src/Pure/Syntax/parser.ML
author wenzelm
Sat, 26 Jan 2008 22:14:07 +0100
changeset 25986 26f1e4c172c3
parent 24245 4ffeb1dd048a
child 25987 bfda3f3beccd
permissions -rw-r--r--
syntax error: reduced spam -- print expected nonterminals instead of terminals; tuned pretty_gram;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
     1
(*  Title:      Pure/Syntax/parser.ML
46
f0f4978af183 *** empty log message ***
wenzelm
parents: 18
diff changeset
     2
    ID:         $Id$
2186
35ade4941904 removed silly message;
wenzelm
parents: 1580
diff changeset
     3
    Author:     Carsten Clasohm, Sonia Mahjoub, and Markus Wenzel, TU Muenchen
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
     4
15979
wenzelm
parents: 15752
diff changeset
     5
General context-free parser for the inner syntax of terms, types, etc.
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
     6
*)
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
     7
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
     8
signature PARSER =
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
     9
sig
1507
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    10
  type gram
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    11
  val empty_gram: gram
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    12
  val extend_gram: gram -> SynExt.xprod list -> gram
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    13
  val make_gram: SynExt.xprod list -> gram
1507
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    14
  val merge_grams: gram -> gram -> gram
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    15
  val pretty_gram: gram -> Pretty.T list
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    16
  datatype parsetree =
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    17
    Node of string * parsetree list |
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    18
    Tip of Lexicon.token
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    19
  val parse: gram -> string -> Lexicon.token list -> parsetree list
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    20
  val branching_level: int ref
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    21
end;
1507
f600215b6ea7 Elimination of fully-functorial style.
paulson
parents: 1438
diff changeset
    22
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    23
structure Parser: PARSER =
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
    24
struct
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    25
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
    26
open Lexicon SynExt;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
    27
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    28
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
    29
(** datatype gram **)
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
    30
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    31
type nt_tag = int;              (*production for the NTs are stored in an array
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    32
                                  so we can identify NTs by their index*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    33
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    34
datatype symb = Terminal of token
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    35
              | Nonterminal of nt_tag * int;              (*(tag, precedence)*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    36
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    37
type nt_gram = ((nt_tag list * token list) *
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    38
                (token option * (symb list * string * int) list) list);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    39
                                     (*(([dependent_nts], [start_tokens]),
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    40
                                        [(start_token, [(rhs, name, prio)])])*)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    41
                              (*depent_nts is a list of all NTs whose lookahead
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    42
                                depends on this NT's lookahead*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    43
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    44
datatype gram =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    45
  Gram of {nt_count: int, prod_count: int,
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    46
           tags: nt_tag Symtab.table,
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    47
           chains: (nt_tag * nt_tag list) list,              (*[(to, [from])]*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    48
           lambdas: nt_tag list,
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    49
           prods: nt_gram Array.array};
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    50
                       (*"tags" is used to map NT names (i.e. strings) to tags;
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    51
                         chain productions are not stored as normal productions
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    52
                         but instead as an entry in "chains";
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    53
                         lambda productions are stored as normal productions
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    54
                         and also as an entry in "lambdas"*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    55
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
    56
fun reverted_tags (tags: nt_tag Symtab.table) =
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
    57
  the o Inttab.lookup (Inttab.make (map swap (Symtab.dest tags)));
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
    58
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    59
val UnknownStart = EndToken;       (*productions for which no starting token is
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    60
                                     known yet are associated with this token*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    61
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    62
(* get all NTs that are connected with a list of NTs
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    63
   (used for expanding chain list)*)
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
    64
fun connected_with _ ([]: nt_tag list) relatives = relatives
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    65
  | connected_with chains (root :: roots) relatives =
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
    66
    let val branches = subtract (op =) relatives (these (AList.lookup (op =) chains root));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    67
    in connected_with chains (branches @ roots) (branches @ relatives) end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
    68
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    69
(* convert productions to grammar;
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    70
   N.B. that the chains parameter has the form [(from, [to])];
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
    71
   prod_count is of type "int option" and is only updated if it is <> NONE*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    72
fun add_prods _ chains lambdas prod_count [] = (chains, lambdas, prod_count)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    73
  | add_prods prods chains lambdas prod_count
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    74
              ((lhs, new_prod as (rhs, name, pri)) :: ps) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    75
    let
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    76
      val chain_from = case (pri, rhs) of (~1, [Nonterminal (id, ~1)]) => SOME id | _ => NONE;
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    77
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
    78
      (*store chain if it does not already exist*)
22573
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
    79
      val (new_chain, chains') = case chain_from of NONE => (NONE, chains) | SOME from_ =>
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
    80
        let val old_tos = these (AList.lookup (op =) chains from_) in
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
    81
          if member (op =) old_tos lhs then (NONE, chains)
22573
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
    82
          else (SOME from_, AList.update (op =) (from_, insert (op =) lhs old_tos) chains)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    83
        end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    84
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    85
      (*propagate new chain in lookahead and lambda lists;
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    86
        added_starts is used later to associate existing
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    87
        productions with new starting tokens*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    88
      val (added_starts, lambdas') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    89
        if is_none new_chain then ([], lambdas) else
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    90
        let (*lookahead of chain's source*)
15979
wenzelm
parents: 15752
diff changeset
    91
            val ((from_nts, from_tks), _) = Array.sub (prods, the new_chain);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    92
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    93
            (*copy from's lookahead to chain's destinations*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    94
            fun copy_lookahead [] added = added
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
    95
              | copy_lookahead (to :: tos) added =
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    96
                let
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
    97
                  val ((to_nts, to_tks), ps) = Array.sub (prods, to);
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
    98
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
    99
                  val new_tks = subtract (op =) to_tks from_tks;  (*added lookahead tokens*)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   100
                in Array.update (prods, to, ((to_nts, to_tks @ new_tks), ps));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   101
                   copy_lookahead tos (if null new_tks then added
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   102
                                       else (to, new_tks) :: added)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   103
                end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   104
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   105
            val tos = connected_with chains' [lhs] [lhs];
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   106
        in (copy_lookahead tos [],
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   107
            gen_union (op =) (if member (op =) lambdas lhs then tos else [], lambdas))
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   108
        end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   109
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   110
      (*test if new production can produce lambda
