@gmod/bam 5.0.5 → 5.0.7

package/src/chunk.ts

@@ -1,12 +1,12 @@
-import VirtualOffset from './virtualOffset'
+import { Offset } from './virtualOffset'
 
 // little class representing a chunk in the index
 export default class Chunk {
   public buffer?: Uint8Array
 
   constructor(
-    public minv: VirtualOffset,
-    public maxv: VirtualOffset,
+    public minv: Offset,
+    public maxv: Offset,
     public bin: number,
     public _fetchedSize?: number,
   ) {}

package/src/csi.ts

@@ -1,15 +1,16 @@
 import { unzip } from '@gmod/bgzf-filehandle'
-import VirtualOffset, { fromBytes } from './virtualOffset'
+import QuickLRU from 'quick-lru'
+
 import Chunk from './chunk'
+import IndexFile from './indexFile'
 import {
-  optimizeChunks,
+  BaseOpts,
   findFirstData,
-  parsePseudoBin,
+  optimizeChunks,
   parseNameBytes,
-  BaseOpts,
+  parsePseudoBin,
 } from './util'
-
-import IndexFile from './indexFile'
+import { VirtualOffset, fromBytes } from './virtualOffset'
 
 const CSI1_MAGIC = 21582659 // CSI\1
 const CSI2_MAGIC = 38359875 // CSI\2
@@ -30,7 +31,7 @@ export default class CSI extends IndexFile {
 
   async lineCount(refId: number, opts?: BaseOpts) {
     const indexData = await this.parse(opts)
-    return indexData.indices[refId]?.stats?.lineCount || 0
+    return indexData.indices(refId)?.stats?.lineCount || 0
   }
 
   async indexCov() {
@@ -94,30 +95,57 @@ export default class CSI extends IndexFile {
     this.minShift = dataView.getInt32(4, true)
     this.depth = dataView.getInt32(8, true)
     this.maxBinNumber = ((1 << ((this.depth + 1) * 3)) - 1) / 7
+    const maxBinNumber = this.maxBinNumber
     const auxLength = dataView.getInt32(12, true)
     const aux = auxLength >= 30 ? this.parseAuxData(bytes, 16) : undefined
     const refCount = dataView.getInt32(16 + auxLength, true)
 
-    type BinIndex = Record<string, Chunk[]>
-
     // read the indexes for each reference sequence
     let curr = 16 + auxLength + 4
     let firstDataLine: VirtualOffset | undefined
-    const indices = new Array<{
-      binIndex: BinIndex
-      stats?: { lineCount: number }
-    }>(refCount)
+    const offsets = [] as number[]
     for (let i = 0; i < refCount; i++) {
+      offsets.push(curr)
+      const binCount = dataView.getInt32(curr, true)
+      curr += 4
+      for (let j = 0; j < binCount; j++) {
+        const bin = dataView.getUint32(curr, true)
+        curr += 4
+        if (bin > this.maxBinNumber) {
+          curr += 28 + 16
+        } else {
+          curr += 8
+          const chunkCount = dataView.getInt32(curr, true)
+          curr += 4
+          for (let k = 0; k < chunkCount; k += 1) {
+            const u = fromBytes(bytes, curr)
+            curr += 8
+            curr += 8
+            firstDataLine = findFirstData(firstDataLine, u)
+          }
+        }
+      }
+    }
+
+    const indicesCache = new QuickLRU<number, ReturnType<typeof getIndices>>({
+      maxSize: 5,
+    })
+
+    function getIndices(refId: number) {
+      let curr = offsets[refId]
+      if (curr === undefined) {
+        return undefined
+      }
       // the binning index
       const binCount = dataView.getInt32(curr, true)
       curr += 4
       const binIndex: Record<string, Chunk[]> = {}
-      let stats // < provided by parsing a pseudo-bin, if present
+      let pseudoBinStats
       for (let j = 0; j < binCount; j++) {
         const bin = dataView.getUint32(curr, true)
         curr += 4
-        if (bin > this.maxBinNumber) {
-          stats = parsePseudoBin(bytes, curr + 28)
+        if (bin > maxBinNumber) {
+          pseudoBinStats = parsePseudoBin(bytes, curr + 28)
           curr += 28 + 16
         } else {
           firstDataLine = findFirstData(firstDataLine, fromBytes(bytes, curr))
@@ -130,20 +158,31 @@ export default class CSI extends IndexFile {
             curr += 8
             const v = fromBytes(bytes, curr)
             curr += 8
-            firstDataLine = findFirstData(firstDataLine, u)
             chunks[k] = new Chunk(u, v, bin)
           }
           binIndex[bin] = chunks
         }
       }
 
