Skip to content

fix(dynamodb): return LastEvaluatedKey when Limit ends the page #72

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
kylecannon wants to merge 1 commit into
base: master
Choose a base branch
from
Open

fix(dynamodb): return LastEvaluatedKey when Limit ends the page #72

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
178 changes: 145 additions & 33 deletions crates/rustack-dynamodb-core/src/provider.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@ use crate::{
state::{DynamoDBServiceState, DynamoDBTable},
storage::{
KeyAttribute, KeySchema, PrimaryKey, SortKeyCondition, SortableAttributeValue,
TableStorage, calculate_item_size, extract_primary_key, partition_key_segment,
TableStorage, calculate_item_size, extract_primary_key, partition_key_segment, take_page,
},
stream::{
ChangeEvent, ChangeEventName, NoopStreamEmitter, NoopStreamLifecycle, StreamEmitter,
Expand Down Expand Up @@ -1662,7 +1662,12 @@ impl RustackDynamoDB {
} else {
let start_pk = extract_primary_key(&table.key_schema, &input.exclusive_start_key)
.map_err(storage_error_to_dynamodb)?;
start_pk.sort_key
// Items of a table without a sort key are stored under the sentinel.
Some(
start_pk
.sort_key
.unwrap_or(SortableAttributeValue::Sentinel),
)
};

let (items, last_key_sort) = table.storage.query(
Expand Down Expand Up @@ -4840,9 +4845,6 @@ fn build_last_evaluated_key(
key
}

/// Maximum response size for a single Query/Scan response (1 MB).
const MAX_RESPONSE_BYTES: u64 = 1_048_576;

/// Result type for GSI query operations.
type GsiQueryResult = (
Vec<HashMap<String, AttributeValue>>,
Expand Down Expand Up @@ -4891,8 +4893,7 @@ fn query_gsi(
);
}

// Apply limit and 1 MB size cap.
let (selected, has_more) = gsi_apply_limit(matching, limit);
let (selected, has_more) = take_page(matching, limit, calculate_item_size);

// Build last_evaluated_key containing both GSI keys and table primary keys.
let last_evaluated_key = if has_more {
Expand Down Expand Up @@ -5011,30 +5012,6 @@ fn gsi_apply_pagination(
}
}

/// Apply count limit and 1 MB size cap, returning `(selected_items, has_more)`.
fn gsi_apply_limit(
items: Vec<HashMap<String, AttributeValue>>,
limit: Option<usize>,
) -> (Vec<HashMap<String, AttributeValue>>, bool) {
let effective_limit = limit.unwrap_or(usize::MAX);
let mut selected: Vec<HashMap<String, AttributeValue>> = Vec::new();
let mut cumulative_size: u64 = 0;

for item in items {
if selected.len() >= effective_limit {
return (selected, true);
}
let item_size = calculate_item_size(&item);
if !selected.is_empty() && cumulative_size + item_size > MAX_RESPONSE_BYTES {
return (selected, true);
}
cumulative_size += item_size;
selected.push(item);
}

(selected, false)
}

/// Build the `last_evaluated_key` for a GSI query response.
///
/// DynamoDB includes both the GSI keys and the table's primary keys in the
Expand Down Expand Up @@ -5087,8 +5064,8 @@ mod tests {
DescribeTableInput, UpdateContinuousBackupsInput, UpdateItemInput,
},
types::{
AttributeDefinition, KeySchemaElement, KeyType, PutRequest, ScalarAttributeType,
WriteRequest,
AttributeDefinition, GlobalSecondaryIndex, KeySchemaElement, KeyType, Projection,
ProjectionType, PutRequest, ScalarAttributeType, WriteRequest,
},
};

