Convert to gf2

GMOD · Dec 11, 2024 · 2b1fa7b · 2b1fa7b
1 parent 546e797
commit 2b1fa7b
Show file tree

Hide file tree

Showing 5 changed files with 61 additions and 137 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,23 +1,13 @@
 ## [3.0.1](https://github.com/GMOD/twobit-js/compare/v2.0.1...v3.0.1) (2024-11-10)
 
-
-
 # [3.0.0](https://github.com/GMOD/twobit-js/compare/v2.0.1...v3.0.0) (2024-11-10)
 
-
-
 # [3.0.0](https://github.com/GMOD/twobit-js/compare/v2.0.1...v3.0.0) (2024-11-10)
 
-
-
 # [3.0.0](https://github.com/GMOD/twobit-js/compare/v2.0.1...v3.0.0) (2024-11-10)
 
-
-
 # [3.0.0](https://github.com/GMOD/twobit-js/compare/v2.0.1...v3.0.0) (2024-11-10)
 
-
-
 ## [2.0.1](https://github.com/GMOD/twobit-js/compare/v2.0.0...v2.0.1) (2024-08-09)
 
 # [2.0.0](https://github.com/GMOD/twobit-js/compare/v1.1.14...v2.0.0) (2024-08-09)

diff --git a/README.md b/README.md
@@ -44,24 +44,24 @@ const seqNames = await t.getSequenceNames()
 
 ##### Table of Contents
 
-*   [constructor](#constructor)
-    *   [Parameters](#parameters)
-*   [getSequenceNames](#getsequencenames)
-*   [getSequenceSizes](#getsequencesizes)
-*   [getSequenceSize](#getsequencesize)
-    *   [Parameters](#parameters-1)
-*   [getSequence](#getsequence)
-    *   [Parameters](#parameters-2)
+- [constructor](#constructor)
+  - [Parameters](#parameters)
+- [getSequenceNames](#getsequencenames)
+- [getSequenceSizes](#getsequencesizes)
+- [getSequenceSize](#getsequencesize)
+  - [Parameters](#parameters-1)
+- [getSequence](#getsequence)
+  - [Parameters](#parameters-2)
 
 #### constructor
 
 ##### Parameters
 
-*   `args` **[object](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Object)**&#x20;
+- `args` **[object](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Object)**&#x20;
 
-    *   `args.filehandle` **Filehandle?** node fs.promises-like filehandle for the .2bit file.
-        Only needs to support `filehandle.read(buffer, offset, length, position)`
-    *   `args.path` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)?** filesystem path for the .2bit file to open
+  - `args.filehandle` **Filehandle?** node fs.promises-like filehandle for the .2bit file.
+    Only needs to support `filehandle.read(buffer, offset, length, position)`
+  - `args.path` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)?** filesystem path for the .2bit file to open
 
 #### getSequenceNames
 
@@ -78,17 +78,17 @@ e.g. a chrom.sizes file, it will be much faster
 
 ##### Parameters
 
-*   `seqName` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** name of the sequence
+- `seqName` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** name of the sequence
 
 Returns **[Promise](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Promise)** for the sequence's length, or undefined if it is not in the file
 
 #### getSequence
 
 ##### Parameters
 
-*   `seqName` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** name of the sequence you want
-*   `regionStart` **[number](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Number)?** optional 0-based half-open start of the sequence region to fetch. (optional, default `0`)
-*   `regionEnd` **[number](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Number)?** optional 0-based half-open end of the sequence region to fetch. defaults to end of the sequence (optional, default `Number.POSITIVE_INFINITY`)
+- `seqName` **[string](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** name of the sequence you want
+- `regionStart` **[number](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Number)?** optional 0-based half-open start of the sequence region to fetch. (optional, default `0`)
+- `regionEnd` **[number](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Number)?** optional 0-based half-open end of the sequence region to fetch. defaults to end of the sequence (optional, default `Number.POSITIVE_INFINITY`)
 
