parser.mly 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231
  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. (* Right-recursive list (less efficient then left-recursive list(x), only use
  46. * when followed by left-recursive list to resolve conflicts) *)
  47. rlist(x):
  48. | { [] }
  49. | rlist(x) x { $1 @ [$2] }
  50. (* Shorthand for comma-separated list *)
  51. %inline clist(x):
  52. lst=separated_list(COMMA, x)
  53. { lst }
  54. basic_type:
  55. | FLOAT { Float }
  56. | INT { Int }
  57. | BOOL { Bool }
  58. program:
  59. decl* EOF
  60. { Program ($1, loc $startpos $endpos) }
  61. decl:
  62. (* function: use location of function name *)
  63. | EXTERN hdr=fun_header SEMICOL
  64. { let (t, n, p, nameloc) = hdr in
  65. FunDec(t, n, p, nameloc) }
  66. | export=boption(EXPORT) hdr=fun_header LBRACE body=fun_body RBRACE
  67. { let (t, n, p, nameloc) = hdr in
  68. FunDef (export, t, n, p, Block body, nameloc) }
  69. (* global variable declaration: use location of variable name *)
  70. | EXTERN ctype=basic_type name=ID SEMICOL
  71. { GlobalDec (ctype, name, loc $startpos $endpos) }
  72. | EXTERN ctype=basic_type LBRACK dims=clist(ID) RBRACK name=ID SEMICOL
  73. { let dimloc = loc $startpos(dims) $endpos(dims) in
  74. let loc = loc $startpos(name) $endpos(name) in
  75. GlobalDec (ArrayDims (ctype, make_dims dimloc dims), name, loc) }
  76. | export=boption(EXPORT) ctype=basic_type name=ID SEMICOL
  77. { let loc = loc $startpos(name) $endpos(name) in
  78. GlobalDef (export, ctype, name, None, loc) }
  79. | export=boption(EXPORT) ctype=basic_type name=ID ASSIGN init=expr SEMICOL
  80. { let loc = loc $startpos(name) $endpos(name) in
  81. GlobalDef (export, ctype, name, Some init, loc) }
  82. | export=boption(EXPORT) ctype=basic_type LBRACK dims=clist(expr) RBRACK
  83. name=ID SEMICOL
  84. { let loc = loc $startpos(name) $endpos(name) in
  85. GlobalDef (export, ArrayDims (ctype, dims), name, None, loc) }
  86. fun_header:
  87. (* function header: use location of function name *)
  88. | ret=basic_type name=ID LPAREN params=clist(param) RPAREN
  89. { (ret, name, params, loc $startpos(name) $endpos(name)) }
  90. | VOID name=ID LPAREN params=clist(param) RPAREN
  91. { (Void, name, params, loc $startpos(name) $endpos(name)) }
  92. param:
  93. (* parameter: use location of parameter name *)
  94. | ctype=basic_type name=ID
  95. { Param (ctype, name, loc $startpos(name) $endpos(name)) }
  96. | ctype=basic_type LBRACK dims=clist(ID) RBRACK name=ID
  97. { let dimloc = loc $startpos(dims) $endpos(dims) in
  98. let loc = loc $startpos(name) $endpos(name) in
  99. Param (ArrayDims (ctype, make_dims dimloc dims), name, loc) }
  100. fun_body:
  101. | rlist(var_dec) local_fun_dec* statement* loption(return_statement)
  102. { VarDecs $1 :: (LocalFuns $2) :: $3 @ $4 }
  103. return_statement:
  104. (* return statement: use location of return value *)
  105. | RETURN value=expr SEMICOL
  106. { [Return (value, loc $startpos(value) $endpos(value))] }
  107. (* function: use location of function name *)
  108. local_fun_dec:
  109. | hdr=fun_header LBRACE body=fun_body RBRACE
  110. { let (t, n, p, nameloc) = hdr in
  111. FunDef (false, t, n, p, Block body, nameloc) }
  112. var_dec:
  113. (* variable declaration: use location of variable name *)
  114. | ctype=basic_type name=ID SEMICOL
  115. { VarDec (ctype, name, None, loc $startpos(name) $endpos(name)) }