-      indices[i] = { binIndex, stats }
+      return {
+        binIndex,
+        stats: pseudoBinStats,
+      }
     }
 
     return {
       csiVersion,
       firstDataLine,
-      indices,
+      indices: (refId: number) => {
+        if (!indicesCache.has(refId)) {
+          const result = getIndices(refId)
+          if (result) {
+            indicesCache.set(refId, result)
+          }
+          return result
+        }
+        return indicesCache.get(refId)
+      },
       refCount,
       csi: true,
       maxBlockSize: 1 << 16,
@@ -162,8 +201,8 @@ export default class CSI extends IndexFile {
     }
 
     const indexData = await this.parse(opts)
-    const ba = indexData.indices[refId]
-    // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+    const ba = indexData.indices(refId)
+
     if (!ba) {
       return []
     }
@@ -177,9 +216,8 @@ export default class CSI extends IndexFile {
     // Find chunks in overlapping bins.  Leaf bins (< 4681) are not pruned
     for (const [start, end] of overlappingBins) {
       for (let bin = start; bin <= end; bin++) {
-        // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
         if (ba.binIndex[bin]) {
-          const binChunks = ba.binIndex[bin]
+          const binChunks = ba.binIndex[bin]!
           for (const c of binChunks) {
             chunks.push(c)
           }
@@ -232,6 +270,6 @@ export default class CSI extends IndexFile {
 
   async hasRefSeq(seqId: number, opts: BaseOpts = {}) {
     const header = await this.parse(opts)
-    return !!header.indices[seqId]?.binIndex
+    return !!header.indices(seqId)?.binIndex
   }
 }

package/src/htsget.ts

@@ -1,8 +1,9 @@
 import { unzip } from '@gmod/bgzf-filehandle'
-import { BaseOpts, BamOpts, concatUint8Array } from './util'
+
 import BamFile, { BAM_MAGIC } from './bamFile'
 import Chunk from './chunk'
 import { parseHeaderText } from './sam'
+import { BamOpts, BaseOpts, concatUint8Array } from './util'
 
 interface HtsgetChunk {
   url: string
@@ -22,7 +23,7 @@ async function concat(arr: HtsgetChunk[], opts?: Record<string, any>) {
         const ret = await res.arrayBuffer()
         return new Uint8Array(ret)
       } else {
-        //remove referer header, it is not even allowed to be specified
+        // remove referer header, it is not even allowed to be specified
         // @ts-expect-error
 
         const { referer, ...rest } = headers

package/src/indexFile.ts

@@ -1,14 +1,15 @@
-import { GenericFilehandle } from 'generic-filehandle2'
 import Chunk from './chunk'
 import { BaseOpts } from './util'
 
+import type { GenericFilehandle } from 'generic-filehandle2'
+
 export default abstract class IndexFile {
   public filehandle: GenericFilehandle
   public renameRefSeq: (s: string) => string
 
   /**
    * @param {filehandle} filehandle
-   * @param {function} [renameRefSeqs]
+   * @param {function} renameRefSeqs
    */
   constructor({
     filehandle,

package/src/long.ts

@@ -3,14 +3,14 @@ export const TWO_PWR_32_DBL = TWO_PWR_16_DBL * TWO_PWR_16_DBL
 
 export function longFromBytesToUnsigned(source: Uint8Array, i = 0) {
   const low =
-    source[i] |
-    (source[i + 1] << 8) |
-    (source[i + 2] << 16) |
-    (source[i + 3] << 24)
+    source[i]! |
+    (source[i + 1]! << 8) |
+    (source[i + 2]! << 16) |
+    (source[i + 3]! << 24)
   const high =
-    source[i + 4] |
-    (source[i + 5] << 8) |
-    (source[i + 6] << 16) |
-    (source[i + 7] << 24)
+    source[i + 4]! |
+    (source[i + 5]! << 8) |
+    (source[i + 6]! << 16) |
+    (source[i + 7]! << 24)
   return (high >>> 0) * TWO_PWR_32_DBL + (low >>> 0)
 }

package/src/nullFilehandle.ts

@@ -0,0 +1,16 @@
+export default class NullFilehandle {
+  public read(): Promise<any> {
+    throw new Error('never called')
+  }
+  public stat(): Promise<any> {
+    throw new Error('never called')
+  }
+
+  public readFile(): Promise<any> {
+    throw new Error('never called')
+  }
+
+  public close(): Promise<any> {
+    throw new Error('never called')
+  }
+}