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   111
        (rhs must either be empty or only consist of lambda NTs)*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   112
      val (new_lambda, lambdas') =
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   113
        if forall (fn (Nonterminal (id, _)) => member (op =) lambdas' id
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   114
                    | (Terminal _) => false) rhs then
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   115
          (true, gen_union (op =) (lambdas', connected_with chains' [lhs] [lhs]))
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   116
        else
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   117
          (false, lambdas');
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   118
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   119
      (*list optional terminal and all nonterminals on which the lookahead
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   120
        of a production depends*)
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   121
      fun lookahead_dependency _ [] nts = (NONE, nts)
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   122
        | lookahead_dependency _ ((Terminal tk) :: _) nts = (SOME tk, nts)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   123
        | lookahead_dependency lambdas ((Nonterminal (nt, _)) :: symbs) nts =
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   124
            if member (op =) lambdas nt then
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   125
              lookahead_dependency lambdas symbs (nt :: nts)
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   126
            else (NONE, nt :: nts);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   127
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   128
      (*get all known starting tokens for a nonterminal*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   129
      fun starts_for_nt nt = snd (fst (Array.sub (prods, nt)));
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   130
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   131
      val token_union = gen_union matching_tokens;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   132
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   133
      (*update prods, lookaheads, and lambdas according to new lambda NTs*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   134
      val (added_starts', lambdas') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   135
        let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   136
          (*propagate added lambda NT*)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   137
          fun propagate_lambda [] added_starts lambdas= (added_starts, lambdas)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   138
            | propagate_lambda (l :: ls) added_starts lambdas =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   139
              let
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   140
                (*get lookahead for lambda NT*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   141
                val ((dependent, l_starts), _) = Array.sub (prods, l);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   142
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   143
                (*check productions whose lookahead may depend on lambda NT*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   144
                fun examine_prods [] add_lambda nt_dependencies added_tks
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   145
                                  nt_prods =
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   146
                      (add_lambda, nt_dependencies, added_tks, nt_prods)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   147
                  | examine_prods ((p as (rhs, _, _)) :: ps) add_lambda
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   148
                      nt_dependencies added_tks nt_prods =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   149
                    let val (tk, nts) = lookahead_dependency lambdas rhs [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   150
                    in
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   151
                      if member (op =) nts l then       (*update production's lookahead*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   152
                      let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   153
                        val new_lambda = is_none tk andalso nts subset lambdas;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   154
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   155
                        val new_tks = subtract (op =) l_starts
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   156
                          ((if is_some tk then [the tk] else []) @
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   157
                            Library.foldl token_union ([], map starts_for_nt nts));
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   158
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   159
                        val added_tks' = token_union (new_tks, added_tks);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   160
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   161
                        val nt_dependencies' = gen_union (op =) (nts, nt_dependencies);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   162
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   163
                        (*associate production with new starting tokens*)
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   164
                        fun copy ([]: token option list) nt_prods = nt_prods
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   165
                          | copy (tk :: tks) nt_prods =
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   166
                            let val old_prods = these (AList.lookup (op =) nt_prods tk);
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   167
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   168
                                val prods' = p :: old_prods;
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   169
                            in nt_prods
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   170
                               |> AList.update (op =) (tk, prods')
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   171
                               |> copy tks
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   172
                            end;
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   173
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   174
                        val nt_prods' =
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   175
                          let val new_opt_tks = map SOME new_tks;
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   176
                          in copy ((if new_lambda then [NONE] else []) @
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   177
                                   new_opt_tks) nt_prods
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   178
                          end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   179
                      in examine_prods ps (add_lambda orelse new_lambda)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   180
                           nt_dependencies' added_tks' nt_prods'
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   181
                      end
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   182
                      else                                  (*skip production*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   183
                        examine_prods ps add_lambda nt_dependencies
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   184
                                      added_tks nt_prods
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   185
                    end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   186
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   187
                (*check each NT whose lookahead depends on new lambda NT*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   188
                fun process_nts [] added_lambdas added_starts =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   189
                      (added_lambdas, added_starts)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   190
                  | process_nts (nt :: nts) added_lambdas added_starts =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   191
                    let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   192
                      val (lookahead as (old_nts, old_tks), nt_prods) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   193
                        Array.sub (prods, nt);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   194
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   195
                      (*existing productions whose lookahead may depend on l*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   196
                      val tk_prods =
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   197
                        (these o AList.lookup (op =) nt_prods)
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   198
                               (SOME (hd l_starts  handle Empty => UnknownStart));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   199
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   200
                      (*add_lambda is true if an existing production of the nt
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   201
                        produces lambda due to the new lambda NT l*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   202
                      val (add_lambda, nt_dependencies, added_tks, nt_prods') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   203
                        examine_prods tk_prods false [] [] nt_prods;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   204
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   205
                      val added_nts = subtract (op =) old_nts nt_dependencies;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   206
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   207
                      val added_lambdas' =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   208
                        if add_lambda then nt :: added_lambdas
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   209
                        else added_lambdas;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   210
                    in Array.update (prods, nt,
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   211
                                   ((added_nts @ old_nts, old_tks @ added_tks),
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   212
                                    nt_prods'));
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   213
                                          (*N.B. that because the tks component
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   214
                                            is used to access existing
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   215
                                            productions we have to add new
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   216
                                            tokens at the _end_ of the list*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   217
