Package google.firestore.v1

map<string, Value>

The result of the aggregation functions, ex: COUNT(*) AS total_docs.

The key is the alias assigned to the aggregation function on input and the size of this map equals the number of aggregation functions in the query.

Value

Values in the array.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

string

The names of the documents to retrieve. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}. The request will fail if any of the document is not a child resource of the given database. Duplicate names will be elided.

DocumentMask

The fields to return. If not set, returns all fields.

If a document has a field that is not present in this mask, that field will not be returned in the response.

bytes

Reads documents in a transaction.

TransactionOptions

Starts a new transaction and reads the documents. Defaults to a read-only transaction. The new transaction ID will be returned as the first response in the stream.

Timestamp

Reads documents as they were at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

bytes

The transaction that was started as part of this request. Will only be set in the first response, and only if BatchGetDocumentsRequest.new_transaction was set in the request.

Timestamp

The time at which the document was read. This may be monotically increasing, in this case the previous documents in the result stream are guaranteed not to have changed between their read_time and this one.

Document

A document that was requested.

string

A document name that was requested but does not exist. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

Write

The writes to apply.

Method does not apply writes atomically and does not guarantee ordering. Each write succeeds or fails independently. You cannot write to the same document more than once per request.

map<string, string>

Labels associated with this batch write.

WriteResult

The result of applying the writes.

This i-th write result corresponds to the i-th write in the request.

Status

The status of applying the writes.

This i-th write status corresponds to the i-th write in the request.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

TransactionOptions

The options for the transaction. Defaults to a read-write transaction.

bytes

The transaction that was started.

bytes

The bytes that encode the bit sequence. May have a length of zero.

int32

The number of bits of the last byte in bitmap to ignore as "padding". If the length of bitmap is zero, then this value must be 0. Otherwise, this value must be between 0 and 7, inclusive.

BitSequence

The bloom filter data.

int32

The number of hashes used by the algorithm.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

Write

The writes to apply.

Always executed atomically and in order.

bytes

If set, applies all writes in this transaction, and commits it.

WriteResult

The result of applying the writes.

This i-th write result corresponds to the i-th write in the request.

Timestamp

The time at which the commit occurred. Any read with an equal or greater read_time is guaranteed to see the effects of the commit.

string

Required. The parent resource. For example: projects/{project_id}/databases/{database_id}/documents or projects/{project_id}/databases/{database_id}/documents/chatrooms/{chatroom_id}

string

Required. The collection ID, relative to parent, to list. For example: chatrooms.

string

The client-assigned document ID to use for this document.

Optional. If not specified, an ID will be assigned by the service.

Document

Required. The document to create. name must not be set.

DocumentMask

The fields to return. If not set, returns all fields.

If the document has a field that is not present in this mask, that field will not be returned in the response.

Value

The values that represent a position, in the order they appear in the order by clause of a query.

Can contain fewer values than specified in the order by clause.

bool

If the position is just before or just after the given values, relative to the sort order defined by the query.

string

Required. The resource name of the Document to delete. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}.

Precondition

An optional precondition on the document. The request will fail if this is set and not met by the target document.

string

The resource name of the document, for example projects/{project_id}/databases/{database_id}/documents/{document_path}.

map<string, Value>

Timestamp

Output only. The time at which the document was created.

This value increases monotonically when a document is deleted then recreated. It can also be compared to values from other documents and the read_time of a query.

Timestamp

Output only. The time at which the document was last changed.

This value is initially set to the create_time then increases monotonically with each change to the document. It can also be compared to values from other documents and the read_time of a query.

Document

The new state of the Document.

If mask is set, contains only fields that were updated or added.

int32

A set of target IDs of targets that match this document.

int32

A set of target IDs for targets that no longer match this document.

string

The resource name of the Document that was deleted.

int32

A set of target IDs for targets that previously matched this entity.

Timestamp

The read timestamp at which the delete was observed.

Greater or equal to the commit_time of the delete.

string

The list of field paths in the mask. See Document.fields for a field path syntax reference.

string

The resource name of the Document that has gone out of view.

int32

A set of target IDs for targets that previously matched this document.

