Compare commits

...

5 Commits

Author SHA1 Message Date
Zhenchi
01081ef97b refactor: unify function registry (Part 2)
Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>
2025-06-11 08:21:11 +00:00
localhost
01796c9cc0 chore: org cli sub command (#6265)
* chore: org cli sub command

* chore: make clippy happy

* chore: fix info command not support absolute path

* chore: fix cli test

* Apply suggestions from code review

Co-authored-by: Weny Xu <wenymedia@gmail.com>

* chore: reorganizing the cli tool

* chore: fix limit issue

* chore: add some doc for cli

* chore: format code

---------

Co-authored-by: Weny Xu <wenymedia@gmail.com>
2025-06-11 03:34:56 +00:00
liyang
9469a8f8f2 ci: add signature information when updating downstream repository (#6282)
Signed-off-by: liyang <daviderli614@gmail.com>
2025-06-10 17:18:29 +00:00
Ruihang Xia
2fabe346a1 fix: null value handling on PromQL's join (#6289)
* fix: null value handling on PromQL's join

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

* update sqlness comment

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

---------

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>
2025-06-10 13:24:09 +00:00
Zhenchi
c26138963e refactor: unify function registry (Part 1) (#6262)
* refactor: unify function registry (Part 1)

Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>

* refactor: simplify via register_scalar

Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>

---------

Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>
2025-06-10 10:11:06 +00:00
59 changed files with 801 additions and 481 deletions

View File

@@ -30,7 +30,7 @@ update_helm_charts_version() {
# Commit the changes.
git add .
git commit -m "chore: Update GreptimeDB version to ${VERSION}"
git commit -s -m "chore: Update GreptimeDB version to ${VERSION}"
git push origin $BRANCH_NAME
# Create a Pull Request.

View File

@@ -26,7 +26,7 @@ update_homebrew_greptime_version() {
# Commit the changes.
git add .
git commit -m "chore: Update GreptimeDB version to ${VERSION}"
git commit -s -m "chore: Update GreptimeDB version to ${VERSION}"
git push origin $BRANCH_NAME
# Create a Pull Request.

49
Cargo.lock generated
View File

@@ -2336,6 +2336,7 @@ dependencies = [
"num-traits",
"once_cell",
"paste",
"promql",
"s2",
"serde",
"serde_json",
@@ -3252,7 +3253,7 @@ dependencies = [
[[package]]
name = "datafusion"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-array 54.2.1",
@@ -3303,7 +3304,7 @@ dependencies = [
[[package]]
name = "datafusion-catalog"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"async-trait",
@@ -3323,7 +3324,7 @@ dependencies = [
[[package]]
name = "datafusion-catalog-listing"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-schema 54.3.1",
@@ -3346,7 +3347,7 @@ dependencies = [
[[package]]
name = "datafusion-common"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3371,7 +3372,7 @@ dependencies = [
[[package]]
name = "datafusion-common-runtime"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"log",
"tokio",
@@ -3380,12 +3381,12 @@ dependencies = [
[[package]]
name = "datafusion-doc"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
[[package]]
name = "datafusion-execution"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"dashmap",
@@ -3403,7 +3404,7 @@ dependencies = [
[[package]]
name = "datafusion-expr"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"chrono",
@@ -3423,7 +3424,7 @@ dependencies = [
[[package]]
name = "datafusion-expr-common"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"datafusion-common",
@@ -3434,7 +3435,7 @@ dependencies = [
[[package]]
name = "datafusion-functions"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-buffer 54.3.1",
@@ -3463,7 +3464,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-aggregate"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3484,7 +3485,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-aggregate-common"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3496,7 +3497,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-nested"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-array 54.2.1",
@@ -3518,7 +3519,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-table"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"async-trait",
@@ -3533,7 +3534,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-window"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"datafusion-common",
"datafusion-doc",
@@ -3549,7 +3550,7 @@ dependencies = [
[[package]]
name = "datafusion-functions-window-common"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"datafusion-common",
"datafusion-physical-expr-common",
@@ -3558,7 +3559,7 @@ dependencies = [
[[package]]
name = "datafusion-macros"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"datafusion-expr",
"quote",
@@ -3568,7 +3569,7 @@ dependencies = [
[[package]]
name = "datafusion-optimizer"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"chrono",
@@ -3586,7 +3587,7 @@ dependencies = [
[[package]]
name = "datafusion-physical-expr"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3609,7 +3610,7 @@ dependencies = [
[[package]]
name = "datafusion-physical-expr-common"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3622,7 +3623,7 @@ dependencies = [
[[package]]
name = "datafusion-physical-optimizer"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-schema 54.3.1",
@@ -3643,7 +3644,7 @@ dependencies = [
[[package]]
name = "datafusion-physical-plan"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"ahash 0.8.11",
"arrow 54.2.1",
@@ -3673,7 +3674,7 @@ dependencies = [
[[package]]
name = "datafusion-sql"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"arrow 54.2.1",
"arrow-array 54.2.1",
@@ -3691,7 +3692,7 @@ dependencies = [
[[package]]
name = "datafusion-substrait"
version = "45.0.0"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=e104c7cf62b11dd5fe41461b82514978234326b4#e104c7cf62b11dd5fe41461b82514978234326b4"
source = "git+https://github.com/waynexia/arrow-datafusion.git?rev=12c0381babd52c681043957e9d6ee083a03f7646#12c0381babd52c681043957e9d6ee083a03f7646"
dependencies = [
"async-recursion",
"async-trait",

View File

@@ -116,15 +116,15 @@ clap = { version = "4.4", features = ["derive"] }
config = "0.13.0"
crossbeam-utils = "0.8"
dashmap = "6.1"
datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-functions = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-physical-plan = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "e104c7cf62b11dd5fe41461b82514978234326b4" }
datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-functions = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-physical-plan = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "12c0381babd52c681043957e9d6ee083a03f7646" }
deadpool = "0.12"
deadpool-postgres = "0.14"
derive_builder = "0.20"

View File

@@ -58,6 +58,7 @@ where
info!("{desc}, average operation cost: {cost:.2} ms");
}
/// Command to benchmark table metadata operations.
#[derive(Debug, Default, Parser)]
pub struct BenchTableMetadataCommand {
#[clap(long)]

View File

@@ -244,6 +244,18 @@ pub enum Error {
#[snafu(implicit)]
location: Location,
},
#[snafu(display("Unsupported memory backend"))]
UnsupportedMemoryBackend {
#[snafu(implicit)]
location: Location,
},
#[snafu(display("File path invalid: {}", msg))]
InvalidFilePath {
msg: String,
#[snafu(implicit)]
location: Location,
},
}
pub type Result<T> = std::result::Result<T, Error>;
@@ -262,6 +274,8 @@ impl ErrorExt for Error {
| Error::ConnectEtcd { .. }
| Error::CreateDir { .. }
| Error::EmptyResult { .. }
| Error::InvalidFilePath { .. }
| Error::UnsupportedMemoryBackend { .. }
| Error::ParseProxyOpts { .. } => StatusCode::InvalidArguments,
Error::StartProcedureManager { source, .. }

View File

@@ -50,6 +50,7 @@ enum ExportTarget {
All,
}
/// Command for exporting data from the GreptimeDB.
#[derive(Debug, Default, Parser)]
pub struct ExportCommand {
/// Server address to connect

View File

@@ -40,6 +40,7 @@ enum ImportTarget {
All,
}
/// Command to import data from a directory into a GreptimeDB instance.
#[derive(Debug, Default, Parser)]
pub struct ImportCommand {
/// Server address to connect

View File

@@ -20,7 +20,7 @@ mod import;
mod meta_snapshot;
use async_trait::async_trait;
use clap::Parser;
use clap::{Parser, Subcommand};
use common_error::ext::BoxedError;
pub use database::DatabaseClient;
use error::Result;
@@ -28,7 +28,7 @@ use error::Result;
pub use crate::bench::BenchTableMetadataCommand;
pub use crate::export::ExportCommand;
pub use crate::import::ImportCommand;
pub use crate::meta_snapshot::{MetaRestoreCommand, MetaSnapshotCommand};
pub use crate::meta_snapshot::{MetaCommand, MetaInfoCommand, MetaRestoreCommand, MetaSaveCommand};
#[async_trait]
pub trait Tool: Send + Sync {
@@ -51,3 +51,19 @@ impl AttachCommand {
unimplemented!("Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373")
}
}
/// Subcommand for data operations like export and import.
#[derive(Subcommand)]
pub enum DataCommand {
Export(ExportCommand),
Import(ImportCommand),
}
impl DataCommand {
pub async fn build(&self) -> std::result::Result<Box<dyn Tool>, BoxedError> {
match self {
DataCommand::Export(cmd) => cmd.build().await,
DataCommand::Import(cmd) => cmd.build().await,
}
}
}

