Fix parser of optional fields in SAM/BAM.

src/cljam/io/bam/decoder.clj

@@ -23,7 +23,7 @@
 (definline parse-tag-single [tag-type ^ByteBuffer bb]
   `(case (long ~tag-type)
     ~(long \Z) (lsb/read-null-terminated-string ~bb)
-    ~(long \A) (.get ~bb)
+    ~(long \A) (char (.get ~bb))
     ~(long \I) (bit-and (.getInt ~bb) 0xffffffff)
     ~(long \i) (.getInt ~bb)
     ~(long \s) (int (.getShort ~bb))
@@ -37,7 +37,7 @@
 (defn- parse-tag-array [^ByteBuffer bb]
   (let [typ (char (.get bb))
         len (.getInt bb)]
-    (->> (for [i (range len)]
+    (->> (for [_ (range len)]
            (case typ
              \c (int (.get bb))
              \C (bit-and (int (.get bb)) 0xff)
@@ -118,7 +118,7 @@
           pos         (inc (lsb/read-int buffer))
           l-read-name (int (lsb/read-ubyte buffer))
           mapq        (lsb/read-ubyte buffer)
-          bin         (lsb/read-ushort buffer)
+          _           (lsb/read-ushort buffer) ; bin
           n-cigar-op  (lsb/read-ushort buffer)
           flag        (lsb/read-ushort buffer)
           l-seq       (lsb/read-int buffer)

src/cljam/io/sam/util.clj

@@ -1,7 +1,7 @@
 (ns cljam.io.sam.util
   "Utilities related to SAM/BAM format."
   (:require [clojure.string :as cstr]
-            [proton.core :refer [as-long as-double]]
+            [proton.core :refer [as-long as-double hex->bytes]]
             cljam.io.protocols
             [cljam.io.util.cigar :refer [count-ref]]
             [cljam.util :refer [ubyte]])
@@ -88,7 +88,7 @@
     \c (as-long val)
     \C (as-long val)
     \f (as-double val)
-    \H nil ;;FIXME
+    \H (hex->bytes val)
     (throw (Exception. "Unrecognized tag type"))))
 
 (defn- parse-optional-fields [options]

test-resources/sam/opts.sam

@@ -0,0 +1,10 @@
+@SQ	SN:ref	LN:45
+q001	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xa:A:p	Xi:i:-100	Xf:f:-1.4e3	Xz:Z:AATTGGCC
+q002	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xh:H:1AE301
+q003	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xc:B:c,-128,0,127
+q004	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	XC:B:C,0,127,255
+q005	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xs:B:s,-32768,0,32767
+q006	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	XS:B:S,0,32767,65535
+q007	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xi:B:i,-2147483648,0,2147483647
+q008	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	XI:B:I,0,2147483647,4294967295
+q009	0	ref	10	60	16M	*	0	0	AAAAGGGGTTTTCCCC	*	Xf:B:f,-0.3,0.0,0.3

test/cljam/io/t_sam.clj

@@ -214,3 +214,34 @@
     (are [f] (thrown? Exception (sam/writer (.getAbsolutePath (cio/file util/temp-dir f))))
       "temp.baam"
       "temp.bai")))
+
+(def test-options
+  [{:Xa {:type "A", :value \p}}
+   {:Xi {:type "i", :value -100}}
+   {:Xf {:type "f", :value -1400.0}}
+   {:Xz {:type "Z", :value "AATTGGCC"}}
+   {:Xh {:type "H", :value (map unchecked-byte [0x1A 0xE3 0x01])}}
+   {:Xc {:type "B", :value "c,-128,0,127"}}
+   {:XC {:type "B", :value "C,0,127,255"}}
+   {:Xs {:type "B", :value "s,-32768,0,32767"}}
+   {:XS {:type "B", :value "S,0,32767,65535"}}
+   {:Xi {:type "B", :value "i,-2147483648,0,2147483647"}}
+   {:XI {:type "B", :value "I,0,2147483647,4294967295"}}
+   {:Xf {:type "B", :value "f,-0.3,0.0,0.3"}}])
+
+(defn bytes-to-seq [t]
+  (if (= (get-in t [(ffirst t) :type]) "H")
+    (update-in t [(ffirst t) :value] seq)
+    t))
+
+(deftest options
+  (testing "sam"
+    (is
+     (= (with-open [r (sam/reader opts-sam-file)]
+          (->> (sam/read-alignments r) (mapcat :options) (map bytes-to-seq) doall))
+        test-options)))
+  (testing "bam"
+    (is
+     (= (with-open [r (sam/reader opts-bam-file)]
+          (->> (sam/read-alignments r) (mapcat :options) (map bytes-to-seq) doall))
+        test-options))))

test/cljam/t_common.clj

@@ -107,6 +107,8 @@
 (def normalize-before-sam-file "test-resources/sam/normalize_before.sam")
 (def normalize-after-sam-file "test-resources/sam/normalize_after.sam")
 
+(def opts-sam-file "test-resources/sam/opts.sam")
+
 ;; ### BAM files
 
 (def test-bam-file "test-resources/bam/test.bam")
@@ -120,6 +122,7 @@
 
 (def normalize-before-bam-file "test-resources/bam/normalize_before.bam")
 (def normalize-after-bam-file "test-resources/bam/normalize_after.bam")
+(def opts-bam-file "test-resources/bam/opts.bam")
 
 ;; ### BAM index files