Timestamp

The read timestamp at which the remove was observed.

Greater or equal to the commit_time of the change/delete/remove.

string

The name of the document to transform.

FieldTransform

The list of transformations to apply to the fields of the document, in order. This must not be empty.

string

The path of the field. See Document.fields for the field path syntax reference.

ServerValue

Sets the field to the given server value.

Value

Adds the given value to the field's current value.

This must be an integer or a double value. If the field is not an integer or double, or if the field does not yet exist, the transformation will set the field to the given value. If either of the given value or the current field value are doubles, both values will be interpreted as doubles. Double arithmetic and representation of double values follow IEEE 754 semantics. If there is positive/negative integer overflow, the field is resolved to the largest magnitude positive/negative integer.

Value

Sets the field to the maximum of its current value and the given value.

This must be an integer or a double value. If the field is not an integer or double, or if the field does not yet exist, the transformation will set the field to the given value. If a maximum operation is applied where the field and the input value are of mixed types (that is - one is an integer and one is a double) the field takes on the type of the larger operand. If the operands are equivalent (e.g. 3 and 3.0), the field does not change. 0, 0.0, and -0.0 are all zero. The maximum of a zero stored value and zero input value is always the stored value. The maximum of any numeric value x and NaN is NaN.

Value

Sets the field to the minimum of its current value and the given value.

This must be an integer or a double value. If the field is not an integer or double, or if the field does not yet exist, the transformation will set the field to the input value. If a minimum operation is applied where the field and the input value are of mixed types (that is - one is an integer and one is a double) the field takes on the type of the smaller operand. If the operands are equivalent (e.g. 3 and 3.0), the field does not change. 0, 0.0, and -0.0 are all zero. The minimum of a zero stored value and zero input value is always the stored value. The minimum of any numeric value x and NaN is NaN.

ArrayValue

Append the given elements in order if they are not already present in the current field value. If the field is not an array, or if the field does not yet exist, it is first set to the empty array.

Equivalent numbers of different types (e.g. 3L and 3.0) are considered equal when checking if a value is missing. NaN is equal to NaN, and Null is equal to Null. If the input contains multiple equivalent values, only the first will be considered.

The corresponding transform_result will be the null value.

ArrayValue

Remove all of the given elements from the array in the field. If the field is not an array, or if the field does not yet exist, it is set to the empty array.

Equivalent numbers of the different types (e.g. 3L and 3.0) are considered equal when deciding whether an element should be removed. NaN is equal to NaN, and Null is equal to Null. This will remove all equivalent values if there are duplicates.

The corresponding transform_result will be the null value.

int64

Total number of results returned, including documents, projections, aggregation results, keys.

Duration

Total time to execute the query in the backend.

int64

Total billable read operations.

Struct

Debugging statistics from the execution of the query. Note that the debugging stats are subject to change as Firestore evolves. It could include: { "indexes_entries_scanned": "1000", "documents_scanned": "20", "billing_details" : { "documents_billable": "20", "index_entries_billable": "1000", "min_query_cost": "0" } }

int32

The target ID to which this filter applies.

int32

The total count of documents that match target_id.

If different from the count of documents in the client that match, the client must manually determine which documents no longer match the target.

The client can use the unchanged_names bloom filter to assist with this determination by testing ALL the document names against the filter; if the document name is NOT in the filter, it means the document no longer matches the target.

BloomFilter

A bloom filter that, despite its name, contains the UTF-8 byte encodings of the resource names of ALL the documents that match target_id, in the form projects/{project_id}/databases/{database_id}/documents/{document_path}.

This bloom filter may be omitted at the server's discretion, such as if it is deemed that the client will not make use of it or if it is too computationally expensive to calculate or transmit. Clients must gracefully handle this field being absent by falling back to the logic used before this field existed; that is, re-add the target without a resume token to figure out which documents in the client's cache are out of sync.

PlanSummary

Planning phase information for the query.