View File

@@ -12,10 +12,11 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::path::Path;
use std::sync::Arc;
use async_trait::async_trait;
use clap::Parser;
use clap::{Parser, Subcommand};
use common_base::secrets::{ExposeSecret, SecretString};
use common_error::ext::BoxedError;
use common_meta::kv_backend::chroot::ChrootKvBackend;
@@ -26,10 +27,50 @@ use meta_srv::bootstrap::create_etcd_client;
use meta_srv::metasrv::BackendImpl;
use object_store::services::{Fs, S3};
use object_store::ObjectStore;
use snafu::ResultExt;
use snafu::{OptionExt, ResultExt};
use crate::error::{KvBackendNotSetSnafu, OpenDalSnafu, S3ConfigNotSetSnafu};
use crate::error::{
InvalidFilePathSnafu, KvBackendNotSetSnafu, OpenDalSnafu, S3ConfigNotSetSnafu,
UnsupportedMemoryBackendSnafu,
};
use crate::Tool;
/// Subcommand for metadata snapshot management.
#[derive(Subcommand)]
pub enum MetaCommand {
#[clap(subcommand)]
Snapshot(MetaSnapshotCommand),
}
impl MetaCommand {
pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
match self {
MetaCommand::Snapshot(cmd) => cmd.build().await,
}
}
}
/// Subcommand for metadata snapshot operations. such as save, restore and info.
#[derive(Subcommand)]
pub enum MetaSnapshotCommand {
/// Export metadata snapshot tool.
Save(MetaSaveCommand),
/// Restore metadata snapshot tool.
Restore(MetaRestoreCommand),
/// Explore metadata from metadata snapshot.
Info(MetaInfoCommand),
}
impl MetaSnapshotCommand {
pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
match self {
MetaSnapshotCommand::Save(cmd) => cmd.build().await,
MetaSnapshotCommand::Restore(cmd) => cmd.build().await,
MetaSnapshotCommand::Info(cmd) => cmd.build().await,
}
}
}
#[derive(Debug, Default, Parser)]
struct MetaConnection {
/// The endpoint of store. one of etcd, pg or mysql.
@@ -91,6 +132,9 @@ impl MetaConnection {
.await
.map_err(BoxedError::new)?)
}
Some(BackendImpl::MemoryStore) => UnsupportedMemoryBackendSnafu
.fail()
.map_err(BoxedError::new),
_ => KvBackendNotSetSnafu { backend: "all" }
.fail()
.map_err(BoxedError::new),
@@ -170,7 +214,7 @@ impl S3Config {
/// It will dump the metadata snapshot to local file or s3 bucket.
/// The snapshot file will be in binary format.
#[derive(Debug, Default, Parser)]
pub struct MetaSnapshotCommand {
pub struct MetaSaveCommand {
/// The connection to the metadata store.
#[clap(flatten)]
connection: MetaConnection,
@@ -196,7 +240,7 @@ fn create_local_file_object_store(root: &str) -> Result<ObjectStore, BoxedError>
Ok(object_store)
}
impl MetaSnapshotCommand {
impl MetaSaveCommand {
pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
let kvbackend = self.connection.build().await?;
let output_dir = &self.output_dir;
@@ -327,3 +371,89 @@ impl Tool for MetaRestoreTool {
}
}
}
/// Explore metadata from metadata snapshot.
#[derive(Debug, Default, Parser)]
pub struct MetaInfoCommand {
/// The s3 config.
#[clap(flatten)]
s3_config: S3Config,
/// The name of the target snapshot file. we will add the file extension automatically.
#[clap(long, default_value = "metadata_snapshot")]
file_name: String,
/// The query string to filter the metadata.
#[clap(long, default_value = "*")]
inspect_key: String,
/// The limit of the metadata to query.
#[clap(long)]
limit: Option<usize>,
}
pub struct MetaInfoTool {
inner: ObjectStore,
source_file: String,
inspect_key: String,
limit: Option<usize>,
}
#[async_trait]
impl Tool for MetaInfoTool {
async fn do_work(&self) -> std::result::Result<(), BoxedError> {
let result = MetadataSnapshotManager::info(
&self.inner,
&self.source_file,
&self.inspect_key,
self.limit,
)
.await
.map_err(BoxedError::new)?;
for item in result {
println!("{}", item);
}
Ok(())
}
}
impl MetaInfoCommand {
fn decide_object_store_root_for_local_store(
file_path: &str,
) -> Result<(&str, &str), BoxedError> {
let path = Path::new(file_path);
let parent = path
.parent()
.and_then(|p| p.to_str())
.context(InvalidFilePathSnafu { msg: file_path })
.map_err(BoxedError::new)?;
let file_name = path
.file_name()
.and_then(|f| f.to_str())
.context(InvalidFilePathSnafu { msg: file_path })
.map_err(BoxedError::new)?;
let root = if parent.is_empty() { "." } else { parent };
Ok((root, file_name))
}
pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
let object_store = self.s3_config.build("").map_err(BoxedError::new)?;
if let Some(store) = object_store {
let tool = MetaInfoTool {
inner: store,
source_file: self.file_name.clone(),
inspect_key: self.inspect_key.clone(),
limit: self.limit,
};
Ok(Box::new(tool))
} else {
let (root, file_name) =
Self::decide_object_store_root_for_local_store(&self.file_name)?;
let object_store = create_local_file_object_store(root)?;
let tool = MetaInfoTool {
inner: object_store,
source_file: file_name.to_string(),
inspect_key: self.inspect_key.clone(),
limit: self.limit,
};
Ok(Box::new(tool))
}
}
}

View File

@@ -146,6 +146,7 @@ mod tests {
let output_dir = tempfile::tempdir().unwrap();
let cli = cli::Command::parse_from([
"cli",
"data",
"export",
"--addr",
"127.0.0.1:4000",

View File

@@ -47,6 +47,7 @@ num = "0.4"
num-traits = "0.2"
once_cell.workspace = true
paste.workspace = true
promql.workspace = true
s2 = { version = "0.0.12", optional = true }
serde.workspace = true
serde_json.workspace = true

View File

@@ -12,11 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
mod geo_path;
mod hll;
mod uddsketch_state;
pub use geo_path::{GeoPathAccumulator, GEO_PATH_NAME};
pub(crate) use hll::HllStateType;
pub use hll::{HllState, HLL_MERGE_NAME, HLL_NAME};
pub use uddsketch_state::{UddSketchState, UDDSKETCH_MERGE_NAME, UDDSKETCH_STATE_NAME};
pub mod approximate;
#[cfg(feature = "geo")]
pub mod geo;
pub mod vector;

View File

@@ -0,0 +1,32 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use crate::function_registry::FunctionRegistry;
pub(crate) mod hll;
mod uddsketch;
pub(crate) struct ApproximateFunction;
impl ApproximateFunction {
pub fn register(registry: &FunctionRegistry) {
// uddsketch
registry.register_aggr(uddsketch::UddSketchState::state_udf_impl());
registry.register_aggr(uddsketch::UddSketchState::merge_udf_impl());
// hll
registry.register_aggr(hll::HllState::state_udf_impl());
registry.register_aggr(hll::HllState::merge_udf_impl());
}
}

View File

@@ -0,0 +1,27 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use crate::function_registry::FunctionRegistry;
mod encoding;
mod geo_path;
pub(crate) struct GeoFunction;
impl GeoFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register_aggr(geo_path::GeoPathAccumulator::uadf_impl());
registry.register_aggr(encoding::JsonPathAccumulator::uadf_impl());
}
}

View File

@@ -19,9 +19,12 @@ use common_error::status_code::StatusCode;
use common_macro::{as_aggr_func_creator, AggrFuncTypeStore};
use common_query::error::{self, InvalidInputStateSnafu, Result};
use common_query::logical_plan::accumulator::AggrFuncTypeStore;
use common_query::logical_plan::{Accumulator, AggregateFunctionCreator};
use common_query::logical_plan::{
create_aggregate_function, Accumulator, AggregateFunctionCreator,
};
use common_query::prelude::AccumulatorCreatorFunction;
use common_time::Timestamp;
use datafusion_expr::AggregateUDF;
use datatypes::prelude::ConcreteDataType;
use datatypes::value::{ListValue, Value};
use datatypes::vectors::VectorRef;
@@ -47,6 +50,16 @@ impl JsonPathAccumulator {
timestamp_type,
}
}
/// Create a new `AggregateUDF` for the `json_encode_path` aggregate function.
pub fn uadf_impl() -> AggregateUDF {
create_aggregate_function(
"json_encode_path".to_string(),
3,
Arc::new(JsonPathEncodeFunctionCreator::default()),
)
.into()
}
}
impl Accumulator for JsonPathAccumulator {

View File

@@ -47,7 +47,7 @@ impl GeoPathAccumulator {
Self::default()
}
pub fn udf_impl() -> AggregateUDF {
pub fn uadf_impl() -> AggregateUDF {
create_udaf(
GEO_PATH_NAME,
// Input types: lat, lng, timestamp

View File

@@ -0,0 +1,29 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use crate::aggrs::vector::product::VectorProduct;
use crate::aggrs::vector::sum::VectorSum;
use crate::function_registry::FunctionRegistry;
mod product;
mod sum;
pub(crate) struct VectorFunction;
impl VectorFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register_aggr(VectorSum::uadf_impl());
registry.register_aggr(VectorProduct::uadf_impl());
}
}

View File

@@ -16,8 +16,11 @@ use std::sync::Arc;
use common_macro::{as_aggr_func_creator, AggrFuncTypeStore};
use common_query::error::{CreateAccumulatorSnafu, Error, InvalidFuncArgsSnafu};
use common_query::logical_plan::{Accumulator, AggregateFunctionCreator};
use common_query::logical_plan::{
create_aggregate_function, Accumulator, AggregateFunctionCreator,
};
use common_query::prelude::AccumulatorCreatorFunction;
use datafusion_expr::AggregateUDF;
use datatypes::prelude::{ConcreteDataType, Value, *};
use datatypes::vectors::VectorRef;
use nalgebra::{Const, DVectorView, Dyn, OVector};
@@ -75,6 +78,16 @@ impl AggregateFunctionCreator for VectorProductCreator {
}
impl VectorProduct {
/// Create a new `AggregateUDF` for the `vec_product` aggregate function.
pub fn uadf_impl() -> AggregateUDF {
create_aggregate_function(
"vec_product".to_string(),
1,
Arc::new(VectorProductCreator::default()),
)
.into()
}
fn inner(&mut self, len: usize) -> &mut OVector<f32, Dyn> {
self.product.get_or_insert_with(|| {
OVector::from_iterator_generic(Dyn(len), Const::<1>, (0..len).map(|_| 1.0))

