Remove file length check pt 2.

eslint.config.mjs

@@ -12,6 +12,7 @@ export default tseslint.config(
       '*.mjs',
       'example/*',
       'src/htscodecs',
+      'coverage',
     ],
   },
   {

src/cramFile/codecs/_base.ts

@@ -41,5 +41,5 @@ export default abstract class CramCodec<
     coreDataBlock: CramFileBlock,
     blocksByContentId: Record<number, CramFileBlock>,
     cursors: Cursors,
-  ): DataTypeMapping[TResult]
+  ): DataTypeMapping[TResult] | undefined
 }

src/cramFile/codecs/byteArrayLength.ts

@@ -32,22 +32,14 @@ export default class ByteArrayStopCodec extends CramCodec<
     cursors: Cursors,
   ) {
     const lengthCodec = this._getLengthCodec()
-    const arrayLength = lengthCodec.decode(
-      slice,
-      coreDataBlock,
-      blocksByContentId,
-      cursors,
-    )
+    const arrayLength =
+      lengthCodec.decode(slice, coreDataBlock, blocksByContentId, cursors) || 0
 
     const dataCodec = this._getDataCodec()
     const data = new Uint8Array(arrayLength)
     for (let i = 0; i < arrayLength; i += 1) {
-      data[i] = dataCodec.decode(
-        slice,
-        coreDataBlock,
-        blocksByContentId,
-        cursors,
-      )
+      data[i] =
+        dataCodec.decode(slice, coreDataBlock, blocksByContentId, cursors) || 0
     }
 
     return data

src/cramFile/codecs/external.ts

@@ -1,5 +1,5 @@
 import CramCodec, { Cursor, Cursors } from './_base'
-import { CramMalformedError, CramUnimplementedError } from '../../errors'
+import { CramUnimplementedError } from '../../errors'
 import { CramFileBlock } from '../file'
 import CramSlice from '../slice'
 import { parseItf8 } from '../util'
@@ -39,13 +39,9 @@ export default class ExternalCodec extends CramCodec<
   ) {
     const { blockContentId } = this.parameters
     const contentBlock = blocksByContentId[blockContentId]
-    if (!contentBlock) {
-      throw new CramMalformedError(
-        `no block found with content ID ${blockContentId}}`,
-      )
-    }
+
     const cursor = cursors.externalBlocks.getCursor(blockContentId)
-    return this._decodeData(contentBlock, cursor)
+    return contentBlock ? this._decodeData(contentBlock, cursor) : undefined
   }
 
   _decodeInt(contentBlock: CramFileBlock, cursor: Cursor) {

src/cramFile/container/index.ts

@@ -21,16 +21,13 @@ export default class CramContainer {
 
     // if there are no records in the container, there will be no compression
     // header
-    if (!containerHeader?.numRecords) {
+    if (!containerHeader.numRecords) {
       return null
     }
     const { majorVersion } = await this.file.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
 
     const block = await this.getFirstBlock()
-    if (block === undefined) {
-      return undefined
-    }
     if (block.contentType !== 'COMPRESSION_HEADER') {
       throw new CramMalformedError(
         `invalid content type ${block.contentType} in compression header block`,
@@ -51,9 +48,6 @@ export default class CramContainer {
 
   async getFirstBlock() {
     const containerHeader = await this.getHeader()
-    if (!containerHeader) {
-      return undefined
-    }
     return this.file.readBlock(containerHeader._endPosition)
   }
 
@@ -78,12 +72,6 @@ export default class CramContainer {
     const { majorVersion } = await this.file.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
     const { cramContainerHeader1, cramContainerHeader2 } = sectionParsers
-    const { size: fileSize } = await this.file.stat()
-
-    if (position >= fileSize) {
-      console.warn(`pos:${position}>=fileSize:${fileSize} in cram container`)
-      return undefined
-    }
 
