59 lines
1.9 KiB
Standard ML
59 lines
1.9 KiB
Standard ML
structure Scanner =
|
|
struct
|
|
structure Dfa = CaseSensitiveDfa
|
|
|
|
val intDfa = Dfa.fromString "[0-9]+"
|
|
val wordDfa = Dfa.fromString "0w[0-9]+"
|
|
val realDfa = "[0-9]+.[0-9]+"
|
|
val idDfa = Dfa.fromString "[a-zA-Z][a-zA-Z0-9_']*"
|
|
|
|
fun skipFormattingChars (str, pos) =
|
|
if pos = String.size str then
|
|
raise Fail "unterminated formatting char sequence"
|
|
else
|
|
let
|
|
val chr = String.sub (str, pos)
|
|
in
|
|
if Char.isSpace chr then skipFormattingChars (str, pos + 1)
|
|
else if chr = #"\\" then pos + 1
|
|
else raise Fail "encountered non-space char in formatting sequence"
|
|
end
|
|
|
|
fun getEscapeChr (str, pos, acc) =
|
|
case String.sub (str, pos) of
|
|
#"a" => (pos + 1, acc ^ "\a")
|
|
| #"b" => (pos + 1, acc ^ "\b")
|
|
| #"t" => (pos + 1, acc ^ "\t")
|
|
| #"n" => (pos + 1, acc ^ "\n")
|
|
| #"v" => (pos + 1, acc ^ "\v")
|
|
| #"f" => (pos + 1, acc ^ "\f")
|
|
| #"r" => (pos + 1, acc ^ "\r")
|
|
| #"\"" => (pos + 1, acc ^ "\"")
|
|
| #"\\" => (pos + 1, acc ^ "\\")
|
|
| chr =>
|
|
if Char.isSpace chr then
|
|
(* formatting characters ahead *)
|
|
let val nextPos = skipFormattingChars (str, pos + 1)
|
|
in (nextPos, acc)
|
|
end
|
|
else
|
|
raise Fail "invalid escape char"
|
|
|
|
fun helpExtractString (pos, str, acc) =
|
|
if pos >= String.size str then
|
|
raise Fail ("unterminated string: [" ^ acc ^ "]\n")
|
|
else
|
|
case String.sub (str, pos) of
|
|
#"\"" => (pos + 1, acc)
|
|
| #"\\" =>
|
|
if pos + 1 >= String.size str then
|
|
raise Fail ("unterminated string: [" ^ acc ^ "\\]\n")
|
|
else
|
|
let val (nextPos, acc) = getEscapeChr (str, pos + 1, acc)
|
|
in helpExtractString (nextPos, str, acc)
|
|
end
|
|
| chr => helpExtractString (pos + 1, str, acc ^ String.implode [chr])
|
|
|
|
fun extractString (pos, str) = helpExtractString (pos, str, "")
|
|
end
|