Expand Down Expand Up @@ -5622,4 +5599,139 @@ mod tests {
)])
);
}

#[test]
fn test_should_return_last_key_when_limit_ends_on_last_gsi_item() {
let provider = RustackDynamoDB::new(DynamoDBConfig::default());
provider
.handle_create_table(CreateTableInput {
table_name: "GsiTable".to_owned(),
key_schema: vec![KeySchemaElement {
attribute_name: "pk".to_owned(),
key_type: KeyType::Hash,
}],
attribute_definitions: vec![
AttributeDefinition {
attribute_name: "pk".to_owned(),
attribute_type: ScalarAttributeType::S,
},
AttributeDefinition {
attribute_name: "gpk".to_owned(),
attribute_type: ScalarAttributeType::S,
},
AttributeDefinition {
attribute_name: "gsk".to_owned(),
attribute_type: ScalarAttributeType::N,
},
],
global_secondary_indexes: vec![GlobalSecondaryIndex {
index_name: "ByGroup".to_owned(),
key_schema: vec![
KeySchemaElement {
attribute_name: "gpk".to_owned(),
key_type: KeyType::Hash,
},
KeySchemaElement {
attribute_name: "gsk".to_owned(),
key_type: KeyType::Range,
},
],
projection: Projection {
projection_type: Some(ProjectionType::All),
non_key_attributes: Vec::new(),
},
provisioned_throughput: None,
}],
billing_mode: Some(BillingMode::PayPerRequest),
..Default::default()
})
.unwrap();
for i in 1..=4 {
provider
.handle_put_item(PutItemInput {
table_name: "GsiTable".to_owned(),
item: HashMap::from([
("pk".to_owned(), AttributeValue::S(format!("item{i}"))),
("gpk".to_owned(), AttributeValue::S("group".to_owned())),
("gsk".to_owned(), AttributeValue::N(i.to_string())),
]),
..Default::default()
})
.unwrap();
}
let query = |exclusive_start_key| {
provider
.handle_query(QueryInput {
table_name: "GsiTable".to_owned(),
index_name: Some("ByGroup".to_owned()),
key_condition_expression: Some("gpk = :g".to_owned()),
expression_attribute_values: HashMap::from([(
":g".to_owned(),
AttributeValue::S("group".to_owned()),
)]),
limit: Some(2),
exclusive_start_key,
..Default::default()
})
.unwrap()
};

let first = query(HashMap::new());
assert_eq!(first.count, 2);
assert!(!first.last_evaluated_key.is_empty());

// Limit stops this page on the last item: DynamoDB still returns a key.
let second = query(first.last_evaluated_key);
assert_eq!(second.count, 2);
assert_eq!(
second.last_evaluated_key,
HashMap::from([
("gpk".to_owned(), AttributeValue::S("group".to_owned())),
("gsk".to_owned(), AttributeValue::N("4".to_owned())),
("pk".to_owned(), AttributeValue::S("item4".to_owned())),
])
);

let third = query(second.last_evaluated_key);
assert_eq!(third.count, 0);
assert!(third.last_evaluated_key.is_empty());
}

#[test]
fn test_should_end_query_on_partition_key_only_table() {
let provider = setup_provider_with_table();
provider
.handle_put_item(PutItemInput {
table_name: "TestTable".to_owned(),
item: HashMap::from([("pk".to_owned(), AttributeValue::S("k1".to_owned()))]),
..Default::default()
})
.unwrap();
let query = |exclusive_start_key| {
provider
.handle_query(QueryInput {
table_name: "TestTable".to_owned(),
key_condition_expression: Some("pk = :p".to_owned()),
expression_attribute_values: HashMap::from([(
":p".to_owned(),
AttributeValue::S("k1".to_owned()),
)]),
limit: Some(1),
exclusive_start_key,
..Default::default()
})
.unwrap()
};

let first = query(HashMap::new());
assert_eq!(first.count, 1);
assert_eq!(
first.last_evaluated_key,
HashMap::from([("pk".to_owned(), AttributeValue::S("k1".to_owned()))])
);

let second = query(first.last_evaluated_key);
assert_eq!(second.count, 0);
assert!(second.last_evaluated_key.is_empty());
}
}
Loading