IBM / cloudant-node-sdk

Cloudant SDK for Node.js
Apache License 2.0
27 stars 13 forks source link

Creating attachment gets corrupted for some of the data types #546

Closed kvv002 closed 2 years ago

kvv002 commented 2 years ago

Describe the bug

Create attachment as per the documentation, the attachment gets uploaded but the content gets corrupted. When I open cloudantDB UI, and I try to view my attachment,

  1. The size of the image is reduced.
  2. When I open/Download the image it says wrong format.

To Reproduce

  1. The below snippet for a pdf works fine and attachment content is not corrupted.

    const ContentType = "application/pdf" // --works
    const pdfb64Data = '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';
    const data = {
    db_name: 'v-products',
    doc_id: 'testing',
    name: 'testingwithfile2.pdf',
    content_type: ContentType,
    data: pdfb64Data 
    }
    data.data = Buffer.from(data.data, 'base64')
    const stream = new Readable()
    stream.push(data.data)
    stream.push(null)
    service.putAttachment({
    db: data.db_name,
    docId: data.doc_id,
    attachmentName: data.name,
    attachment: stream,
    contentType: data.content_type,
    rev: '109-9511957b4c370887269a7ec9a5f847a9'
    }).then(response => {
    console.log('CREATE SUCCESS')
    }).catch((err) => {
    console.log('--------INSIDE CATCHHHHHH-----', err)
    })
  2. If I try to upload an image the same way as above, the image in the cloudant gets corrupted, below is my sample base64 image snippet

    const ContentType = 'image/png';
    const content3 = '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'
    const data = {
    db_name: 'v-products',
    doc_id: 'testing',
    name: 'testingwithfile2.png',
    content_type: ContentType,
    data: content3
    }
    data.data = Buffer.from(data.data, 'base64')
    const stream = new Readable()
    stream.push(data.data)
    stream.push(null)
    service.putAttachment({
    db: data.db_name,
    docId: data.doc_id,
    attachmentName: data.name,
    attachment: stream,
    contentType: data.content_type,
    rev: '109-9511957b4c370887269a7ec9a5f847a9'
    }).then(response => {
    console.log('CREATE SUCCESS')
    }).catch((err) => {
    console.log('--------INSIDE CATCHHHHHH-----', err)
    })
  3. upload by using fs.readFile directly, and it has similar behavior as above.

    const ContentType = 'image/jpg'
    const data = {
    db_name: 'v-products',
    doc_id: 'testing',
    name: 'testingwithjpgscript2.jpg',
    content_type: ContentType
    }
    const modelFilePath = path.join(path.join(__dirname, 'image'), 'Blue_morpho_butterfly.jpg')
    fs.readFile(modelFilePath, function (err, file) {
    console.log('----file read', file)
    console.log('---err--', err)
    service.putAttachment({
    db: data.db_name,
    docId: data.doc_id,
    attachmentName: data.name,
    attachment: file,
    contentType: data.content_type,
    rev: '117-11eeb0b8f618be2068508969df7b66bf'
    }).then(response => {
    console.log('CREATE SUCCESS')
    }).catch((err) => {
    console.log('--------INSIDE CATCHHHHHH-----', err)
    })
    })
  4. uploaded using old cloudant driver, and It seems to be working fine.

    const cloudantDriver = require('@cloudant/cloudant')
    const path = require('path')
    const fs = require('fs')
    this.cloudant
    this.cloudant = cloudantDriver({
    url: CLOUDANT_URL,
    plugins: [{ iamauth: { iamApiKey: CLOUDANT_APIKEY } }]
    })
    const data = {
    db_name: 'v-products',
    doc_id: 'testing',
    name: 'JPGOLDSDK.jpg',
    content_type: 'image/jpg'
    }
    const db = this.cloudant.db.use(data.db_name)
    const modelFilePath = path.join(path.join(__dirname, 'image'), 'Blue_morpho_butterfly.jpg')
    fs.readFile(modelFilePath, function (err, file) {
    data.data = file
    const reqOptions = {
        rev : '129-0607166bab7c506cb861c01d237a1141'
    }
    
    db.attachment.insert(data.doc_id, data.name, data.data, data.content_type, reqOptions, function(err,res){
        console.log('----errr-------------',err)
        console.log('-----res-------', res)
    })
    })

    Expected behavior The content of the data should not be corrupted for any file type.

Must gather (please complete the following information):

ricellis commented 2 years ago

I was able to reproduce the issue for both a pdf and a png. I was able to workaround by disabling request body compression and I suggest using that as a workaround for now.

I'll keep investigating the cause of the problem and try to identify if it is a server or client issue with compression.

ricellis commented 2 years ago

I am not able to reproduce the problem using compression with either curl or the Java SDK, so it seems to be specific to the handling of compressed bodies in the Node SDK core or axios. That said we have request body compression on by default and other [non-attachment] endpoints have certainly been working OK so there must be something specific about the attachment operations that trigger it.

ricellis commented 2 years ago

It looks like the node-sdk-core assumes utf-8 during compression which would explain why it works OK for JSON content, but not these binary attachment types (and I guess why it could work for some PDF files). I've asked the node-sdk-core team for their thoughts on the problem and I expect a fix will have to come via the core in due course.

ricellis commented 2 years ago

Just FYI, we're still waiting for an upsteam fix for this.

ricellis commented 2 years ago

I made a contribution upstream https://github.com/IBM/node-sdk-core/pull/182, which we consumed here via #612 and released in version 0.0.23. It fixes the case where attachment is provided as a ReadableStream.

The case where an attachment is provided as a Buffer is still broken - the previous workaround of disabling compression still works or if compression is required a new workaround is to convert Buffer to ReadableStream before using the putAttachment operation.

I will leave this issue open pending a fix for the latter case as well.

ricellis commented 2 years ago

I made a PR upstream to fix the Buffer case as well https://github.com/IBM/node-sdk-core/pull/184

ricellis commented 2 years ago

Second part of the fix came in via https://github.com/IBM/cloudant-node-sdk/pull/616 and will be included in the next release.

ricellis commented 2 years ago

This is fully fixed since version 0.0.24.