View File

@@ -16,8 +16,11 @@ use std::sync::Arc;
use common_macro::{as_aggr_func_creator, AggrFuncTypeStore};
use common_query::error::{CreateAccumulatorSnafu, Error, InvalidFuncArgsSnafu};
use common_query::logical_plan::{Accumulator, AggregateFunctionCreator};
use common_query::logical_plan::{
create_aggregate_function, Accumulator, AggregateFunctionCreator,
};
use common_query::prelude::AccumulatorCreatorFunction;
use datafusion_expr::AggregateUDF;
use datatypes::prelude::{ConcreteDataType, Value, *};
use datatypes::vectors::VectorRef;
use nalgebra::{Const, DVectorView, Dyn, OVector};
@@ -25,6 +28,7 @@ use snafu::ensure;
use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
/// The accumulator for the `vec_sum` aggregate function.
#[derive(Debug, Default)]
pub struct VectorSum {
sum: Option<OVector<f32, Dyn>>,
@@ -74,6 +78,16 @@ impl AggregateFunctionCreator for VectorSumCreator {
}
impl VectorSum {
/// Create a new `AggregateUDF` for the `vec_sum` aggregate function.
pub fn uadf_impl() -> AggregateUDF {
create_aggregate_function(
"vec_sum".to_string(),
1,
Arc::new(VectorSumCreator::default()),
)
.into()
}
fn inner(&mut self, len: usize) -> &mut OVector<f32, Dyn> {
self.sum
.get_or_insert_with(|| OVector::zeros_generic(Dyn(len), Const::<1>))

View File

@@ -0,0 +1,63 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
use datafusion_expr::ScalarUDF;
use crate::function::{FunctionContext, FunctionRef};
use crate::scalars::udf::create_udf;
/// A factory for creating `ScalarUDF` that require a function context.
#[derive(Clone)]
pub struct ScalarFunctionFactory {
name: String,
factory: Arc<dyn Fn(FunctionContext) -> ScalarUDF + Send + Sync>,
}
impl ScalarFunctionFactory {
/// Returns the name of the function.
pub fn name(&self) -> &str {
&self.name
}
/// Returns a `ScalarUDF` when given a function context.
pub fn provide(&self, ctx: FunctionContext) -> ScalarUDF {
(self.factory)(ctx)
}
}
impl From<ScalarUDF> for ScalarFunctionFactory {
fn from(df_udf: ScalarUDF) -> Self {
let name = df_udf.name().to_string();
let func = Arc::new(move |_ctx| df_udf.clone());
Self {
name,
factory: func,
}
}
}
impl From<FunctionRef> for ScalarFunctionFactory {
fn from(func: FunctionRef) -> Self {
let name = func.name().to_string();
let func = Arc::new(move |ctx: FunctionContext| {
create_udf(func.clone(), ctx.query_ctx, ctx.state)
});
Self {
name,
factory: func,
}
}
}

View File

@@ -16,11 +16,14 @@
use std::collections::HashMap;
use std::sync::{Arc, RwLock};
use datafusion_expr::AggregateUDF;
use once_cell::sync::Lazy;
use crate::admin::AdminFunction;
use crate::function::{AsyncFunctionRef, FunctionRef};
use crate::scalars::aggregate::{AggregateFunctionMetaRef, AggregateFunctions};
use crate::aggrs::approximate::ApproximateFunction;
use crate::aggrs::vector::VectorFunction as VectorAggrFunction;
use crate::function::{AsyncFunctionRef, Function, FunctionRef};
use crate::function_factory::ScalarFunctionFactory;
use crate::scalars::date::DateFunction;
use crate::scalars::expression::ExpressionFunction;
use crate::scalars::hll_count::HllCalcFunction;
@@ -31,25 +34,30 @@ use crate::scalars::matches_term::MatchesTermFunction;
use crate::scalars::math::MathFunction;
use crate::scalars::timestamp::TimestampFunction;
use crate::scalars::uddsketch_calc::UddSketchCalcFunction;
use crate::scalars::vector::VectorFunction;
use crate::scalars::vector::VectorFunction as VectorScalarFunction;
use crate::system::SystemFunction;
#[derive(Default)]
pub struct FunctionRegistry {
functions: RwLock<HashMap<String, FunctionRef>>,
scalar_functions: RwLock<HashMap<String, ScalarFunctionFactory>>,
async_functions: RwLock<HashMap<String, AsyncFunctionRef>>,
aggregate_functions: RwLock<HashMap<String, AggregateFunctionMetaRef>>,
aggregate_functions: RwLock<HashMap<String, AggregateUDF>>,
}
impl FunctionRegistry {
pub fn register(&self, func: FunctionRef) {
pub fn register(&self, func: impl Into<ScalarFunctionFactory>) {
let func = func.into();
let _ = self
.functions
.scalar_functions
.write()
.unwrap()
.insert(func.name().to_string(), func);
}
pub fn register_scalar(&self, func: impl Function + 'static) {
self.register(Arc::new(func) as FunctionRef);
}
pub fn register_async(&self, func: AsyncFunctionRef) {
let _ = self
.async_functions
@@ -58,6 +66,14 @@ impl FunctionRegistry {
.insert(func.name().to_string(), func);
}
pub fn register_aggr(&self, func: AggregateUDF) {
let _ = self
.aggregate_functions
.write()
.unwrap()
.insert(func.name().to_string(), func);
}
pub fn get_async_function(&self, name: &str) -> Option<AsyncFunctionRef> {
self.async_functions.read().unwrap().get(name).cloned()
}
@@ -71,27 +87,20 @@ impl FunctionRegistry {
.collect()
}
pub fn register_aggregate_function(&self, func: AggregateFunctionMetaRef) {
let _ = self
.aggregate_functions
.write()
pub fn get_scalar_function(&self, name: &str) -> Option<ScalarFunctionFactory> {
self.scalar_functions.read().unwrap().get(name).cloned()
}
pub fn scalar_functions(&self) -> Vec<ScalarFunctionFactory> {
self.scalar_functions
.read()
.unwrap()
.insert(func.name(), func);
.values()
.cloned()
.collect()
}
pub fn get_aggr_function(&self, name: &str) -> Option<AggregateFunctionMetaRef> {
self.aggregate_functions.read().unwrap().get(name).cloned()
}
pub fn get_function(&self, name: &str) -> Option<FunctionRef> {
self.functions.read().unwrap().get(name).cloned()
}
pub fn functions(&self) -> Vec<FunctionRef> {
self.functions.read().unwrap().values().cloned().collect()
}
pub fn aggregate_functions(&self) -> Vec<AggregateFunctionMetaRef> {
pub fn aggregate_functions(&self) -> Vec<AggregateUDF> {
self.aggregate_functions
.read()
.unwrap()
@@ -112,9 +121,6 @@ pub static FUNCTION_REGISTRY: Lazy<Arc<FunctionRegistry>> = Lazy::new(|| {
UddSketchCalcFunction::register(&function_registry);
HllCalcFunction::register(&function_registry);
// Aggregate functions
AggregateFunctions::register(&function_registry);
// Full text search function
MatchesFunction::register(&function_registry);
MatchesTermFunction::register(&function_registry);
@@ -127,15 +133,26 @@ pub static FUNCTION_REGISTRY: Lazy<Arc<FunctionRegistry>> = Lazy::new(|| {
JsonFunction::register(&function_registry);
// Vector related functions
VectorFunction::register(&function_registry);
VectorScalarFunction::register(&function_registry);
VectorAggrFunction::register(&function_registry);
// Geo functions
#[cfg(feature = "geo")]
crate::scalars::geo::GeoFunctions::register(&function_registry);
#[cfg(feature = "geo")]
crate::aggrs::geo::GeoFunction::register(&function_registry);
// Ip functions
IpFunctions::register(&function_registry);
// Approximate functions
ApproximateFunction::register(&function_registry);
// PromQL aggregate functions
for aggr in promql::functions::aggr_funcs() {
function_registry.register_aggr(aggr);
}
Arc::new(function_registry)
});
@@ -147,12 +164,11 @@ mod tests {
#[test]
fn test_function_registry() {
let registry = FunctionRegistry::default();
let func = Arc::new(TestAndFunction);
assert!(registry.get_function("test_and").is_none());
assert!(registry.functions().is_empty());
registry.register(func);
let _ = registry.get_function("test_and").unwrap();
assert_eq!(1, registry.functions().len());
assert!(registry.get_scalar_function("test_and").is_none());
assert!(registry.scalar_functions().is_empty());
registry.register_scalar(TestAndFunction);
let _ = registry.get_scalar_function("test_and").unwrap();
assert_eq!(1, registry.scalar_functions().len());
}
}

View File

@@ -18,13 +18,14 @@
mod admin;
mod flush_flow;
mod macros;
pub mod scalars;
mod system;
pub mod aggr;
pub mod aggrs;
pub mod function;
pub mod function_factory;
pub mod function_registry;
pub mod handlers;
pub mod helper;
pub mod scalars;
pub mod state;
pub mod utils;

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pub mod aggregate;
pub(crate) mod date;
pub mod expression;
#[cfg(feature = "geo")]

