Pipeline used for converting the file to Markdown if the file is a
document (i.e., a file with pdf, doc[x] or ppt[x] extension). The value
identifies the pipeline release and and MUST have the format
{namespaceID}/{pipelineID}@{version}.
The pipeline recipe MUST have the following variable and output fields:
variable:
document_input:
title: document-input
description: Upload a document (PDF/DOCX/DOC/PPTX/PPT)
type: file
The convert_result output should be a list of strings, one per page.
output:
convert_result:
title: convert-result
value: ${merge-markdown-refinement.output.results[0]}
Other variable and output fields will be ignored.
The pipeline will be executed first, falling back to the catalog's
conversion pipelines if the conversion doesn't yield a non-empty result
(see the catalog creation endpoint documentation).
For non-document catalog files, the conversion pipeline is deterministic
(such files are typically trivial to convert and don't require a dedicated
pipeline to improve the conversion performance).
