How can I transform a large discriminated union tree to a readable form?
https://stackoverflow.com/questions/3817819
-
26-09-2019 - |
italiano
english
français
española
中国
日本の
العربية
Deutsch
한국어
Português
RussianQuestion
The following type is clearly quite large so manually writing the code to convert this to a readable form would be tedious. I would like to know the simplest way to display the tree in a readable form.
type Element =
| Nil
| Token of Token
| Expression of Element * Element
| ExpressionNoIn of Element * Element
| AssignmentExpression of Element * AssignmentOperator * Element
| AssignmentExpressionNoIn of Element * AssignmentOperator * Element
| ConditionalExpression of Element * Element * Element
| ConditionalExpressionNoIn of Element * Element * Element
| LogicalORExpression of Element * Element
| LogicalORExpressionNoIn of Element * Element
| LogicalANDExpression of Element * Element
| LogicalANDExpressionNoIn of Element * Element
| BitwiseORExpression of Element * Element
| BitwiseORExpressionNoIn of Element * Element
| BitwiseXORExpression of Element * Element
| BitwiseXORExpressionNoIn of Element * Element
| BitwiseANDExpression of Element * Element
| BitwiseANDExpressionNoIn of Element * Element
| EqualityExpression of Element * EqualityOperator * Element
| EqualityExpressionNoIn of Element * EqualityOperator * Element
| RelationalExpression of Element * RelationalOperator * Element
| RelationalExpressionNoIn of Element * RelationalOperator * Element
| ShiftExpression of Element * BitwiseShiftOperator * Element
| AdditiveExpression of Element * AdditiveOperator * Element
| MultiplicativeExpression of Element * MultiplicativeOperator * Element
| UnaryExpression of UnaryOperator * Element
| PostfixExpression of Element * PostfixOperator
| MemberExpression of Element * Element
| Arguments of Element * Element
| ArgumentList of Element
| CallExpression of Element * Element
| NewExpression of NewOperator * Element
| LeftHandSideExpression of Element
| PrimaryExpression of Element
| ObjectLiteral of Element
| PropertyNameAndValueList of Element * Element
| PropertyAssignment of Element * Element * Element
| PropertyName of Element
| PropertySetParameterList of Element
| ArrayLiteral of Element * Element
| Elision of Element * Element
| ElementList of Element * Element * Element
| Statement of Element
| Block of Element
| StatementList of Element * Element
| VariableStatement of Element
| VariableDeclarationList of Element * Element
| VariableDeclarationListNoIn of Element * Element
| VariableDeclaration of Element * Element
| VariableDeclarationNoIn of Element * Element
| Initialiser of Element
| InitialiserNoIn of Element
| EmptyStatement
| ExpressionStatement of Element
| IfStatement of Element * Element * Element
| IterationStatement of Element * Element * Element * Element
| ContinueStatement of Element
| BreakStatement of Element
| ReturnStatement of Element
| WithStatement of Element * Element
| SwitchStatement of Element * Element
| CaseBlock of Element * Element * Element
| CaseClauses of Element * Element
| CaseClause of Element * Element
| DefaultClause of Element
| LabelledStatement of Element * Element
| ThrowStatement of Element
| TryStatement of Element * Element * Element
| Catch of Element * Element
| Finally of Element
| DebuggerStatement
| FunctionDeclaration of Element * Element * Element
| FunctionExpression of Element * Element * Element
| FormalParameterList of Element * Element
| FunctionBody of Element
| SourceElement of Element
| SourceElements of Element * Element
| Program of Element
Here is an example of how this might be displayed. (It is a bit different as I produced it a while ago.)