ExecutionStats

Aggregated stats from the execution of the query. Only present when ExplainOptions.analyze is set to true.

bool

Optional. Whether to execute this query.

When false (the default), the query will be planned, returning only metrics from the planning stages.

When true, the query will be planned and executed, returning the full query results along with both planning and execution stage metrics.

string

Required. The resource name of the Document to get. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}.

DocumentMask

The fields to return. If not set, returns all fields.

If the document has a field that is not present in this mask, that field will not be returned in the response.

bytes

Reads the document in a transaction.

Timestamp

Reads the version of the document at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

string

Required. The parent document. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}. For example: projects/my-project/databases/my-database/documents/chatrooms/my-chatroom

int32

The maximum number of results to return.

string

A page token. Must be a value from ListCollectionIdsResponse.

Timestamp

Reads documents as they were at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

string

The collection ids.

string

A page token that may be used to continue the list.

string

Required. The parent resource name. In the format: projects/{project_id}/databases/{database_id}/documents or projects/{project_id}/databases/{database_id}/documents/{document_path}.

For example: projects/my-project/databases/my-database/documents or projects/my-project/databases/my-database/documents/chatrooms/my-chatroom

string

Optional. The collection ID, relative to parent, to list.

For example: chatrooms or messages.

This is optional, and when not provided, Firestore will list documents from all collections under the provided parent.

int32

Optional. The maximum number of documents to return in a single response.

Firestore may return fewer than this value.

string

Optional. A page token, received from a previous ListDocuments response.

Provide this to retrieve the subsequent page. When paginating, all other parameters (with the exception of page_size) must match the values set in the request that generated the page token.

string

Optional. The optional ordering of the documents to return.

For example: priority desc, __name__ desc.

This mirrors the ORDER BY used in Firestore queries but in a string representation. When absent, documents are ordered based on __name__ ASC.

DocumentMask

Optional. The fields to return. If not set, returns all fields.

If a document has a field that is not present in this mask, that field will not be returned in the response.

bool

If the list should show missing documents.

A document is missing if it does not exist, but there are sub-documents nested underneath it. When true, such missing documents will be returned with a key but will not have fields, create_time, or update_time set.

Requests with show_missing may not specify where or order_by.

bytes

Perform the read as part of an already active transaction.

Timestamp

Perform the read at the provided time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

Document

The Documents found.

string

A token to retrieve the next page of documents.

If this field is omitted, there are no subsequent pages.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

map<string, string>

Labels associated with this target change.

Target

A target to add to this stream.

int32

The ID of a target to remove from this stream.

TargetChange

Targets have changed.

DocumentChange

A Document has changed.

DocumentDelete

A Document has been deleted.

DocumentRemove

A Document has been removed from a target (because it is no longer relevant to that target).

ExistenceFilter

A filter to apply to the set of documents previously returned for the given target.

Returned when documents may have been removed from the given target, but the exact documents are unknown.

map<string, Value>

The map's fields.

The map keys represent field names. Field names matching the regular expression __.*__ are reserved. Reserved field names are forbidden except in certain documented contexts. The map keys, represented as UTF-8, must not exceed 1,500 bytes and cannot be empty.

string

Required. The parent resource name. In the format: projects/{project_id}/databases/{database_id}/documents. Document resource names are not supported; only database resource names can be specified.

int64

The desired maximum number of partition points. The partitions may be returned across multiple pages of results. The number must be positive. The actual number of partitions returned may be fewer.

For example, this may be set to one fewer than the number of parallel queries to be run, or in running a data pipeline job, one fewer than the number of workers or compute instances available.

string

The next_page_token value returned from a previous call to PartitionQuery that may be used to get an additional set of results. There are no ordering guarantees between sets of results. Thus, using multiple sets of results will require merging the different result sets.

For example, two subsequent calls using a page_token may return:

• cursor B, cursor M, cursor Q
• cursor A, cursor U, cursor W

To obtain a complete result set ordered with respect to the results of the query supplied to PartitionQuery, the results sets should be merged: cursor A, cursor B, cursor M, cursor Q, cursor U, cursor W

int32

The maximum number of partitions to return in this call, subject to partition_count.