View File

@@ -1,89 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//! # Deprecate Warning:
//!
//! This module is deprecated and will be removed in the future.
//! All UDAF implementation here are not maintained and should
//! not be used before they are refactored into the `src/aggr`
//! version.
use std::sync::Arc;
use common_query::logical_plan::AggregateFunctionCreatorRef;
use crate::function_registry::FunctionRegistry;
use crate::scalars::vector::product::VectorProductCreator;
use crate::scalars::vector::sum::VectorSumCreator;
/// A function creates `AggregateFunctionCreator`.
/// "Aggregator" *is* AggregatorFunction. Since the later one is long, we named an short alias for it.
/// The two names might be used interchangeably.
type AggregatorCreatorFunction = Arc<dyn Fn() -> AggregateFunctionCreatorRef + Send + Sync>;
/// `AggregateFunctionMeta` dynamically creates AggregateFunctionCreator.
#[derive(Clone)]
pub struct AggregateFunctionMeta {
name: String,
args_count: u8,
creator: AggregatorCreatorFunction,
}
pub type AggregateFunctionMetaRef = Arc<AggregateFunctionMeta>;
impl AggregateFunctionMeta {
pub fn new(name: &str, args_count: u8, creator: AggregatorCreatorFunction) -> Self {
Self {
name: name.to_string(),
args_count,
creator,
}
}
pub fn name(&self) -> String {
self.name.to_string()
}
pub fn args_count(&self) -> u8 {
self.args_count
}
pub fn create(&self) -> AggregateFunctionCreatorRef {
(self.creator)()
}
}
pub(crate) struct AggregateFunctions;
impl AggregateFunctions {
pub fn register(registry: &FunctionRegistry) {
registry.register_aggregate_function(Arc::new(AggregateFunctionMeta::new(
"vec_sum",
1,
Arc::new(|| Arc::new(VectorSumCreator::default())),
)));
registry.register_aggregate_function(Arc::new(AggregateFunctionMeta::new(
"vec_product",
1,
Arc::new(|| Arc::new(VectorProductCreator::default())),
)));
#[cfg(feature = "geo")]
registry.register_aggregate_function(Arc::new(AggregateFunctionMeta::new(
"json_encode_path",
3,
Arc::new(|| Arc::new(super::geo::encoding::JsonPathEncodeFunctionCreator::default())),
)));
}
}

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
mod date_add;
mod date_format;
mod date_sub;
@@ -27,8 +26,8 @@ pub(crate) struct DateFunction;
impl DateFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(DateAddFunction));
registry.register(Arc::new(DateSubFunction));
registry.register(Arc::new(DateFormatFunction));
registry.register_scalar(DateAddFunction);
registry.register_scalar(DateSubFunction);
registry.register_scalar(DateFormatFunction);
}
}

View File

@@ -17,8 +17,6 @@ mod ctx;
mod is_null;
mod unary;
use std::sync::Arc;
pub use binary::scalar_binary_op;
pub use ctx::EvalContext;
pub use unary::scalar_unary_op;
@@ -30,6 +28,6 @@ pub(crate) struct ExpressionFunction;
impl ExpressionFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(IsNullFunction));
registry.register_scalar(IsNullFunction);
}
}

View File

@@ -12,11 +12,9 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
pub(crate) mod encoding;
mod geohash;
mod h3;
mod helpers;
pub(crate) mod helpers;
mod measure;
mod relation;
mod s2;
@@ -29,57 +27,57 @@ pub(crate) struct GeoFunctions;
impl GeoFunctions {
pub fn register(registry: &FunctionRegistry) {
// geohash
registry.register(Arc::new(geohash::GeohashFunction));
registry.register(Arc::new(geohash::GeohashNeighboursFunction));
registry.register_scalar(geohash::GeohashFunction);
registry.register_scalar(geohash::GeohashNeighboursFunction);
// h3 index
registry.register(Arc::new(h3::H3LatLngToCell));
registry.register(Arc::new(h3::H3LatLngToCellString));
registry.register_scalar(h3::H3LatLngToCell);
registry.register_scalar(h3::H3LatLngToCellString);
// h3 index inspection
registry.register(Arc::new(h3::H3CellBase));
registry.register(Arc::new(h3::H3CellIsPentagon));
registry.register(Arc::new(h3::H3StringToCell));
registry.register(Arc::new(h3::H3CellToString));
registry.register(Arc::new(h3::H3CellCenterLatLng));
registry.register(Arc::new(h3::H3CellResolution));
registry.register_scalar(h3::H3CellBase);
registry.register_scalar(h3::H3CellIsPentagon);
registry.register_scalar(h3::H3StringToCell);
registry.register_scalar(h3::H3CellToString);
registry.register_scalar(h3::H3CellCenterLatLng);
registry.register_scalar(h3::H3CellResolution);
// h3 hierarchical grid
registry.register(Arc::new(h3::H3CellCenterChild));
registry.register(Arc::new(h3::H3CellParent));
registry.register(Arc::new(h3::H3CellToChildren));
registry.register(Arc::new(h3::H3CellToChildrenSize));
registry.register(Arc::new(h3::H3CellToChildPos));
registry.register(Arc::new(h3::H3ChildPosToCell));
registry.register(Arc::new(h3::H3CellContains));
registry.register_scalar(h3::H3CellCenterChild);
registry.register_scalar(h3::H3CellParent);
registry.register_scalar(h3::H3CellToChildren);
registry.register_scalar(h3::H3CellToChildrenSize);
registry.register_scalar(h3::H3CellToChildPos);
registry.register_scalar(h3::H3ChildPosToCell);
registry.register_scalar(h3::H3CellContains);
// h3 grid traversal
registry.register(Arc::new(h3::H3GridDisk));
registry.register(Arc::new(h3::H3GridDiskDistances));
registry.register(Arc::new(h3::H3GridDistance));
registry.register(Arc::new(h3::H3GridPathCells));
registry.register_scalar(h3::H3GridDisk);
registry.register_scalar(h3::H3GridDiskDistances);
registry.register_scalar(h3::H3GridDistance);
registry.register_scalar(h3::H3GridPathCells);
// h3 measurement
registry.register(Arc::new(h3::H3CellDistanceSphereKm));
registry.register(Arc::new(h3::H3CellDistanceEuclideanDegree));
registry.register_scalar(h3::H3CellDistanceSphereKm);
registry.register_scalar(h3::H3CellDistanceEuclideanDegree);
// s2
registry.register(Arc::new(s2::S2LatLngToCell));
registry.register(Arc::new(s2::S2CellLevel));
registry.register(Arc::new(s2::S2CellToToken));
registry.register(Arc::new(s2::S2CellParent));
registry.register_scalar(s2::S2LatLngToCell);
registry.register_scalar(s2::S2CellLevel);
registry.register_scalar(s2::S2CellToToken);
registry.register_scalar(s2::S2CellParent);
// spatial data type
registry.register(Arc::new(wkt::LatLngToPointWkt));
registry.register_scalar(wkt::LatLngToPointWkt);
// spatial relation
registry.register(Arc::new(relation::STContains));
registry.register(Arc::new(relation::STWithin));
registry.register(Arc::new(relation::STIntersects));
registry.register_scalar(relation::STContains);
registry.register_scalar(relation::STWithin);
registry.register_scalar(relation::STIntersects);
// spatial measure
registry.register(Arc::new(measure::STDistance));
registry.register(Arc::new(measure::STDistanceSphere));
registry.register(Arc::new(measure::STArea));
registry.register_scalar(measure::STDistance);
registry.register_scalar(measure::STDistanceSphere);
registry.register_scalar(measure::STArea);
}
}

View File

@@ -37,7 +37,7 @@ macro_rules! ensure_columns_len {
};
}
pub(super) use ensure_columns_len;
pub(crate) use ensure_columns_len;
macro_rules! ensure_columns_n {
($columns:ident, $n:literal) => {
@@ -58,7 +58,7 @@ macro_rules! ensure_columns_n {
};
}
pub(super) use ensure_columns_n;
pub(crate) use ensure_columns_n;
macro_rules! ensure_and_coerce {
($compare:expr, $coerce:expr) => {{
@@ -72,4 +72,4 @@ macro_rules! ensure_and_coerce {
}};
}
pub(super) use ensure_and_coerce;
pub(crate) use ensure_and_coerce;

View File

@@ -16,7 +16,6 @@
use std::fmt;
use std::fmt::Display;
use std::sync::Arc;
use common_query::error::{DowncastVectorSnafu, InvalidFuncArgsSnafu, Result};
use common_query::prelude::{Signature, Volatility};
@@ -27,7 +26,7 @@ use datatypes::vectors::{BinaryVector, MutableVector, UInt64VectorBuilder, Vecto
use hyperloglogplus::HyperLogLog;
use snafu::OptionExt;
use crate::aggr::HllStateType;
use crate::aggrs::approximate::hll::HllStateType;
use crate::function::{Function, FunctionContext};
use crate::function_registry::FunctionRegistry;
@@ -44,7 +43,7 @@ pub struct HllCalcFunction;
impl HllCalcFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(HllCalcFunction));
registry.register_scalar(HllCalcFunction);
}
}
@@ -117,6 +116,8 @@ impl Function for HllCalcFunction {
#[cfg(test)]
mod tests {
use std::sync::Arc;
use datatypes::vectors::BinaryVector;
use super::*;

View File

@@ -17,8 +17,6 @@ mod ipv4;
mod ipv6;
mod range;
use std::sync::Arc;
use cidr::{Ipv4ToCidr, Ipv6ToCidr};
use ipv4::{Ipv4NumToString, Ipv4StringToNum};
use ipv6::{Ipv6NumToString, Ipv6StringToNum};
@@ -31,15 +29,15 @@ pub(crate) struct IpFunctions;
impl IpFunctions {
pub fn register(registry: &FunctionRegistry) {
// Register IPv4 functions
registry.register(Arc::new(Ipv4NumToString));
registry.register(Arc::new(Ipv4StringToNum));
registry.register(Arc::new(Ipv4ToCidr));
registry.register(Arc::new(Ipv4InRange));
registry.register_scalar(Ipv4NumToString);
registry.register_scalar(Ipv4StringToNum);
registry.register_scalar(Ipv4ToCidr);
registry.register_scalar(Ipv4InRange);
// Register IPv6 functions
registry.register(Arc::new(Ipv6NumToString));
registry.register(Arc::new(Ipv6StringToNum));
registry.register(Arc::new(Ipv6ToCidr));
registry.register(Arc::new(Ipv6InRange));
registry.register_scalar(Ipv6NumToString);
registry.register_scalar(Ipv6StringToNum);
registry.register_scalar(Ipv6ToCidr);
registry.register_scalar(Ipv6InRange);
}
}