<Expression>
<AssignmentExpression>
<ConditionalExpression>
<LogicalORExpression>
<LogicalORExpression>
<LogicalANDExpression>
<BitwiseORExpression>
<BitwiseXORExpression>
<BitwiseANDExpression>
<EqualityExpression>
<EqualityExpression>
<RelationalExpression>
<ShiftExpression>
<AdditiveExpression>
<MultiplicativeExpression>
<MultiplicativeExpression>
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="i" Line="9" Column="13" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
<TokenNode Value="%" Line="9" Column="15" />
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="3" Line="9" Column="17" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
</AdditiveExpression>
</ShiftExpression>
</RelationalExpression>
</EqualityExpression>
<TokenNode Value="===" Line="9" Column="19" />
<RelationalExpression>
<ShiftExpression>
<AdditiveExpression>
<MultiplicativeExpression>
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="0" Line="9" Column="23" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
</AdditiveExpression>
</ShiftExpression>
</RelationalExpression>
</EqualityExpression>
</BitwiseANDExpression>
</BitwiseXORExpression>
</BitwiseORExpression>
</LogicalANDExpression>
</LogicalORExpression>
<TokenNode Value="||" Line="9" Column="25" />
<LogicalANDExpression>
<BitwiseORExpression>
<BitwiseXORExpression>
<BitwiseANDExpression>
<EqualityExpression>
<EqualityExpression>
<RelationalExpression>
<ShiftExpression>
<AdditiveExpression>
<MultiplicativeExpression>
<MultiplicativeExpression>
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="i" Line="9" Column="28" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
<TokenNode Value="%" Line="9" Column="30" />
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="5" Line="9" Column="32" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
</AdditiveExpression>
</ShiftExpression>
</RelationalExpression>
</EqualityExpression>
<TokenNode Value="===" Line="9" Column="34" />
<RelationalExpression>
<ShiftExpression>
<AdditiveExpression>
<MultiplicativeExpression>
<UnaryExpression>
<PostfixExpression>
<LeftHandSideExpression>
<NewExpression>
<MemberExpression>
<PrimaryExpression>
<TokenNode Value="0" Line="9" Column="38" />
</PrimaryExpression>
</MemberExpression>
</NewExpression>
</LeftHandSideExpression>
</PostfixExpression>
</UnaryExpression>
</MultiplicativeExpression>
</AdditiveExpression>
</ShiftExpression>
</RelationalExpression>
</EqualityExpression>
</BitwiseANDExpression>
</BitwiseXORExpression>
</BitwiseORExpression>
</LogicalANDExpression>
</LogicalORExpression>
</ConditionalExpression>
</AssignmentExpression>
</Expression>
Solution
If you want to write generic union processing code that will not need to list all the union cases, then you'll probably need to use F# reflection API. Here is a simple example.
The formatUnion function uses F# reflection. It assumes that the type parameter 'T is a union type and uses GetUnionFields to get a tuple containing the name of the current case and arguments. It prints the current case name and iterates over all the arguments. If some of the arguments is value of type 'T (meaning that it is recursive union), we recursively print information about the value:
let rec formatUnion indent (value:'T) = //'
// Get name and arguments of the current union case
let info, args = Reflection.FSharpValue.GetUnionFields(value, typeof<'T>) //'
// Print current name (with some indentation)
printfn "%s%s" indent info.Name
for a in args do
match box a with
| :? 'T as v ->
// Recursive use of the same union type..
formatUnion (indent + " ") v
| _ -> ()
The following example runs the function on a very simple union value:
type Element = | Nil | And of Element * Element | Or of Element * Element
formatUnion "" (And(Nil, Or(Nil, Nil)))
// Here is the expected output:
// And
// Nil
// Or
// Nil
// Nil
As a side-note, I think that you could largely simplify your discriminated union by having cases for BinaryOperator and UnaryOperator (with one additional parameter) instead of listing all the element types explicitly. Then you could probably implement the function directly, because it would be quite simple. Something like:
type BinaryOperator = LogicalOr | LogicalAnd | BitwiseOr // ...
type UnaryOperator = Statement | Block | Initializer // ...
type Element =
| BinaryOperator of BinaryOperator * Element * Element
| UnaryOperator of UnaryOperator * Element
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