  116. | ctype=basic_type name=ID ASSIGN init=expr SEMICOL
  117. { VarDec (ctype, name, Some init, loc $startpos(name) $endpos(name)) }
  118. | ctype=basic_type LBRACK dims=clist(expr) RBRACK name=ID SEMICOL
  119. { let loc = loc $startpos(name) $endpos(name) in
  120. VarDec (ArrayDims (ctype, dims), name, None, loc) }
  121. | ctype=basic_type LBRACK dims=clist(expr) RBRACK name=ID ASSIGN
  122. init=expr SEMICOL
  123. { let loc = loc $startpos(name) $endpos(name) in
  124. VarDec (ArrayDims (ctype, dims), name, Some init, loc) }
  125. statement:
  126. (* assignment: use location of assigned variable name *)
  127. | name=ID ASSIGN value=expr SEMICOL
  128. { Assign (name, None, value, loc $startpos(name) $endpos(name)) }
  129. | name=ID LBRACK dims=clist(expr) brk=RBRACK ASSIGN value=expr SEMICOL
  130. { Assign (name, Some dims, value, loc $startpos(name) $endpos(brk)) }
  131. | name=ID LPAREN args=clist(expr) RPAREN SEMICOL
  132. { Expr (FunCall (name, make_args args, loc $startpos(name) $endpos(name))) }
  133. (* if-statements and (do-)while-loops: use location of condition *)
  134. | IF LPAREN cond=expr RPAREN body=block
  135. { If (cond, Block body, loc $startpos $endpos) } %prec IF
  136. | IF LPAREN c=expr RPAREN t=block ELSE f=block
  137. { IfElse (c, Block t, Block f, loc $startpos(c) $endpos(c)) } %prec ELSE
  138. | WHILE LPAREN cond=expr RPAREN body=block
  139. { While (cond, Block body, loc $startpos(cond) $endpos(cond)) }
  140. | DO body=block WHILE LPAREN cond=expr RPAREN SEMICOL
  141. { DoWhile (cond, Block body, loc $startpos(cond) $endpos(cond)) }
  142. (* for-loop: use location of counter id *)
  143. | FOR LPAREN INT cnt=ID ASSIGN start=expr COMMA stop=expr RPAREN body=block
  144. { let loc = loc $startpos(cnt) $endpos(cnt) in
  145. For (cnt, start, stop, Const (IntVal 1, []), Block body, loc) }
  146. | FOR LPAREN INT cnt=ID ASSIGN start=expr COMMA stop=expr COMMA step=expr
  147. RPAREN body=block
  148. { let loc = loc $startpos(cnt) $endpos(cnt) in
  149. For (cnt, start, stop, step, Block body, loc) }
  150. block:
  151. | LBRACE stats=statement* RBRACE { stats }
  152. | stat=statement { [stat] }
  153. expr:
  154. | name=ID LPAREN args=clist(expr) RPAREN
  155. { FunCall (name, make_args args, loc $startpos $endpos) }
  156. | LPAREN expr RPAREN { $2 }
  157. | ID { Var ($1, None, loc $startpos $endpos) }
  158. | l=expr op=binop r=expr { Binop (op, l, r, loc $startpos $endpos) }
  159. | SUB expr { Monop (Neg, $2, loc $startpos $endpos) } %prec NEG
  160. | NOT expr { Monop (Not, $2, loc $startpos $endpos) }
  161. | LPAREN basic_type RPAREN expr { TypeCast ($2, $4, loc $startpos $endpos) } %prec CAST
  162. | FLOAT_CONST { Const (FloatVal $1, loc $startpos $endpos) }
  163. | INT_CONST { Const (IntVal $1, loc $startpos $endpos) }
  164. | BOOL_CONST { Const (BoolVal $1, loc $startpos $endpos) }
  165. | ID array_const { Var ($1, Some $2, loc $startpos $endpos) }
  166. | array_const { ArrayConst ($1, loc $startpos $endpos) }
  167. %inline binop:
  168. | ADD { Add }
  169. | SUB { Sub }
  170. | MUL { Mul }
  171. | DIV { Div }
  172. | MOD { Mod }
  173. | EQ { Eq }
  174. | NE { Ne }
  175. | LT { Lt }
  176. | LE { Le }
  177. | GT { Gt }
  178. | GE { Ge }
  179. | AND { And }
  180. | OR { Or }
  181. array_const:
  182. LBRACK values=clist(expr) RBRACK
  183. { values }
  184. %%