The metadata lacks the base64 encoding of the image.

[mggger]

version: 0.17.1

please see the picture, and api call is

const res = await client.general.partition({
        partitionParameters: {
            files: {
                content: data,
                fileName: filename,
            },
            strategy: 'hi_res',
            languages: ['eng'],
            splitPdfPage: true,
            splitPdfAllowFailed: true,
            splitPdfConcurrencyLevel: 15,
            extractImageBlockTypes: ["Image","Table"]
        }
    });

return res.elements;

[awalker4]

Hi @mggger we identified a bug where extractImageBlockTypes was not sent by the SDK correctly. This is fixed in #122 and a v0.18.1 should be published shortly.

[mggger]

@awalker4 Thanks!

[awalker4]

0.18.1 is published!

[spencermize]

I'm still struggling with this on 0.18.2.

Here are my parameters:

  const result = await this.client.general.partition({
    partitionParameters: {
      files: {
        content: buffer,
        fileName: filename,
      },
      maxCharacters: 1000,
      strategy: (i've tried multple),
      includePageBreaks: true,
      splitPdfPage: false,
      extractImageBlockTypes: ['Image', 'Table'],
    },
  })

Response:

[giladrom]

From what I can tell, the python SDK has a specific "extract_images_in_pdf" whereas the JS SDK does not. I'm also missing all base64 information in the reply I get. Most likely isn't being generated.

[mchamoudadev]

Hey team, is there any progress in the JS/TS SDK? The image information isn't exposed in the API response — it only returns tables and plain text.

Please correct this.