@gmod/cram 2.0.0 → 2.0.2

package/src/cramFile/slice/index.ts

@@ -199,7 +199,7 @@ export default class CramSlice {
       containerHeader._endPosition + this.containerPosition,
     )
     if (header === undefined) {
-      throw new Error()
+      throw new Error('block header undefined')
     }
     if (header.contentType === 'MAPPED_SLICE_HEADER') {
       const content = parseItem(
@@ -233,7 +233,7 @@ export default class CramSlice {
     for (let i = 0; i < blocks.length; i += 1) {
       const block = await this.file.readBlock(blockPosition)
       if (block === undefined) {
-        throw new Error()
+        throw new Error('block undefined')
       }
       blocks[i] = block
       blockPosition = blocks[i]._endPosition
@@ -270,7 +270,7 @@ export default class CramSlice {
     // read the slice header
     const sliceHeader = (await this.getHeader()).parsedContent
     if (!isMappedSliceHeader(sliceHeader)) {
-      throw new Error()
+      throw new Error('slice header not mapped')
     }
 
     if (sliceHeader.refSeqId < 0) {
@@ -279,11 +279,9 @@ export default class CramSlice {
 
     const compressionScheme = await this.container.getCompressionScheme()
     if (compressionScheme === undefined) {
-      throw new Error()
+      throw new Error('compression scheme undefined')
     }
 
-    // console.log(JSON.stringify(sliceHeader, null, '  '))
-
     if (sliceHeader.refBaseBlockId >= 0) {
       const refBlock = await this.getBlockByContentId(
         sliceHeader.refBaseBlockId,
@@ -349,12 +347,12 @@ export default class CramSlice {
 
     const compressionScheme = await this.container.getCompressionScheme()
     if (compressionScheme === undefined) {
-      throw new Error()
+      throw new Error('compression scheme undefined')
     }
 
     const sliceHeader = await this.getHeader()
     if (sliceHeader === undefined) {
-      throw new Error()
+      throw new Error('slice header undefined')
     }
 
     const blocksByContentId = await this._getBlocksContentIdIndex()
@@ -503,22 +501,23 @@ export default class CramSlice {
             : undefined
         const compressionScheme = await this.container.getCompressionScheme()
         if (compressionScheme === undefined) {
-          throw new Error()
+          throw new Error('compression scheme undefined')
         }
         const refRegions: Record<
           string,
           { id: number; start: number; end: number; seq: string | null }
-        > = {} // seqId => { start, end, seq }
+        > = {}
 
-        // iterate over the records to find the spans of the reference sequences we need to fetch
-        for (let i = 0; i < records.length; i += 1) {
+        // iterate over the records to find the spans of the reference
+        // sequences we need to fetch
+        for (const record of records) {
           const seqId =
-            singleRefId !== undefined ? singleRefId : records[i].sequenceId
+            singleRefId !== undefined ? singleRefId : record.sequenceId
           let refRegion = refRegions[seqId]
           if (!refRegion) {
             refRegion = {
               id: seqId,
-              start: records[i].alignmentStart,
+              start: record.alignmentStart,
               end: -Infinity,
               seq: null,
             }
@@ -526,21 +525,25 @@ export default class CramSlice {
           }
 
           const end =
-            records[i].alignmentStart +
-            (records[i].lengthOnRef || records[i].readLength) -
+            record.alignmentStart +
+            (record.lengthOnRef || record.readLength) -
             1
           if (end > refRegion.end) {
             refRegion.end = end
           }
-          if (records[i].alignmentStart < refRegion.start) {
-            refRegion.start = records[i].alignmentStart
+          if (record.alignmentStart < refRegion.start) {
+            refRegion.start = record.alignmentStart
           }
         }
 
