Terence Parr
University of San Francisco
parrt AT cs.usfca.edu
Copyright 2003-2005
http://www.stringtemplate.org (BSD license)

C# version created by Kunle Odutola
kunle UNDERSCORE odutola AT hotmail.com
Copyright 2005-2006
(ST# - C# StringTemplate released under BSD License)

Python version created by Marq Kole
marq DOT kole AT xs4all DOT nl
Copyright 2003-2006

Contents

• Introduction
  • Motivation And Philosophy
  • StringTemplate language flavor
• Defining Templates
  • Creating Templates With Code
  • Loading Templates From Files
  • Loading Templates relative to an implementation specific location
  • Caching
• Group Files
  • Supergroups and interfaces
  • Maps
  • Group file format
  • Group loaders
  • Formal argument default values
  • Formal argument error handling
  • Newline handling
• Group interfaces
• Expressions
  • Attribute References
  • Template References
  • Attribute operators
  • Template Application
  • Conditionally Included Subtemplates (IF statements)
• Functionality Summary
• Object Rendering
• Template And Attribute Lookup Rules
  • Template lookup
  • Attribute scoping rules
• Setting the Expression Delimiters
• Template inheritance
• Template regions
• Auto-indentation
• Automatic line wrapping
• Output Filters
• StringTemplate Grammars
• Debugging
• Acknowledgements

Related material

• DRAFT A Functional Language For Generating Structured Text
• Internationalization and Localization of Web Applications In Action
• Language Translation Using ANTLR and StringTemplate
• Intelligent Web Site Page Generation
• The Role of Template Engines in Code Generation

It is highly recommended that you read the (academically-oriented) paper, Enforcing Model-View Separation in Template Engines.

The StringTemplates distribution includes many unit tests that also represent a useful set of examples. The tests are defined in:

Please see the 3.0 Release Notes and changes and bugs page. That page generally discusses the Java version of StringTemplate but, some of the information it contains might apply to other implementations.

Introduction

Most programs that emit source code or other text output are unstructured blobs of generation logic interspersed with print statements. The primary reason is the lack of suitable tools and formalisms. The proper formalism is that of an output grammar because you are not generating random characters--you are generating sentences in an output language. This is analogous to using a grammar to describe the structure of input sentences. Rather than building a parser by hand, most programmers will use a parser generator. Similarly, we need some form of unparser generator to generate text. The most convenient manifestation of the output grammar is a template engine such as StringTemplate.

A template engine is a simply a code generator that emits text using templates, which are really just "documents with holes" in them where you can stick values. StringTemplate breaks up your template into chunks of text and attribute expressions, which are by default enclosed in dollar signs $attribute-expression$ (to make them easy to see in HTML files). StringTemplate ignores everything outside of attribute expressions, treating it as just text to spit out when you call:

For example, the following template has two chunks, a literal and a reference to attribute name:

Hello, $name$

Using templates in code is very easy. Here is the requisite example that prints "Hello, World":

import org.antlr.stringtemplate.*;

StringTemplate hello = new StringTemplate("Hello, $name$");
hello.setAttribute("name", "World");
System.out.println(hello.toString());

using Antlr.StringTemplate;

StringTemplate hello = new StringTemplate("Hello, $name$");
hello.SetAttribute("name", "World");
Console.Out.WriteLine(hello.ToString());

import stringtemplate3

hello = stringtemplate3.StringTemplate("Hello, $name$")
hello["name"] = "World"
print str(hello)

StringTemplate is not a "system" or "engine" or "server"; it is a library with two primary classes of interest: StringTemplate and StringTemplateGroup. You can directly create a StringTemplate in code, you can load a template from a file, and you can load a single file with many templates (a template group file).

Motivation And Philosophy

StringTemplate was born and evolved during the development of http://www.jGuru.com. The need for such dynamically-generated web pages has led to the development of numerous other template engines in an attempt to make web application development easier, improve flexibility, reduce maintenance costs, and allow parallel code and HTML development. These enticing benefits, which have driven the proliferation of template engines, derive entirely from a single principle: separating the specification of a page's business logic and data computations from the specification of how a page displays such information.

These template engines are in a sense are a reaction to the completely entangled specifications encouraged by JSP (Java Server Pages), ASP (Active Server Pages) and, even ASP.NET. With separate encapsulated specifications, template engines promote component reuse, pluggable site "looks", single-points-of-change for common components, and high overall system clarity. In the code generation realm, model-view separation guarantees retargetability.

The normal imperative programming language features like setting variables, loops, arithmetic expressions, arbitrary method calls into the model, etc... are not only unnecessary, but they are very specifically what is wrong with ASP/JSP. Recall that ASP/JSP (and ASP.NET) allow arbitrary code expressions and statements, allowing programmers to incorporate computations and logic in their templates. A quick scan of template engines reveals an unfortunate truth--all but a few are Turing-complete languages just like ASP/JSP/ASP.NET. One can argue that they are worse than ASP/JSP/ASP.NET because they use languages peculiar to that template engine. Many tool builders have clearly lost sight of the original problem we were all trying to solve. We programmers often get caught up in cool implementations, but we should focus on what should be built not what can be built.

The fact that StringTemplate does not allow such things as assignments (no side-effects) should make you suspicious of engines that do allow it. The templates in ANTLR v3's code generator are vastly more complicated than the sort of templates typically used in web pages creation with other template engines yet, there hasn't been a situation where assignments were needed. If your template looks like a program, it probably is--you have totally entangled your model and view.

After examining hundreds of template files that I created over years of jGuru.com (and now in ANTLR v3) development, I found that I needed only the following four basic canonical operations (with some variations):

• attribute reference; e.g., $phoneNumber$
• template reference (like #include or macro expansion); e.g., $searchbox()$
• conditional include of subtemplate (an IF statement); e.g., $if(title)$<title>$title$</title>$endif$
• template application to list of attributes; e.g., $names:bold()$

where template references can be recursive.

Language theory supports my premise that even a minimal StringTemplate engine with only these features is very powerful--such an engine can generate the context-free languages (see Enforcing Strict Model-View Separation in Template Engines); e.g., most programming languages are context-free as are any XML pages whose form can be expressed with a DTD.

While providing all sorts of dangerous features like assignment that promote the use of computations and logic in templates, many engines miss the key elements. Certain language semantics are absolutely required for generative programming and language translation. One is recursion. A template engine without recursion seems unlikely to be capable of generating recursive output structures such as nested tables or nested code blocks.