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
pub mod json_get;
mod json_is;
mod json_path_exists;
@@ -33,23 +32,23 @@ pub(crate) struct JsonFunction;
impl JsonFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(JsonToStringFunction));
registry.register(Arc::new(ParseJsonFunction));
registry.register_scalar(JsonToStringFunction);
registry.register_scalar(ParseJsonFunction);
registry.register(Arc::new(JsonGetInt));
registry.register(Arc::new(JsonGetFloat));
registry.register(Arc::new(JsonGetString));
registry.register(Arc::new(JsonGetBool));
registry.register_scalar(JsonGetInt);
registry.register_scalar(JsonGetFloat);
registry.register_scalar(JsonGetString);
registry.register_scalar(JsonGetBool);
registry.register(Arc::new(JsonIsNull));
registry.register(Arc::new(JsonIsInt));
registry.register(Arc::new(JsonIsFloat));
registry.register(Arc::new(JsonIsString));
registry.register(Arc::new(JsonIsBool));
registry.register(Arc::new(JsonIsArray));
registry.register(Arc::new(JsonIsObject));
registry.register_scalar(JsonIsNull);
registry.register_scalar(JsonIsInt);
registry.register_scalar(JsonIsFloat);
registry.register_scalar(JsonIsString);
registry.register_scalar(JsonIsBool);
registry.register_scalar(JsonIsArray);
registry.register_scalar(JsonIsObject);
registry.register(Arc::new(json_path_exists::JsonPathExistsFunction));
registry.register(Arc::new(json_path_match::JsonPathMatchFunction));
registry.register_scalar(json_path_exists::JsonPathExistsFunction);
registry.register_scalar(json_path_match::JsonPathMatchFunction);
}
}

View File

@@ -38,11 +38,11 @@ use crate::function_registry::FunctionRegistry;
///
/// Usage: matches(`<col>`, `<pattern>`) -> boolean
#[derive(Clone, Debug, Default)]
pub(crate) struct MatchesFunction;
pub struct MatchesFunction;
impl MatchesFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(MatchesFunction));
registry.register_scalar(MatchesFunction);
}
}

View File

@@ -77,7 +77,7 @@ pub struct MatchesTermFunction;
impl MatchesTermFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(MatchesTermFunction));
registry.register_scalar(MatchesTermFunction);
}
}

View File

@@ -18,7 +18,6 @@ mod pow;
mod rate;
use std::fmt;
use std::sync::Arc;
pub use clamp::{ClampFunction, ClampMaxFunction, ClampMinFunction};
use common_query::error::{GeneralDataFusionSnafu, Result};
@@ -39,13 +38,13 @@ pub(crate) struct MathFunction;
impl MathFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(ModuloFunction));
registry.register(Arc::new(PowFunction));
registry.register(Arc::new(RateFunction));
registry.register(Arc::new(RangeFunction));
registry.register(Arc::new(ClampFunction));
registry.register(Arc::new(ClampMinFunction));
registry.register(Arc::new(ClampMaxFunction));
registry.register_scalar(ModuloFunction);
registry.register_scalar(PowFunction);
registry.register_scalar(RateFunction);
registry.register_scalar(RangeFunction);
registry.register_scalar(ClampFunction);
registry.register_scalar(ClampMinFunction);
registry.register_scalar(ClampMaxFunction);
}
}

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
mod to_unixtime;
use to_unixtime::ToUnixtimeFunction;
@@ -23,6 +22,6 @@ pub(crate) struct TimestampFunction;
impl TimestampFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(ToUnixtimeFunction));
registry.register_scalar(ToUnixtimeFunction);
}
}

View File

@@ -16,7 +16,6 @@
use std::fmt;
use std::fmt::Display;
use std::sync::Arc;
use common_query::error::{DowncastVectorSnafu, InvalidFuncArgsSnafu, Result};
use common_query::prelude::{Signature, Volatility};
@@ -44,7 +43,7 @@ pub struct UddSketchCalcFunction;
impl UddSketchCalcFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(UddSketchCalcFunction));
registry.register_scalar(UddSketchCalcFunction);
}
}

View File

@@ -17,10 +17,8 @@ mod distance;
mod elem_product;
mod elem_sum;
pub mod impl_conv;
pub(crate) mod product;
mod scalar_add;
mod scalar_mul;
pub(crate) mod sum;
mod vector_add;
mod vector_dim;
mod vector_div;
@@ -30,37 +28,34 @@ mod vector_norm;
mod vector_sub;
mod vector_subvector;
use std::sync::Arc;
use crate::function_registry::FunctionRegistry;
pub(crate) struct VectorFunction;
impl VectorFunction {
pub fn register(registry: &FunctionRegistry) {
// conversion
registry.register(Arc::new(convert::ParseVectorFunction));
registry.register(Arc::new(convert::VectorToStringFunction));
registry.register_scalar(convert::ParseVectorFunction);
registry.register_scalar(convert::VectorToStringFunction);
// distance
registry.register(Arc::new(distance::CosDistanceFunction));
registry.register(Arc::new(distance::DotProductFunction));
registry.register(Arc::new(distance::L2SqDistanceFunction));
registry.register_scalar(distance::CosDistanceFunction);
registry.register_scalar(distance::DotProductFunction);
registry.register_scalar(distance::L2SqDistanceFunction);
// scalar calculation
registry.register(Arc::new(scalar_add::ScalarAddFunction));
registry.register(Arc::new(scalar_mul::ScalarMulFunction));
registry.register_scalar(scalar_add::ScalarAddFunction);
registry.register_scalar(scalar_mul::ScalarMulFunction);
// vector calculation
registry.register(Arc::new(vector_add::VectorAddFunction));
registry.register(Arc::new(vector_sub::VectorSubFunction));
registry.register(Arc::new(vector_mul::VectorMulFunction));
registry.register(Arc::new(vector_div::VectorDivFunction));
registry.register(Arc::new(vector_norm::VectorNormFunction));
registry.register(Arc::new(vector_dim::VectorDimFunction));
registry.register(Arc::new(vector_kth_elem::VectorKthElemFunction));
registry.register(Arc::new(vector_subvector::VectorSubvectorFunction));
registry.register(Arc::new(elem_sum::ElemSumFunction));
registry.register(Arc::new(elem_product::ElemProductFunction));
registry.register_scalar(vector_add::VectorAddFunction);
registry.register_scalar(vector_sub::VectorSubFunction);
registry.register_scalar(vector_mul::VectorMulFunction);
registry.register_scalar(vector_div::VectorDivFunction);
registry.register_scalar(vector_norm::VectorNormFunction);
registry.register_scalar(vector_dim::VectorDimFunction);
registry.register_scalar(vector_kth_elem::VectorKthElemFunction);
registry.register_scalar(vector_subvector::VectorSubvectorFunction);
registry.register_scalar(elem_sum::ElemSumFunction);
registry.register_scalar(elem_product::ElemProductFunction);
}
}

View File

@@ -36,13 +36,13 @@ pub(crate) struct SystemFunction;
impl SystemFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(BuildFunction));
registry.register(Arc::new(VersionFunction));
registry.register(Arc::new(CurrentSchemaFunction));
registry.register(Arc::new(DatabaseFunction));
registry.register(Arc::new(SessionUserFunction));
registry.register(Arc::new(ReadPreferenceFunction));
registry.register(Arc::new(TimezoneFunction));
registry.register_scalar(BuildFunction);
registry.register_scalar(VersionFunction);
registry.register_scalar(CurrentSchemaFunction);
registry.register_scalar(DatabaseFunction);
registry.register_scalar(SessionUserFunction);
registry.register_scalar(ReadPreferenceFunction);
registry.register_scalar(TimezoneFunction);
registry.register_async(Arc::new(ProcedureStateFunction));
PGCatalogFunction::register(registry);
}

View File

@@ -16,8 +16,6 @@ mod pg_get_userbyid;
mod table_is_visible;
mod version;
use std::sync::Arc;
use pg_get_userbyid::PGGetUserByIdFunction;
use table_is_visible::PGTableIsVisibleFunction;
use version::PGVersionFunction;
@@ -35,8 +33,8 @@ pub(super) struct PGCatalogFunction;
impl PGCatalogFunction {
pub fn register(registry: &FunctionRegistry) {
registry.register(Arc::new(PGTableIsVisibleFunction));
registry.register(Arc::new(PGGetUserByIdFunction));
registry.register(Arc::new(PGVersionFunction));
registry.register_scalar(PGTableIsVisibleFunction);
registry.register_scalar(PGGetUserByIdFunction);
registry.register_scalar(PGVersionFunction);
}
}

View File

@@ -14,6 +14,7 @@
pub mod file;
use std::borrow::Cow;
use std::fmt::{Display, Formatter};
use std::path::{Path, PathBuf};
use std::time::Instant;
@@ -271,6 +272,49 @@ impl MetadataSnapshotManager {
Ok((filename.to_string(), num_keyvalues as u64))
}
fn format_output(key: Cow<'_, str>, value: Cow<'_, str>) -> String {
format!("{} => {}", key, value)
}
pub async fn info(
object_store: &ObjectStore,
file_path: &str,
query_str: &str,
limit: Option<usize>,
) -> Result<Vec<String>> {
let path = Path::new(file_path);
let file_name = path
.file_name()
.and_then(|s| s.to_str())
.context(InvalidFilePathSnafu { file_path })?;
let filename = FileName::try_from(file_name)?;
let data = object_store
.read(file_path)
.await
.context(ReadObjectSnafu { file_path })?;
let document = Document::from_slice(&filename.extension.format, &data.to_bytes())?;
let metadata_content = document.into_metadata_content()?.values();
let mut results = Vec::with_capacity(limit.unwrap_or(256));
for kv in metadata_content {
let key_str = String::from_utf8_lossy(&kv.key);
if let Some(prefix) = query_str.strip_suffix('*') {
if key_str.starts_with(prefix) {
let value_str = String::from_utf8_lossy(&kv.value);
results.push(Self::format_output(key_str, value_str));
}
} else if key_str == query_str {
let value_str = String::from_utf8_lossy(&kv.value);
results.push(Self::format_output(key_str, value_str));
}
if results.len() == limit.unwrap_or(usize::MAX) {
break;
}
}
Ok(results)
}
}
#[cfg(test)]

View File

@@ -111,6 +111,11 @@ impl MetadataContent {
pub fn into_iter(self) -> impl Iterator<Item = KeyValue> {
self.values.into_iter()
}
/// Returns the key-value pairs as a vector.
pub fn values(self) -> Vec<KeyValue> {
self.values
}
}
/// The key-value pair of the backup file.