         // fetch the `seq` for all of the ref regions
         await Promise.all(
           Object.values(refRegions).map(async refRegion => {
-            if (refRegion.id !== -1 && refRegion.start <= refRegion.end) {
+            if (
+              refRegion.id !== -1 &&
+              refRegion.start <= refRegion.end &&
+              this.file.fetchReferenceSequenceCallback
+            ) {
               refRegion.seq = await this.file.fetchReferenceSequenceCallback(
                 refRegion.id,
                 refRegion.start,
@@ -551,13 +554,13 @@ export default class CramSlice {
         )
 
         // now decorate all the records with them
-        for (let i = 0; i < records.length; i += 1) {
+        for (const record of records) {
           const seqId =
-            singleRefId !== undefined ? singleRefId : records[i].sequenceId
+            singleRefId !== undefined ? singleRefId : record.sequenceId
           const refRegion = refRegions[seqId]
-          if (refRegion && refRegion.seq) {
+          if (refRegion?.seq) {
             const seq = refRegion.seq
-            records[i].addReferenceSequence(
+            record.addReferenceSequence(
               { ...refRegion, seq },
               compressionScheme,
             )

package/src/cramFile/util.ts

@@ -176,5 +176,5 @@ export function tinyMemoize(_class: any, methodName: any) {
 }
 
 export function sequenceMD5(seq: string) {
-  return md5(seq.toUpperCase().replace(/[^\x21-\x7e]/g, ''))
+  return md5(seq.toUpperCase().replaceAll(/[^\u0021-\u007e]/g, ''))
 }

package/src/index.ts

@@ -1,5 +1,3 @@
-import CramFile, { CramRecord } from './cramFile'
-import IndexedCramFile from './indexedCramFile'
-import CraiIndex from './craiIndex'
-
-export { CramFile, IndexedCramFile, CraiIndex, CramRecord }
+export { default as CramFile, CramRecord } from './cramFile'
+export { default as CraiIndex } from './craiIndex'
+export { default as IndexedCramFile } from './indexedCramFile'

package/src/indexedCramFile.ts

@@ -6,13 +6,13 @@ import { SeqFetch } from './cramFile/file'
 import { Filehandle } from './cramFile/filehandle'
 import { Slice } from './craiIndex'
 
-export type CramFileSource = {
+export interface CramFileSource {
   cramFilehandle?: Filehandle
   cramUrl?: string
   cramPath?: string
 }
 
-export type CramIndexLike = {
+export interface CramIndexLike {
   getEntriesForRange: (
     seqId: number,
     start: number,
@@ -44,17 +44,16 @@ export default class IndexedCramFile {
       | { cram: CramFile }
       | ({
           cram?: undefined
-          seqFetch: SeqFetch
-          checkSequenceMD5: boolean
+          seqFetch?: SeqFetch
+          checkSequenceMD5?: boolean
           cacheSize?: number
         } & CramFileSource)
     ),
   ) {
     // { cram, index, seqFetch /* fasta, fastaIndex */ }) {
-    if (args.cram) {
-      this.cram = args.cram
-    } else {
-      this.cram = new CramFile({
+    this.cram =
+      args.cram ??
+      new CramFile({
         url: args.cramUrl,
         path: args.cramPath,
         filehandle: args.cramFilehandle,
@@ -62,7 +61,6 @@ export default class IndexedCramFile {
         checkSequenceMD5: args.checkSequenceMD5,
         cacheSize: args.cacheSize,
       })
-    }
 