Another distinctive StringTemplate language feature lacking in other engines is lazy-evaluation. StringTemplate's attributes are lazily evaluated in the sense that referencing attribute "a" does not actually invoke the data lookup mechanism until the template is asked to render itself to text. Lazy evaluation is surprising useful in both the web and code generation worlds because such order decoupling allows code to set attributes when it is convenient or efficient not necessarily before a template that references those attributes is created. For example, a complicated web page may consist of many nested templates many of which reference $userName$, but the value of userName does not need to be set by the model until right before the entire page is rendered to text via ToString(). You can build up the complicated page, setting attribute values in any convenient order.

StringTemplate implements a "poor man's" form of lazy evaluation by simply requiring that all attributes be computed a priori. That is, all attributes must be computed and pushed into a template before it is written to text; this is the so-called "push method" whereas most template engines use the "pull method". The pull method appears more conventional because programmers mistakenly regard templates as programs, but pulling attributes introduces order-of-computation dependencies. Imagine a simple web page that displays a list of names (using some mythical Java-based template engine notation):

<html>
<body>
<ol>
$foreach n in names$
  <li>$n$</li>
$end$
</ol>
There are $numberNames$ names.
</body>
</html>

Using the pull method, the reference to names invokes model.getNames(), which presumably loads a list of names from the database. The reference to numberNames invokes model.getNumberNames() which necessarily uses the internal data structure computed by getNames() to compute names.size() or whatever. Now, suppose a designer moves the numberNames reference to the <title> tag, which is before the reference to names in the foreach statement. The names will not yet have been loaded, yielding a null pointer exception at worst or a blank title at best. You have to anticipate these dependencies and have getNumberNames() invoke getNames() because of a change in the template.

I'm stunned that other template engine authors with whom I've spoken think this is ok. Any time I can get the computer to do something automatically for me that removes an entire class of programming errors, I'll take it!. Automatic garbage collection is the obvious analogy here.

The pull method requires that programmers do a topological sort in their minds anticipating any order that a programmer or designer could induce. To ensure attribute computation safety (i.e., avoid hidden dependency landmines), I have shown trivially in my academic paper that pull reduces to push in the worst case. With a complicated mesh of templates, you will miss a dependency, thus, creating a really nasty, difficult-to-find bug.

StringTemplate mission

When developing StringTemplate, I recalled Frederick Brook's book, "Mythical Man Month", where he identified conceptual integrity as a crucial product ingredient. For example, in UNIX everything is a stream. My concept, if you will, is strict model-view separation. My mission statement is therefore:

"StringTemplate shall be as simple, consistent, and powerful as possible without sacrificing strict model-view separation."

I ruthlessly evaluate all potential features and functionality against this standard. Over the years, however, I have made certain concessions to practicality that one could consider as infringing ever-so-slightly into potential model-view entanglement. That said, StringTemplate still seems to enforce separation while providing excellent functionality.

I let my needs dictate the language and tool feature set. The tool evolved as my needs evolved. I have done almost no feature "backtracking". Further, I have worked really hard to make this little language self-consistent and consistent with existing syntax/metaphors from other languages. There are very few special cases and attribute/template scoping rules make a lot of sense even if they are unfamiliar or strange at first glance. Everything in the language exists to solve a very real need.

StringTemplate language flavor

Just so you know, I've never been a big fan of functional languages and I laughed really hard when I realized (while writing the academic paper) that I had implemented a functional language. The nature of the problem simply dictated a particular solution. We are generating sentences in an output language so we should use something akin to a grammar. Output grammars are inconvenient so tool builders created template engines. Restricted template engines that enforce the universally-agreed-upon goal of strict model-view separation also look remarkably like output grammars as I have shown. So, the very nature of the language generation problem dictates the solution: a template engine that is restricted to support a mutually-recursive set of templates with side-effect-free and order-independent attribute references.

Defining Templates

Creating Templates With Code

Here is a simple example that creates and uses a template on the fly:

StringTemplate query = new StringTemplate("SELECT $column$ FROM $table$;");
query.setAttribute("column", "name");
query.setAttribute("table", "User");

StringTemplate query = new StringTemplate("SELECT $column$ FROM $table$;");
query.SetAttribute("column", "name");
query.SetAttribute("table", "User");

query = stringtemplate3.StringTemplate("SELECT $column$ FROM $table$;")
query["column"] = "name"
query["table"] = "User"

where StringTemplate considers anything in $...$ to be something it needs to pay attention to. By setting attributes, you are "pushing" values into the template for use when the template is printed out. The attribute values are set by referencing their names. Invoking toString() on query would yield

SELECT name FROM User;

You can set an attribute multiple times, which simply means that the attribute is multi-valued. For example, adding another value to the attribute named column as shown below makes the attribute multi-valued:

StringTemplate query = new StringTemplate("SELECT $column$ FROM $table$;");
query.setAttribute("column", "name");
query.setAttribute("column", "email");
query.setAttribute("table", "User");

StringTemplate query = new StringTemplate("SELECT $column$ FROM $table$;");
query.SetAttribute("column", "name");
query.SetAttribute("column", "email");
query.SetAttribute("table", "User");

query = stringtemplate3.StringTemplate("SELECT $column$ FROM $table$;")
query["column"] = "name"
query["column"] = "email"
query["table"] = "User"

Invoking toString() on query would now yield

SELECT nameemail FROM User;

Ooops...there is no separator between the multiple values. If you want a comma, say, between the column names, then change the template to record that formatting information:

StringTemplate query = new StringTemplate("SELECT $column; separator=\",\"$ FROM $table$;");
query.setAttribute("column", "name");
query.setAttribute("column", "email");
query.setAttribute("table", "User");

StringTemplate query = new StringTemplate("SELECT $column; separator=\",\"$ FROM $table$;");
query.SetAttribute("column", "name");
query.SetAttribute("column", "email");
query.SetAttribute("table", "User");

query = stringtemplate3.StringTemplate("SELECT $column; separator=\",\"$ FROM $table$;")
query["column"] = "name"
query["column"] = "email"
query["table"] = "User"

Note that the right-hand-side of the separator specification in this case is a string literal; therefore, we have escaped the double-quotes as the template is specified in a string. In general, the right-hand-side can be any attribute expression. Invoking toString() on query would now yield

SELECT name,email FROM User;

Attributes can be any object at all. StringTemplate calls toString() on each object as it writes the template out. The separator is not used unless the attribute is multi-valued.