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   218
                       if null added_tks then
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   219
                         process_nts nts added_lambdas' added_starts
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   220
                       else
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   221
                         process_nts nts added_lambdas'
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   222
                                      ((nt, added_tks) :: added_starts)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   223
                    end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   224
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   225
                val (added_lambdas, added_starts') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   226
                  process_nts dependent [] added_starts;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   227
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   228
                val added_lambdas' = subtract (op =) lambdas added_lambdas;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   229
              in propagate_lambda (ls @ added_lambdas') added_starts'
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   230
                                  (added_lambdas' @ lambdas)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   231
              end;
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   232
        in propagate_lambda (subtract (op =) lambdas lambdas') added_starts lambdas' end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   233
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   234
      (*insert production into grammar*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   235
      val (added_starts', prod_count') =
15979
wenzelm
parents: 15752
diff changeset
   236
        if is_some chain_from then (added_starts', prod_count)  (*don't store chain production*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   237
        else let
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   238
          (*lookahead tokens of new production and on which
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   239
            NTs lookahead depends*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   240
          val (start_tk, start_nts) = lookahead_dependency lambdas' rhs [];
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   241
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   242
          val start_tks = Library.foldl token_union
15979
wenzelm
parents: 15752
diff changeset
   243
                          (if is_some start_tk then [the start_tk] else [],
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   244
                           map starts_for_nt start_nts);
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   245
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   246
          val opt_starts = (if new_lambda then [NONE]
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   247
                            else if null start_tks then [SOME UnknownStart]
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   248
                            else []) @ (map SOME start_tks);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   249
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   250
          (*add lhs NT to list of dependent NTs in lookahead*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   251
          fun add_nts [] = ()
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   252
            | add_nts (nt :: nts) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   253
              let val ((old_nts, old_tks), ps) = Array.sub (prods, nt);
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   254
              in if member (op =) old_nts lhs then ()
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   255
                 else Array.update (prods, nt, ((lhs :: old_nts, old_tks), ps))
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   256
              end;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   257
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   258
          (*add new start tokens to chained NTs' lookahead list;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   259
            also store new production for lhs NT*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   260
          fun add_tks [] added prod_count = (added, prod_count)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   261
            | add_tks (nt :: nts) added prod_count =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   262
              let
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   263
                val ((old_nts, old_tks), nt_prods) = Array.sub (prods, nt);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   264
20951
868120282837 gen_rem(s) abandoned in favour of remove / subtract
haftmann
parents: 20854
diff changeset
   265
                val new_tks = subtract matching_tokens old_tks start_tks;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   266
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   267
                (*store new production*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   268
                fun store [] prods is_new =
15979
wenzelm
parents: 15752
diff changeset
   269
                      (prods, if is_some prod_count andalso is_new then
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   270
                                Option.map (fn x => x+1) prod_count
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   271
                              else prod_count, is_new)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   272
                  | store (tk :: tks) prods is_new =
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   273
                    let val tk_prods = (these o AList.lookup (op =) prods) tk;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   274
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   275
                        (*if prod_count = NONE then we can assume that
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   276
                          grammar does not contain new production already*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   277
                        val (tk_prods', is_new') =
15979
wenzelm
parents: 15752
diff changeset
   278
                          if is_some prod_count then
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   279
                            if member (op =) tk_prods new_prod then (tk_prods, false)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   280
                            else (new_prod :: tk_prods, true)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   281
                          else (new_prod :: tk_prods, true);
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   282
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   283
                        val prods' = prods
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   284
                          |> is_new' ? AList.update (op =) (tk: token option, tk_prods');
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   285
                    in store tks prods' (is_new orelse is_new') end;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   286
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   287
                val (nt_prods', prod_count', changed) =
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   288
                  if nt = lhs then store opt_starts nt_prods false
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   289
                              else (nt_prods, prod_count, false);
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   290
              in if not changed andalso null new_tks then ()
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   291
                 else Array.update (prods, nt, ((old_nts, old_tks @ new_tks),
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   292
                                                nt_prods'));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   293
                 add_tks nts (if null new_tks then added
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   294
                              else (nt, new_tks) :: added) prod_count'
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   295
              end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   296
        in add_nts start_nts;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   297
           add_tks (connected_with chains' [lhs] [lhs]) [] prod_count
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   298
        end;
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   299
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   300
      (*associate productions with new lookaheads*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   301
      val dummy =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   302
        let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   303
          (*propagate added start tokens*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   304
          fun add_starts [] = ()
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   305
            | add_starts ((changed_nt, new_tks) :: starts) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   306
              let
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   307
                (*token under which old productions which
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   308
                  depend on changed_nt could be stored*)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   309
                val key =
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   310
                 case find_first (not o member (op =) new_tks)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   311
                                 (starts_for_nt changed_nt) of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   312
                      NONE => SOME UnknownStart
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   313
                    | t => t;
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   314
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   315
                (*copy productions whose lookahead depends on changed_nt;
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   316
                  if key = SOME UnknownToken then tk_prods is used to hold
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   317
                  the productions not copied*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   318
                fun update_prods [] result = result
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   319
                  | update_prods ((p as (rhs, _: string, _: nt_tag)) :: ps)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   320
                      (tk_prods, nt_prods) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   321
                    let
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   322
                      (*lookahead dependency for production*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   323
                      val (tk, depends) = lookahead_dependency lambdas' rhs [];
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   324
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   325
                      (*test if this production has to be copied*)
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   326
