1 files changed, 310 insertions, 0 deletions

diff --git a/lib/nix/parser.mly b/lib/nix/parser.mly
new file mode 100644
index 0000000..dc1638d
--- /dev/null
+++ b/lib/nix/parser.mly
@@ -0,0 +1,310 @@
+/* Tokens with data */
+%token <string> INT
+%token <string> FLOAT
+/* A path */
+%token <string> PATH
+/* Search path, enclosed in <> */
+%token <string> SPATH
+/* Home path, starts with ~ */
+%token <string> HPATH
+%token <string> URI
+%token <string> STR_START
+%token <string> STR_MID
+%token STR_END
+%token <string> ISTR_START
+%token <string> ISTR_MID
+%token <int> ISTR_END
+%token <string> ID
+/* Tokens that stand for themselves */
+%token SELECT "."
+%token QMARK "?"
+%token CONCAT "++"
+%token NOT "!"
+%token MERGE "//"
+%token ASSIGN "="
+%token LT "<"
+%token LTE "<="
+%token GT ">"
+%token GTE ">="
+%token EQ "=="
+%token NEQ "!="
+%token AND "&&"
+%token OR "||"
+%token IMPL "->"
+%token AQUOTE_OPEN "${"
+%token AQUOTE_CLOSE "}$"
+%token LBRACE "{"
+%token RBRACE "}"
+%token LBRACK "["
+%token RBRACK "]"
+%token PLUS "+"
+%token MINUS "-"
+%token TIMES "*"
+%token SLASH "/"
+%token LPAREN "("
+%token RPAREN ")"
+%token COLON ":"
+%token SEMICOLON ";"
+%token COMMA ","
+%token ELLIPSIS "..."
+%token AS "@"
+/* Keywords */
+%token WITH "with"
+%token REC "rec"
+%token LET "let"
+%token IN "in"
+%token INHERIT "inherit"
+%token IF "if"
+%token THEN "then"
+%token ELSE "else"
+%token ASSERT "assert"
+%token ORDEF "or"
+
+/* End of input */
+%token EOF
+
+%{
+  open Types
+%}
+
+%start <Types.expr> main
+
+%%
+
+main:
+| e = expr0 EOF
+    { e }
+
+expr0:
+| "if"; e1 = expr0; "then"; e2 = expr0; "else"; e3 = expr0
+    { Cond (e1, e2, e3) }
+| "with"; e1 = expr0; ";"; e2 = expr0
+    { With (e1, e2) }
+| "assert"; e1 = expr0; ";"; e2 = expr0
+    { Assert (e1, e2) }
+| "let"; xs = delimited("{", list(binding), "}")
+    { SetLet xs }
+| "let"; xs = list(binding); "in"; e = expr0
+    { Let (xs, e) }
+| l = lambda
+    { Val l }
+| e = expr1
+    { e }
+
+/* Rules expr1-expr14 are almost direct translation of the operator
+   precedence table:
+   https://nixos.org/nix/manual/#sec-language-operators */
+
+%inline binary_expr(Lhs, Op, Rhs):
+| lhs = Lhs; op = Op; rhs = Rhs
+    { BinaryOp (op, lhs, rhs) }
+
+expr1:
+| e = binary_expr(expr2, "->" {Impl}, expr1)
+| e = expr2
+    { e }
+
+expr2:
+| e = binary_expr(expr2, "||" {Or}, expr3)
+| e = expr3
+    { e }
+
+expr3:
+| e = binary_expr(expr3, "&&" {And}, expr4)
+| e = expr4
+    { e }
+
+%inline expr4_ops:
+| "==" { Eq }
+| "!=" { Neq }
+
+expr4:
+| e = binary_expr(expr5, expr4_ops, expr5)
+| e = expr5
+    { e }
+
+%inline expr5_ops:
+| "<" { Lt }
+| ">" { Gt }
+| "<=" { Lte }
+| ">=" { Gte }
+
+expr5:
+| e = binary_expr(expr6, expr5_ops, expr6)
+| e = expr6
+    { e }
+
+expr6:
+| e = binary_expr(expr7, "//" {Merge}, expr6)
+| e = expr7
+    { e }
+
+expr7:
+| e = preceded("!", expr7)
+    { UnaryOp (Not, e) }
+| e = expr8
+    { e }
+
+%inline expr8_ops:
+| "+" { Plus }
+| "-" { Minus }
+
+expr8:
+| e = binary_expr(expr8, expr8_ops, expr9)
+| e = expr9
+    { e }
+
+%inline expr9_ops:
