Documentation Index
Fetch the complete documentation index at: https://docs.trelent.com/llms.txt
Use this file to discover all available pages before exploring further.
Use this quickstart to go from a file on your machine to Markdown output in three steps. It relies on the file connector, which processes files you upload directly to the Data Ingestion API.
Prerequisites
TRELENT_DATA_INGESTION_API_URL pointing to your deploymentTRELENT_DATA_INGESTION_API_TOKEN token with the FILE:UPLOAD permission- A PDF, video, or Office document to ingest (for example
sample.pdf)
Installation
bun add @trelent/data-ingestion
Upload a file
Upload a file to the API. It stores the object under your account and returns a file ID that stays valid until the upload expires.import { DataIngestionClient } from "@trelent/data-ingestion";
import { readFileSync } from "fs";
const client = new DataIngestionClient();
const fileBuffer = readFileSync("sample.pdf");
const blob = new Blob([fileBuffer], { type: "application/pdf" });
const upload = await client.uploadFile(blob, "sample.pdf", { expiresInDays: 30 });
console.log("Uploaded file:", upload.id);
Files expire after 30 days by default. Adjust expiresInDays if you need a shorter retention period.
Start a job with the file connector
Use the returned file ID to create a job. The connector definition references one or more uploaded files and tells the service to pull inputs from managed storage.import { DataIngestionClient } from "@trelent/data-ingestion";
import type { JobInput } from "@trelent/data-ingestion";
const client = new DataIngestionClient();
const job: JobInput = {
connector: {
type: "file_upload",
file_ids: [upload.id],
},
output: {
type: "s3-signed-url",
expires_minutes: 120,
},
};
const response = await client.submitJob(job);
console.log("Job submitted:", response.job_id);
On success, the API responds with a job_id. Capture that ID for status polling.
Poll job status and retrieve outputs
Query the job status until it becomes completed. When complete, the response includes delivery pointers (signed URLs or bucket locations) for generated Markdown and images.import { DataIngestionClient } from "@trelent/data-ingestion";
const client = new DataIngestionClient();
const status = await client.getJobStatus(response.job_id, {
includeMarkdown: false,
includeFileMetadata: true,
});
console.log("Status:", status.status);
if (status.delivery) {
console.log("Deliveries:", Object.keys(status.delivery));
}
If you need to reuse uploads across multiple jobs, call listFiles() to list every file ID still within its retention window.
If a file fails to process, check status.errors for details. Errors are keyed by the input identifier (file ID).
Installation
pip install trelent-data-ingestion
Upload a file
Upload a file to the API. It stores the object under your account and returns a file ID that stays valid until the upload expires.from pathlib import Path
from trelent_data_ingestion_sdk import DataIngestionClient
client = DataIngestionClient()
file_data = Path("sample.pdf").read_bytes()
upload = client.upload_file(file_data, "sample.pdf", content_type="application/pdf", expires_in_days=30)
print("Uploaded file:", upload.id)
Files expire after 30 days by default. Adjust expires_in_days if you need a shorter retention period.
Start a job with the file connector
Use the returned file ID to create a job. The connector definition references one or more uploaded files and tells the service to pull inputs from managed storage.from trelent_data_ingestion_sdk import DataIngestionClient, JobInput, FileUploadConnector, S3SignedUrlOutput
client = DataIngestionClient()
job = JobInput(
connector=FileUploadConnector(file_ids=[upload.id]),
output=S3SignedUrlOutput(expires_minutes=120),
)
response = client.submit_job(job)
print("Job submitted:", response.job_id)
On success, the API responds with a job_id. Capture that ID for status polling.
Poll job status and retrieve outputs
Query the job status until it becomes completed. When complete, the response includes delivery pointers (signed URLs or bucket locations) for generated Markdown and images.from trelent_data_ingestion_sdk import DataIngestionClient
client = DataIngestionClient()
status = client.get_job_status(
response.job_id,
include_markdown=False,
include_file_metadata=True,
)
print("Status:", status.status)
if status.delivery:
print("Deliveries:", list(status.delivery.keys()))
If you need to reuse uploads across multiple jobs, call list_files() to list every file ID still within its retention window.
If a file fails to process, check status.errors for details. Errors are keyed by the input identifier (file ID).