parser.mly 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226
  1. %{
  2. (**
  3. * Parser for the CiviC language.
  4. *
  5. * Note that some shift/reduce conflicts exist, but these need not be solved
  6. * since menhir can automatically resolve them
  7. *)
  8. open Lexing
  9. open Types
  10. let loc start stop = [Loc (Util.loc_from_lexpos start stop)]
  11. let rec make_dims dimloc = function
  12. | [] -> []
  13. | dim :: tail -> Dim (dim, dimloc) :: (make_dims dimloc tail)
  14. let rec make_args = function
  15. | [] -> []
  16. | h::t -> Arg h :: (make_args t)
  17. %}
  18. (* Tokens *)
  19. %token LPAREN RPAREN LBRACK RBRACK LBRACE RBRACE SEMICOL COMMA
  20. %token NOT ADD SUB MUL DIV MOD
  21. %token EQ NE LT LE GT GE
  22. %token AND OR
  23. %token ASSIGN IF ELSE WHILE DO FOR RETURN EXTERN EXPORT
  24. %token INT BOOL FLOAT VOID
  25. %token EOF
  26. %token <bool> BOOL_CONST
  27. %token <float> FLOAT_CONST
  28. %token <int> INT_CONST
  29. %token <string> ID
  30. (* Precedence *)
  31. (*%right ASSIGN*)
  32. %left OR
  33. %left AND
  34. %left EQ NE
  35. %left LT LE GT GE
  36. %left ADD SUB
  37. %left MUL DIV MOD
  38. %right NOT NEG CAST
  39. %nonassoc IF
  40. %nonassoc ELSE
  41. (* Start symbol *)
  42. %type <Types.node> program
  43. %start program
  44. %%
  45. basic_type:
  46. | FLOAT { Float }
  47. | INT { Int }
  48. | BOOL { Bool }
  49. program:
  50. decl*; EOF
  51. { Program ($1, loc $startpos $endpos) }
  52. decl:
  53. (* function: use location of function name *)
  54. | EXTERN; hdr=fun_header; SEMICOL
  55. { let (t, n, p, nameloc) = hdr in
  56. FunDec(t, n, p, nameloc) }
  57. | export=boption(EXPORT); hdr=fun_header; LBRACE; body=fun_body; RBRACE
  58. { let (t, n, p, nameloc) = hdr in
  59. FunDef (export, t, n, p, Block body, nameloc) }
  60. (* global variable declaration: use location of variable name *)
  61. | EXTERN; ctype=basic_type; name=ID; SEMICOL
  62. { GlobalDec (ctype, name, loc $startpos $endpos) }
  63. | EXTERN; ctype=basic_type;
  64. LBRACK; dims=separated_list(COMMA, ID); RBRACK;
  65. name=ID; SEMICOL
  66. { let dimloc = loc $startpos(dims) $endpos(dims) in
  67. let loc = loc $startpos(name) $endpos(name) in
  68. GlobalDec (ArrayDims (ctype, make_dims dimloc dims), name, loc) }
  69. | export=boption(EXPORT); ctype=basic_type; name=ID; SEMICOL
  70. { let loc = loc $startpos(name) $endpos(name) in
  71. GlobalDef (export, ctype, name, None, loc) }
  72. | export=boption(EXPORT); ctype=basic_type; name=ID; ASSIGN; init=expr; SEMICOL
  73. { let loc = loc $startpos(name) $endpos(name) in
  74. GlobalDef (export, ctype, name, Some init, loc) }
  75. | export=boption(EXPORT); ctype=basic_type;
  76. LBRACK; dims=separated_list(COMMA, expr); RBRACK;
  77. name=ID; SEMICOL
  78. { let loc = loc $startpos(name) $endpos(name) in
  79. GlobalDef (export, ArrayDims (ctype, dims), name, None, loc) }
  80. fun_header:
  81. (* function header: use location of function name *)
  82. | ret=basic_type; name=ID; LPAREN; params=separated_list(COMMA, param); RPAREN
  83. { (ret, name, params, loc $startpos(name) $endpos(name)) }
  84. | VOID; name=ID; LPAREN; params=separated_list(COMMA, param); RPAREN
  85. { (Void, name, params, loc $startpos(name) $endpos(name)) }
  86. param:
  87. (* parameter: use location of parameter name *)
  88. | ctype=basic_type; name=ID
  89. { Param (ctype, name, loc $startpos(name) $endpos(name)) }
  90. | ctype=basic_type; LBRACK; dims=separated_list(COMMA, ID); RBRACK; name=ID
  91. { let dimloc = loc $startpos(dims) $endpos(dims) in
  92. let loc = loc $startpos(name) $endpos(name) in
  93. Param (ArrayDims (ctype, make_dims dimloc dims), name, loc) }
  94. fun_body:
  95. | var_dec* local_fun_dec* statement* loption(return_statement)
  96. { VarDecs $1 :: (LocalFuns $2) :: $3 @ $4 }
  97. return_statement:
  98. (* return statement: use location of return value *)
  99. | RETURN; value=expr; SEMICOL
  100. { [Return (value, loc $startpos(value) $endpos(value))] }