Loading Templates From Files

To load a template from the disk you must use a StringTemplateGroup that will manage all the templates you load, caching them so you do not waste time talking to the disk for each template fetch request (you can change it to not cache; see below). You may have multiple template groups. Here is a simple example that loads the previous SQL template from a file /tmp/theQuery.st:

SELECT $column; separator=","$ FROM $table$;

The code below creates a StringTemplateGroup called myGroup rooted at /tmp so that requests for template theQuery forces a load of file /tmp/theQuery.st.

StringTemplateGroup group = new StringTemplateGroup("myGroup", "/tmp");
StringTemplate query = group.getInstanceOf("theQuery");
query.setAttribute("column", "name");
query.setAttribute("column", "email");
query.setAttribute("table", "User");

StringTemplateGroup group = new StringTemplateGroup("myGroup", "/tmp");
StringTemplate query = group.GetInstanceOf("theQuery");
query.SetAttribute("column", "name");
query.SetAttribute("column", "email");
query.SetAttribute("table", "User");

group = stringtemplate3.StringTemplateGroup("myGroup", "/tmp")
query = group.getInstanceOf("theQuery")
query["column"] = "name"
query["column"] = "email"
query["table"] = "User"

If you have a directory hierarchy of templates such as file /tmp/jguru/bullet.st, you would reference them relative to the root; in this case, you would ask for template jguru/bullet().

Note StringTemplate strips whitespace from the front and back of all loaded template files. You can add, for example, <\n> at the end of the file to get an extra carriage return.

Loading Templates relative to an implementation specific location

// Look for templates in CLASSPATH as resources
StringTemplateGroup group = new StringTemplateGroup("mygroup");
StringTemplate st = group.getInstanceOf("com/mycompany/server/templates/page");

// Look for templates relative to assembly location
StringTemplateGroup group = new StringTemplateGroup("mygroup", (string)null);
StringTemplate st = group.GetInstanceOf("templates/page");

If page.st references, say, searchbox template, it must be fully qualified as:
<font size=2>SEARCH</font>: $com/mycompany/server/templates/page/searchbox()$

This is inconvenient and ST may add the invoking template's path prefix automatically in the future.

Caching

By default templates are loaded from disk just once. During development, however, it is convenient to turn caching off. Also, you may want to turn off caching so that you can quickly update a running site. You can set a simple refresh interval using StringTemplateGroup.setRefreshInterval(...). When the interval is reached, all templates are thrown out. Set interval to 0 to refresh constantly (no caching). Set the interval to a huge number like Integer.MAX_INT or Int32.MaxValue to have no refreshing at all.

StringTemplateGroup group = new StringTemplateGroup("myGroup", "/tmp");
group.setRefreshInterval(0);  // no caching
group.setRefreshInterval(Integer.MAX\_INT);  // no refreshing

StringTemplateGroup group = new StringTemplateGroup("myGroup", "/tmp");
group.setRefreshInterval(0);  // no caching
group.setRefreshInterval(Int32.MaxValue);  // no refreshing

group = stringtemplate3.StringTemplateGroup("myGroup", "/tmp")
group.refreshInterval = 0  # no caching
group.refreshInterval = sys.maxint # no refreshing

Group Files

StringTemplate 2.0 introduced the notion of a group file that has two main attractions. First, it allows you to define lots of small templates more conveniently because they may all be defined within a single file. Second, you may specify formal template arguments that help StringTemplate detect errors (such as setting unknown attributes) and make the templates easier to read. Here is a sample group file with two templates, vardef and method, that could be used to generate C files:

group simple;

vardef(type,name) ::= "<type> <name>;"

method(type,name,args) ::= <<
<type> <name>(<args; separator=",">) {
  <statements; separator="\n">
}
>>

All groups use <...> delimiters by default. Single line templates are enclosed in double quotes while multi-line templates are enclosed in double angle-brackets. Every template must define arguments even if the formal argument list is blank.

Using templates in a group file is straightforward. The StringTemplateGroup class has a number of constructors, one of which allows you to pass in a string or file or whatever:

String templates = "group simple; vardef(type,name) ..."; // templates from above
// Use the constructor that accepts a Reader
StringTemplateGroup group = new StringTemplateGroup(new StringReader(templates));
StringTemplate t = group.getInstanceOf("vardef");
t.setAttribute("type", "int");
t.setAttribute("name", "foo");
System.out.println(t);

String templates = "group simple; vardef(type,name) ..."; // templates from above
// Use the constructor that accepts a System.IO.TextReader
StringTemplateGroup group = new StringTemplateGroup(new StringReader(templates));
StringTemplate t = group.GetInstanceOf("vardef");
t.SetAttribute("type", "int");
t.SetAttribute("name", "foo");
Console.Out.WriteLine(t);

templates = "group simple; vardef(type,name) ..."; # templates from above
# Use the constructor that accepts a Reader
group = stringtemplate3.StringTemplateGroup(file=StringIO(templates))
t = group.getInstanceOf("vardef")
t["type"] = "int"
t["name"] = "foo"
print str(t)

The output would be: "int foo;".

Supergroups and interfaces

Groups may derive from other groups, thus, inheriting all of the templates from the supergroup. Group inheritance provides an appropriate model whereby a variation on a code generation target may be defined by describing how it differs from a previously defined target. Considering Java 1.4 versus 1.5, a Java1_5 group could specify how to alter the main Java (1.4) group templates in order to use generics and enumerated types.

Group inheritance would not yield its full potential without template polymorphism. A parser template instantiated via the Java1_5 group should always look for templates in Java1_5 rather than the Java supergroup even though that template is lexically defined within group Java.

Templates with the same name in a subgroup override templates in a supergroup just as in class inheritance. ST does not support overloaded templates so group inheritance does not take formal arguments into consideration.

The supergroup for a group may be changed dynamically using the setSuperGroup() method. If, however, a group must always derive from another group, use the following syntax:

group mygroup : supergroup;
...

If your group must satisfy a particular interface (see Group interfaces) and use the following syntax:

group mygroup implements anInterface, andAnotherInterface;
...

or

group mygroup : supergroup implements anInterface;
...

if the group derives from a supergroup and implements an interface.