For example, if partition_count = 10 and page_size = 8, the first call to PartitionQuery will return up to 8 partitions and a next_page_token if more results exist. A second call to PartitionQuery will return up to 2 partitions, to complete the total of 10 specified in partition_count.

StructuredQuery

A structured query. Query must specify collection with all descendants and be ordered by name ascending. Other filters, order bys, limits, offsets, and start/end cursors are not supported.

Timestamp

Reads documents as they were at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

Cursor

Partition results. Each partition is a split point that can be used by RunQuery as a starting or end point for the query results. The RunQuery requests must be made with the same query supplied to this PartitionQuery request. The partition cursors will be ordered according to same ordering as the results of the query supplied to PartitionQuery.

For example, if a PartitionQuery request returns partition cursors A and B, running the following three queries will return the entire result set of the original query:

• query, end_at A
• query, start_at A, end_at B
• query, start_at B

An empty result may indicate that the query has too few results to be partitioned, or that the query is not yet supported for partitioning.

string

A page token that may be used to request an additional set of results, up to the number specified by partition_count in the PartitionQuery request. If blank, there are no more results.

Struct

The indexes selected for the query. For example: [ {"query_scope": "Collection", "properties": "(foo ASC, name ASC)"}, {"query_scope": "Collection", "properties": "(bar ASC, name ASC)"} ]

bool

When set to true, the target document must exist. When set to false, the target document must not exist.

Timestamp

When set, the target document must exist and have been last updated at that time. Timestamp must be microsecond aligned.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}.

bytes

Required. The transaction to roll back.

string

Required. The parent resource name. In the format: projects/{project_id}/databases/{database_id}/documents or projects/{project_id}/databases/{database_id}/documents/{document_path}. For example: projects/my-project/databases/my-database/documents or projects/my-project/databases/my-database/documents/chatrooms/my-chatroom

ExplainOptions

Optional. Explain options for the query. If set, additional query statistics will be returned. If not, only query results will be returned.

StructuredAggregationQuery

An aggregation query.

bytes

Run the aggregation within an already active transaction.

The value here is the opaque transaction ID to execute the query in.

TransactionOptions

Starts a new transaction as part of the query, defaulting to read-only.

The new transaction ID will be returned as the first response in the stream.

Timestamp

Executes the query at the given timestamp.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

AggregationResult

A single aggregation result.

Not present when reporting partial progress.

bytes

The transaction that was started as part of this request.

Only present on the first response when the request requested to start a new transaction.

Timestamp

The time at which the aggregate result was computed. This is always monotonically increasing; in this case, the previous AggregationResult in the result stream are guaranteed not to have changed between their read_time and this one.

If the query returns no results, a response with read_time and no result will be sent, and this represents the time at which the query was run.

ExplainMetrics

Query explain metrics. This is only present when the RunAggregationQueryRequest.explain_options is provided, and it is sent only once with the last response in the stream.

string

Required. The parent resource name. In the format: projects/{project_id}/databases/{database_id}/documents or projects/{project_id}/databases/{database_id}/documents/{document_path}. For example: projects/my-project/databases/my-database/documents or projects/my-project/databases/my-database/documents/chatrooms/my-chatroom

ExplainOptions

Optional. Explain options for the query. If set, additional query statistics will be returned. If not, only query results will be returned.

StructuredQuery

A structured query.

bytes

Run the query within an already active transaction.

The value here is the opaque transaction ID to execute the query in.

TransactionOptions

Starts a new transaction and reads the documents. Defaults to a read-only transaction. The new transaction ID will be returned as the first response in the stream.

Timestamp

Reads documents as they were at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

bytes

The transaction that was started as part of this request. Can only be set in the first response, and only if RunQueryRequest.new_transaction was set in the request. If set, no other fields will be set in this response.

Document

A query result, not set when reporting partial progress.

Timestamp

The time at which the document was read. This may be monotonically increasing; in this case, the previous documents in the result stream are guaranteed not to have changed between their read_time and this one.