  101. (* function: use location of function name *)
  102. local_fun_dec:
  103. hdr=fun_header; LBRACE; body=fun_body; RBRACE
  104. { let (t, n, p, nameloc) = hdr in
  105. FunDef (false, t, n, p, Block body, nameloc) }
  106. var_dec:
  107. (* variable declaration: use location of variable name *)
  108. | ctype=basic_type; name=ID; SEMICOL
  109. { VarDec (ctype, name, None, loc $startpos(name) $endpos(name)) }
  110. | ctype=basic_type; name=ID; ASSIGN; init=expr; SEMICOL
  111. { VarDec (ctype, name, Some init, loc $startpos(name) $endpos(name)) }
  112. | ctype=basic_type; LBRACK; dims=separated_list(COMMA, expr); RBRACK;
  113. name=ID; SEMICOL
  114. { let loc = loc $startpos(name) $endpos(name) in
  115. VarDec (ArrayDims (ctype, dims), name, None, loc) }
  116. | ctype=basic_type; LBRACK; dims=separated_list(COMMA, expr); RBRACK;
  117. name=ID; ASSIGN; init=expr; SEMICOL
  118. { let loc = loc $startpos(name) $endpos(name) in
  119. VarDec (ArrayDims (ctype, dims), name, Some init, loc) }
  120. statement:
  121. (* assignment: use location of assigned variable name *)
  122. | name=ID; ASSIGN; value=expr; SEMICOL
  123. { Assign (name, None, value, loc $startpos(name) $endpos(name)) }
  124. | name=ID; LBRACK; dims=separated_list(COMMA, expr); brk=RBRACK;
  125. ASSIGN; value=expr; SEMICOL
  126. { Assign (name, Some dims, value, loc $startpos(name) $endpos(brk)) }
  127. | name=ID; LPAREN; args=separated_list(COMMA, expr); RPAREN; SEMICOL
  128. { Expr (FunCall (name, make_args args, loc $startpos(name) $endpos(name))) }
  129. (* if-statements and (do-)while-loops: use location of condition *)
  130. | IF; LPAREN; cond=expr; RPAREN; body=block
  131. { If (cond, Block body, loc $startpos $endpos) } %prec IF
  132. | IF; LPAREN; c=expr; RPAREN; t=block; ELSE; f=block
  133. { IfElse (c, Block t, Block f, loc $startpos(c) $endpos(c)) } %prec ELSE
  134. | WHILE; LPAREN; cond=expr; RPAREN; body=block
  135. { While (cond, Block body, loc $startpos(cond) $endpos(cond)) }
  136. | DO; body=block; WHILE; LPAREN; cond=expr; RPAREN; SEMICOL
  137. { DoWhile (cond, Block body, loc $startpos(cond) $endpos(cond)) }
  138. (* for-loop: use location of counter id *)
  139. | FOR; LPAREN; INT; cnt=ID; ASSIGN; start=expr; COMMA; stop=expr; RPAREN;
  140. body=block
  141. { let loc = loc $startpos(cnt) $endpos(cnt) in
  142. For (cnt, start, stop, Const (IntVal 1, []), Block body, loc) }
  143. | FOR; LPAREN; INT; cnt=ID; ASSIGN; start=expr; COMMA; stop=expr; COMMA;
  144. step=expr; RPAREN; body=block
  145. { let loc = loc $startpos(cnt) $endpos(cnt) in
  146. For (cnt, start, stop, step, Block body, loc) }
  147. block:
  148. | LBRACE; stats=statement*; RBRACE { stats }
  149. | stat=statement { [stat] }
  150. expr:
  151. | name=ID; LPAREN; args=separated_list(COMMA, expr); RPAREN
  152. { FunCall (name, make_args args, loc $startpos $endpos) }
  153. | LPAREN; expr; RPAREN { $2 }
  154. | ID { Var ($1, None, loc $startpos $endpos) }
  155. | l=expr; op=binop; r=expr { Binop (op, l, r, loc $startpos $endpos) }
  156. | SUB; expr { Monop (Neg, $2, loc $startpos $endpos) } %prec NEG
  157. | NOT; expr { Monop (Not, $2, loc $startpos $endpos) }
  158. | LPAREN; basic_type; RPAREN; expr { TypeCast ($2, $4, loc $startpos $endpos) } %prec CAST
  159. | FLOAT_CONST { Const (FloatVal $1, loc $startpos $endpos) }
  160. | INT_CONST { Const (IntVal $1, loc $startpos $endpos) }
  161. | BOOL_CONST { Const (BoolVal $1, loc $startpos $endpos) }
  162. | ID; array_const { Var ($1, Some $2, loc $startpos $endpos) }
  163. | array_const { ArrayConst ($1, loc $startpos $endpos) }
  164. %inline binop:
  165. | ADD { Add }
  166. | SUB { Sub }
  167. | MUL { Mul }
  168. | DIV { Div }
  169. | MOD { Mod }
  170. | EQ { Eq }
  171. | NE { Ne }
  172. | LT { Lt }
  173. | LE { Le }
  174. | GT { Gt }
  175. | GE { Ge }
  176. | AND { And }
  177. | OR { Or }
  178. array_const:
  179. LBRACK; values=separated_list(COMMA, expr); RBRACK
  180. { values }
  181. %%