java - The Ceylon Typechecker: How to obtain the typed syntax tree? -

im trying programmatically use / embed ceylon typechecker analyse ceylon source code. in use case want access information compiler consume. im not going compile , im not going add dependency on compiler.

it seem me main.main entry point in ceylon/typechecker/src/main/main.java not appropriate entry point use case (obtaining typed tree , attached models), because information, collected visitors in 3 checker passes discarded, , errors printed.

so, question is:

how can parse , typecheck compilation unit , obtain 1. typed syntax tree, , 2. associated model objects of types analysis visitors encounter in tree, linked tree.

edited:

there (and is) confusion on side 3 different asts there are.

in readme in ceylon /ceylon.ast said:

¨ ... ceylon.ast.core – ceylon classes represent ceylon ast. pure ceylon (backend-independent). ... ceylon.ast.redhat – transforms ceylon.ast.core ast + redhat compiler (ceylon-spec ast, , contains functions compile ceylon.ast.core ast code string (using redhat compiler) ... ¨.

so there 3 asts: 1. 1 generated antlr, 2. ceylon.ast.core, , 3. ceylon.ast.redhat. why?

in short, you'll want to:

• configure typecheckerbuilder source files want typecheck,
• obtain typechecker builder (builder.typechecker),
• invoke typechecker (typechecker.process()),
• process results available typechecker.phasedunits. specifically, typechecker.getphasedunits().getphasedunits() give list<phasedunit>, , each phasedunit, can call getcompilationunit() obtain tree.compilationunit, root of ast. ast nodes include getters model objects.

for detailed example, can review code dart backend, working forwards , backwards the call process() in compiledart() function.

see testcompile example code calls compiledart().