View File

@@ -19,12 +19,11 @@ use std::time::Duration;
use api::region::RegionResponse;
use async_trait::async_trait;
use common_error::ext::BoxedError;
use common_function::function::FunctionRef;
use common_function::scalars::aggregate::AggregateFunctionMetaRef;
use common_function::function_factory::ScalarFunctionFactory;
use common_query::Output;
use common_runtime::runtime::{BuilderBuild, RuntimeTrait};
use common_runtime::Runtime;
use datafusion_expr::LogicalPlan;
use datafusion_expr::{AggregateUDF, LogicalPlan};
use query::dataframe::DataFrame;
use query::planner::LogicalPlanner;
use query::query_engine::{DescribeResult, QueryEngineState};
@@ -76,9 +75,9 @@ impl QueryEngine for MockQueryEngine {
unimplemented!()
}
fn register_aggregate_function(&self, _func: AggregateFunctionMetaRef) {}
fn register_aggregate_function(&self, _func: AggregateUDF) {}
fn register_function(&self, _func: FunctionRef) {}
fn register_scalar_function(&self, _func: ScalarFunctionFactory) {}
fn read_table(&self, _table: TableRef) -> query::error::Result<DataFrame> {
unimplemented!()

View File

@@ -17,7 +17,7 @@ use std::collections::BTreeMap;
use std::sync::Arc;
use common_error::ext::BoxedError;
use common_function::function::FunctionContext;
use common_function::function::{FunctionContext, FunctionRef};
use datafusion_substrait::extensions::Extensions;
use datatypes::data_type::ConcreteDataType as CDT;
use query::QueryEngine;
@@ -108,9 +108,13 @@ impl FunctionExtensions {
/// register flow-specific functions to the query engine
pub fn register_function_to_query_engine(engine: &Arc<dyn QueryEngine>) {
engine.register_function(Arc::new(TumbleFunction::new("tumble")));
engine.register_function(Arc::new(TumbleFunction::new(TUMBLE_START)));
engine.register_function(Arc::new(TumbleFunction::new(TUMBLE_END)));
let tumble_fn = Arc::new(TumbleFunction::new("tumble")) as FunctionRef;
let tumble_start_fn = Arc::new(TumbleFunction::new(TUMBLE_START)) as FunctionRef;
let tumble_end_fn = Arc::new(TumbleFunction::new(TUMBLE_END)) as FunctionRef;
engine.register_scalar_function(tumble_fn.into());
engine.register_scalar_function(tumble_start_fn.into());
engine.register_scalar_function(tumble_end_fn.into());
}
#[derive(Debug)]

View File

@@ -282,14 +282,15 @@ mod tests {
use std::sync::Arc;
use api::v1::SemanticType;
use common_function::function_registry::FUNCTION_REGISTRY;
use common_function::scalars::udf::create_udf;
use common_function::function::FunctionRef;
use common_function::function_factory::ScalarFunctionFactory;
use common_function::scalars::matches::MatchesFunction;
use common_function::scalars::matches_term::MatchesTermFunction;
use datafusion::functions::string::lower;
use datafusion_common::Column;
use datafusion_expr::expr::ScalarFunction;
use datafusion_expr::ScalarUDF;
use datatypes::schema::ColumnSchema;
use session::context::QueryContext;
use store_api::metadata::{ColumnMetadata, RegionMetadataBuilder};
use store_api::storage::RegionId;
@@ -317,19 +318,17 @@ mod tests {
}
fn matches_func() -> Arc<ScalarUDF> {
Arc::new(create_udf(
FUNCTION_REGISTRY.get_function("matches").unwrap(),
QueryContext::arc(),
Default::default(),
))
Arc::new(
ScalarFunctionFactory::from(Arc::new(MatchesFunction) as FunctionRef)
.provide(Default::default()),
)
}
fn matches_term_func() -> Arc<ScalarUDF> {
Arc::new(create_udf(
FUNCTION_REGISTRY.get_function("matches_term").unwrap(),
QueryContext::arc(),
Default::default(),
))
Arc::new(
ScalarFunctionFactory::from(Arc::new(MatchesTermFunction) as FunctionRef)
.provide(Default::default()),
)
}
#[test]

View File

@@ -13,20 +13,17 @@
// limitations under the License.
use clap::Parser;
use cli::{
BenchTableMetadataCommand, ExportCommand, ImportCommand, MetaRestoreCommand,
MetaSnapshotCommand, Tool,
};
use cli::{BenchTableMetadataCommand, DataCommand, MetaCommand, Tool};
use common_error::ext::BoxedError;
#[derive(Parser)]
pub enum SubCommand {
// Attach(AttachCommand),
Bench(BenchTableMetadataCommand),
Export(ExportCommand),
Import(ImportCommand),
MetaSnapshot(MetaSnapshotCommand),
MetaRestore(MetaRestoreCommand),
#[clap(subcommand)]
Data(DataCommand),
#[clap(subcommand)]
Meta(MetaCommand),
}
impl SubCommand {
@@ -34,10 +31,8 @@ impl SubCommand {
match self {
// SubCommand::Attach(cmd) => cmd.build().await,
SubCommand::Bench(cmd) => cmd.build().await,
SubCommand::Export(cmd) => cmd.build().await,
SubCommand::Import(cmd) => cmd.build().await,
SubCommand::MetaSnapshot(cmd) => cmd.build().await,
SubCommand::MetaRestore(cmd) => cmd.build().await,
SubCommand::Data(cmd) => cmd.build().await,
SubCommand::Meta(cmd) => cmd.build().await,
}
}
}

View File

@@ -34,6 +34,7 @@ pub use changes::Changes;
use datafusion::arrow::array::{ArrayRef, Float64Array, TimestampMillisecondArray};
use datafusion::error::DataFusionError;
use datafusion::physical_plan::ColumnarValue;
use datafusion_expr::{AggregateUDF, ScalarUDF};
pub use deriv::Deriv;
pub use extrapolate_rate::{Delta, Increase, Rate};
pub use holt_winters::HoltWinters;
@@ -44,6 +45,39 @@ pub use quantile_aggr::{quantile_udaf, QUANTILE_NAME};
pub use resets::Resets;
pub use round::Round;
/// Range functions for PromQL.
pub fn range_funcs() -> Vec<ScalarUDF> {
vec![
IDelta::<false>::scalar_udf(),
IDelta::<true>::scalar_udf(),
Rate::scalar_udf(),
Increase::scalar_udf(),
Delta::scalar_udf(),
Resets::scalar_udf(),
Changes::scalar_udf(),
Deriv::scalar_udf(),
Round::scalar_udf(),
AvgOverTime::scalar_udf(),
MinOverTime::scalar_udf(),
MaxOverTime::scalar_udf(),
SumOverTime::scalar_udf(),
CountOverTime::scalar_udf(),
LastOverTime::scalar_udf(),
AbsentOverTime::scalar_udf(),
PresentOverTime::scalar_udf(),
StddevOverTime::scalar_udf(),
StdvarOverTime::scalar_udf(),
QuantileOverTime::scalar_udf(),
PredictLinear::scalar_udf(),
HoltWinters::scalar_udf(),
]
}
/// Aggregate functions for PromQL.
pub fn aggr_funcs() -> Vec<AggregateUDF> {
vec![quantile_udaf()]
}
/// Extracts an array from a `ColumnarValue`.
///
/// If the `ColumnarValue` is a scalar, it converts it to an array of size 1.

