feat: ingest jsonbench data through pipeline (#7312)

Signed-off-by: luofucong <luofc@foxmail.com>
This commit is contained in:
LFC
2026-01-05 20:12:34 +08:00
committed by GitHub
parent 2d756b24c8
commit 522ca99cd6
25 changed files with 595 additions and 251 deletions

View File

@@ -895,7 +895,7 @@ pub fn is_column_type_value_eq(
.unwrap_or(false)
}
fn encode_json_value(value: JsonValue) -> v1::JsonValue {
pub fn encode_json_value(value: JsonValue) -> v1::JsonValue {
fn helper(json: JsonVariant) -> v1::JsonValue {
let value = match json {
JsonVariant::Null => None,

View File

@@ -17,8 +17,8 @@ use std::collections::HashMap;
use arrow_schema::extension::{EXTENSION_TYPE_METADATA_KEY, EXTENSION_TYPE_NAME_KEY};
use datatypes::schema::{
COMMENT_KEY, ColumnDefaultConstraint, ColumnSchema, FULLTEXT_KEY, FulltextAnalyzer,
FulltextBackend, FulltextOptions, INVERTED_INDEX_KEY, SKIPPING_INDEX_KEY, SkippingIndexOptions,
SkippingIndexType,
FulltextBackend, FulltextOptions, INVERTED_INDEX_KEY, Metadata, SKIPPING_INDEX_KEY,
SkippingIndexOptions, SkippingIndexType,
};
use greptime_proto::v1::{
Analyzer, FulltextBackend as PbFulltextBackend, SkippingIndexType as PbSkippingIndexType,
@@ -36,6 +36,14 @@ const INVERTED_INDEX_GRPC_KEY: &str = "inverted_index";
/// Key used to store skip index options in gRPC column options.
const SKIPPING_INDEX_GRPC_KEY: &str = "skipping_index";
const COLUMN_OPTION_MAPPINGS: [(&str, &str); 5] = [
(FULLTEXT_GRPC_KEY, FULLTEXT_KEY),
(INVERTED_INDEX_GRPC_KEY, INVERTED_INDEX_KEY),
(SKIPPING_INDEX_GRPC_KEY, SKIPPING_INDEX_KEY),
(EXTENSION_TYPE_NAME_KEY, EXTENSION_TYPE_NAME_KEY),
(EXTENSION_TYPE_METADATA_KEY, EXTENSION_TYPE_METADATA_KEY),
];
/// Tries to construct a `ColumnSchema` from the given `ColumnDef`.
pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
let data_type = ColumnDataTypeWrapper::try_new(
@@ -131,6 +139,21 @@ pub fn try_as_column_def(column_schema: &ColumnSchema, is_primary_key: bool) ->
})
}
/// Collect the [ColumnOptions] into the [Metadata] that can be used in, for example, [ColumnSchema].
pub fn collect_column_options(column_options: Option<&ColumnOptions>) -> Metadata {
let Some(ColumnOptions { options }) = column_options else {
return Metadata::default();
};
let mut metadata = Metadata::with_capacity(options.len());
for (x, y) in COLUMN_OPTION_MAPPINGS {
if let Some(v) = options.get(x) {
metadata.insert(y.to_string(), v.clone());
}
}
metadata
}
/// Constructs a `ColumnOptions` from the given `ColumnSchema`.
pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<ColumnOptions> {
let mut options = ColumnOptions::default();