                      val update = member (op =) depends changed_nt;
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   327
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   328
                      (*test if production could already be associated with
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   329
                        a member of new_tks*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   330
                      val lambda = length depends > 1 orelse
15979
wenzelm
parents: 15752
diff changeset
   331
                                   not (null depends) andalso is_some tk
20664
ffbc5a57191a member (op =);
wenzelm
parents: 19482
diff changeset
   332
                                   andalso member (op =) new_tks (the tk);
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   333
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   334
                      (*associate production with new starting tokens*)
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   335
                      fun copy ([]: token list) nt_prods = nt_prods
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   336
                        | copy (tk :: tks) nt_prods =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   337
                          let
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   338
                            val tk_prods = (these o AList.lookup (op =) nt_prods) (SOME tk);
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   339
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   340
                            val tk_prods' =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   341
                              if not lambda then p :: tk_prods
20854
f9cf9e62d11c insert replacing ins ins_int ins_string
haftmann
parents: 20664
diff changeset
   342
                              else insert (op =) p tk_prods;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   343
                                      (*if production depends on lambda NT we
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   344
                                        have to look for duplicates*)
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   345
                         in
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   346
                           nt_prods
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   347
                           |> AList.update (op =) (SOME tk, tk_prods')
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   348
                           |> copy tks
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   349
                         end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   350
                      val result =
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   351
                        if update then
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   352
                          (tk_prods, copy new_tks nt_prods)
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   353
                        else if key = SOME UnknownStart then
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   354
                          (p :: tk_prods, nt_prods)
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   355
                        else (tk_prods, nt_prods);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   356
                    in update_prods ps result end;
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   357
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   358
                (*copy existing productions for new starting tokens*)
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   359
                fun process_nts [] added = added
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   360
                  | process_nts (nt :: nts) added =
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   361
                    let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   362
                      val (lookahead as (old_nts, old_tks), nt_prods) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   363
                        Array.sub (prods, nt);
377
ab8917806779 lookaheads are now computed faster (during the grammar is built)
clasohm
parents: 373
diff changeset
   364
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   365
                      val tk_prods = (these o AList.lookup (op =) nt_prods) key;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   366
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   367
                      (*associate productions with new lookahead tokens*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   368
                      val (tk_prods', nt_prods') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   369
                        update_prods tk_prods ([], nt_prods);
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   370
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   371
                      val nt_prods' =
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   372
                        nt_prods'
21565
bd28361f4c5b simplified '?' operator;
wenzelm
parents: 20951
diff changeset
   373
                        |> (key = SOME UnknownStart) ? AList.update (op =) (key, tk_prods')
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   374
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   375
                      val added_tks =
20951
868120282837 gen_rem(s) abandoned in favour of remove / subtract
haftmann
parents: 20854
diff changeset
   376
                        subtract matching_tokens old_tks new_tks;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   377
                    in if null added_tks then
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   378
                         (Array.update (prods, nt, (lookahead, nt_prods'));
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   379
                          process_nts nts added)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   380
                       else
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   381
                         (Array.update (prods, nt,
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   382
                            ((old_nts, added_tks @ old_tks), nt_prods'));
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   383
                          process_nts nts ((nt, added_tks) :: added))
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   384
                    end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   385
1175
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   386
                val ((dependent, _), _) = Array.sub (prods, changed_nt);
1b15a4b1a4f7 added comments; fixed a bug; reduced memory usage slightly
clasohm
parents: 1147
diff changeset
   387
              in add_starts (starts @ (process_nts dependent [])) end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   388
        in add_starts added_starts' end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   389
  in add_prods prods chains' lambdas' prod_count ps end;
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   390
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   391
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   392
(* pretty_gram *)
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   393
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   394
fun pretty_gram (Gram {tags, prods, chains, ...}) =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   395
  let
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   396
    fun pretty_name name = [Pretty.str (name ^ " =")];
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   397
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   398
    val nt_name = reverted_tags tags;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   399
14600
ba51bc239716 Replaced quote by Pretty.quote / Library.quote, since quote now refers to Symbol.quote
berghofe
parents: 13057
diff changeset
   400
    fun pretty_symb (Terminal (Token s)) = Pretty.quote (Pretty.str s)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   401
      | pretty_symb (Terminal tok) = Pretty.str (str_of_token tok)
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   402
      | pretty_symb (Nonterminal (tag, p)) = Pretty.str (nt_name tag ^ "[" ^ string_of_int p ^ "]");
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   403
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   404
    fun pretty_const "" = []
14600
ba51bc239716 Replaced quote by Pretty.quote / Library.quote, since quote now refers to Symbol.quote
berghofe
parents: 13057
diff changeset
   405
      | pretty_const c = [Pretty.str ("=> " ^ Library.quote c)];
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   406
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   407
    fun pretty_pri p = [Pretty.str ("(" ^ string_of_int p ^ ")")];
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   408
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   409
    fun pretty_prod name (symbs, const, pri) =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   410
      Pretty.block (Pretty.breaks (pretty_name name @
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   411
        map pretty_symb symbs @ pretty_const const @ pretty_pri pri));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   412
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   413
    fun pretty_nt (name, tag) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   414
      let
22573
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
   415
        fun prod_of_chain from_ = ([Nonterminal (from_, ~1)], "", ~1);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   416
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   417
        val nt_prods =
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   418
          Library.foldl (gen_union op =) ([], map snd (snd (Array.sub (prods, tag)))) @
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   419
          map prod_of_chain ((these o AList.lookup (op =) chains) tag);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   420
      in map (pretty_prod name) nt_prods end;
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   421
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   422
  in maps pretty_nt (sort_wrt fst (Symtab.dest tags)) end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   423
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   424
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   425
(** Operations on gramars **)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   426
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   427
(*The mother of all grammars*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   428
val empty_gram = Gram {nt_count = 0, prod_count = 0,
4487
9b4c1db5aca1 renamed Symtab.null to Symtab.empty;
wenzelm
parents: 2229
diff changeset
   429
                       tags = Symtab.empty, chains = [], lambdas = [],
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   430
                       prods = Array.array (0, (([], []), []))};
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   431
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   432
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   433
(*Invert list of chain productions*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   434
fun inverse_chains [] result = result
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   435
  | inverse_chains ((root, branches: nt_tag list) :: cs) result =
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   436