View File

@@ -40,8 +40,8 @@ pub struct QuantileAccumulator {
/// Create a quantile `AggregateUDF` for PromQL quantile operator,
/// which calculates φ-quantile (0 ≤ φ ≤ 1) over dimensions
pub fn quantile_udaf() -> Arc<AggregateUDF> {
Arc::new(create_udaf(
pub fn quantile_udaf() -> AggregateUDF {
create_udaf(
QUANTILE_NAME,
// Input type: (φ, values)
vec![DataType::Float64, DataType::Float64],
@@ -63,7 +63,7 @@ pub fn quantile_udaf() -> Arc<AggregateUDF> {
)]
.into(),
)]),
))
)
}
impl QuantileAccumulator {

View File

@@ -25,8 +25,7 @@ use async_trait::async_trait;
use common_base::Plugins;
use common_catalog::consts::is_readonly_schema;
use common_error::ext::BoxedError;
use common_function::function::FunctionRef;
use common_function::scalars::aggregate::AggregateFunctionMetaRef;
use common_function::function_factory::ScalarFunctionFactory;
use common_query::{Output, OutputData, OutputMeta};
use common_recordbatch::adapter::RecordBatchStreamAdapter;
use common_recordbatch::{EmptyRecordBatchStream, SendableRecordBatchStream};
@@ -35,7 +34,9 @@ use datafusion::physical_plan::analyze::AnalyzeExec;
use datafusion::physical_plan::coalesce_partitions::CoalescePartitionsExec;
use datafusion::physical_plan::ExecutionPlan;
use datafusion_common::ResolvedTableReference;
use datafusion_expr::{DmlStatement, LogicalPlan as DfLogicalPlan, LogicalPlan, WriteOp};
use datafusion_expr::{
AggregateUDF, DmlStatement, LogicalPlan as DfLogicalPlan, LogicalPlan, WriteOp,
};
use datatypes::prelude::VectorRef;
use datatypes::schema::Schema;
use futures_util::StreamExt;
@@ -454,14 +455,14 @@ impl QueryEngine for DatafusionQueryEngine {
/// `SELECT "my_UDAF"(x)` will look for an aggregate named `"my_UDAF"`
///
/// So it's better to make UDAF name lowercase when creating one.
fn register_aggregate_function(&self, func: AggregateFunctionMetaRef) {
self.state.register_aggregate_function(func);
fn register_aggregate_function(&self, func: AggregateUDF) {
self.state.register_aggr_function(func);
}
/// Register an UDF function.
/// Register an scalar function.
/// Will override if the function with same name is already registered.
fn register_function(&self, func: FunctionRef) {
self.state.register_function(func);
fn register_scalar_function(&self, func: ScalarFunctionFactory) {
self.state.register_scalar_function(func);
}
fn read_table(&self, table: TableRef) -> Result<DataFrame> {

View File

@@ -18,12 +18,7 @@ use std::sync::Arc;
use arrow_schema::DataType;
use catalog::table_source::DfTableSourceProvider;
use common_function::aggr::{
GeoPathAccumulator, HllState, UddSketchState, GEO_PATH_NAME, HLL_MERGE_NAME, HLL_NAME,
UDDSKETCH_MERGE_NAME, UDDSKETCH_STATE_NAME,
};
use common_function::scalars::udf::create_udf;
use common_query::logical_plan::create_aggregate_function;
use common_function::function::FunctionContext;
use datafusion::common::TableReference;
use datafusion::datasource::cte_worktable::CteWorkTable;
use datafusion::datasource::file_format::{format_as_file_type, FileFormatFactory};
@@ -151,38 +146,21 @@ impl ContextProvider for DfContextProviderAdapter {
}
fn get_function_meta(&self, name: &str) -> Option<Arc<ScalarUDF>> {
self.engine_state.udf_function(name).map_or_else(
self.engine_state.scalar_function(name).map_or_else(
|| self.session_state.scalar_functions().get(name).cloned(),
|func| {
Some(Arc::new(create_udf(
func,
self.query_ctx.clone(),
self.engine_state.function_state(),
)))
Some(Arc::new(func.provide(FunctionContext {
query_ctx: self.query_ctx.clone(),
state: self.engine_state.function_state(),
})))
},
)
}
fn get_aggregate_meta(&self, name: &str) -> Option<Arc<AggregateUDF>> {
if name == UDDSKETCH_STATE_NAME {
return Some(Arc::new(UddSketchState::state_udf_impl()));
} else if name == UDDSKETCH_MERGE_NAME {
return Some(Arc::new(UddSketchState::merge_udf_impl()));
} else if name == HLL_NAME {
return Some(Arc::new(HllState::state_udf_impl()));
} else if name == HLL_MERGE_NAME {
return Some(Arc::new(HllState::merge_udf_impl()));
} else if name == GEO_PATH_NAME {
return Some(Arc::new(GeoPathAccumulator::udf_impl()));
}
self.engine_state.aggregate_function(name).map_or_else(
self.engine_state.aggr_function(name).map_or_else(
|| self.session_state.aggregate_functions().get(name).cloned(),
|func| {
Some(Arc::new(
create_aggregate_function(func.name(), func.args_count(), func.create()).into(),
))
},
|func| Some(Arc::new(func)),
)
}
@@ -213,13 +191,13 @@ impl ContextProvider for DfContextProviderAdapter {
}
fn udf_names(&self) -> Vec<String> {
let mut names = self.engine_state.udf_names();
let mut names = self.engine_state.scalar_names();
names.extend(self.session_state.scalar_functions().keys().cloned());
names
}
fn udaf_names(&self) -> Vec<String> {
let mut names = self.engine_state.udaf_names();
let mut names = self.engine_state.aggr_names();
names.extend(self.session_state.aggregate_functions().keys().cloned());
names
}

View File

@@ -1948,7 +1948,7 @@ impl PromPlanner {
token::T_QUANTILE => {
let q = Self::get_param_value_as_f64(op, param)?;
non_col_args.push(lit(q));
quantile_udaf()
Arc::new(quantile_udaf())
}
token::T_AVG => avg_udaf(),
token::T_COUNT_VALUES | token::T_COUNT => count_udaf(),
@@ -2444,7 +2444,7 @@ impl PromPlanner {
LogicalPlanBuilder::from(left)
.alias(left_table_ref)
.context(DataFusionPlanningSnafu)?
.join(
.join_detailed(
right,
JoinType::Inner,
(
@@ -2458,6 +2458,7 @@ impl PromPlanner {
.collect::<Vec<_>>(),
),
None,
true,
)
.context(DataFusionPlanningSnafu)?
.build()

View File

@@ -22,14 +22,13 @@ use std::sync::Arc;
use async_trait::async_trait;
use catalog::CatalogManagerRef;
use common_base::Plugins;
use common_function::function::FunctionRef;
use common_function::function_factory::ScalarFunctionFactory;
use common_function::function_registry::FUNCTION_REGISTRY;
use common_function::handlers::{
FlowServiceHandlerRef, ProcedureServiceHandlerRef, TableMutationHandlerRef,
};
use common_function::scalars::aggregate::AggregateFunctionMetaRef;
use common_query::Output;
use datafusion_expr::LogicalPlan;
use datafusion_expr::{AggregateUDF, LogicalPlan};
use datatypes::schema::Schema;
pub use default_serializer::{DefaultPlanDecoder, DefaultSerializer};
use session::context::QueryContextRef;
@@ -79,11 +78,11 @@ pub trait QueryEngine: Send + Sync {
///
/// # Panics
/// Will panic if the function with same name is already registered.
fn register_aggregate_function(&self, func: AggregateFunctionMetaRef);
fn register_aggregate_function(&self, func: AggregateUDF);
/// Register a SQL function.
/// Register a scalar function.
/// Will override if the function with same name is already registered.
fn register_function(&self, func: FunctionRef);
fn register_scalar_function(&self, func: ScalarFunctionFactory);
/// Create a DataFrame from a table.
fn read_table(&self, table: TableRef) -> Result<DataFrame>;
@@ -154,8 +153,8 @@ impl QueryEngineFactory {
/// Register all functions implemented by GreptimeDB
fn register_functions(query_engine: &Arc<DatafusionQueryEngine>) {
for func in FUNCTION_REGISTRY.functions() {
query_engine.register_function(func);
for func in FUNCTION_REGISTRY.scalar_functions() {
query_engine.register_scalar_function(func);
}
for accumulator in FUNCTION_REGISTRY.aggregate_functions() {

View File

@@ -15,9 +15,8 @@
use std::sync::Arc;
use common_error::ext::BoxedError;
use common_function::aggr::{GeoPathAccumulator, HllState, UddSketchState};
use common_function::function::FunctionContext;
use common_function::function_registry::FUNCTION_REGISTRY;
use common_function::scalars::udf::create_udf;
use common_query::error::RegisterUdfSnafu;
use common_query::logical_plan::SubstraitPlanDecoder;
use datafusion::catalog::CatalogProviderList;
@@ -29,11 +28,6 @@ use datafusion::execution::{FunctionRegistry, SessionStateBuilder};
use datafusion::logical_expr::LogicalPlan;
use datafusion_expr::UserDefinedLogicalNode;
use greptime_proto::substrait_extension::MergeScan as PbMergeScan;
use promql::functions::{
quantile_udaf, AbsentOverTime, AvgOverTime, Changes, CountOverTime, Delta, Deriv, IDelta,
Increase, LastOverTime, MaxOverTime, MinOverTime, PresentOverTime, Rate, Resets, Round,
StddevOverTime, StdvarOverTime, SumOverTime,
};
use prost::Message;
use session::context::QueryContextRef;
use snafu::ResultExt;
@@ -118,49 +112,41 @@ impl SubstraitPlanDecoder for DefaultPlanDecoder {
let mut session_state = SessionStateBuilder::new_from_existing(self.session_state.clone())
.with_catalog_list(catalog_list)
.build();
// Substrait decoder will look up the UDFs in SessionState, so we need to register them
// Note: the query context must be passed to set the timezone
// We MUST register the UDFs after we build the session state, otherwise the UDFs will be lost
// if they have the same name as the default UDFs or their alias.
// e.g. The default UDF `to_char()` has an alias `date_format()`, if we register a UDF with the name `date_format()`
// before we build the session state, the UDF will be lost.
for func in FUNCTION_REGISTRY.functions() {
let udf = Arc::new(create_udf(
func.clone(),
self.query_ctx.clone(),
Default::default(),
));
session_state
.register_udf(udf)
.context(RegisterUdfSnafu { name: func.name() })?;
let _ = session_state.register_udaf(Arc::new(UddSketchState::state_udf_impl()));
let _ = session_state.register_udaf(Arc::new(UddSketchState::merge_udf_impl()));
let _ = session_state.register_udaf(Arc::new(HllState::state_udf_impl()));
let _ = session_state.register_udaf(Arc::new(HllState::merge_udf_impl()));
let _ = session_state.register_udaf(Arc::new(GeoPathAccumulator::udf_impl()));
let _ = session_state.register_udaf(quantile_udaf());
let _ = session_state.register_udf(Arc::new(IDelta::<false>::scalar_udf()));
let _ = session_state.register_udf(Arc::new(IDelta::<true>::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Rate::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Increase::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Delta::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Resets::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Changes::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Deriv::scalar_udf()));
let _ = session_state.register_udf(Arc::new(Round::scalar_udf()));
let _ = session_state.register_udf(Arc::new(AvgOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(MinOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(MaxOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(SumOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(CountOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(LastOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(AbsentOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(PresentOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(StddevOverTime::scalar_udf()));
let _ = session_state.register_udf(Arc::new(StdvarOverTime::scalar_udf()));
// TODO(ruihang): add quantile_over_time, predict_linear, holt_winters, round
// Scalar functions
for func in FUNCTION_REGISTRY.scalar_functions() {
let udf = func.provide(FunctionContext {
query_ctx: self.query_ctx.clone(),
state: Default::default(),
});
session_state
.register_udf(Arc::new(udf))
.context(RegisterUdfSnafu { name: func.name() })?;
}
// PromQL range functions
for func in promql::functions::range_funcs() {
let name = func.name().to_string();
session_state
.register_udf(Arc::new(func))
.context(RegisterUdfSnafu { name })?;
}
// Aggregate functions
for func in FUNCTION_REGISTRY.aggregate_functions() {
let name = func.name().to_string();
session_state
.register_udaf(Arc::new(func))
.context(RegisterUdfSnafu { name })?;
}
let logical_plan = DFLogicalSubstraitConvertor
.decode(message, session_state)
.await

