list_datasets()
Discover all datasets visible to you — public datasets plus any private datasets you’ve been granted access to.
datasets = await client.list_datasets()
for ds in datasets:
print(f"{ds.name} (ID: {ds.id})")
print(f" Price: {ds.price_per_chunk} {ds.currency}/chunk")
print(f" Queryable: {ds.queryable}")
print(f" Subscribed: {ds.subscribed}")
Dataset Fields
| Field | Type | Description |
|---|
id | str | Dataset UUID |
seller_name | str | Name of the seller |
name | str | Dataset name |
description | str | What the dataset contains |
price_per_chunk | str | Price per result, 8 decimal places |
visibility | str | "public" or "private" |
metadata_schema | list[dict] | None | Available filter fields |
queryable | bool | True if ready to query (connector wired) |
subscribed | bool | True if you have access |
query()
Search across one or more datasets. Provide either text (natural language) or vector (pre-computed embedding).
Text query (recommended)
response = await client.query(
text="machine learning transformers",
dataset_ids=["uuid-1", "uuid-2"],
top_k=10,
)
print(f"Query ID: {response.query_id}")
for result in response.results:
print(f"[{result.score:.4f}] {result.metadata}")
Pre-computed vector
response = await client.query(
vector=[0.012, -0.034, 0.056, ...],
dataset_ids=["uuid-1"],
top_k=10,
)
All datasets must use the same embedding model when using pre-computed vectors. The server validates vector dimensions match.
response = await client.query(
text="privacy regulations",
dataset_ids=["ds-aaa", "ds-bbb"],
filters={
"ds-aaa": {"year": {"$gte": 2023}},
"ds-bbb": {
"$and": [
{"category": {"$eq": "legal"}},
{"status": {"$in": ["published", "reviewed"]}},
]
},
},
)
Response Fields
| Field | Type | Description |
|---|
query_id | str | UUID for audit trail |
results | list[QueryResult] | Ranked results |
warnings | list[str] | Non-fatal issues (e.g., a dataset timed out) |
QueryResult:
| Field | Type | Description |
|---|
dataset_id | str | Which dataset this came from |
id | str | Vector ID in the seller’s DB |
score | float | Relevance score (higher = better) |
metadata | dict | Key-value pairs from the seller’s vector DB |
embedding_model | str | None | Only present in multi-model queries |
Error Handling
from datagate import (
DatagateClient,
DatagateError,
AuthenticationError,
InsufficientBalanceError,
ValidationError,
NotFoundError,
)
try:
response = await client.query(text="...", dataset_ids=["..."])
except AuthenticationError:
print("Invalid API key")
except InsufficientBalanceError as e:
print(f"Balance: {e.balance}, need: {e.estimated_cost}")
except ValidationError as e:
print(f"Bad request: {e.message}")
except DatagateError as e:
print(f"API error ({e.status_code}): {e.message}")
| Exception | Status | When |
|---|
ValidationError | 400 | Bad input |
AuthenticationError | 401 | Invalid credential |
InsufficientBalanceError | 402 | Balance too low |
ForbiddenError | 403 | Wrong role |
NotFoundError | 404 | Resource not found |
ServerError | 500 | Server error |