If the query returns no results, a response with read_time and no document will be sent, and this represents the time at which the query was run.

int32

The number of results that have been skipped due to an offset between the last response and the current response.

ExplainMetrics

Query explain metrics. This is only present when the RunQueryRequest.explain_options is provided, and it is sent only once with the last response in the stream.

bool

If present, Firestore has completely finished the request and no more documents will be returned.

Aggregation

Optional. Series of aggregations to apply over the results of the structured_query.

Requires:

• A minimum of one and maximum of five aggregations per query.

StructuredQuery

Nested structured query.

string

Optional. Optional name of the field to store the result of the aggregation into.

If not provided, Firestore will pick a default name following the format field_<incremental_id++>. For example:

AGGREGATE
  COUNT_UP_TO(1) AS count_up_to_1,
  COUNT_UP_TO(2),
  COUNT_UP_TO(3) AS count_up_to_3,
  COUNT(*)
OVER (
  ...
);

becomes:

AGGREGATE
  COUNT_UP_TO(1) AS count_up_to_1,
  COUNT_UP_TO(2) AS field_1,
  COUNT_UP_TO(3) AS count_up_to_3,
  COUNT(*) AS field_2
OVER (
  ...
);

Requires:

• Must be unique across all aggregation aliases.
• Conform to document field name limitations.

Count

Count aggregator.

Sum

Sum aggregator.

Avg

Average aggregator.

FieldReference

The field to aggregate on.

Int64Value

Optional. Optional constraint on the maximum number of documents to count.

This provides a way to set an upper bound on the number of documents to scan, limiting latency, and cost.

Unspecified is interpreted as no bound.

High-Level Example:

AGGREGATE COUNT_UP_TO(1000) OVER ( SELECT * FROM k );

Requires:

• Must be greater than zero when present.

FieldReference

The field to aggregate on.

Projection

Optional sub-set of the fields to return.

This acts as a DocumentMask over the documents returned from a query. When not set, assumes that the caller wants all fields returned.

CollectionSelector

The collections to query.

Filter

The filter to apply.

Order

The order to apply to the query results.

Firestore allows callers to provide a full ordering, a partial ordering, or no ordering at all. In all cases, Firestore guarantees a stable ordering through the following rules:

• The order_by is required to reference all fields used with an inequality filter.
• All fields that are required to be in the order_by but are not already present are appended in lexicographical ordering of the field name.
• If an order on __name__ is not specified, it is appended by default.

Fields are appended with the same sort direction as the last order specified, or 'ASCENDING' if no order was specified. For example:

• ORDER BY a becomes ORDER BY a ASC, __name__ ASC
• ORDER BY a DESC becomes ORDER BY a DESC, __name__ DESC
• WHERE a > 1 becomes WHERE a > 1 ORDER BY a ASC, __name__ ASC
• WHERE __name__ > ... AND a > 1 becomes WHERE __name__ > ... AND a > 1 ORDER BY a ASC, __name__ ASC

Cursor

A potential prefix of a position in the result set to start the query at.

The ordering of the result set is based on the ORDER BY clause of the original query.

SELECT * FROM k WHERE a = 1 AND b > 2 ORDER BY b ASC, __name__ ASC;

This query's results are ordered by (b ASC, __name__ ASC).

Cursors can reference either the full ordering or a prefix of the location, though it cannot reference more fields than what are in the provided ORDER BY.

Continuing off the example above, attaching the following start cursors will have varying impact:

• START BEFORE (2, /k/123): start the query right before a = 1 AND b > 2 AND __name__ > /k/123.
• START AFTER (10): start the query right after a = 1 AND b > 10.

Unlike OFFSET which requires scanning over the first N results to skip, a start cursor allows the query to begin at a logical position. This position is not required to match an actual result, it will scan forward from this position to find the next document.

Requires:

• The number of values cannot be greater than the number of fields specified in the ORDER BY clause.

Cursor

A potential prefix of a position in the result set to end the query at.

This is similar to START_AT but with it controlling the end position rather than the start position.