+| "*" { Mult }
+| "/" { Div }
+
+expr9:
+| e = binary_expr(expr9, expr9_ops, expr10)
+| e = expr10
+    { e }
+
+expr10:
+| e = binary_expr(expr11, "++" {Concat}, expr10)
+| e = expr11
+    { e }
+
+expr11:
+| e = expr12 "?" p = attr_path
+    { Test (e, p) }
+| e = expr12
+    { e }
+
+expr12:
+| e = preceded("-", expr13)
+    { UnaryOp (Negate, e) }
+| e = expr13
+    { e }
+
+expr13:
+| f = expr13; arg = expr14
+    { Apply (f, arg) }
+| e = expr14
+    { e }
+
+%inline selectable:
+| s = set
+    { Val s }
+| id = ID
+    { Id id }
+| e = delimited("(", expr0, ")")
+    { e }
+
+expr14:
+| e = selectable; "."; p = attr_path; o = option(preceded("or", expr14))
+    { Select (e, p, o) }
+| e = atomic_expr; "or"
+    { Apply (e, Id "or") }
+| e = atomic_expr
+    { e }
+
+atomic_expr:
+| id = ID
+    { Id id }
+| v = value
+    { Val v }
+| e = delimited("(", expr0, ")")
+    { e }
+
+attr_path:
+| p = separated_nonempty_list(".", attr_path_component)
+    { p }
+
+attr_path_component:
+| "or"
+    { Id "or" }
+| id = ID
+    { Id id }
+| e = delimited("${", expr0, "}$")
+    { Aquote e }
+| s = str
+    { Val s }
+
+value:
+| s = str
+    { s }
+| s = istr
+    { s }
+| i = INT
+    {Int i}
+| f = FLOAT
+    { Float f }
+| p = PATH
+    { Path p }
+| sp = SPATH
+    { SPath sp }
+| hp = HPATH
+    { HPath hp }
+| uri = URI
+    { Uri uri }
+| l = nixlist
+    { l }
+| s = set
+    { s }
+
+%inline str_mid(X):
+| xs = list(pair(delimited("${", expr0, "}$"), X)) { xs }
+
+/* Double-quoted string */
+str:
+| start = STR_START; mids = str_mid(STR_MID); STR_END
+    { Str (start, mids) }
+
+/* Indented string */
+istr:
+| start = ISTR_START; mids = str_mid(ISTR_MID); i = ISTR_END
+    { IStr (i, start, mids) }
+
+/* Lists and sets */
+nixlist:
+| xs = delimited("[", list(expr14), "]")
+    { List xs }
+
+empty_set:
+| "{"; "}" {}
+
+set:
+| empty_set
+    { AttSet (Nonrec, []) }
+| xs = delimited("{", nonempty_list(binding), "}")
+    { AttSet (Nonrec, xs) }
+| xs = preceded("rec", delimited("{", list(binding), "}"))
+    { AttSet (Rec, xs) }
+
+binding:
+| kv = terminated(separated_pair(attr_path, "=", expr0), ";")
+    { let (k, v) = kv in AttrPath (k, v) }
+| xs = delimited("inherit", pair(option(delimited("(", expr0, ")")), list(attr_path_component)), ";")
+    { let (prefix, ids) = xs in Inherit (prefix, ids) }
+
+lambda:
+| id = ID; "@"; p = param_set; ":"; e = expr0
+    { Lambda (ParamSet (Some id, p), e) }
+| p = param_set; "@"; id = ID; ":"; e = expr0
+    { Lambda (ParamSet (Some id, p), e) }
+| p = param_set; ":"; e = expr0
+    { Lambda (ParamSet (None, p), e) }
+| id = ID; ":"; e = expr0
+    { Lambda (Alias id, e) }
+
+%inline param_set:
+| empty_set
+    { ([], Exact) }
+| "{"; "..."; "}"
+    { ([], Loose) }
+| ps = delimited("{", pair(pair(params, ","?), boption("...")), "}")
+    { let ((ps, _), ellipsis) = ps in (ps, if ellipsis then Loose else Exact) }
+
+params:
+| p = param
+    { [p] }
+| ps = params; ","; p = param
+    { ps @ [p] }
+
+%inline param:
+p = pair(ID, option(preceded("?", expr0)))
+    { p }