theory Antiquotes
imports "../Base"
begin


section {* Useful Document Antiquotations *}

text {*
  {\bf Problem:} 
  How to keep your ML-code inside a document synchronised with the actual code?\smallskip

  {\bf Solution:} This can be achieved using document antiquotations.\smallskip

  Document antiquotations can be used for ensuring consistent type-setting of
  various entities in a document. They can also be used for sophisticated
  \LaTeX-hacking.

  Below we give the code for two antiquotations that can be used to typeset
  ML-code and also to check whether the given code actually compiles. This
  provides a sanity check for the code and also allows one to keep documents
  in sync with other code, for example Isabelle.

  We first describe the antiquotation @{text "@{ML_checked \"\<dots>\"}"}. This
  antiquotation takes a piece of code as argument; this code is then checked
  by sending the ML-expression @{text [quotes] "val _ = \<dots>"} containing the
  given code to the ML-compiler (i.e.~the function @{ML "ML_Context.eval_in"}
  in the snippet below). The code for @{text "@{ML_checked \"\<dots>\"}"} is as
  follows:

*}

ML %linenumbers {*fun ml_val txt = "val _ = " ^ txt

fun output_ml ml src ctxt txt =
  (ML_Context.eval_in (SOME ctxt) false Position.none (ml txt); 
  ThyOutput.output_list (fn _ => fn s => Pretty.str s) src ctxt 
                                            (space_explode "\n" txt))

val _ = ThyOutput.add_commands
 [("ML_checked", ThyOutput.args (Scan.lift Args.name) (output_ml ml_val))]
*}

text {*

  Note that the parser @{ML "(Scan.lift Args.name)"} in line 9 parses a string, 
  in this case the code. This code is send to the ML-compiler in the line 4.  
  If the code is ``approved'' by the compiler, then the output function @{ML
  "ThyOutput.output_list (fn _ => fn s => Pretty.str s)"} in the next line pretty prints the
  code. This function expects that the code is a list of strings where each
  string correspond to a line (therefore the @{ML "(space_explode \"\\n\" txt)" for txt} 
  which produces this list).  There are a number of options for antiquotations
  that are observed by @{ML ThyOutput.output_list} when printing the code (for
  example @{text "[display]"}, @{text "[quotes]"} and @{text "[source]"}).

  \begin{readmore}
  For more information about options of antiquotations see \rsccite{sec:antiq}).
  \end{readmore}

  Since we used the argument @{ML "Position.none"}, the compiler cannot give specific 
  information about the line number where an error might have occurred. We 
  can improve this code slightly by writing 

*}

ML {*
fun output_ml ml src ctxt (txt,pos) =
  (ML_Context.eval_in (SOME ctxt) false pos (ml txt);
  ThyOutput.output_list (fn _ => fn s => Pretty.str s) src ctxt 
                                            (space_explode "\n" txt))

val _ = ThyOutput.add_commands
 [("ML_checked", ThyOutput.args 
       (Scan.lift (OuterParse.position Args.name)) (output_ml ml_val))]
*}

text {*
  (FIXME: say something about OuterParse.position)

  We can now write in a document @{text "@{ML_checked \"2 + 3\"}"} in order to
  obtain @{ML_checked "2 + 3"} and be sure that this code compiles until
  somebody changes the definition of @{ML "(op +)"}.


  The second antiquotation extends the first by allowing to also give
  hints what the result of the ML-code is and check consistency of these
  hints. For this we use the antiquotation  @{text "@{ML_response \"\<dots>\" \"\<dots>\"}"}
  whose first argument is the ML-code and the second is the result. 

  In the antiquotation @{text "@{ML_checked \"\<dots>\"}"} we send the expresion 
  @{text [quotes] "val _ = \<dots>"} to the compiler. Now we will use the hints
  to construct a pattern for the @{text "_"}. To add some convenince we allow
  the user to give partial hints using @{text "\<dots>"}, which however need to
  be replaced by @{text "_"} before sending the code to the compiler. The
  function 

*}

ML {*
fun ml_pat (rhs, pat) =
let val pat' = implode (map (fn "\<dots>" => "_" | s => s) (Symbol.explode pat))
in "val " ^ pat' ^ " = " ^ rhs end;
*}

text {* 
  will do this. Next we like to add a response indicator to the result using:
*}


ML {*
fun add_response_indicator txt =
  map (fn s => "> " ^ s) (space_explode "\n" txt)
*}

text {* 
  The rest of the code of the antiquotation is
  *}

ML {*
fun output_ml_response ml src ctxt ((lhs,pat),pos) = 
  (ML_Context.eval_in (SOME ctxt) false pos (ml (lhs,pat));
  let val txt = (space_explode "\n" lhs) @ (add_response_indicator pat)
  in ThyOutput.output_list (fn _ => fn s => Pretty.str s) src ctxt txt end)

val _ = ThyOutput.add_commands
 [("ML_response", 
     ThyOutput.args 
      (Scan.lift (OuterParse.position (Args.name -- Args.name))) 
      (output_ml_response ml_pat))]
*}

text {*
  This extended antiquotation allows us to write 
  @{text [display] "@{ML_response [display] \"true andalso false\" \"false\"}"}
  to obtain

@{ML_response [display] "true andalso false" "false"} 

  or 

@{text [display] "@{ML_response [display] \"let val i = 3 in (i * i,\"foo\") end\" \"(9,\<dots>)\"}"}
  
  to obtain

@{ML_response [display] "let val i = 3 in (i * i,\"foo\") end" "(9,\<dots>)"} 

  In both cases, the check by the compiler ensures that code and result match. A limitation
  of this antiquotation is that the hints can only be given for results that can actually
  be constructed as a pattern. This excludes values that are abstract datatypes, like 
  theorems or cterms.

*}



end