yacc-parser.mly 8.74 KB
Newer Older
1 2
/* This is the crude version of the parser. It is meant to be processed
   by ocamlyacc. Its existence is necessary for bootstrapping. It is kept
3 4 5 6
   in sync with [fancy-parser], with a few differences:
   1. [fancy-parser] exploits many features of Menhir;
   2. [fancy-parser] performs slightly more refined error handling;
   3. [fancy-parser] supports anonymous rules. */
7 8 9 10 11 12 13 14 15 16 17 18 19 20

%{

open ConcreteSyntax
open Syntax
open Positions

%}

%token TOKEN TYPE LEFT RIGHT NONASSOC START PREC PUBLIC COLON BAR EOF EQUAL 
%token INLINE LPAREN RPAREN COMMA QUESTION STAR PLUS PARAMETER
%token <string Positions.located> LID UID 
%token <Stretch.t> HEADER
%token <Stretch.ocamltype> OCAMLTYPE
21
%token <Stretch.t Lazy.t> PERCENTPERCENT
22
%token <Syntax.identifier option array -> Syntax.action> ACTION
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
%start grammar
%type <ConcreteSyntax.grammar> grammar

/* These declarations solve a shift-reduce conflict in favor of
   shifting: when the declaration of a non-terminal symbol begins with
   a leading bar, it is understood as an (insignificant) leading
   optional bar, *not* as an empty right-hand side followed by a bar.
   This ambiguity arises due to the existence of a new notation for
   letting several productions share a single semantic action. */

%nonassoc no_optional_bar
%nonassoc BAR

%%

/* ------------------------------------------------------------------------- */
/* A grammar consists of declarations and rules, followed by an optional
   trailer, which we do not parse. */

grammar:
  declarations PERCENTPERCENT rules trailer
    { 
      { 
	pg_filename          = ""; (* filled in by the caller *)
	pg_declarations      = List.rev $1;
	pg_rules	     = $3;
	pg_trailer           = $4 
      }
    }

trailer:
  EOF
    { None }
| PERCENTPERCENT /* followed by actual trailer */
    { Some (Lazy.force $1) }

/* ------------------------------------------------------------------------- */
/* A declaration is an %{ Objective Caml header %}, or a %token, %start,
   %type, %left, %right, or %nonassoc declaration. */

declarations:
  /* epsilon */
    { [] }
| declarations declaration
    { $2 @ $1 }

declaration:
| HEADER /* lexically delimited by %{ ... %} */
    { [ unknown_pos (DCode $1) ] }

| TOKEN optional_ocamltype terminals
    { List.map (Positions.map (fun terminal -> DToken ($2, terminal))) $3 }

| START nonterminals
    { List.map (Positions.map (fun nonterminal -> DStart nonterminal)) $2 }

79
| TYPE OCAMLTYPE actuals
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
    { List.map (Positions.map (fun nt -> DType ($2, nt)))
        (List.map Parameters.with_pos $3) }

| START OCAMLTYPE nonterminals
    /* %start <ocamltype> foo is syntactic sugar for %start foo %type <ocamltype> foo */
    { Misc.mapd (fun ntloc -> 
        Positions.mapd (fun nt -> DStart nt, DType ($2, ParameterVar ntloc)) ntloc) $3 }

| priority_keyword symbols
    { let prec = ParserAux.current_token_precedence (rhs_start_pos 1) (rhs_end_pos 1) in
      List.map (Positions.map (fun symbol -> DTokenProperties (symbol, $1, prec))) $2 }

| PARAMETER OCAMLTYPE
    { [ unknown_pos (DParameter $2) ] }

optional_ocamltype:
  /* epsilon */
    { None }
| OCAMLTYPE /* lexically delimited by angle brackets */
    { Some $1 }

priority_keyword:
  LEFT
    { LeftAssoc }
| RIGHT
    { RightAssoc }
| NONASSOC
    { NonAssoc }

/* ------------------------------------------------------------------------- */
/* A symbol is a terminal or nonterminal symbol. One would like to
   require nonterminal symbols to begin with a lowercase letter, so as
   to lexically distinguish them from terminal symbols, which must
   begin with an uppercase letter. However, for compatibility with
   ocamlyacc, this is impossible. It can be required only for
   nonterminal symbols that are also start symbols. */

symbols:
  /* epsilon */
    { [] }
| symbols optional_comma symbol
    { $3 :: $1 }

symbol:
  LID
    { $1 }
| UID
    { $1 }

optional_comma:
  /* epsilon */
    { () }
| COMMA
    { () }

/* ------------------------------------------------------------------------- */
/* Terminals must begin with an uppercase letter. Nonterminals that are
   declared to be start symbols must begin with a lowercase letter. */

terminals:
  /* epsilon */
    { [] }
| terminals optional_comma UID
    { $3 :: $1 }

nonterminals:
  /* epsilon */
    { [] }
| nonterminals LID
    { $2 :: $1 }

/* ------------------------------------------------------------------------- */
/* A rule defines a symbol. It is optionally declared %public, and optionally
   carries a number of formal parameters. The right-hand side of the definition
   consists of a list of production groups. */