Requires:

• The number of values cannot be greater than the number of fields specified in the ORDER BY clause.

int32

The number of documents to skip before returning the first result.

This applies after the constraints specified by the WHERE, START AT, & END AT but before the LIMIT clause.

Requires:

• The value must be greater than or equal to zero if specified.

Int32Value

The maximum number of results to return.

Applies after all other constraints.

Requires:

• The value must be greater than or equal to zero if specified.

FindNearest

Optional. A potential nearest neighbors search.

Applies after all other filters and ordering.

Finds the closest vector embeddings to the given query vector.

string

The collection ID. When set, selects only collections with this ID.

bool

When false, selects only collections that are immediate children of the parent specified in the containing RunQueryRequest. When true, selects all descendant collections.

Operator

The operator for combining multiple filters.

Filter

The list of filters to combine.

Requires:

• At least one filter is present.

FieldReference

The field to filter by.

Operator

The operator to filter by.

Value

The value to compare to.

string

A reference to a field in a document.

Requires:

• MUST be a dot-delimited (.) string of segments, where each segment conforms to document field name limitations.

CompositeFilter

A composite filter.

FieldFilter

A filter on a document field.

UnaryFilter

A filter that takes exactly one argument.

FieldReference

Required. An indexed vector field to search upon. Only documents which contain vectors whose dimensionality match the query_vector can be returned.

Value

Required. The query vector that we are searching on. Must be a vector of no more than 2048 dimensions.

DistanceMeasure

Required. The distance measure to use, required.

Int32Value

Required. The number of nearest neighbors to return. Must be a positive integer of no more than 1000.

FieldReference

The field to order by.

Direction

The direction to order by. Defaults to ASCENDING.

FieldReference

The fields to return.

If empty, all fields are returned. To only return the name of the document, use ['__name__'].

Operator

The unary operator to apply.

FieldReference

The field to which to apply the operator.

int32

The target ID that identifies the target on the stream. Must be a positive number and non-zero.

If target_id is 0 (or unspecified), the server will assign an ID for this target and return that in a TargetChange::ADD event. Once a target with target_id=0 is added, all subsequent targets must also have target_id=0. If an AddTarget request with target_id != 0 is sent to the server after a target with target_id=0 is added, the server will immediately send a response with a TargetChange::Remove event.

Note that if the client sends multiple AddTarget requests without an ID, the order of IDs returned in TargetChage.target_ids are undefined. Therefore, clients should provide a target ID instead of relying on the server to assign one.

If target_id is non-zero, there must not be an existing active target on this stream with the same ID.

bool

If the target should be removed once it is current and consistent.

Int32Value

The number of documents that last matched the query at the resume token or read time.

This value is only relevant when a resume_type is provided. This value being present and greater than zero signals that the client wants ExistenceFilter.unchanged_names to be included in the response.

QueryTarget

A target specified by a query.

DocumentsTarget

A target specified by a set of document names.

bytes

A resume token from a prior TargetChange for an identical target.

Using a resume token with a different target is unsupported and may fail.

Timestamp

Start listening after a specific read_time.

The client must know the state of matching documents at this time.

string

The names of the documents to retrieve. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}. The request will fail if any of the document is not a child resource of the given database. Duplicate names will be elided.

string

The parent resource name. In the format: projects/{project_id}/databases/{database_id}/documents or projects/{project_id}/databases/{database_id}/documents/{document_path}. For example: projects/my-project/databases/my-database/documents or projects/my-project/databases/my-database/documents/chatrooms/my-chatroom

StructuredQuery

A structured query.

TargetChangeType

The type of change that occurred.

int32

The target IDs of targets that have changed.

If empty, the change applies to all targets.

The order of the target IDs is not defined.

Status

The error that resulted in this change, if applicable.

bytes

A token that can be used to resume the stream for the given target_ids, or all targets if target_ids is empty.

Not set on every target change.

Timestamp

The consistent read_time for the given target_ids (omitted when the target_ids are not at a consistent snapshot).