package/src/record.ts

@@ -77,7 +77,7 @@ export default class BamRecord {
   get name() {
     let str = ''
     for (let i = 0; i < this.read_name_length - 1; i++) {
-      str += String.fromCharCode(this.byteArray[this.b0 + i])
+      str += String.fromCharCode(this.byteArray[this.b0 + i]!)
     }
     return str
   }
@@ -93,12 +93,15 @@ export default class BamRecord {
     const blockEnd = this.bytes.end
     const tags = {} as Record<string, unknown>
     while (p < blockEnd) {
-      const tag = String.fromCharCode(this.byteArray[p], this.byteArray[p + 1])
-      const type = String.fromCharCode(this.byteArray[p + 2])
+      const tag = String.fromCharCode(
+        this.byteArray[p]!,
+        this.byteArray[p + 1]!,
+      )
+      const type = String.fromCharCode(this.byteArray[p + 2]!)
       p += 3
 
       if (type === 'A') {
-        tags[tag] = String.fromCharCode(this.byteArray[p])
+        tags[tag] = String.fromCharCode(this.byteArray[p]!)
         p += 1
       } else if (type === 'i') {
         tags[tag] = this.#dataView.getInt32(p, true)
@@ -124,7 +127,7 @@ export default class BamRecord {
       } else if (type === 'Z' || type === 'H') {
         const value = []
         while (p <= blockEnd) {
-          const cc = this.byteArray[p++]
+          const cc = this.byteArray[p++]!
           if (cc !== 0) {
             value.push(String.fromCharCode(cc))
           } else {
@@ -133,7 +136,7 @@ export default class BamRecord {
         }
         tags[tag] = value.join('')
       } else if (type === 'B') {
-        const cc = this.byteArray[p++]
+        const cc = this.byteArray[p++]!
         const Btype = String.fromCharCode(cc)
         const limit = this.#dataView.getInt32(p, true)
         p += 4
@@ -143,7 +146,7 @@ export default class BamRecord {
             for (let k = 0; k < limit; k++) {
               const cigop = this.#dataView.getInt32(p, true)
               const lop = cigop >> 4
-              const op = CIGAR_DECODER[cigop & 0xf]
+              const op = CIGAR_DECODER[cigop & 0xf]!
               value.push(lop + op)
               p += 4
             }
@@ -162,7 +165,7 @@ export default class BamRecord {
             for (let k = 0; k < limit; k++) {
               const cigop = this.#dataView.getUint32(p, true)
               const lop = cigop >> 4
-              const op = CIGAR_DECODER[cigop & 0xf]
+              const op = CIGAR_DECODER[cigop & 0xf]!
               value.push(lop + op)
               p += 4
             }
@@ -318,7 +321,7 @@ export default class BamRecord {
       for (let c = 0; c < numCigarOps; ++c) {
         cigop = this.#dataView.getInt32(p, true)
         lop = cigop >> 4
-        op = CIGAR_DECODER[cigop & 0xf]
+        op = CIGAR_DECODER[cigop & 0xf]!
         CIGAR.push(lop + op)
         // soft clip, hard clip, and insertion don't count toward the length on
         // the reference
@@ -357,14 +360,13 @@ export default class BamRecord {
   }
 
   get seq() {
-    const { byteArray } = this.bytes
     const p = this.b0 + this.read_name_length + this.num_cigar_ops * 4
     const seqBytes = this.num_seq_bytes
     const len = this.seq_length
     const buf = []
     let i = 0
     for (let j = 0; j < seqBytes; ++j) {
-      const sb = byteArray[p + j]
+      const sb = this.byteArray[p + j]!
       buf.push(SEQRET_DECODER[(sb & 0xf0) >> 4])
       i++
       if (i < len) {
@@ -442,7 +444,7 @@ export default class BamRecord {
       if (k.startsWith('_') || k === 'bytes') {
         continue
       }
-      //@ts-ignore
+      // @ts-ignore
       data[k] = this[k]
     }
 

package/src/util.ts

@@ -1,6 +1,6 @@
 import Chunk from './chunk'
 import { longFromBytesToUnsigned } from './long'
-import VirtualOffset from './virtualOffset'
+import { Offset, VirtualOffset } from './virtualOffset'
 
 export function timeout(ms: number) {
   return new Promise(resolve => setTimeout(resolve, ms))
@@ -27,7 +27,7 @@ export function checkAbortSignal(signal?: AbortSignal) {
     // console.log('bam aborted!')
     if (typeof DOMException === 'undefined') {
       const e = new Error('aborted')
-      //@ts-ignore
+      // @ts-ignore
       e.code = 'ERR_ABORTED'
       throw e
     } else {
@@ -69,7 +69,7 @@ export function makeOpts(obj: AbortSignal | BaseOpts = {}): BaseOpts {
   return 'aborted' in obj ? ({ signal: obj } as BaseOpts) : obj
 }
 