    let fun add ([]: nt_tag list) result = result
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   437
          | add (id :: ids) result =
17192
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   438
            let val old = (these o AList.lookup (op =) result) id;
0cfbf76ed313 introduced AList.*
haftmann
parents: 16668
diff changeset
   439
            in add ids (AList.update (op =) (id, root :: old) result) end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   440
    in inverse_chains cs (add branches result) end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   441
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   442
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   443
(*Add productions to a grammar*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   444
fun extend_gram gram [] = gram
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   445
  | extend_gram (Gram {nt_count, prod_count, tags, chains, lambdas, prods})
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   446
                xprods =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   447
  let
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   448
    (*Get tag for existing nonterminal or create a new one*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   449
    fun get_tag nt_count tags nt =
17412
e26cb20ef0cc TableFun/Symtab: curried lookup and update;
wenzelm
parents: 17314
diff changeset
   450
      case Symtab.lookup tags nt of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   451
        SOME tag => (nt_count, tags, tag)
17412
e26cb20ef0cc TableFun/Symtab: curried lookup and update;
wenzelm
parents: 17314
diff changeset
   452
      | NONE => (nt_count+1, Symtab.update_new (nt, nt_count) tags,
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   453
                 nt_count);
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   454
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   455
    (*Convert symbols to the form used by the parser;
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   456
      delimiters and predefined terms are stored as terminals,
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   457
      nonterminals are converted to integer tags*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   458
    fun symb_of [] nt_count tags result = (nt_count, tags, rev result)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   459
      | symb_of ((Delim s) :: ss) nt_count tags result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   460
          symb_of ss nt_count tags ((Terminal (Token s)) :: result)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   461
      | symb_of ((Argument (s, p)) :: ss) nt_count tags result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   462
          let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   463
            val (nt_count', tags', new_symb) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   464
              case predef_term s of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   465
                NONE =>
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   466
                  let val (nt_count', tags', s_tag) = get_tag nt_count tags s;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   467
                  in (nt_count', tags', Nonterminal (s_tag, p)) end
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   468
              | SOME tk => (nt_count, tags, Terminal tk);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   469
          in symb_of ss nt_count' tags' (new_symb :: result) end
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   470
      | symb_of (_ :: ss) nt_count tags result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   471
          symb_of ss nt_count tags result;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   472
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   473
    (*Convert list of productions by invoking symb_of for each of them*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   474
    fun prod_of [] nt_count prod_count tags result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   475
          (nt_count, prod_count, tags, result)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   476
      | prod_of ((XProd (lhs, xsymbs, const, pri)) :: ps)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   477
                nt_count prod_count tags result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   478
        let val (nt_count', tags', lhs_tag) = get_tag nt_count tags lhs;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   479
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   480
            val (nt_count'', tags'', prods) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   481
              symb_of xsymbs nt_count' tags' [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   482
        in prod_of ps nt_count'' (prod_count+1) tags''
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   483
                   ((lhs_tag, (prods, const, pri)) :: result)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   484
        end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   485
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   486
    val (nt_count', prod_count', tags', xprods') =
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   487
      prod_of xprods nt_count prod_count tags [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   488
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   489
    (*Copy array containing productions of old grammar;
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   490
      this has to be done to preserve the old grammar while being able
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   491
      to change the array's content*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   492
    val prods' =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   493
      let fun get_prod i = if i < nt_count then Array.sub (prods, i)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   494
                           else (([], []), []);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   495
      in Array.tabulate (nt_count', get_prod) end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   496
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   497
    val fromto_chains = inverse_chains chains [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   498
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   499
    (*Add new productions to old ones*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   500
    val (fromto_chains', lambdas', _) =
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   501
      add_prods prods' fromto_chains lambdas NONE xprods';
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   502
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   503
    val chains' = inverse_chains fromto_chains' [];
1177
58e4d9221db7 removed debugging output
clasohm
parents: 1175
diff changeset
   504
  in Gram {nt_count = nt_count', prod_count = prod_count', tags = tags',
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   505
           chains = chains', lambdas = lambdas', prods = prods'}
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   506
  end;
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   507
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   508
val make_gram = extend_gram empty_gram;
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   509
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   510
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   511
(*Merge two grammars*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   512
fun merge_grams gram_a gram_b =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   513
  let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   514
    (*find out which grammar is bigger*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   515
    val (Gram {nt_count = nt_count1, prod_count = prod_count1, tags = tags1,
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   516
               chains = chains1, lambdas = lambdas1, prods = prods1},
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   517
         Gram {nt_count = nt_count2, prod_count = prod_count2, tags = tags2,
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   518
               chains = chains2, lambdas = lambdas2, prods = prods2}) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   519
      let val Gram {prod_count = count_a, ...} = gram_a;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   520
          val Gram {prod_count = count_b, ...} = gram_b;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   521
      in if count_a > count_b then (gram_a, gram_b)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   522
                              else (gram_b, gram_a)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   523
      end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   524
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   525
    (*get existing tag from grammar1 or create a new one*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   526
    fun get_tag nt_count tags nt =
17412
e26cb20ef0cc TableFun/Symtab: curried lookup and update;
wenzelm
parents: 17314
diff changeset
   527
      case Symtab.lookup tags nt of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   528
        SOME tag => (nt_count, tags, tag)
17412
e26cb20ef0cc TableFun/Symtab: curried lookup and update;
wenzelm
parents: 17314
diff changeset
   529
      | NONE => (nt_count+1, Symtab.update_new (nt, nt_count) tags,
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   530
                nt_count)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   531
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   532
    val ((nt_count1', tags1'), tag_table) =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   533
      let val tag_list = Symtab.dest tags2;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   534
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   535
          val table = Array.array (nt_count2, ~1);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   536
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   537
          fun store_tag nt_count tags ~1 = (nt_count, tags)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   538
            | store_tag nt_count tags tag =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   539
              let val (nt_count', tags', tag') =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   540
                   get_tag nt_count tags
15979
wenzelm
parents: 15752
diff changeset
   541
                     (fst (the (find_first (fn (n, t) => t = tag) tag_list)));
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   542
              in Array.update (table, tag, tag');
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   543
                 store_tag nt_count' tags' (tag-1)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   544
              end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   545
      in (store_tag nt_count1 tags1 (nt_count2-1), table) end;
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   546
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   547
    (*convert grammar2 tag to grammar1 tag*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   548
    fun convert_tag tag = Array.sub (tag_table, tag);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   549
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   550
