dify (Version 1.10.1) throws an error when uploading pptx files to the pipeline knowledge base via API

Discussion
1

dify(Version 1.10.1) 通过知识库api上传pptx文件到流水线知识库报错invalid_param但是message为空

调用接口：https://api.dify.ai/v1/datasets/{dataset_id}/document/create-by-file

传递的参数data：

{
    "doc_form":"hierarchical_model",
    "indexing_technique":"high_quality",
    "process_rule": {
        "mode":"hierarchical",
        "rules": {
            "pre_processing_rules": [
                {
                    "id":"remove_extra_spaces",
                    "enabled": true
                }
            ],
            "segmentation": {"max_tokens": 1000, "separator": "\n\n", "chunk_overlap": 0},
            "parent_mode":"paragraph",
            "subchunk_segmentation": {
                "separator":"\n",
                "max_tokens": 256,
                "chunk_overlap":0
            }
        }
    }
}

file：pptx文件

返回的信息

{
	"code": "invalid_param",
	"message": "",
	"status": 400
}

我使用以上方式上传docx等文件时没有问题，上传pptx格式的文件返回invalid_param，但是messgae为空，导致我不知道到底是遇到了什么问题，尝试各种方法修改参数都无法解决

2

The API upload interface does not support pptx.

You might wonder, since the dataset in the pipeline already defines support for pptx format, why the interface does not support it.

Through my personal testing, I found that when uploading knowledge base documents via the API interface, it does not go through the pipeline process.

1 Like
3

I’m also facing this issue now—the API upload document doesn’t go through the pipeline. Is there any good solution for this at the moment?