-export function optimizeChunks(chunks: Chunk[], lowest?: VirtualOffset) {
+export function optimizeChunks(chunks: Chunk[], lowest?: Offset) {
   const mergedChunks: Chunk[] = []
   let lastChunk: Chunk | undefined
 
@@ -133,7 +133,7 @@ export function parseNameBytes(
       if (currNameStart < i) {
         let refName = ''
         for (let j = currNameStart; j < i; j++) {
-          refName += String.fromCharCode(namesBytes[j])
+          refName += String.fromCharCode(namesBytes[j]!)
         }
         refName = renameRefSeq(refName)
         refIdToName[currRefId] = refName
@@ -162,3 +162,11 @@ export function concatUint8Array(args: Uint8Array[]) {
   }
   return mergedArray
 }
+
+export async function gen2array<T>(gen: AsyncIterable<T[]>): Promise<T[]> {
+  let out: T[] = []
+  for await (const x of gen) {
+    out = out.concat(x)
+  }
+  return out
+}

package/src/virtualOffset.ts

@@ -1,4 +1,11 @@
-export default class VirtualOffset {
+export interface Offset {
+  blockPosition: number
+  dataPosition: number
+  toString(): string
+  compareTo(arg: Offset): number
+}
+
+export class VirtualOffset {
   public blockPosition: number
   public dataPosition: number
   constructor(blockPosition: number, dataPosition: number) {
@@ -15,20 +22,6 @@ export default class VirtualOffset {
       this.blockPosition - b.blockPosition || this.dataPosition - b.dataPosition
     )
   }
-
-  static min(...args: VirtualOffset[]) {
-    let min
-    let i = 0
-    for (; !min; i += 1) {
-      min = args[i]
-    }
-    for (; i < args.length; i += 1) {
-      if (min.compareTo(args[i]) > 0) {
-        min = args[i]
-      }
-    }
-    return min
-  }
 }
 export function fromBytes(bytes: Uint8Array, offset = 0, bigendian = false) {
   if (bigendian) {
@@ -36,12 +29,12 @@ export function fromBytes(bytes: Uint8Array, offset = 0, bigendian = false) {
   }
 
   return new VirtualOffset(
-    bytes[offset + 7] * 0x10000000000 +
-      bytes[offset + 6] * 0x100000000 +
-      bytes[offset + 5] * 0x1000000 +
-      bytes[offset + 4] * 0x10000 +
-      bytes[offset + 3] * 0x100 +
-      bytes[offset + 2],
-    (bytes[offset + 1] << 8) | bytes[offset],
+    bytes[offset + 7]! * 0x10000000000 +
+      bytes[offset + 6]! * 0x100000000 +
+      bytes[offset + 5]! * 0x1000000 +
+      bytes[offset + 4]! * 0x10000 +
+      bytes[offset + 3]! * 0x100 +
+      bytes[offset + 2]!,
+    (bytes[offset + 1]! << 8) | bytes[offset]!,
   )
 }

package/dist/errors.d.ts

@@ -1,23 +0,0 @@
-export declare class BamError extends Error {
-}
-/** Error caused by encountering a part of the BAM spec that has not yet been implemented */
-export declare class BamUnimplementedError extends Error {
-}
-/** An error caused by malformed data.  */
-export declare class BamMalformedError extends BamError {
-}
-/**
- * An error caused by attempting to read beyond the end of the defined data.
- */
-export declare class BamBufferOverrunError extends BamMalformedError {
-}
-/**
- * An error caused by data being too big, exceeding a size limit.
- */
-export declare class BamSizeLimitError extends BamError {
-}
-/**
- * An invalid argument was supplied to a bam-js method or object.
- */
-export declare class BamArgumentError extends BamError {
-}