     // parse the container header. do it in 2 pieces because you cannot tell
     // how much to buffer until you read numLandmarks
@@ -93,13 +81,6 @@ export default class CramContainer {
     )
     const header1 = parseItem(bytes1, cramContainerHeader1.parser)
     const numLandmarksSize = itf8Size(header1.numLandmarks)
-    if (position + header1.length >= fileSize) {
-      // header indicates container goes beyond fileSize
-      console.warn(
-        `container at ${position} is beyond fileSize:${fileSize}, skipping`,
-      )
-      return undefined
-    }
 
     const bytes2 = await this.file.read(
       cramContainerHeader2.maxLength(header1.numLandmarks),
@@ -116,12 +97,12 @@ export default class CramContainer {
       )
     }
 
-    const completeHeader = Object.assign(header1, header2, {
+    return {
+      ...header1,
+      ...header2,
       _size: header1._size + header2._size - numLandmarksSize,
       _endPosition: header1._size + header2._size - numLandmarksSize + position,
-    })
-
-    return completeHeader
+    }
   }
 }
 

src/cramFile/file.ts

@@ -102,12 +102,6 @@ export default class CramFile {
     }
   }
 
-  // can just stat this object like a filehandle
-  stat() {
-    return this.file.stat()
-  }
-
-  // can just stat this object like a filehandle
   read(length: number, position: number) {
     return this.file.read(length, position)
   }
@@ -133,20 +127,17 @@ export default class CramFile {
     }
 
     const firstBlock = await firstContainer.getFirstBlock()
-    if (firstBlock === undefined) {
-      return parseHeaderText('')
-    } else {
-      const content = firstBlock.content
-      const dataView = new DataView(content.buffer)
-      const headerLength = dataView.getInt32(0, true)
-      const textStart = 4
-      const decoder = new TextDecoder('utf8')
-      const text = decoder.decode(
-        content.subarray(textStart, textStart + headerLength),
-      )
-      this.header = text
-      return parseHeaderText(text)
-    }
+
+    const content = firstBlock.content
+    const dataView = new DataView(content.buffer)
+    const headerLength = dataView.getInt32(0, true)
+    const textStart = 4
+    const decoder = new TextDecoder('utf8')
+    const text = decoder.decode(
+      content.subarray(textStart, textStart + headerLength),
+    )
+    this.header = text
+    return parseHeaderText(text)
   }
 
   async getHeaderText() {
@@ -158,35 +149,26 @@ export default class CramFile {
     const { majorVersion } = await this.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
     let position = sectionParsers.cramFileDefinition.maxLength
-    const { size: fileSize } = await this.file.stat()
-    const { cramContainerHeader1 } = sectionParsers
 
     // skip with a series of reads to the proper container
     let currentContainer: CramContainer | undefined
     for (let i = 0; i <= containerNumber; i++) {
       // if we are about to go off the end of the file
       // and have not found that container, it does not exist
-      if (position + cramContainerHeader1.maxLength + 8 >= fileSize) {
-        return undefined
-      }
+      // if (position + cramContainerHeader1.maxLength + 8 >= fileSize) {
+      //   return undefined
+      // }
 
       currentContainer = this.getContainerAtPosition(position)
       const currentHeader = await currentContainer.getHeader()
-      if (!currentHeader) {
-        throw new CramMalformedError(
-          `container ${containerNumber} not found in file`,
-        )
-      }
+
       // if this is the first container, read all the blocks in the container
       // to determine its length, because we cannot trust the container
       // header's given length due to a bug somewhere in htslib
       if (i === 0) {
         position = currentHeader._endPosition
         for (let j = 0; j < currentHeader.numBlocks; j++) {
           const block = await this.readBlock(position)
-          if (block === undefined) {
-            return undefined
-          }
           position = block._endPosition
         }
       } else {
@@ -219,39 +201,41 @@ export default class CramFile {
 
