GcsSource

Cloud Storage location for the input content.

JSON representation 
{
  "inputUris": [
    string
  ],
  "sourceFormat": enum (SourceFormat)
}
Fields
inputUris[]

string

Required. Cloud Storage URIs to input files. How the URIs are read depends on the method: see ImportDocumentsRequest.gcs_source and PurgeDocumentsRequest.gcs_source.

The whole request, including every URI, must not exceed 2,000,000 bytes. Split a larger set of URIs across multiple requests.
sourceFormat

enum (SourceFormat)

Optional. Format of the input files. Used by ImportDocuments; PurgeDocuments ignores it.

SourceFormat

SourceFormat specifies the format of the input content. This enum can be used in GcsSource to notify the service how to parse the input URIs.

Enums
SOURCE_FORMAT_UNSPECIFIED

Default value. Each input URI names a file to import, or a pattern that matches files to import (see ImportDocumentsRequest.gcs_source). Each file is imported as a separate Document.

An UNSTRUCTURED DataStore accepts PDF (.pdf), plain text (.txt), Word (.docx), PowerPoint (.pptx) and Excel (.xlsx, .xlsm) files of up to 200 MB (200,000,000 bytes). A VIDEO DataStore accepts video files of up to 600 GB that the service can decode and that are at most 15 hours long. A file that doesn't meet these requirements is reported in the error report rather than failing the import.
SOURCE_FORMAT_DOCUMENT_JSON

Each input URI names a JSON Lines file: one Document per line, in the JSON format of the Document resource. Each line must set name, which gives the Document ID, and content.gcs_uri, and can set metadata, which must conform to the DataStore's documentMetadataSchema. Patterns are not supported in the input URIs.

The files that the lines reference must meet the same requirements as for SOURCE_FORMAT_UNSPECIFIED. Lines that can't be parsed or that don't conform are reported in the error report and skipped.