 Returns **[Promise](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/Promise)** for a string of sequence bases
 

diff --git a/package.json b/package.json
@@ -39,8 +39,7 @@
     "biojs"
   ],
   "dependencies": {
-    "buffer": "^6.0.3",
-    "generic-filehandle": "^3.0.0"
+    "generic-filehandle2": "^0.0.1"
   },
   "devDependencies": {
     "@typescript-eslint/eslint-plugin": "^8.0.1",
@@ -50,7 +49,7 @@
     "eslint": "^9.7.0",
     "eslint-plugin-import": "^2.26.0",
     "eslint-plugin-unicorn": "^56.0.0",
-    "prettier": "^3.3.3",
+    "prettier": "^3.4.2",
     "rimraf": "^6.0.1",
     "standard-changelog": "^6.0.0",
     "typescript": "^5.5.3",

diff --git a/src/twoBitFile.ts b/src/twoBitFile.ts
@@ -1,5 +1,4 @@
-import { LocalFile, GenericFilehandle } from 'generic-filehandle'
-import { Buffer } from 'buffer'
+import { LocalFile, GenericFilehandle } from 'generic-filehandle2'
 
 const TWOBIT_MAGIC = 0x1a412743
 
@@ -9,9 +8,9 @@ const byteTo4Bases = [] as string[]
 for (let index = 0; index < 256; index++) {
   byteTo4Bases.push(
     twoBit[(index >> 6) & 3] +
-    twoBit[(index >> 4) & 3] +
-    twoBit[(index >> 2) & 3] +
-    twoBit[index & 3],
+      twoBit[(index >> 4) & 3] +
+      twoBit[(index >> 2) & 3] +
+      twoBit[index & 3],
   )
 }
 
@@ -46,17 +45,11 @@ export default class TwoBitFile {
   }
 
   async _detectEndianness() {
-    const returnValue = await this.filehandle.read(
-      Buffer.allocUnsafe(8),
-      0,
-      8,
-      0,
-    )
-    const { buffer } = returnValue
-    if (buffer.readInt32LE(0) === TWOBIT_MAGIC) {
-      this.version = buffer.readInt32LE(4)
-    } else if (buffer.readInt32BE(0) === TWOBIT_MAGIC) {
-      throw new Error('big endian not supported')
+    const buffer = await this.filehandle.read(8, 0)
+    const dataView = new DataView(buffer.buffer)
+    const magic = dataView.getInt32(0, true)
+    if (magic === TWOBIT_MAGIC) {
+      this.version = dataView.getInt32(0, true)
     } else {
       throw new Error('not a 2bit file')
     }
@@ -75,14 +68,7 @@ export default class TwoBitFile {
   async _getHeader() {
     await this._detectEndianness()
 
-    const { buffer } = await this.filehandle.read(
-      Buffer.allocUnsafe(16),
-      0,
-      16,
-      0,
-    )
-
-    const b = buffer
+    const b = await this.filehandle.read(16, 0)
     const le = true
     const dataView = new DataView(b.buffer, b.byteOffset, b.length)
     let offset = 0
@@ -119,28 +105,21 @@ export default class TwoBitFile {
     const header = await this.getHeader()
     const maxIndexLength =
       8 + header.sequenceCount * (1 + 256 + (this.version === 1 ? 8 : 4))
-    const { buffer } = await this.filehandle.read(
-      Buffer.allocUnsafe(maxIndexLength),
-      0,
-      maxIndexLength,
-      8,
-    )
+    const b = await this.filehandle.read(maxIndexLength, 8)
 
     const le = true
-    const b = buffer
     const dataView = new DataView(b.buffer, b.byteOffset, b.length)
     let offset = 0
     const sequenceCount = dataView.getUint32(offset, le)
     offset += 4
     // const reserved = dataView.getUint32(offset, le)
     offset += 4
     const indexData = []
+    const decoder = new TextDecoder('utf8')
     for (let i = 0; i < sequenceCount; i++) {
       const nameLength = dataView.getUint8(offset)
       offset += 1
-      const name = buffer
-        .subarray(offset, offset + nameLength)
-        .toString() as string
+      const name = decoder.decode(b.subarray(offset, offset + nameLength))
       offset += nameLength
       if (header.version === 1) {
         const dataOffset = Number(dataView.getBigUint64(offset, le))
@@ -159,16 +138,16 @@ export default class TwoBitFile {
   }
 
   /**
-   * @returns {Promise} for an array of string sequence names that are found in the file
+   * @returns for an array of string sequence names that are found in the file
    */
   async getSequenceNames() {
     const index = await this.getIndex()
     return Object.keys(index)
   }
 