     if (!(this.cram instanceof CramFile)) {
       throw new Error('invalid arguments: no cramfile')
@@ -128,12 +126,12 @@ export default class IndexedCramFile {
     if (opts.viewAsPairs) {
       const readNames: Record<string, number> = {}
       const readIds: Record<string, number> = {}
-      for (let i = 0; i < ret.length; i += 1) {
-        const name = ret[i].readName
+      for (const read of ret) {
+        const name = read.readName
         if (name === undefined) {
-          throw new Error()
+          throw new Error('readName undefined')
         }
-        const id = ret[i].uniqueId
+        const id = read.uniqueId
         if (!readNames[name]) {
           readNames[name] = 0
         }
@@ -147,11 +145,10 @@ export default class IndexedCramFile {
         }
       })
       const matePromises = []
-      for (let i = 0; i < ret.length; i += 1) {
-        const cramRecord = ret[i]
+      for (const cramRecord of ret) {
         const name = cramRecord.readName
         if (name === undefined) {
-          throw new Error()
+          throw new Error('readName undefined')
         }
         if (
           unmatedPairs[name] &&
@@ -169,9 +166,9 @@ export default class IndexedCramFile {
         }
       }
       const mateBlocks = await Promise.all(matePromises)
-      let mateChunks = []
-      for (let i = 0; i < mateBlocks.length; i += 1) {
-        mateChunks.push(...mateBlocks[i])
+      let mateChunks = [] as Slice[]
+      for (const block of mateBlocks) {
+        mateChunks.push(...block)
       }
       // filter out duplicates
       mateChunks = mateChunks
@@ -182,7 +179,7 @@ export default class IndexedCramFile {
         )
 
       const mateRecordPromises = []
-      const mateFeatPromises: Array<Promise<CramRecord[]>> = []
+      const mateFeatPromises: Promise<CramRecord[]>[] = []
 
       const mateTotalSize = mateChunks
         .map(s => s.sliceBytes)
@@ -202,10 +199,9 @@ export default class IndexedCramFile {
         mateRecordPromises.push(recordPromise)
         const featPromise = recordPromise.then(feats => {
           const mateRecs = []
-          for (let i = 0; i < feats.length; i += 1) {
-            const feature = feats[i]
+          for (const feature of feats) {
             if (feature.readName === undefined) {
-              throw new Error()
+              throw new Error('readName undefined')
             }
             if (unmatedPairs[feature.readName] && !readIds[feature.uniqueId]) {
               mateRecs.push(feature)

package/src/io/index.ts

@@ -28,4 +28,6 @@ function open(
   throw new Error('no url, path, or filehandle provided, cannot open')
 }
 
-export { LocalFile, RemoteFile, fromUrl, open }
+export { fromUrl, open }
+
+export { LocalFile, RemoteFile } from 'generic-filehandle'

package/src/sam.ts

@@ -1,20 +1,34 @@
-export type HeaderDataItem = {
-  tag: string
-  data: Array<{ tag: string; value: string }>
-}
-
-export function parseHeaderText(text: string): HeaderDataItem[] {
+export function parseHeaderText(text: string) {
   const lines = text.split(/\r?\n/)
-  const data: HeaderDataItem[] = []
-  lines.forEach(line => {
+  const data: {
+    tag: string
+    data: {
+      tag: string
+      value: string | undefined
+    }[]
+  }[] = []
+  for (const line of lines) {
     const [tag, ...fields] = line.split(/\t/)
-    const parsedFields = fields.map(f => {
-      const [fieldTag, value] = f.split(':', 2)
-      return { tag: fieldTag, value }
-    })
     if (tag) {
-      data.push({ tag: tag.substr(1), data: parsedFields })
+      data.push({
+        tag: tag.slice(1),
+        data: fields.map(f => {
+          const r = f.indexOf(':')
+          return r !== -1
+            ? {
+                tag: f.slice(0, r),
+                value: f.slice(r + 1),
+              }
+            : // @CO lines are not comma separated.
+              // See "samtools view -H c2\#pad.3.0.cram"
+              // so, just store value tag itself
+              {
+                tag: f,
+                value: undefined,
+              }
+        }),
+      })
     }
-  })
+  }
   return data
 }

package/src/typings/binary-parser.d.ts

@@ -1,5 +1,5 @@
 declare module '@gmod/binary-parser' {
-  export type Options = {
+  export interface Options {
     stripNull?: boolean
     formatter?: (item: any) => any
     length?: number | string | ((this: { $parent: unknown }) => void)

package/src/unzip.ts

@@ -1,2 +1 @@
-import { gunzipSync } from 'zlib'
-export { gunzipSync as unzip }
+export { gunzipSync as unzip } from 'zlib'