Metadata

In JSON mode, LlamaParse will return a data structure representing the parsed object. This is useful for further processing or analysis.

To use this mode, set the result type to “json”.

API

curl -X 'POST' \
  'https://api.cloud.llamaindex.ai/api/v1/parsing/job/<job_id>/result/json'  \
  -H 'accept: application/json' \
  -H 'Content-Type: multipart/form-data' \
  -H "Authorization: Bearer $LLAMA_CLOUD_API_KEY"

Result format

{
    "pages": [
        ..page objects..
    ],
    "job_metadata": {
        "job_pages": int,
        "job_is_cache_hit": boolean
    }
}

Page objects

Within page objects, the following keys may be present depending on your document.

• page: The page number of the document.
• text: The text extracted from the page.
• md: The markdown version of the extracted text.
• images: Any images extracted from the page.
• items: An array of heading, text and table objects in the order they appear on the page.

Retrieving images

Images are returned as an array of image objects, of the form:

{
    "name": "img_p2_5.png",
    "height": 718,
    "width": 251
}

You can retrieve the image extracted directly using the value of the name, like this:

API

curl -X 'POST' \
  'https://api.cloud.llamaindex.ai/api/v1/parsing/job/<job_id>/result/image/<name>'  \
  -H 'accept: application/json' \
  -H 'Content-Type: multipart/form-data' \
  -H "Authorization: Bearer $LLAMA_CLOUD_API_KEY" \
  --output "file.png"

Note the additional --output argument to curl to get the binary saved to a file.

Slide speaker notes

For certain presentation formats, if a slide contains speaker notes, the speaker notes will be extracted and returned in the slideSpeakerNotes entry for the page:

[
  {
    "page": 1,
    "text": "Hello\nSlide with notes",
    "slideSpeakerNotes": "This is a speaker note",
    ...
  },
...

Currently supported formats for slide speaker notes extraction:

• .pptx (PowerPoint 2007+)

For these formats, slide speaker notes extraction is supported in all parse modes except parse_page_with_lvm.