Maps

There are situations where you need to translate a string in one language to a string in another language. For example, you might want to translate integer to int when translating Pascal to C. You could pass a Map or IDictionary (e.g. hashtable) from the model into the templates, but then you have output literals in your model! The only solution is to have StringTemplate support mappings. For example, here is how ANTLR v3 knows how to initialize local variables to their default values:

typeInitMap ::= [
        "int":"0",
        "long":"0",
        "float":"0.0",
        "double":"0.0",
        "boolean":"false",
        "byte":"0",
        "short":"0",
        "char":"0",
        default:"null" // anything other than an atomic type
]

To use the map in a template, refer to it as you would an attribute For example, <typeInitMap.int> which returns "0". If your type name is an attribute not a constant like int, then use an indirect field access: <typeInitMap.(typeName)>.

Map strings are actually templates that can refer to attributes that will become visible via dynamic scoping of attributes once the map entry has been embedded within a template. This is useful for referencing things like attribute username from within map values. That attribute will eventually become visible when the map a value is embedded within, say, a page template.

Large strings, such as those with newlines, can be specified with the usual large template delimiters from the group file format: <<...>>.

The default and other mappings can have empty values (implying no value). if no key is matched by the map then an empty value is returned, which is the same as using "default :" explicitly. The keyword key is available if you would like to refer to the key that maps to this value. This is particularly useful if you would like to filter certain words but otherwise leave a value unchanged; use default : key to return the key unmolested if it is not found in the map.

Maps are defined in the group's scope and are visible if no attribute hides them. For example, if you define a formal argument called typeInitMap in template foo then foo cannot see the map defined in the group (though you could pass it in as another parameter). If a name is not an attribute and it's not in the group's maps table, then the super group is consulted etc... You may not redefine a map and it may not have the same name as a template in that group. The default value is used if you use a key as a property that doesn't exist. For example <typeInitMap.foo> returns "null". The default clause must be at the end of the map.

You'll note that the square brackets will denote data structure in other areas too such as [a,b,c,...] which makes a singe multi-valued attribute out of other attributes so you can iterate across them.

Group file format

group
    :   "group" ID ( ':' ID  )?  ( "implements" ID (',' ID  )* )? ';'
        ( template | mapdef )+
    ;

template
    :   (   '@' ID '.' ID
        |   ID
        )
        '(' (args)? ')' "::="
        (   STRING      // "..."
        |   BIGSTRING   // <<...>>
        )
    |   ID "::=" ID     // alias one template to another
    ;

args:   arg ( ',' arg )*
    ;

arg :   ID '=' STRING               // x="..."
    |   ID '=' ANONYMOUS_TEMPLATE   // x={...}
    ;

mapdef
    :   ID "::=" map
    ;

map :   '['
            keyValuePair (',' keyValuePair)*
            ( ',' "default" ':' keyValue )?
        ']'
    ;

keyValuePair
    :   STRING ':' keyValue
    ;

keyValue
    :   BIGSTRING
    |   STRING
    |   "key"
    |
    ;

Both /* ... */ and single-line // ... comments are allowed outside of templates. Inside templates, you must use <!...!>.

An aside: All along, during my website construction days, I kept in mind that any text output follows a format and, thus, output sentences conform to a language. Consequently, a grammar should describe the output rather than a bunch of ad hoc print statements in code. This helped me formalize the study of templates because I could compare templates (output grammars) to well established ideas from formal language theory and context-free grammars. This allowed me to show, among other things, that StringTemplate can easily generate any document describable with an XML DTD even though it is deliberately limited. The group file format should look very much like a grammar to you.

Scoping rules and attribute look-up See the scoping rules section for information on how formal arguments affect attribute look up.

Group files have a .stg file extension.

Group loaders

When group files derive from another group, StringTemplate has to know how to load that group and its supergroups. StringTemplate 2.3 introduces the StringTemplateGroupLoader interface to describe objects that know how to load groups and interfaces.

public interface StringTemplateGroupLoader {
    /** Load the group called groupName from somewhere.  Return null
     *  if no group is found.
     */
    public StringTemplateGroup loadGroup(String groupName);

    /** Load a group with a specified superGroup.  Groups with
     *  region definitions must know their supergroup to find templates
     *  during parsing.
     */
    public StringTemplateGroup loadGroup(String groupName,
                                         StringTemplateGroup superGroup);


    /** Load the interface called interfaceName from somewhere.  Return null
     *  if no interface is found.
     */
    public StringTemplateGroupInterface loadInterface(String interfaceName);
}

public interface IStringTemplateGroupLoader
{
    /// <summary>
    /// Loads the named StringTemplateGroup instance from somewhere.
    /// </summary>
    StringTemplateGroup LoadGroup(string groupName);

    /// <summary>
    /// Loads the named StringTemplateGroup instance with the specified supergroup from somewhere.
    /// </summary>
    StringTemplateGroup LoadGroup(string groupName, StringTemplateGroup superGroup);

    /// <summary>
    /// Loads the named StringTemplateGroup instance with the specified supergroup from somewhere.
    /// Configure to use lexer of specified type.
    /// </summary>
    StringTemplateGroup LoadGroup(string groupName, StringTemplateGroup superGroup, Type lexer);

    /// <summary>
    /// Loads the named StringTemplateGroupInterface instance from somewhere.
    /// </summary>
    StringTemplateGroupInterface LoadInterface(string interfaceName);
}

class StringTemplateGroupLoader(object):
    def loadGroup(self, groupName, superGroup=None):
        raise NotImplementedError

    def loadInterface(self, interfaceName):
        raise NotImplementedError

By default, there are two implementations: PathGroupLoader and CommonGroupLoader. PathGroupLoader is a simple loader that looks only in the directory(ies) you specify in the ctor (Note that you can specify the char encoding). CommonGroupLoader, on the other hand, is a loader that also looks in the directory(ies) you specify in the ctor, but it uses the classpath rather than absolute dirs so it can be used when the ST application is jar'd up. Use Static method:

StringTemplateGroup.registerGroupLoader(loader);

StringTemplateGroup.registerGroupLoader(loader)

to specify a loader. For example, here is how ANTLR loads its templates:

// get a group loader containing main templates dir and target subdir
String templateDirs =
    classpathTemplateRootDirectoryName+":"+
    classpathTemplateRootDirectoryName+"/"+language;
StringTemplateGroupLoader loader =
    new CommonGroupLoader(templateDirs.toString(),
                          ErrorManager.getStringTemplateErrorListener());
StringTemplateGroup.registerGroupLoader(loader);

// first load main language template
StringTemplateGroup coreTemplates =
    StringTemplateGroup.loadGroup(language);

In order to use the group file format inheritance specifier, group sub : sup, you must specify a loader.

