From b632667ce57af89691407bb8668e1512775278ae Mon Sep 17 00:00:00 2001
From: Calvin <calvin@EESI>
Date: Fri, 15 Mar 2013 15:26:20 -0400
Subject: nbc added

---
 src/nbc/genome.sml | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 src/nbc/genome.sml

(limited to 'src/nbc/genome.sml')

diff --git a/src/nbc/genome.sml b/src/nbc/genome.sml
new file mode 100644
index 0000000..81f3183
--- /dev/null
+++ b/src/nbc/genome.sml
@@ -0,0 +1,33 @@
+signature GENOME = sig
+	exception Bad
+	type t
+	val load: string * int -> t
+	val get: t * string -> int option
+end
+
+structure Genome :> GENOME = struct
+	exception Bad
+	fun |> (x, f) = f x
+	infix |>
+
+	type t = (string, int) HashTable.hash_table
+	fun load (gname, order) =
+		let
+			val h = HashTable.mkTable
+				(HashString.hashString, op =)
+				(1024 * 1024, Fail "")
+		in
+			Options.genomeText (order, gname) |> Gzip.openIn |> Misc.sequenceLines
+			|> Sequence.map (fn s => (
+				 case Misc.split2 s of
+					SOME (count, nmer) => (
+						nmer
+						, case Int.fromString count of
+							NONE => raise Bad
+							| SOME x => x
+					) | NONE => raise Bad
+			)) |> Sequence.app (HashTable.insert h)
+			; h
+		end
+	fun get (h, nmer) = HashTable.find h nmer
+end
-- 
cgit v1.2.3