The stream is guaranteed to send a read_time with target_ids empty whenever the entire stream reaches a new consistent snapshot. ADD, CURRENT, and RESET messages are guaranteed to (eventually) result in a new consistent snapshot (while NO_CHANGE and REMOVE messages are not).

For a given stream, read_time is guaranteed to be monotonically increasing.

ReadOnly

The transaction can only be used for read operations.

ReadWrite

The transaction can be used for both read and write operations.

Timestamp

Reads documents at the given time.

This must be a microsecond precision timestamp within the past one hour, or if Point-in-Time Recovery is enabled, can additionally be a whole minute timestamp within the past 7 days.

bytes

An optional transaction to retry.

Document

Required. The updated document. Creates the document if it does not already exist.

DocumentMask

The fields to update. None of the field paths in the mask may contain a reserved name.

If the document exists on the server and has fields not referenced in the mask, they are left unchanged. Fields referenced in the mask, but not present in the input document, are deleted from the document on the server.

DocumentMask

The fields to return. If not set, returns all fields.

If the document has a field that is not present in this mask, that field will not be returned in the response.

Precondition

An optional precondition on the document. The request will fail if this is set and not met by the target document.

NullValue

A null value.

bool

A boolean value.

int64

An integer value.

double

A double value.

Timestamp

A timestamp value.

Precise only to microseconds. When stored, any additional precision is rounded down.

string

A string value.

The string, represented as UTF-8, must not exceed 1 MiB - 89 bytes. Only the first 1,500 bytes of the UTF-8 representation are considered by queries.

bytes

A bytes value.

Must not exceed 1 MiB - 89 bytes. Only the first 1,500 bytes are considered by queries.

string

A reference to a document. For example: projects/{project_id}/databases/{database_id}/documents/{document_path}.

LatLng

A geo point value representing a point on the surface of Earth.

ArrayValue

An array value.

Cannot directly contain another array value, though can contain a map which contains another array.

MapValue

A map value.

DocumentMask

The fields to update in this write.

This field can be set only when the operation is update. If the mask is not set for an update and the document exists, any existing data will be overwritten. If the mask is set and the document on the server has fields not covered by the mask, they are left unchanged. Fields referenced in the mask, but not present in the input document, are deleted from the document on the server. The field paths in this mask must not contain a reserved field name.

FieldTransform

The transforms to perform after update.

This field can be set only when the operation is update. If present, this write is equivalent to performing update and transform to the same document atomically and in order.

Precondition

An optional precondition on the document.

The write will fail if this is set and not met by the target document.

Document

A document to write.

string

A document name to delete. In the format: projects/{project_id}/databases/{database_id}/documents/{document_path}.

DocumentTransform

Applies a transformation to a document.

string

Required. The database name. In the format: projects/{project_id}/databases/{database_id}. This is only required in the first message.

string

The ID of the write stream to resume. This may only be set in the first message. When left empty, a new write stream will be created.

Write

The writes to apply.

Always executed atomically and in order. This must be empty on the first request. This may be empty on the last request. This must not be empty on all other requests.

bytes

A stream token that was previously sent by the server.

The client should set this field to the token from the most recent WriteResponse it has received. This acknowledges that the client has received responses up to this token. After sending this token, earlier tokens may not be used anymore.

The server may close the stream if there are too many unacknowledged responses.

Leave this field unset when creating a new stream. To resume a stream at a specific point, set this field and the stream_id field.

Leave this field unset when creating a new stream.

map<string, string>

Labels associated with this write request.

string

The ID of the stream. Only set on the first message, when a new stream was created.

bytes

A token that represents the position of this response in the stream. This can be used by a client to resume the stream at this point.

This field is always set.

WriteResult

The result of applying the writes.

This i-th write result corresponds to the i-th write in the request.

Timestamp

The time at which the commit occurred. Any read with an equal or greater read_time is guaranteed to see the effects of the write.

Timestamp

The last update time of the document after applying the write. Not set after a delete.

If the write did not actually change the document, this will be the previous update_time.

Value

The results of applying each DocumentTransform.FieldTransform, in the same order.