    (*convert chain list to raw productions*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   551
    fun mk_chain_prods [] result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   552
      | mk_chain_prods ((to, froms) :: cs) result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   553
        let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   554
          val to_tag = convert_tag to;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   555
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   556
          fun make [] result = result
22573
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
   557
            | make (from_ :: froms) result = make froms ((to_tag,
2ac646ab2f6c avoid clash with Alice keywords;
wenzelm
parents: 21565
diff changeset
   558
                ([Nonterminal (convert_tag from_, ~1)], "", ~1)) :: result);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   559
        in mk_chain_prods cs (make froms [] @ result) end;
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   560
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   561
    val chain_prods = mk_chain_prods chains2 [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   562
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   563
    (*convert prods2 array to productions*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   564
    fun process_nt ~1 result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   565
      | process_nt nt result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   566
        let
23909
6e4fba2ea7d0 avoid polymorphic equality;
wenzelm
parents: 22573
diff changeset
   567
          val nt_prods = Library.foldl (gen_union op =)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   568
                             ([], map snd (snd (Array.sub (prods2, nt))));
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   569
          val lhs_tag = convert_tag nt;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   570
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   571
          (*convert tags in rhs*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   572
          fun process_rhs [] result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   573
            | process_rhs (Terminal tk :: rhs) result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   574
                process_rhs rhs (result @ [Terminal tk])
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   575
            | process_rhs (Nonterminal (nt, prec) :: rhs) result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   576
                process_rhs rhs
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   577
                            (result @ [Nonterminal (convert_tag nt, prec)]);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   578
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   579
          (*convert tags in productions*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   580
          fun process_prods [] result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   581
            | process_prods ((rhs, id, prec) :: ps) result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   582
                process_prods ps ((lhs_tag, (process_rhs rhs [], id, prec))
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   583
                                  :: result);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   584
        in process_nt (nt-1) (process_prods nt_prods [] @ result) end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   585
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   586
    val raw_prods = chain_prods @ process_nt (nt_count2-1) [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   587
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   588
    val prods1' =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   589
      let fun get_prod i = if i < nt_count1 then Array.sub (prods1, i)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   590
                           else (([], []), []);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   591
      in Array.tabulate (nt_count1', get_prod) end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   592
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   593
    val fromto_chains = inverse_chains chains1 [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   594
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   595
    val (fromto_chains', lambdas', SOME prod_count1') =
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   596
      add_prods prods1' fromto_chains lambdas1 (SOME prod_count1) raw_prods;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   597
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   598
    val chains' = inverse_chains fromto_chains' [];
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   599
  in Gram {nt_count = nt_count1', prod_count = prod_count1',
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   600
           tags = tags1', chains = chains', lambdas = lambdas',
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   601
           prods = prods1'}
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   602
  end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   603
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   604
1438
3cc22794ce71 added comments
clasohm
parents: 1177
diff changeset
   605
(** Parser **)
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   606
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   607
datatype parsetree =
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   608
  Node of string * parsetree list |
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   609
  Tip of token;
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   610
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   611
type state =
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   612
  nt_tag * int *                (*identification and production precedence*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   613
  parsetree list *              (*already parsed nonterminals on rhs*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   614
  symb list *                   (*rest of rhs*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   615
  string *                      (*name of production*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   616
  int;                          (*index for previous state list*)
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   617
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   618
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   619
(*Get all rhss with precedence >= minPrec*)
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   620
fun getRHS minPrec = List.filter (fn (_, _, prec:int) => prec >= minPrec);
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   621
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   622
(*Get all rhss with precedence >= minPrec and < maxPrec*)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   623
fun getRHS' minPrec maxPrec =
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   624
  List.filter (fn (_, _, prec:int) => prec >= minPrec andalso prec < maxPrec);
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   625
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   626
(*Make states using a list of rhss*)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   627
fun mkStates i minPrec lhsID rhss =
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   628
  let fun mkState (rhs, id, prodPrec) = (lhsID, prodPrec, [], rhs, id, i);
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   629
  in map mkState rhss end;
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   630
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   631
(*Add parse tree to list and eliminate duplicates
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   632
  saving the maximum precedence*)
16668
fdb4992cf1d2 avoid polyeq;
wenzelm
parents: 15979
diff changeset
   633
fun conc (t: parsetree list, prec:int) [] = (NONE, [(t, prec)])
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   634
  | conc (t, prec) ((t', prec') :: ts) =
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   635
      if t = t' then
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   636
        (SOME prec', if prec' >= prec then (t', prec') :: ts
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   637
                     else (t, prec) :: ts)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   638
      else
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   639
        let val (n, ts') = conc (t, prec) ts
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   640
        in (n, (t', prec') :: ts') end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   641
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   642
(*Update entry in used*)
16668
fdb4992cf1d2 avoid polyeq;
wenzelm
parents: 15979
diff changeset
   643
fun update_trees ((B: nt_tag, (i, ts)) :: used) (A, t) =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   644
  if A = B then
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   645
    let val (n, ts') = conc t ts
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   646
    in ((A, (i, ts')) :: used, n) end
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   647
  else
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   648
    let val (used', n) = update_trees used (A, t)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   649
    in ((B, (i, ts)) :: used', n) end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   650
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   651
(*Replace entry in used*)
16668
fdb4992cf1d2 avoid polyeq;
wenzelm
parents: 15979
diff changeset
   652
fun update_prec (A: nt_tag, prec) used =
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   653
  let fun update ((hd as (B, (_, ts))) :: used, used') =
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   654
        if A = B
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   655
        then used' @ ((A, (prec, ts)) :: used)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   656
        else update (used, hd :: used')
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   657
  in update (used, []) end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   658
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   659
fun getS A maxPrec Si =
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   660
  List.filter
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   661
    (fn (_, _, _, Nonterminal (B, prec) :: _, _, _)
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   662
          => A = B andalso prec <= maxPrec
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   663
      | _ => false) Si;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   664
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   665
fun getS' A maxPrec minPrec Si =
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   666
  List.filter
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   667
    (fn (_, _, _, Nonterminal (B, prec) :: _, _, _)
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   668
          => A = B andalso prec > minPrec andalso prec <= maxPrec
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   669
      | _ => false) Si;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   670
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   671
fun getStates Estate i ii A maxPrec =
15570
8d8c70b41bab Move towards standard functions.