   /**
-   * @returns {Promise} for an object listing the lengths of all sequences like
-   * `{seqName: length, ...}`.
+   * @returns object listing the lengths of all sequences like `{seqName:
+   * length, ...}`.
    *
    * note: this is a relatively slow operation especially if there are many
    * refseqs in the file, if you can get this information from a different file
@@ -177,9 +156,9 @@ export default class TwoBitFile {
   async getSequenceSizes() {
     const index = await this.getIndex()
     const seqNames = Object.keys(index)
-    const sizes = await Promise.all(Object.values(index).map(offset =>
-      this._getSequenceSize(offset),
-    ))
+    const sizes = await Promise.all(
+      Object.values(index).map(offset => this._getSequenceSize(offset)),
+    )
     const returnObject = {} as Record<string, number>
     for (const [index_, seqName] of seqNames.entries()) {
       returnObject[seqName] = sizes[index_]
@@ -188,30 +167,22 @@ export default class TwoBitFile {
   }
 
   /**
-   * @param {string} seqName name of the sequence
-   * @returns {Promise} for the sequence's length, or undefined if it is not in the file
+   * @param seqName name of the sequence
+   *
+   * @returns sequence length, or undefined if it is not in the file
    */
   async getSequenceSize(seqName: string) {
     const index = await this.getIndex()
     const offset = index[seqName]
-    if (!offset) {
-      return undefined
-    }
-    return this._getSequenceSize(offset)
+    return offset ? this._getSequenceSize(offset) : undefined
   }
 
   async _getSequenceSize(offset: number) {
     return this._record1(offset).then(f => f.dnaSize)
   }
 
   async _record1(offset2: number, len = 8) {
-    const { buffer } = await this.filehandle.read(
-      Buffer.allocUnsafe(len),
-      0,
-      len,
-      offset2,
-    )
-    const b = buffer
+    const b = await this.filehandle.read(len, offset2)
     const le = true
     let offset = 0
     const dataView = new DataView(b.buffer, b.byteOffset, b.length)
@@ -224,13 +195,7 @@ export default class TwoBitFile {
   }
 
   async _record2(offset2: number, len: number) {
-    const { buffer } = await this.filehandle.read(
-      Buffer.allocUnsafe(len),
-      0,
-      len,
-      offset2,
-    )
-    const b = buffer
+    const b = await this.filehandle.read(len, offset2)
     const le = true
     let offset = 0
     const dataView = new DataView(b.buffer, b.byteOffset, b.length)
@@ -257,13 +222,7 @@ export default class TwoBitFile {
     }
   }
   async _record3(offset2: number, len: number) {
-    const { buffer } = await this.filehandle.read(
-      Buffer.allocUnsafe(len),
-      0,
-      len,
-      offset2,
-    )
-    const b = buffer
+    const b = await this.filehandle.read(len, offset2)
     const le = true
     let offset = 0
     const dataView = new DataView(b.buffer, b.byteOffset, b.length)
@@ -314,10 +273,15 @@ export default class TwoBitFile {
   }
 
   /**
-   * @param {string} seqName name of the sequence you want
-   * @param {number} [regionStart] optional 0-based half-open start of the sequence region to fetch.
-   * @param {number} [regionEnd] optional 0-based half-open end of the sequence region to fetch. defaults to end of the sequence
-   * @returns {Promise} for a string of sequence bases
+   * @param seqName name of the sequence you want
+   *
+   * @param [regionStart] optional 0-based half-open start of the sequence
+   * region to fetch.
+   *
+   * @param [regionEnd] optional 0-based half-open end of the sequence region
+   * to fetch. defaults to end of the sequence
+   *
+   * @returns for a string of sequence bases
    */
   async getSequence(
     seqName: string,
@@ -353,14 +317,10 @@ export default class TwoBitFile {
       record.maskBlocks.sizes,
     )
 
-    const baseBytes = Buffer.allocUnsafe(
-      Math.ceil((regionEnd - regionStart) / 4) + 1,
-    )
+    const baseBytesLen = Math.ceil((regionEnd - regionStart) / 4) + 1
     const baseBytesOffset = Math.floor(regionStart / 4)
-    const { buffer } = await this.filehandle.read(
-      baseBytes,
-      0,
-      baseBytes.length,
+    const buffer = await this.filehandle.read(
+      baseBytesLen,
       record.dnaPosition + baseBytesOffset,
     )