rules:
  /* epsilon */
    { [] }
| rules rule
    { $2 :: $1 }

rule:
  flags
  symbol
  optional_formal_parameters
166
  COLON
167 168
  optional_bar
  production_group production_groups
169
    {
170 171 172 173 174 175
      let public, inline = $1 in
      { pr_public_flag = public; 
	pr_inline_flag = inline;
	pr_nt          = Positions.value $2;
	pr_positions   = [ Positions.position $2 ];
	pr_parameters  = $3;
176
	pr_branches    = List.flatten ($6 :: List.rev $7)
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
      }
    }

flags:
  /* epsilon */
    { false, false }
| PUBLIC
    { true, false }
| INLINE
    { false, true }
| PUBLIC INLINE
    { true, true }
| INLINE PUBLIC
    { true, true }

/* ------------------------------------------------------------------------- */
/* Parameters are surroundered with parentheses and delimited by commas.
   The syntax of actual parameters allows applications, whereas the syntax
   of formal parameters does not. It also allows use of the "?", "+", and
   "*" shortcuts. */

optional_formal_parameters:
  /* epsilon */
    { [] }
| LPAREN formal_parameters RPAREN
    { $2 }

formal_parameters:
  symbol
    { [ Positions.value $1 ] }
| symbol COMMA formal_parameters
    { Positions.value $1 :: $3 }

210
optional_actuals:
211 212
  /* epsilon */
    { [] }
213
| LPAREN actuals_comma RPAREN
214 215
    { $2 }

216 217
actuals_comma:
  actual 
218
    { [ $1 ] }
219
| actual COMMA actuals_comma
220 221
    { $1 :: $3 }

222 223
actual:
  symbol optional_actuals
224
    { Parameters.app $1 $2 }
225
| actual modifier
226
    { ParameterApp ($2, [ $1 ]) }
227

228
actuals:
229 230
  /* epsilon */
    { [] }
231
| actuals optional_comma actual
232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
    { $3::$1 }

optional_bar:
  /* epsilon */ %prec no_optional_bar
    { () }
| BAR
    { () }

/* ------------------------------------------------------------------------- */
/* The "?", "+", and "*" modifiers are short-hands for applications of
   certain parameterized nonterminals, defined in the standard library. */

modifier:
  QUESTION
    { unknown_pos "option" }
| PLUS
    { unknown_pos "nonempty_list" }
| STAR
    { unknown_pos "list" }

/* ------------------------------------------------------------------------- */
/* A production group consists of a list of productions, followed by a
   semantic action and an optional precedence specification. */

production_groups:
  /* epsilon */
    { [] }
| production_groups BAR production_group
    { $3 :: $1 }

production_group:
  productions ACTION /* action is lexically delimited by braces */ optional_precedence
264
    {
265
      let productions, action, oprec2 = $1, $2, $3 in
266 267 268 269
      (* If multiple productions share a single semantic action, check
         that all of them bind the same names. *)
      ParserAux.check_production_group productions;
      (* Then, *)
270
      List.map (fun (producers, oprec1, rprec, pos) ->
271 272 273 274 275
        (* Replace [$i] with [_i]. *)
        let pr_producers = ParserAux.normalize_producers producers in
        (* Distribute the semantic action. Also, check that every [$i]
           is within bounds. *)
        let pr_action = action (ParserAux.producer_names producers) in
276
	{
277 278
	  pr_producers;
	  pr_action;
279 280 281
	  pr_branch_shift_precedence  = ParserAux.override pos oprec1 oprec2;
	  pr_branch_reduce_precedence = rprec;
	  pr_branch_position          = pos
282 283
	})
      productions
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
    }

optional_precedence:
  /* epsilon */
    { None }
| PREC symbol
    { Some $2 }

/* ------------------------------------------------------------------------- */
/* A production is a list of producers, optionally followed by a
   precedence declaration. Lists of productions are nonempty and
   separated with bars. */

productions:
  production
    { [ $1 ] }
| production bar_productions
    { $1 :: $2 }

bar_productions:
  BAR production
    { [ $2 ] }
| BAR production bar_productions
    { $2 :: $3 }

production:
  producers optional_precedence
    { List.rev $1,
      $2,
      ParserAux.current_reduce_precedence(),
      Positions.lex_join (symbol_start_pos()) (symbol_end_pos())
    }

producers:
  /* epsilon */ 
    { [] }
| producers producer
    { $2 :: $1 }

/* ------------------------------------------------------------------------- */
/* A producer is an actual parameter, possibly preceded by a
   binding. */

producer:
328
| actual
329
    { Positions.lex_join (symbol_start_pos()) (symbol_end_pos()), None, $1 }
330
| LID EQUAL actual
331
    { Positions.lex_join (symbol_start_pos()) (symbol_end_pos()), Some $1, $3 }
332 333 334

%%