Bunzip

package.json

@@ -40,13 +40,16 @@
     "biojs"
   ],
   "dependencies": {
+    "@foxglove/wasm-bz2": "^0.1.1",
     "bz2": "^1.0.1",
     "bzip2": "^0.1.1",
+    "bzip2-wasm": "^1.0.1",
     "crc": "^4.3.2",
     "generic-filehandle2": "^1.0.0",
     "md5": "^2.2.1",
     "pako": "^1.0.4",
     "quick-lru": "^4.0.1",
+    "seek-bzip": "^2.0.0",
     "xz-decompress": "^0.2.1"
   },
   "devDependencies": {

src/cramFile/declare.d.ts

@@ -1 +1 @@
-declare module 'bz2'
+declare module 'bzip2'

src/cramFile/file.ts

@@ -1,4 +1,9 @@
-import { decompress } from 'bz2'
+// import bzip2 from 'bzip2'
+// import BZip2 from 'bzip2-wasm'
+// import { decompress } from 'bz2'
+
+import Bunzip from 'seek-bzip'
+
 import crc32 from 'crc/calculators/crc32'
 import QuickLRU from 'quick-lru'
 import { XzReadableStream } from 'xz-decompress'
@@ -280,10 +285,15 @@ export default class CramFile {
     inputBuffer: Uint8Array,
     uncompressedSize: number,
   ) {
+    // console.log({ compressionMethod })
     if (compressionMethod === 'gzip') {
-      return unzip(inputBuffer)
+      const ret = unzip(inputBuffer)
+      if (ret[0] === 24) {
+        // console.log(ret.slice(0, 500).join(','))
+      }
+      return ret
     } else if (compressionMethod === 'bzip2') {
-      return decompress(inputBuffer)
+      return Bunzip.decode(inputBuffer)
     } else if (compressionMethod === 'lzma') {
       const decompressedResponse = new Response(
         new XzReadableStream(bufferToStream(inputBuffer)),

src/cramFile/slice/decodeRecord.ts

@@ -13,6 +13,7 @@ import {
 import CramSlice, { SliceHeader } from './index'
 import { CramFileBlock } from '../file'
 import { isMappedSliceHeader } from '../sectionParsers'
+let kk = 0
 
 /**
  * given a Buffer, read a string up to the first null character
@@ -371,6 +372,7 @@ export default function decodeRecord(
 
     // mapping quality
     mappingQuality = decodeDataSeries('MQ')!
+
     if (CramFlagsDecoder.isPreservingQualityScores(cramFlags)) {
       qualityScores = new Array(readLength)
       for (let i = 0; i < qualityScores.length; i++) {
@@ -414,3 +416,12 @@ export default function decodeRecord(
     tags,
   }
 }
+function quals(quals: number[]) {
+  if (!quals || quals.length === 0) {
+    return '*'
+  }
+
+  return quals
+    .map(q => String.fromCharCode(Math.min(Math.max(q, 0), 93) + 33))
+    .join('')
+}

src/htscodecs/arith_gen.ts

@@ -34,7 +34,8 @@
  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-import { decompress } from 'bz2'
+import Bunzip from 'seek-bzip'
+
 import RangeCoder from './arith_sh'
 import ByteModel from './byte_model'
 import IOStream from './iostream'
@@ -153,19 +154,7 @@ export default class RangeCoderGen {
   // ----------------------------------------------------------------------
   // External codec
   decodeExt(stream, n_out) {
-    return decompress(stream.buf.slice(stream.pos))
-    // const bits = bzip2.array(stream.buf.slice(stream.pos))
-    // let size = bzip2.header(bits)
-    // let chunk
-    // const chunks = []
-    // do {
-    //   chunk = bzip2.decompress(bits, size)
-    //   if (chunk !== -1) {
-    //     chunks.push(chunk)
-    //     size -= chunk.length
-    //   }
-    // } while (chunk !== -1)
-    // return concatUint8Array(chunks)
+    return Bunzip.decode(stream.buf.slice(stream.pos))
   }
 
   // ----------------------------------------------------------------------