skalberg
parents: 15531
diff changeset
   672
  List.filter
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   673
    (fn (_, _, _, Nonterminal (B, prec) :: _, _, _)
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   674
          => A = B andalso prec <= maxPrec
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   675
      | _ => false)
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   676
    (Array.sub (Estate, ii));
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   677
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   678
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   679
fun movedot_term (A, j, ts, Terminal a :: sa, id, i) c =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   680
  if valued_token c then
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   681
    (A, j, ts @ [Tip c], sa, id, i)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   682
  else (A, j, ts, sa, id, i);
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   683
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   684
fun movedot_nonterm ts (A, j, tss, Nonterminal _ :: sa, id, i) =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   685
  (A, j, tss @ ts, sa, id, i);
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   686
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   687
fun movedot_lambda _ [] = []
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   688
  | movedot_lambda (B, j, tss, Nonterminal (A, k) :: sa, id, i) ((t, ki) :: ts) =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   689
      if k <= ki then
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   690
        (B, j, tss @ t, sa, id, i) ::
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   691
          movedot_lambda (B, j, tss, Nonterminal (A, k) :: sa, id, i) ts
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   692
      else movedot_lambda (B, j, tss, Nonterminal (A, k) :: sa, id, i) ts;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   693
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   694
13057
805de10ca485 increased treshold for "this expression could be extremely ambigous" warning
kleing
parents: 12785
diff changeset
   695
val branching_level = ref 600;                   (*trigger value for warnings*)
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   696
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   697
(*get all productions of a NT and NTs chained to it which can
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   698
  be started by specified token*)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   699
fun prods_for prods chains include_none tk nts =
15979
wenzelm
parents: 15752
diff changeset
   700
let
wenzelm
parents: 15752
diff changeset
   701
    fun token_assoc (list, key) =
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   702
      let fun assoc [] result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   703
            | assoc ((keyi, pi) :: pairs) result =
15979
wenzelm
parents: 15752
diff changeset
   704
                if is_some keyi andalso matching_tokens (the keyi, key)
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   705
                   orelse include_none andalso is_none keyi then
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   706
                  assoc pairs (pi @ result)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   707
                else assoc pairs result;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   708
      in assoc list [] end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   709
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   710
    fun get_prods [] result = result
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   711
      | get_prods (nt :: nts) result =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   712
        let val nt_prods = snd (Array.sub (prods, nt));
15979
wenzelm
parents: 15752
diff changeset
   713
        in get_prods nts ((token_assoc (nt_prods, tk)) @ result) end;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   714
in get_prods (connected_with chains nts nts) [] end;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   715
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   716
23941
6234185a2e2e avoid global reference warned'';
wenzelm
parents: 23909
diff changeset
   717
fun PROCESSS warned prods chains Estate i c states =
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   718
let
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   719
fun all_prods_for nt = prods_for prods chains true c [nt];
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   720
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   721
fun processS used [] (Si, Sii) = (Si, Sii)
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   722
  | processS used (S :: States) (Si, Sii) =
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   723
      (case S of
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   724
        (_, _, _, Nonterminal (nt, minPrec) :: _, _, _) =>
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   725
          let                                       (*predictor operation*)
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   726
            val (used', new_states) =
17314
04e21a27c0ad introduces some modern-style AList operations
haftmann
parents: 17221
diff changeset
   727
              (case AList.lookup (op =) used nt of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   728
                SOME (usedPrec, l) =>       (*nonterminal has been processed*)
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   729
                  if usedPrec <= minPrec then
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   730
                                      (*wanted precedence has been processed*)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   731
                    (used, movedot_lambda S l)
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   732
                  else            (*wanted precedence hasn't been parsed yet*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   733
                    let
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   734
                      val tk_prods = all_prods_for nt;
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   735
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   736
                      val States' = mkStates i minPrec nt
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   737
                                      (getRHS' minPrec usedPrec tk_prods);
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   738
                    in (update_prec (nt, minPrec) used,
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   739
                        movedot_lambda S l @ States')
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   740
                    end
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   741
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   742
              | NONE =>           (*nonterminal is parsed for the first time*)
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   743
                  let val tk_prods = all_prods_for nt;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   744
                      val States' = mkStates i minPrec nt
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   745
                                      (getRHS minPrec tk_prods);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   746
                  in ((nt, (minPrec, [])) :: used, States') end);
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   747
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   748
            val dummy =
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   749
              if not (!warned) andalso
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   750
                 length (new_states @ States) > (!branching_level) then
19305
5c16895d548b avoid polymorphic equality;
wenzelm
parents: 19046
diff changeset
   751
                (warning "Currently parsed expression could be extremely ambiguous.";
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   752
                 warned := true)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   753
              else ();
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   754
          in
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   755
            processS used' (new_states @ States) (S :: Si, Sii)
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   756
          end
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   757
      | (_, _, _, Terminal a :: _, _, _) =>               (*scanner operation*)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   758
          processS used States
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   759
            (S :: Si,
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   760
              if matching_tokens (a, c) then movedot_term S c :: Sii else Sii)
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   761
      | (A, prec, ts, [], id, j) =>                   (*completer operation*)
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   762
          let val tt = if id = "" then ts else [Node (id, ts)] in
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   763