   /**
    * @returns {Promise[number]} the number of containers in the file
+   *
+   * note: this is currently used only in unit tests, and after removing file
+   * length check, relies on a try catch to read return an error to break
    */
   async containerCount(): Promise<number | undefined> {
     const { majorVersion } = await this.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
-    const { size: fileSize } = await this.file.stat()
-    const { cramContainerHeader1 } = sectionParsers
 
     let containerCount = 0
     let position = sectionParsers.cramFileDefinition.maxLength
-    while (position + cramContainerHeader1.maxLength + 8 < fileSize) {
-      const currentHeader =
-        await this.getContainerAtPosition(position).getHeader()
-      if (!currentHeader) {
-        break
-      }
-      // if this is the first container, read all the blocks in the container,
-      // because we cannot trust the container header's given length due to a
-      // bug somewhere in htslib
-      if (containerCount === 0) {
-        position = currentHeader._endPosition
-        for (let j = 0; j < currentHeader.numBlocks; j++) {
-          const block = await this.readBlock(position)
-          if (block === undefined) {
-            return undefined
+    try {
+      // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+      while (true) {
+        const currentHeader =
+          await this.getContainerAtPosition(position).getHeader()
+
+        // if this is the first container, read all the blocks in the container,
+        // because we cannot trust the container header's given length due to a
+        // bug somewhere in htslib
+        if (containerCount === 0) {
+          position = currentHeader._endPosition
+          for (let j = 0; j < currentHeader.numBlocks; j++) {
+            const block = await this.readBlock(position)
+            position = block._endPosition
           }
-          position = block._endPosition
+        } else {
+          // otherwise, just traverse to the next container using the container's
+          // length
+          position += currentHeader._size + currentHeader.length
         }
-      } else {
-        // otherwise, just traverse to the next container using the container's
-        // length
-        position += currentHeader._size + currentHeader.length
+        containerCount += 1
       }
-      containerCount += 1
+    } catch (e) {
+      containerCount--
+      /* do nothing */
     }
 
     return containerCount
@@ -265,11 +249,6 @@ export default class CramFile {
     const { majorVersion } = await this.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
     const { cramBlockHeader } = sectionParsers
-    const { size: fileSize } = await this.file.stat()
-
-    if (position + cramBlockHeader.maxLength >= fileSize) {
-      return undefined
-    }
 
     const buffer = await this.file.read(cramBlockHeader.maxLength, position)
     return parseItem(buffer, cramBlockHeader.parser, 0, position)
@@ -287,16 +266,7 @@ export default class CramFile {
     size = section.maxLength,
     preReadBuffer?: Uint8Array,
   ) {
-    let buffer: Uint8Array
-    if (preReadBuffer) {
-      buffer = preReadBuffer
-    } else {
-      const { size: fileSize } = await this.file.stat()
-      if (position + size >= fileSize) {
-        return undefined
-      }
-      buffer = await this.file.read(size, position)
-    }
+    const buffer = preReadBuffer ?? (await this.file.read(size, position))
     const data = parseItem(buffer, section.parser, 0, position)
     if (data._size !== size) {
       throw new CramMalformedError(
@@ -356,9 +326,6 @@ export default class CramFile {
     const { majorVersion } = await this.getDefinition()
     const sectionParsers = getSectionParsers(majorVersion)
     const blockHeader = await this.readBlockHeader(position)
-    if (blockHeader === undefined) {
-      return undefined
-    }
     const blockContentPosition = blockHeader._endPosition
 
     const d = await this.file.read(
@@ -386,9 +353,6 @@ export default class CramFile {
         sectionParsers.cramBlockCrc32,
         blockContentPosition + blockHeader.compressedSize,
       )
-      if (crc === undefined) {
-        return undefined
-      }
       block.crc32 = crc.crc32
 
       // check the block data crc32

src/cramFile/record.ts

@@ -271,7 +271,7 @@ export default class CramRecord {
 
     this.readGroupId = readGroupId
     this.readName = readName
-    this.sequenceId = sequenceId
+    this.sequenceId = sequenceId!
     this.uniqueId = uniqueId
     this.templateSize = templateSize
     this.alignmentStart = alignmentStart