Formal argument default values

Sometimes it is convenient to have default values for formal arguments that are used when no value is set by the model. For example, when generating a parser in Java from ANTLR, I want the super class of the generated object to be Parser unless the ANTLR user uses an option to set the super class to some custom class. For example, here is a partial parser template definition:

parser(name, rules, superClass="Parser") ::= ...

Any argument may be given a default value by following the name with an equals sign and a string or an anonymous template.

Formal argument error handling

When using a group file format to specify templates, you must specify the formal arguments for that template. If you try to set an attribute via setAttribute that is not specifically formally defined in that template, you will get the following exception:

If you reference an attribute that is not formally defined in that template or any enclosing template, you also get the same exception.

Newline handling

The first newline following the << in a template definition is ignored as it is usually used just to get the first line of text for the template at the start of a line. In other words, if you want to have a blank line at the start of your template, use:

foo() ::= <<

2nd line is not blank, but first is
>>

or

foo() ::= <<<\n>
same as before; newline then this line
>>

The last newline before the >> is also ignored and is not included in the output. To add a final newline, add an extra line or <\n> before the >>:

foo() ::= <<
rodent

>>

or

foo() ::= <<
rodent<\n>
>>

The following template:

foo() ::= <<
rodent
>>

on the other hand, is identical to

foo() ::= "rodent"

Group Interfaces

To promote retargetable code generators, ST supports interface implementation a la Java interfaces where a template group that implements an interface must implement all templates in the interface and with the proper argument lists. The interface is the published, executable documentation for building back-ends for the code generator and has proven to be an excellent way to inform programmers responsible for the various targets of changes to the requirements.

The developers of the ANTLR code generation targets always have the same two questions: Initially they ask, "What is the set of templates I have to define for my target?'' and then, during development, they ask, "Has a change to the code generation logic forced any changes to the requirements of my template library?"

Originally, the answer to the first question involved abstracting the list of templates and their formal arguments from the existing Java target. The answer to the second question involved using a difference tool to point out changes in the Java target from repository check-in to check-in. Without a way to formally notify target developers and to automatically catch logic-template mismatches, bugs creep in that become apparent only when the stale template definitions are exercised by the code generator. This situation is analogous to programs in dynamically typed languages like Python where method signature changes can leave landmines in unexercised code. In short, there were no good answers.

ST now supports group interfaces that describe a collection of template signatures, names and formal arguments, in a manner analogous to Java interfaces. Interfaces clearly identify the set of all templates that a target must define as well as the attributes they operate on. The first question regarding the required set of templates now has a good answer.

Interfaces also provide a form of type safety whereby a target is examined upon code generator startup to see that it satisfies the interface. Here is a piece of the ANTLR main target interface:

interface ANTLRCore;
parser(name, scopes, tokens, tokenNames, rules,
       numRules, cyclicDFAs, bitsets, ASTLabelType,
       superClass, labelType, members);
rule(ruleName, ruleDescriptor, block, emptyRule,
       description, exceptions);
/** What file extension to use; e.g., ".java" */
codeFileExtension();
...

All of the various targets then implement the interface; e.g.,

group Java implements ANTLRCore;

The code generator, which loads target templates, notifies developers of any inconsistencies immediately upon startup effectively answering the second question regarding notification of template library changes. Group interfaces provide excellent documentation, promote consistency, and reduce hidden bugs.

Interfaces look exactly like groups except that they don't have template implementations for the template declarations although they must have the complete parameter list. Further, a template may be defined as optional using the optional keyword:

optional headerFile(actionScope, actions, docComment, recognizer, ...);

Expressions

Attribute References

Named attributes

The most common thing in a template besides plain text is a simple named attribute reference such as:

Your email: $email$

The template will look up the value of email and insert it into the output stream when you ask the template to print itself out. If email has no value, then it evaluates to the empty string and nothing is printed out for that attribute expression. When working with group files, if email is not defined in the formal parameter list of an enclosing template, an exception is thrown.

If the attribute is multi-value such as an instance of a list, the elements are emitted without separator one after the other. If there are null values in the list, these are ignored by default. Given template $values$ with attribute values=9,6,null,2,null then the output would be:

962

To use a separator in between those multiple values, use the separator option:

$values; separator=", "$

The output would be:

9, 6, 2

To emit a special value for each null element in a list, use the null option:

$values; null="-1", separator=", "$

Again using values=9,6,null,2,null then the output would be:

9, 6, -1, 2, -1

Property references

If a named attribute is an aggregate with a property or a simple data field, you may reference that property using attribute.property. For example:

Your name: $person.name$
Your email: $person.email$

StringTemplate ignores the actual object type stored in attribute person and simply looks for one of the following via reflection (in search order):

An exception is thrown if that property is not defined on the target object.

Because the type is ignored, you can pass in whatever existing aggregate (class) you have such as User or Person:

User u = database.lookupPerson("parrt@jguru.com");
st.setAttribute("person", u);

User u = database.LookupPerson("parrt@jguru.com");
st.SetAttribute("person", u);

u = database.lookupPerson("parrt@jguru.com")
st["person"] = u

Or, if a suitable aggregate doesn't exist, you can make a connector or "glue" object and pass that in instead:

st.setAttribute("person", new Connector());

st.SetAttribute("person", new Connector());

st["person"] = Connector()

where Connector is defined as:

public class Connector {
    public String getName() { return "Terence"; }
    public String getEmail() { return "parrt@jguru.com"; }
}

public class Connector {
    public string Name  { get {return "Terence";} }
    public string Email { get { return "parrt@jguru.com";} }
}

class Connector(object):
    def getName(self):
        return "Terence"

    def getEmail(self):
        return "parrt@jguru.com"

The ability to reference aggregrate properties saves you the trouble of having to pull out the properties with code like this:

User u = database.lookupPerson("parrt@jguru.com");
st.setAttribute("name", u.getName());
st.setAttribute("email", u.getEmail());

User u = database.lookupPerson("parrt@jguru.com");
st.SetAttribute("name", u.Name);
st.SetAttribute("email", u.Email);

u = database.lookupPerson("parrt@jguru.com")
st["name"] = u.getName()
st["email"] = u.getEmail()

and having template:

Your name: $name$
Your email: $email$

The latter is more widely applicable and totally decoupled from code and logic; i.e., it's "better" but much less convenient. Be very careful that the property methods do not have any side-effects like updating a counter or whatever. This breaks the rule of order of evaluation independence.