View File

@@ -19,11 +19,10 @@ use std::sync::{Arc, RwLock};
use async_trait::async_trait;
use catalog::CatalogManagerRef;
use common_base::Plugins;
use common_function::function::FunctionRef;
use common_function::function_factory::ScalarFunctionFactory;
use common_function::handlers::{
FlowServiceHandlerRef, ProcedureServiceHandlerRef, TableMutationHandlerRef,
};
use common_function::scalars::aggregate::AggregateFunctionMetaRef;
use common_function::state::FunctionState;
use common_telemetry::warn;
use datafusion::dataframe::DataFrame;
@@ -37,7 +36,7 @@ use datafusion::physical_optimizer::sanity_checker::SanityCheckPlan;
use datafusion::physical_optimizer::PhysicalOptimizerRule;
use datafusion::physical_plan::ExecutionPlan;
use datafusion::physical_planner::{DefaultPhysicalPlanner, ExtensionPlanner, PhysicalPlanner};
use datafusion_expr::LogicalPlan as DfLogicalPlan;
use datafusion_expr::{AggregateUDF, LogicalPlan as DfLogicalPlan};
use datafusion_optimizer::analyzer::count_wildcard_rule::CountWildcardRule;
use datafusion_optimizer::analyzer::{Analyzer, AnalyzerRule};
use datafusion_optimizer::optimizer::Optimizer;
@@ -70,8 +69,8 @@ pub struct QueryEngineState {
df_context: SessionContext,
catalog_manager: CatalogManagerRef,
function_state: Arc<FunctionState>,
udf_functions: Arc<RwLock<HashMap<String, FunctionRef>>>,
aggregate_functions: Arc<RwLock<HashMap<String, AggregateFunctionMetaRef>>>,
scalar_functions: Arc<RwLock<HashMap<String, ScalarFunctionFactory>>>,
aggr_functions: Arc<RwLock<HashMap<String, AggregateUDF>>>,
extension_rules: Vec<Arc<dyn ExtensionAnalyzerRule + Send + Sync>>,
plugins: Plugins,
}
@@ -186,10 +185,10 @@ impl QueryEngineState {
procedure_service_handler,
flow_service_handler,
}),
aggregate_functions: Arc::new(RwLock::new(HashMap::new())),
aggr_functions: Arc::new(RwLock::new(HashMap::new())),
extension_rules,
plugins,
udf_functions: Arc::new(RwLock::new(HashMap::new())),
scalar_functions: Arc::new(RwLock::new(HashMap::new())),
}
}
@@ -222,38 +221,28 @@ impl QueryEngineState {
self.session_state().optimize(&plan)
}
/// Register an udf function.
/// Will override if the function with same name is already registered.
pub fn register_function(&self, func: FunctionRef) {
let name = func.name().to_string();
let x = self
.udf_functions
.write()
.unwrap()
.insert(name.clone(), func);
if x.is_some() {
warn!("Already registered udf function '{name}'");
}
}
/// Retrieve the udf function by name
pub fn udf_function(&self, function_name: &str) -> Option<FunctionRef> {
self.udf_functions
/// Retrieve the scalar function by name
pub fn scalar_function(&self, function_name: &str) -> Option<ScalarFunctionFactory> {
self.scalar_functions
.read()
.unwrap()
.get(function_name)
.cloned()
}
/// Retrieve udf function names.
pub fn udf_names(&self) -> Vec<String> {
self.udf_functions.read().unwrap().keys().cloned().collect()
/// Retrieve scalar function names.
pub fn scalar_names(&self) -> Vec<String> {
self.scalar_functions
.read()
.unwrap()
.keys()
.cloned()
.collect()
}
/// Retrieve the aggregate function by name
pub fn aggregate_function(&self, function_name: &str) -> Option<AggregateFunctionMetaRef> {
self.aggregate_functions
pub fn aggr_function(&self, function_name: &str) -> Option<AggregateUDF> {
self.aggr_functions
.read()
.unwrap()
.get(function_name)
@@ -261,8 +250,8 @@ impl QueryEngineState {
}
/// Retrieve aggregate function names.
pub fn udaf_names(&self) -> Vec<String> {
self.aggregate_functions
pub fn aggr_names(&self) -> Vec<String> {
self.aggr_functions
.read()
.unwrap()
.keys()
@@ -270,6 +259,21 @@ impl QueryEngineState {
.collect()
}
/// Register an scalar function.
/// Will override if the function with same name is already registered.
pub fn register_scalar_function(&self, func: ScalarFunctionFactory) {
let name = func.name().to_string();
let x = self
.scalar_functions
.write()
.unwrap()
.insert(name.clone(), func);
if x.is_some() {
warn!("Already registered scalar function '{name}'");
}
}
/// Register an aggregate function.
///
/// # Panics
@@ -278,10 +282,10 @@ impl QueryEngineState {
/// Panicking consideration: currently the aggregated functions are all statically registered,
/// user cannot define their own aggregate functions on the fly. So we can panic here. If that
/// invariant is broken in the future, we should return an error instead of panicking.
pub fn register_aggregate_function(&self, func: AggregateFunctionMetaRef) {
let name = func.name();
pub fn register_aggr_function(&self, func: AggregateUDF) {
let name = func.name().to_string();
let x = self
.aggregate_functions
.aggr_functions
.write()
.unwrap()
.insert(name.clone(), func);

View File

@@ -16,11 +16,12 @@ use std::fmt::Debug;
use std::marker::PhantomData;
use std::sync::Arc;
use common_function::scalars::aggregate::AggregateFunctionMeta;
use common_macro::{as_aggr_func_creator, AggrFuncTypeStore};
use common_query::error::{CreateAccumulatorSnafu, Result as QueryResult};
use common_query::logical_plan::accumulator::AggrFuncTypeStore;
use common_query::logical_plan::{Accumulator, AggregateFunctionCreator};
use common_query::logical_plan::{
create_aggregate_function, Accumulator, AggregateFunctionCreator,
};
use common_query::prelude::*;
use common_recordbatch::{RecordBatch, RecordBatches};
use datatypes::prelude::*;
@@ -207,11 +208,14 @@ where
let engine = new_query_engine_with_table(testing_table);
engine.register_aggregate_function(Arc::new(AggregateFunctionMeta::new(
"my_sum",
1,
Arc::new(|| Arc::new(MySumAccumulatorCreator::default())),
)));
engine.register_aggregate_function(
create_aggregate_function(
"my_sum".to_string(),
1,
Arc::new(MySumAccumulatorCreator::default()),
)
.into(),
);
let sql = format!("select MY_SUM({column_name}) as my_sum from {table_name}");
let batches = exec_selection(engine, &sql).await;

View File

@@ -675,11 +675,16 @@ insert into cache_miss_with_null_label values
Affected Rows: 4
-- SQLNESS SORT_RESULT 3 1
-- null!=null, so it will returns the empty set.
tql eval (3, 4, '1s') cache_hit_with_null_label / (cache_miss_with_null_label + cache_hit_with_null_label);
++
++
+-------+------------+---------------------+---------------------------------------------------------------------------------------------------------------+
| job | null_label | ts | lhs.greptime_value / rhs.cache_miss_with_null_label.greptime_value + cache_hit_with_null_label.greptime_value |
+-------+------------+---------------------+---------------------------------------------------------------------------------------------------------------+
| read | | 1970-01-01T00:00:03 | 0.5 |
| read | | 1970-01-01T00:00:04 | 0.75 |
| write | | 1970-01-01T00:00:03 | 0.5 |
| write | | 1970-01-01T00:00:04 | 0.6666666666666666 |
+-------+------------+---------------------+---------------------------------------------------------------------------------------------------------------+
-- SQLNESS SORT_RESULT 3 1
tql eval (3, 4, '1s') cache_hit_with_null_label / ignoring(null_label) (cache_miss_with_null_label + ignoring(null_label) cache_hit_with_null_label);

View File

@@ -325,7 +325,6 @@ insert into cache_miss_with_null_label values
(4000, "write", null, 2.0);
-- SQLNESS SORT_RESULT 3 1
-- null!=null, so it will returns the empty set.
tql eval (3, 4, '1s') cache_hit_with_null_label / (cache_miss_with_null_label + cache_hit_with_null_label);
-- SQLNESS SORT_RESULT 3 1