
39
ExpandoMetaClass
• A DSL for MetaClasses!
• MoneyAmount.metaClass.constructor = { ... }Number.metaClass.getDollars = { ... }Distance.metaClass.toMeters = { ... }Distance.metaClass.static.create = { ... }
• To avoid repetition of Type.metaClass, you can pass a closure to metaClass { ... }
• The delegate variable in closure represents the
current instance, and it the default parameter
40
The Builder pattern
The Groovy MarkupBuilder
•def mkp = new MarkupBuilder()mkp.html { head { title “Groovy in Action” } body { div(width: ‘100’) { p(class: ‘para) { span “Best book ever!” } } }}
42
A builder for HR
• softskills { ideas { capture 2 formulate 3 } ...}knowhow { languages { java 4 groovy 5 } ...}
43
A builder for HR
• softskills { ideas { capture 2 formulate 3 } ...}knowhow { languages { java 4 groovy 5 } ...}
43
Builders
• Builders are...
–a mechanism for creating any tree-structered graph
–the realization of the GoF builder pattern at the syntax level in Groovy
–simply a clever use of chained method invocation, closures, parentheses omission, and use of the GroovyObject methods
• Existing builders
–XML, Object graph, Swing, Ant, JMX, and more...
44
The clever trick
• GroovyObject#invokeMethod() is used to catch all non-existing method calls in the context of the builder
• The nesting of closures visually shows the level of nesting / depth in the tree
• builder.m1(attr1:1, attr2:2, { builder.m2(...,
{...}) } becomes equivalent to
builder.m1(attr1:1, attr2:2) { m2(...) {...} }
thanks to parens omission
45
Adding properties to numbers
• Three possible approaches
–create a Category
•a category is a kind of decorator for default MCs
–create a custom MetaClass
•a full-blown MC class to implement and to set on the
POGO instance
–use ExpandoMetaClass
•friendlier DSL approach but with a catch
46
With a Category
• class DistanceCategory { static Distance getMeters(Integer self) { new Distance(self, Unit.METERS) }}
use(DistanceCategory) { 100.meters}
• Interesting scope: thread-bound & lexical
• But doesn’t work across the hierarchy of classes
–ie. subclasses won’t benefit from the new property
47
With an ExpandoMetaClass
• Number.metaClass.getMeters = {-> new Distance(delegate, Unit.METERS) }
100.meters
• Works for the class hierarchy for POJOs, and a flag exists to make it work for POGOs too
• But the catch is it’s really a global change, so beware EMC enhancements collisions
48
Compile-time metaprogramming
• Groovy 1.6 introduced AST Transformations
• Compile-time == No runtime performance penalty!
Transformation
49
AST Transformations
• Two kinds of transformations
–Global transformations
•applicable to all compilation units
–Local transformations
•applicable to marked program elements
•using specific marker annotations
50
Example #1: @Singleton
• Let’s revisit this evil (anti-)pattern! public class Evil { public static final Evil instance = new Evil (); private Evil () {} Evil getInstance() { return instance; } }
• In Groovy! @Singleton class Evil {}
• Also a “lazy” version! @Singleton(lazy = true) class Evil {}
51
• You can delegate to fields of your classes
–class Employee { def doTheWork() { “done” }}class Manager { @Delegate Employee slave = new Employee()}def god = new Manager()assert god.doTheWork() == “done”
• Damn manager who will get all the praise...
52
Example #2: @DelegateNot just for managers!
Global transformations
• Implement ASTTransformation
• Annotate the transfo specifying a compilation phase
• @GroovyASTTransformation(phase=CompilePhase.CONVERSION)public class MyTransformation implements ASTTransformation { public void visit(ASTNode[] nodes, SourceUnit unit) { ... }}
• For discovery, create the file META-INF/services/org.codehaus.groovy.transform.ASTTransformation
• Add the fully qualified name of the class in that file
53
Local transformations
• Same approach as Globale transformations
• But you don’t need the META-INF file
• Instead create an annotation to specify on which element the transformation should apply
• @Retention(RetentionPolicy.SOURCE)@Target([ElementType.METHOD])@GroovyASTTransformationClass( ["fqn.MyTransformation"])public @interface WithLogging {...}
54
Example: the Spock framework
• Changing the semantics of the original code
• But keeping a valid Groovy syntax
• @Speckclass HelloSpock { def "can you figure out what I'm up to?"() { expect: name.size() == size
where: name << ["Kirk", "Spock", "Scotty"] size << [4, 5, 6] }}
55
• The context and the usual issues we face
• Some real-life examples of Domain-Specific Languages
• Groovy’s DSL capabilities
• Integrating a DSL in your application
• Considerations to remember when designing your own DSL
Agenda
56
• Java 6’s javax.script.* APIs (aka JSR-223)
• Spring’s language namespace
• Groovy’s own mechanisms
• But a key idea is to externalize those DSL programs
–DSL programs can have their own lifecycle
–no need to redeploy an application because of a rule change
–business people won’t see the technical code
57
Various integration mechanisms