Indirect property names

Sometimes the property name is itself variable, in which case you need to use indirect property access notation:

$person.(propertyName)$

where propertyName is an attribute whose value is the name of a property to fetch from person. Using the examples from above, propertyName could hold the value of either name or email.

propertyName may actually be an expression instead of a simple attribute name.

Map key/value pair access

StringTemplate a = new StringTemplate("$user.name$, $user.phone$");
HashMap user = new HashMap();
user.put("name", "Terence");
user.put("phone", "none-of-your-business");
a.setAttribute("user", user);
String results = a.toString();

StringTemplate a = new StringTemplate("$user.name$, $user.phone$");
Hashtable user = new Hashtable();
user.Add("name", "Terence");
user.Add("phone", "none-of-your-business");
a.SetAttribute("user", user);
string results = a.ToString();

a = stringtemplate3.StringTemplate("$user.name$, $user.phone$")
user = {}
user["name"] = "Terence"
user["phone"] = "none-of-your-business"
a["user"] = user
results = str(a)

StringTemplate interprets Map objects to have two predefined properties: keys and values that yield a list of all keys and the list of all values, respectively. When applying a template to a map, StringTemplate iterates over the values so that <aMap> is a shorthand for <aMap.values>. Similarly <aMap.keys> walks over the keys. You can list all of the elements in a map like this:

<aMap.keys:{k| <k> maps to <aMap.(k)>}>.

Note the use of the indirect property reference <aMap.(k)>, which says to take the value of the k as the key in the lookup. Clearly without the parentheses the normal map lookup mechanism would treat k as a literal and try to look up k in the map.  Also note that the map must have keys that are Strings for indirect property referencing to work, because the key is first rendered into a string by ST and then that is used to look up the value in the map.

Difficult property names

Some property names cause parse errors because of clashes with built in keywords or because they do not match the rules for IDs as used by String Template. These difficult property names can be dealt with by quoting the property name in combination with the indirect property construct:

$person.("first")$       --- Build in keyword
$person.("1")$           --- non ID conforment name

Difficult properties names are quite likely to occur when dealing with maps. Map keys can be defined using arbitrary strings, including spaces and syntax characters used to defined templates themselves.

Be careful that the keys are the appropriate type. If person keys are Integer, $person.("1")$ won't work as Strings are never Integers.

Automatic aggregate creation

Creating one-off data aggregates is a pain, you have to define a new class just to associate two pieces of data. StringTemplate makes it easy to group data during setAttribute() calls. You may pass in an aggregrate attribute name to setAttribute() with the data to aggregate:

StringTemplate st = new StringTemplate("$items:{$it.(\"last\")$, $it.(\"first\")$\n}$");
st.setAttribute("items.{first,last}", "John", "Smith");
st.setAttribute("items.{first,last}", "Baron", "Von Munchhausen");
String expecting =
        "Smith, John\n" +
        "Von Munchhausen, Baron\n";

StringTemplate st = new StringTemplate("$items:{$it.(\"last\")$, $it.(\"first\")$\n}$");
st.SetAttribute("items.{first,last}", "John", "Smith");
st.SetAttribute("items.{first,last}", "Baron", "Von Munchhausen");
string expecting = "Smith, John\n" +
                   "Von Munchhausen, Baron\n";

st = stringtemplate3.StringTemplate("$items:{$it.(\"last\")$, $it.(\"first\")$\n}$")
st.setAttribute("items.{first,last}", "John", "Smith")
st.setAttribute("items.{first,last}", "Baron", "Von Munchhausen")
expecting = \
   "Smith, John\n" + \
   "Von Munchhausen, Baron\n"

Note that the template, st, expects the items to be aggregates with properties first and last. By using attribute name

items.{first,last}

You are telling StringTemplate to take the following two arguments as properties first and last.

The various overloads of the setAttribute() method can handle from 1 to 5 arguments. The C# version uses variable-length argument list (using params keyword).

List construction

As of v2.2, you may combine multiple attributes into a single multi-valued attribute in a syntax similar to the group map feature. Catenate attributes by placing them in square brackets in a comma-separated list. For example,

$[mine,yours]$

creates a new multi-valued attribute (a list) with both elements - all of mine first then all of yours. This feature is handy when the model happens to group attributes differently than you need to access them in the view. This ability to rearrange attributes is consistent with model-view separation because the template cannot alter the data structure nor test its values - the template is merely looking at the data from a new perspective.

Naturally you may combine the list construction with template application:

$[mine,yours]:{ v | ...}$

Note that this is very different from

$mine,yours:{ x,y | ...}$

which iterates max(n,m) times where n and m are the lengths of mine and yours, respectively. The [mine,yours] version iterates n+m times.

Template References

You may reference other templates to have them included just like the C language preprocessor #include construct behaves. For example, if you are building a web page (page.st) that has a search box, you might want the search box stored in a separate template file, say, searchbox.st. This has two advantages:

• You can reuse the template over and over (no cut/paste)
• You can change one template and all search boxes change on the whole site.

Using method call syntax, just reference the foreign template:

<html>
<body>
...
$searchbox()$
...
</body>
</html>

The invoking code would still just create the overall page and the enclosing page template would automatically create an instance of the referenced template and insert it:

StringTemplateGroup group = new StringTemplateGroup("webpages", "/usr/local/site/templates");
StringTemplate page = group.getInstanceOf("page");

StringTemplateGroup group = new StringTemplateGroup("webpages", "C:/Inetpub/wwwroot/site/templates");
StringTemplate page = group.GetInstanceOf("page");

group = stringtemplate3.StringTemplateGroup("webpages", "/usr/local/site/templates")
page = group.getInstanceOf("page")

If the template you want to reference, say searchbox, is in a subdirectory of the StringTemplateGroup root directory called misc, then you must reference the template as: misc/searchbox().

The included template may access attributes. How can you set the attribute of an included template? There are two ways: inheriting attributes and passing parameters.

Accessing Attributes Of Enclosing Template

Any included template can reference the attributes of the enclosing template instance. So if searchbox references an attribute called resource:

<form ...>
...
<input type=hidden name=resource value=$resource$>
...
</form>

you could set attribute resource in the enclosing template page object:

StringTemplate page = group.getInstanceOf("page");
page.setAttribute("resource", "faqs");

StringTemplate page = group.GetInstanceOf("page");
page.SetAttribute("resource", "faqs");

page = group.getInstanceOf("page")
page["resource"] = "faqs"

This "inheritance" (dynamic scoping really) of attributes feature is particularly handy for setting generally useful attributes like siteFontTag in the outermost body template and being able to reference it in any nested template in the body.