package/dist/errors.js

@@ -1,33 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.BamArgumentError = exports.BamSizeLimitError = exports.BamBufferOverrunError = exports.BamMalformedError = exports.BamUnimplementedError = exports.BamError = void 0;
-class BamError extends Error {
-}
-exports.BamError = BamError;
-/** Error caused by encountering a part of the BAM spec that has not yet been implemented */
-class BamUnimplementedError extends Error {
-}
-exports.BamUnimplementedError = BamUnimplementedError;
-/** An error caused by malformed data.  */
-class BamMalformedError extends BamError {
-}
-exports.BamMalformedError = BamMalformedError;
-/**
- * An error caused by attempting to read beyond the end of the defined data.
- */
-class BamBufferOverrunError extends BamMalformedError {
-}
-exports.BamBufferOverrunError = BamBufferOverrunError;
-/**
- * An error caused by data being too big, exceeding a size limit.
- */
-class BamSizeLimitError extends BamError {
-}
-exports.BamSizeLimitError = BamSizeLimitError;
-/**
- * An invalid argument was supplied to a bam-js method or object.
- */
-class BamArgumentError extends BamError {
-}
-exports.BamArgumentError = BamArgumentError;
-//# sourceMappingURL=errors.js.map

package/dist/errors.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"errors.js","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":";;;AAAA,MAAa,QAAS,SAAQ,KAAK;CAAG;AAAtC,4BAAsC;AAEtC,4FAA4F;AAC5F,MAAa,qBAAsB,SAAQ,KAAK;CAAG;AAAnD,sDAAmD;AAEnD,0CAA0C;AAC1C,MAAa,iBAAkB,SAAQ,QAAQ;CAAG;AAAlD,8CAAkD;AAElD;;GAEG;AACH,MAAa,qBAAsB,SAAQ,iBAAiB;CAAG;AAA/D,sDAA+D;AAE/D;;GAEG;AACH,MAAa,iBAAkB,SAAQ,QAAQ;CAAG;AAAlD,8CAAkD;AAElD;;GAEG;AACH,MAAa,gBAAiB,SAAQ,QAAQ;CAAG;AAAjD,4CAAiD"}

package/esm/errors.d.ts

@@ -1,23 +0,0 @@
-export declare class BamError extends Error {
-}
-/** Error caused by encountering a part of the BAM spec that has not yet been implemented */
-export declare class BamUnimplementedError extends Error {
-}
-/** An error caused by malformed data.  */
-export declare class BamMalformedError extends BamError {
-}
-/**
- * An error caused by attempting to read beyond the end of the defined data.
- */
-export declare class BamBufferOverrunError extends BamMalformedError {
-}
-/**
- * An error caused by data being too big, exceeding a size limit.
- */
-export declare class BamSizeLimitError extends BamError {
-}
-/**
- * An invalid argument was supplied to a bam-js method or object.
- */
-export declare class BamArgumentError extends BamError {
-}

package/esm/errors.js

@@ -1,24 +0,0 @@
-export class BamError extends Error {
-}
-/** Error caused by encountering a part of the BAM spec that has not yet been implemented */
-export class BamUnimplementedError extends Error {
-}
-/** An error caused by malformed data.  */
-export class BamMalformedError extends BamError {
-}
-/**
- * An error caused by attempting to read beyond the end of the defined data.
- */
-export class BamBufferOverrunError extends BamMalformedError {
-}
-/**
- * An error caused by data being too big, exceeding a size limit.
- */
-export class BamSizeLimitError extends BamError {
-}
-/**
- * An invalid argument was supplied to a bam-js method or object.
- */
-export class BamArgumentError extends BamError {
-}
-//# sourceMappingURL=errors.js.map

package/esm/errors.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"errors.js","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA,MAAM,OAAO,QAAS,SAAQ,KAAK;CAAG;AAEtC,4FAA4F;AAC5F,MAAM,OAAO,qBAAsB,SAAQ,KAAK;CAAG;AAEnD,0CAA0C;AAC1C,MAAM,OAAO,iBAAkB,SAAQ,QAAQ;CAAG;AAElD;;GAEG;AACH,MAAM,OAAO,qBAAsB,SAAQ,iBAAiB;CAAG;AAE/D;;GAEG;AACH,MAAM,OAAO,iBAAkB,SAAQ,QAAQ;CAAG;AAElD;;GAEG;AACH,MAAM,OAAO,gBAAiB,SAAQ,QAAQ;CAAG"}

package/src/errors.ts

@@ -1,22 +0,0 @@
-export class BamError extends Error {}
-
-/** Error caused by encountering a part of the BAM spec that has not yet been implemented */
-export class BamUnimplementedError extends Error {}
-
-/** An error caused by malformed data.  */
-export class BamMalformedError extends BamError {}
-
-/**
- * An error caused by attempting to read beyond the end of the defined data.
- */
-export class BamBufferOverrunError extends BamMalformedError {}
-
-/**
- * An error caused by data being too big, exceeding a size limit.
- */
-export class BamSizeLimitError extends BamError {}
-
-/**
- * An invalid argument was supplied to a bam-js method or object.
- */
-export class BamArgumentError extends BamError {}