Java 6’s javax.script.* API
• Groovy 1.6 provides its own implementation of the javax.script.* API
• ScriptEngineManager mgr = new ScriptEngineManager();ScriptEngine engine = mgr.getEngineByName(“Groovy”);
String result = (String)engine.eval(“2+3”);
58
Spring’s lang namespace
• POGOs (Plain Old Groovy Objects) can be pre-compiled as any POJO and used interchangeably with POJOs in a Spring application
• But Groovy scripts & classes can be loaded at runtime through the <lang:groovy/> namespace and tag
• Reloadable on change
• Customizable through a custom MetaClass
• <lang:groovy id="events" script-source="classpath:dsl/eventsChart.groovy" customizer-ref="eventsMetaClass" />
59
Groovy’s own mechanisms
• Eval
–for evaluating simple expressions
• GroovyShell
–for more complex scripts and DSLs
• GroovyClassLoader
–the most powerful mechanism
60
Eval
• Simple mechanism to evaluate math-like formulas
• Eval.me ( ‘3*4’)Eval.x (1, ‘3*x + 4’)Eval.xy (1, 2, ‘x + y’)Eval.xyz(1, 2, 3, ‘x * y - z’)
61
• A Binding provides a context of execution
–can implement lazy evaluation if needed
• A base script class can be specified
• def binding = new Binding()binding.mass = 22.3binding.velocity = 10.6def shell = new GroovyShell(binding)shell.evaluate(“mass * velocity ** 2 / 2”)
62
GroovyShell
GroovyClassLoader
• Most powerful mechanism
–could also visit or change the AST
–scripts & classes can be loaded from elsewhere
–more control on compilation
• GroovyClassLoader gcl = new GroovyClassLoader();Class clazz = gcl.parseClass( new File(“f.groovy”));GroovyObject instance = (GroovyObject)clazz.newInstance();instance.setMetaClass(customMC);
63
Externalize business rules
• Although Groovy DSLs can be embedded in normal Groovy classes, you should externalize them
• Store them elsewhere
–in a database, an XML file, etc.
• Benefits
–Business rules are not entangled in technical application code
–Business rules can have their own lifecycle, without requiring application redeployments
64
• The context and the usual issues we face
• Some real-life examples of Domain-Specific Languages
• Groovy’s DSL capabilities
• Integrating a DSL in your application
• Considerations to remember when designing your own DSL
Agenda
65
Start small, with key concepts
Beware overengineering!
Grow your language progressively
Get your hands dirty
Play with the end-users
Let your DSL fly, it’s not yours, it’s theirs!
Tight feedback loop
Iterative process
Stay humble.
You can’t get it right the first time.
Don’t design alone at your deskInvolve the end users from the start
Playing it safein a sandbox
Various levels of sandboxing
• Groovy supports the usual Java Security Managers
• Use metaprogramming tricks to prevent calling / instantiating certain classes
• Create a special GroovyClassLoader AST code visitor to filter only the nodes of the AST you want to keep
–ArithmeticShell in Groovy’s samples
73
Test, test, test!
• Don’t just test for nominal cases
–Explicitly test for errors!
• Ensure end-users get meaningful error messages
74
• Summary
• Questions & Answers
Agenda
75
Summary
• Groovy’s a great fit for Domain-Specific Languages
–Malleable & flexible syntax
–Full object-orientation
• Metaprogramming capabilities
–Runtime metaprogramming
–Compile-time metaprogramming
• Groovy’s very often used for mission-critical DSLs
76
I kan haz my cheezburgr naw?Or do ya reely haz keshtionz?
?
Appendix
78
• http://www.flickr.com/photos/wheatfields/420088151/sizes/l/
• http://www.flickr.com/photos/therefromhere/518053737/sizes/l/
• http://www.flickr.com/photos/romainguy/230416692/sizes/l/
• http://www.flickr.com/photos/addictive_picasso/2874279971/sizes/l/
• http://www.flickr.com/photos/huangjiahui/3127634297/sizes/l/
• http://www.flickr.com/photos/25831000@N08/3064515804/sizes/o/
• http://www.flickr.com/photos/lanier67/3147696168/sizes/l/
• http://www.flickr.com/photos/ktb/4916063/sizes/o/
• http://www.flickr.com/photos/nathonline/918128338/sizes/l/
• http://www.flickr.com/photos/kevinsteele/39300193/sizes/l/
• http://commons.wikimedia.org/wiki/File:Brueghel-tower-of-babel.jpg
• http://commons.wikimedia.org/wiki/File:Platypus.jpg
• http://www.flickr.com/photos/joaomoura/2317171808/sizes/l/
• http://www.flickr.com/photos/wiccked/132687067/
• http://www.flickr.com/photos/xcbiker/386876546/sizes/l/
• http://www.flickr.com/photos/pietel/152403711/sizes/o/
79
• http://www.flickr.com/photos/forezt/192554677/sizes/o/
• http://keremkosaner.files.wordpress.com/2008/04/
softwaredevelopment.gif
• http://www.jouy.inra.fr
• http://www.flickr.com/photos/ejpphoto/408101818/sizes/o/
• http://www.flickr.com/photos/solaro/2127576608/sizes/l/
• http://www.flickr.com/photos/biggreymare/2846899405/sizes/l/
• http://www.flickr.com/photos/timsamoff/252370986/sizes/l/
• http://www.flickr.com/photos/29738009@N08/2975466425/sizes/l/
• http://www.flickr.com/photos/howie_berlin/180121635/sizes/o/
• http://www.flickr.com/photos/yogi/1281980605/sizes/l/
• http://www.flickr.com/photos/dorseygraphics/1336468896/sizes/l/
80
top related