Passing Parameters To Another Template

Another, more obvious, way to set the attributes of an included template is to pass in values as parameters, making them look like C macro invocations rather than includes. The syntax looks like a set of attribute assignments:

<html>
<body>
...
$searchbox(resource="faqs")$
...
</body>
</html>

where I am setting the attribute of the included searchbox to be the
string literal "faqs".

The right-hand-side of the assignment may be any expression such as an attribute reference or even a reference to another template like this:

$boldMe(item=copyrightNotice())$

You may also use an anonymous template such as:

$bold(it={$firstName$ $lastName$})$

which first computes the template argument and then assigns it to it.

If you are using StringTemplate groups, then you have formal parameters and for those templates with a sole formal argument, you can pass just an expression instead of doing an assignment to the argument name. For example, if you do $bold(name)$ and bold has one formal argument called item, then item gets the value of name just as if you had said {$bold(item=name)$}.

Allowing enclosing attributes to pass through

When template x calls template y, the formal arguments of y hide any x arguments of the same because the formal parameters force you to define values. This prevents surprises and makes it easy to ensure any parameter value is empty unless you specifically set it for that template. The problem is that you need to factor templates sometimes and want to refine behavior with a subclass or just invoke another shared template but invoking y as <y()> hides all of x's parameters with the same name. Use <y(...)> syntax to indicate y should inherit all values even those with the same name. <y(name="foo", ...)> would set one arg, but the others are inherited whereas <y(name="foo")> only has name set; all other arguments of template y are empty. You can set manually with:

Argument evaluation scope

The right-hand-side of the argument assignments are evaluated within the scope of the enclosing template whereas the left-hand-side attribute name is the name of an attribute in the target template. Template invocations like $bold(item=item)$ actually make sense because the item on the right is evaluated in a different scope.

Attribute operators

StringTemplate provides a number of operators that you can apply to attributes to get a new view of that data: first, rest, last, length, strip.

Sometimes you need to treat the first or last element of multi-valued attribute differently than the others. For example, if you have a list of integers in an attribute and you need to generate code to sum those numbers, you could start like this:

<numbers:{ n | sum += <n>;}>

You need to define sum, however:

int sum = 0;
<numbers:{ n | sum += <n>;}>

What if numbers is empty though? No need to create the sum definition so you could do this:

<if(numbers)>int sum = 0;<endif>
<numbers:{ n | sum += <n>;}>

A more specific strategy (and one that generates slightly better code as it avoids an unnecessary initialization to 0) is the following:

<first(numbers):{ n | int sum = <n>;}>
<rest(numbers):{ n | sum += <n>;}>

where first(numbers) results in the first value of attribute numbers if any and rest(numbers) results all values in numbers but the first value.

The other operator available to you is last, which naturally results in the last value of a multi-valued attribute.

Special cases:

• operations on empty attributes yields an empty value

• rest of a single-valued attribute yields an empty value

• tail of a single-valued attribute yields the same as first, the attribute value

You may find it handy to use another operator sometimes: plus "string concatenate". operator. For example, you may want to compute an argument to a template using a literal and an attribute:

...$link(url="/faq/view?ID="+faqid, title=faqtitle)$...

where faqid and faqtitle are attributes you have set for
the template that referenced link.

Terence says I'm a little uncomfortable with this catenation operation. Please use a template instead: ...$link(url={/faq/view?ID=$faqid$}, title=faqtitle)$...

In order to emit the number of attributes in a single or multi-value attribute, use the length operator:

int data[$length(x)$] = { $x; separator=", "$ };

In this example, with x=5,2,9 the following would be emitted:

int data[3] = { 5, 2, 9 };

Null values are counted by length but you can use the strip operator to return a new view of your list without null values:

int data[$length(strip(x))] = { $x; separator=", "$ };

Template Application

Imagine a simple template called bold:

<b>$item$</b>

Just as with template link described above, you can reference it from a template by invoking it like a method call:

$bold(item=name)$

What if you want something bold and italicized? You could simply nest the template reference:

$bold(item=italics(item=name))$

(or $bold(italics(name))$ if you're using group file format and have formal parameters). Template italics is defined as:

<i>$item$</i>

using a different attribute with the same name, item; the attributes have different values just like you would expect if these template references where method calls in say Java or C# and, item was a local variable. Parameters and attribute references are scoped like a programming language.

Think about what you are really trying to say here. You want to say "make name italics and then make it bold", or "apply italics to the name and then apply bold." There is an "apply template" syntax that is a literal translation:

$name:italics():bold()$

where the templates are applied in the order specified from left to right. This is much more clear, particularly if you had three templates to apply:

$name:courierFont():italics():bold()$

For this syntax to work, however, the applied templates have to reference a standard attribute because you are not setting the attribute in a parameter assignment. In general for syntax expr:template(), an attribute called it is set to the value of expr. So, the definition of bold (and analogously italics), would have to be:

<b>$it$</b>

to pick up the value of name in our examples above.

As of 2.2 StringTemplate, you can avoid using it as a default parameter by using formal arguments. For expression $x:y()$, StringTemplate will assign the value of x to it and any sole formal argument of y. For example, if y is:

y(item) ::= "_$item$_"

then item would also have the value of x.

If the attribute to which you are applying a template is null (i.e., missing), then the application is not done as there is no work to do. Optionally, you can specify what string template should display when the attribute is null a using the null option:

$name:bold(); null="n/a"$

That is equivalent to the following conditional:

$if(name)$$name:bold()$$else$n/a$endif$

Applying Templates To Multi-Valued Attributes

Where template application really shines though is when an attribute is multi-valued. One of the most common web page generation issues is making lists of items either as bullet lists or table rows etc... Applying a template to a multi-valued attribute means that you want the template applied to each of the values.

Consider a list of names (i.e., you set attribute names multiple times) that you want in a bullet list. If you have a template called listItem:

<li>$it$</li>

then you can do this:

<ul>
$names:listItem()$
</ul>

and each name will appear as a bullet item. For example, if you set names to "Terence", "Tom", and "Kunle", then you would see:

<ul>
<li>Terence</li>
<li>Tom</li>
<li>Kunle</li>
</ul>

in the output.