            if j = i then                             (*lambda production?*)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   764
              let
697
40f72ab196f8 changed warning for extremely ambiguous expressions
clasohm
parents: 682
diff changeset
   765
                val (used', O) = update_trees used (A, (tt, prec));
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   766
              in
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   767
                case O of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   768
                  NONE =>
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   769
                    let val Slist = getS A prec Si;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   770
                        val States' = map (movedot_nonterm tt) Slist;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   771
                    in processS used' (States' @ States) (S :: Si, Sii) end
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   772
                | SOME n =>
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   773
                    if n >= prec then processS used' States (S :: Si, Sii)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   774
                    else
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   775
                      let val Slist = getS' A prec n Si;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   776
                          val States' = map (movedot_nonterm tt) Slist;
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   777
                      in processS used' (States' @ States) (S :: Si, Sii) end
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   778
              end
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   779
            else
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   780
              let val Slist = getStates Estate i j A prec
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   781
              in processS used (map (movedot_nonterm tt) Slist @ States)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   782
                          (S :: Si, Sii)
682
c36f49c76d22 added warning message
clasohm
parents: 624
diff changeset
   783
              end
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   784
          end)
682
c36f49c76d22 added warning message
clasohm
parents: 624
diff changeset
   785
in processS [] states ([], []) end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   786
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   787
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   788
fun produce warned prods tags chains stateset i indata prev_token =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   789
  (case Array.sub (stateset, i) of
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   790
    [] =>
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   791
      let
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   792
        val toks =
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   793
          if prev_token = EndToken then indata
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   794
          else prev_token :: indata;
373
68400ea32f7b fixed a bug in syntax_error, added "Building new grammar" message;
clasohm
parents: 372
diff changeset
   795
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   796
        val nt_name = reverted_tags tags;
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   797
        val nts =
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   798
          fold (fn (_, _, _, Nonterminal nt :: _, _, _) => insert (op =) nt | _ => I)
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   799
            (Array.sub (stateset, i - 1)) []
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   800
          |> map (fn (a, prec) => nt_name a ^ "(" ^ signed_string_of_int prec ^ ")");
373
68400ea32f7b fixed a bug in syntax_error, added "Building new grammar" message;
clasohm
parents: 372
diff changeset
   801
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   802
        val msg =
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   803
          (if null toks then Pretty.str "Inner syntax error: unexpected end of input"
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   804
           else
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   805
             Pretty.block (Pretty.str "Inner syntax error at: \"" ::
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   806
               Pretty.breaks (map (Pretty.str o str_of_token) (#1 (split_last toks))) @
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   807
               [Pretty.str "\""])) ::
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   808
          (if null nts then []
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   809
           else [Pretty.block
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   810
            (Pretty.str "Expected syntax categories: " :: Pretty.commas (map Pretty.str nts))]);
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   811
      in error (Pretty.string_of (Pretty.chunks msg)) end
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   812
  | s =>
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   813
    (case indata of
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   814
       [] => Array.sub (stateset, i)
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   815
     | c :: cs =>
23941
6234185a2e2e avoid global reference warned'';
wenzelm
parents: 23909
diff changeset
   816
       let val (si, sii) = PROCESSS warned prods chains stateset i c s;
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   817
       in Array.update (stateset, i, si);
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   818
          Array.update (stateset, i + 1, sii);
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   819
          produce warned prods tags chains stateset (i + 1) cs c
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   820
       end));
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   821
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   822
19482
9f11af8f7ef9 tuned basic list operators (flat, maps, map_filter);
wenzelm
parents: 19305
diff changeset
   823
fun get_trees l = map_filter (fn (_, _, [pt], _, _, _) => SOME pt | _ => NONE)
2229
64acb485ecce Eta-expansion of a function definition, for value polymorphism
paulson
parents: 2186
diff changeset
   824
                            l;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   825
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   826
fun earley prods tags chains startsymbol indata =
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   827
  let
17412
e26cb20ef0cc TableFun/Symtab: curried lookup and update;
wenzelm
parents: 17314
diff changeset
   828
    val start_tag = case Symtab.lookup tags startsymbol of
15531
08c8dad8e399 Deleted Library.option type.
skalberg
parents: 14981
diff changeset
   829
                       SOME tag => tag
15752
8cdc6249044b added make_gram;
wenzelm
parents: 15570
diff changeset
   830
                     | NONE   => error ("parse: Unknown startsymbol " ^
624
33b9b5da3e6f made major changes to grammar;
clasohm
parents: 552
diff changeset
   831
                                        quote startsymbol);
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   832
    val S0 = [(~1, 0, [], [Nonterminal (start_tag, 0), Terminal EndToken],
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   833
               "", 0)];
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   834
    val s = length indata + 1;
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   835
    val Estate = Array.array (s, []);
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   836
  in
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   837
    Array.update (Estate, 0, S0);
25986
26f1e4c172c3 syntax error: reduced spam -- print expected nonterminals instead of terminals;
wenzelm
parents: 24245
diff changeset
   838
    get_trees (produce (ref false) prods tags chains Estate 0 indata EndToken)
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   839
  end;
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   840
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   841
1147
57b5f55bf879 removed 'raw' productions from gram datatype; replaced mk_gram by add_prods;
clasohm
parents: 890
diff changeset
   842
fun parse (Gram {tags, prods, chains, ...}) start toks =
624
33b9b5da3e6f made major changes to grammar;
clasohm
parents: 552
diff changeset
   843
let val r =
24245
4ffeb1dd048a Lexicon.tokenize: do not appen EndToken yet;
wenzelm
parents: 23941
diff changeset
   844
  (case earley prods tags chains start (toks @ [Lexicon.EndToken]) of
237
a7d3e712767a MAJOR INTERNAL CHANGE: extend and merge operations of syntax tables
wenzelm
parents: 46
diff changeset
   845
    [] => sys_error "parse: no parse trees"
330
2fda15dd1e0f changed the way a grammar is generated to allow the new parser to work;
clasohm
parents: 258
diff changeset
   846
  | pts => pts);
624
33b9b5da3e6f made major changes to grammar;
clasohm
parents: 552
diff changeset
   847
in r end
18
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   848
c9ec452ff08f lots of internal cleaning and tuning;
wenzelm
parents:
diff changeset
   849
end;