Whenever you apply a template to an attribute or multi-valued attribute, the default attribute it is set. Similarly, attributes i and i0 (since v3.0) of type integer are set to the value's index number starting from 1 (i0 starts from 0). For example, if you wanted to make your own style of numbered list, you could reference i to get the index:

$names:numberedListItem()$

where template numberedListItem is defined as:

$i$. $it$<br>

In this case, the output would be:

1. Terence<br>
2. Tom<br>
3. Kunle<br>

If there is only one attribute value, then i will be 1. However, if template numberedListItem is defined as:

$i0$. $it$<br>

The output would be:

0. Terence<br>
1. Tom<br>
2. Kunle<br>

As when invoking templates ala "includes", a single formal argument is also set to the iterated value. For example, you could define numberedListItem as follows in a StringTemplateGroup file:

numberedListItem(item) ::= "$i$. $item$<br>"

Templates are not applied to null values in multi-valued attributes. StringTemplate behaves as if those values simply did not exist in the list. To emit a special string or template for each null value, use the null option:

$names:bold(); null="n/a"$

which will emit "n/a" for any null value in attribute names.

Applying Multiple Templates To Multi-Valued Attributes

The result of applying a template to a multi-valued attribute is another multi-valued attribute containing the results of the application. You may apply another template to the results of the first template application, which comes in handy when you need to format the elements of a list before they go into the list. For example, to bold the elements of a list do the following (given the appropriate template definitions from above):

$names:bold():listItem()$

If you actually want to apply a template to the combined (string) result of a previous template application, enclose the previous application in parenthesis. The parenthesis will force immediate evaluation of the template application, resulting in a string. For example,

$(names:bold()):listItem()$

results in a single list item full of a bunch of bolded names. Without the parenthesis, you get a list of items that are bolded.

Applying Alternating Templates To Multi-Valued Attributes

When generating lists of things, you often need to change the color or other formatting instructions depending on the list position. For example, you might want to alternate the color of the background for the elements of a list. The easiest and most natural way to specify this is with an alternating list of templates to apply to an expression of the form: $expr:t1(),t2(),...,tN()$. To make an alternating list of blue and green names, you might say:

$names:blueListItem(),greenListItem()$

where presumably blueListItem template is an HTML <table> or something that lets you change background color. names[0] would get blueListItem() applied to it, names[1] would get greenListItem(), and names[2] would get blueListItem() again, etc...

If names is single-valued, then blueListItem() is applied and that's it.

Applying Anonymous Templates

Some templates are so simple or so unlikely to be reused that it seems a waste of time making a separate template file and then referencing it. StringTemplate provides anonymous subtemplates to handle this case. The templates are anonymous in the sense that they are not named; they are directly applied in a single instance.

For example, to show a name list do the following:

<ul>
$names:{<li>$it$</li>}$
</ul>

where anything enclosed in curlies is an anonymous subtemplate if, of course, it's within an attribute expression. Note that in the subtemplate, I must enclose the it reference in the template expression delimiters. You have started a new template exactly like the surrounding template and you must distinguish between text and attribute expressions.

You can apply multiple templates very conveniently. Here is the bold list of names again with anonymous templates:

<ul>
$names:{<b>$it$</b>}:{<li>$it$</li>}$
</ul>

The output would look like:

<ul>
<li><b>Terence</b></li>
<li><b>Tom</b></li>
<li><b>Kunle</b></li>
</ul>

Anonymous templates work on single-valued attributes as well.

As of 2.2, you may define formal arguments on anonymous templates even if you are not using StringTemplate groups. This syntax is borrowed from SmallTalk though it is identical in function to lambda of Python. Use a comma-separated list of argument names followed by the '|' "pipe" symbol. Any single whitespace character immediately following the pipe is ignored. The following example bolds the names in a list using an argument to avoid the monotonous use of it:

<ul>
$names:{ n | <b>$n$</b>}$
</ul>

Clearly only one argument may be defined in this situation: the iterated value of a single list.

Anonymous template application to multiple attributes

In some cases, the model may present data to the view as separate columns of data rather than as a single list of objects, such as multi-valued attributes names and phones rather than a single users multi-valued attribute. As of 2.2, you may iterate over multiple attributes:

$names,phones:{ n,p | $n$: $p$}$

An error is generated if you have too many arguments for the number of attributes. Iteration proceeds while at least one of the attributes (names or phones, in this case) has values.

Indirect template references

Sometimes the name of the template you would like to include is itself a variable. So, rather than using "<item:format()>" you want the name of the template, format, to be a variable rather than a literal. Just enclose the template name in parenthesis to indicate you want the immediate value of that attribute and then add () like a normal template invocation and you get "<item:(someFormat)()>", which means "look up attribute someFormat and use its value as a template name; appy to item." This deliberately looks similar to the C function call indirection through a function pointer (e.g., "(*fp)()" where fp is a pointer to a function). A better way to look at it though is that the (someFormat) implies immediately evaluate someFormat and use as the template name.

Usually this "variable template" situation occurs when you have a list of items to format and each element may require a different template. Rather than have the controller code create a bunch of instances, one could consider it better to have StringTemplate do the creation--the controller just names what format to use.

If StringTemplate did not have a map definition, you could simulate its functionality. Consider generating a list of C# declarations that are initialized to 0, false, null, etc... You could define a template for int, Object, Array, etc... declarations and then pass in an aggregate object that has the variable declaration object and the format. In a template group file you might have:

group Java;

file(variables,methods) ::= <<
<variables:{ v | <v.decl:(v.format)()>}; separator="\n">
<methods>
\>>
intdecl(decl) ::= "int <decl.name> = 0;"
intarray(decl) ::= "int[] <decl.name> = null;"

Your code might look like:

StringTemplateGroup group =
        new StringTemplateGroup(new StringReader(templates),
                AngleBracketTemplateLexer.class);
StringTemplate f = group.getInstanceOf("file");
f.setAttribute("variables.{decl,format}", new Decl("i","int"), "intdecl");
f.setAttribute("variables.{decl,format}", new Decl("a","int-array"), "intarray");
System.out.println("f="+f);
String expecting = ""+newline+newline;

StringTemplateGroup group =
        new StringTemplateGroup(new StringReader(templates),
                typeof(AngleBracketTemplateLexer));
StringTemplate f = group.GetInstanceOf("file");
f.setAttribute("variables.{decl,format}", new Decl("i","int"), "intdecl");
f.setAttribute("variables.{decl,format}", new Decl("a","int-array"), "intarray");
Console.Out.WriteLine("f="+f);
string expecting = ""+newline+newline;

group = stringtemplate3.StringTemplateGroup(file=StringIO(templates), lexer="a