feat: impl COPY a query resultset to external file (#5250)

* feat: impl COPY a query resultset to external file

* chore: add more tests for parse `copy_table_to`

* chore: add more tests for parse `copy_table_to`
This commit is contained in:
Kould
2025-01-06 04:32:51 +00:00
committed by GitHub
parent 513569ed5d
commit 2ad50332cb
9 changed files with 296 additions and 14 deletions
+115 -2
View File
@@ -16,11 +16,15 @@ use std::collections::HashMap;
use snafu::ResultExt;
use sqlparser::keywords::Keyword;
use sqlparser::tokenizer::Token;
use sqlparser::tokenizer::Token::Word;
use crate::error::{self, Result};
use crate::parser::ParserContext;
use crate::statements::copy::{CopyDatabase, CopyDatabaseArgument, CopyTable, CopyTableArgument};
use crate::statements::copy::{
CopyDatabase, CopyDatabaseArgument, CopyQueryTo, CopyQueryToArgument, CopyTable,
CopyTableArgument,
};
use crate::statements::statement::Statement;
use crate::util::parse_option_string;
@@ -32,7 +36,11 @@ impl ParserContext<'_> {
pub(crate) fn parse_copy(&mut self) -> Result<Statement> {
let _ = self.parser.next_token();
let next = self.parser.peek_token();
let copy = if let Word(word) = next.token
let copy = if next.token == Token::LParen {
let copy_query = self.parse_copy_query_to()?;
crate::statements::copy::Copy::CopyQueryTo(copy_query)
} else if let Word(word) = next.token
&& word.keyword == Keyword::DATABASE
{
let _ = self.parser.next_token();
@@ -126,6 +134,40 @@ impl ParserContext<'_> {
}
}
fn parse_copy_query_to(&mut self) -> Result<CopyQueryTo> {
self.parser
.expect_token(&Token::LParen)
.with_context(|_| error::UnexpectedSnafu {
expected: "'('",
actual: self.peek_token_as_string(),
})?;
let query = self.parse_query()?;
self.parser
.expect_token(&Token::RParen)
.with_context(|_| error::UnexpectedSnafu {
expected: "')'",
actual: self.peek_token_as_string(),
})?;
self.parser
.expect_keyword(Keyword::TO)
.context(error::SyntaxSnafu)?;
let (with, connection, location, limit) = self.parse_copy_parameters()?;
if limit.is_some() {
return error::InvalidSqlSnafu {
msg: "limit is not supported",
}
.fail();
}
Ok(CopyQueryTo {
query: Box::new(query),
arg: CopyQueryToArgument {
with: with.into(),
connection: connection.into(),
location,
},
})
}
fn parse_copy_parameters(&mut self) -> Result<(With, Connection, String, Option<u64>)> {
let location =
self.parser
@@ -453,4 +495,75 @@ mod tests {
stmt.connection.to_str_map()
);
}
#[test]
fn test_copy_query_to() {
let sql = "COPY (SELECT * FROM tbl WHERE ts > 10) TO 'tbl_file.parquet' WITH (FORMAT = 'parquet') CONNECTION (FOO='Bar', ONE='two')";
let stmt =
ParserContext::create_with_dialect(sql, &GreptimeDbDialect {}, ParseOptions::default())
.unwrap()
.pop()
.unwrap();
let Copy(crate::statements::copy::Copy::CopyQueryTo(stmt)) = stmt else {
unreachable!()
};
let query = ParserContext::create_with_dialect(
"SELECT * FROM tbl WHERE ts > 10",
&GreptimeDbDialect {},
ParseOptions::default(),
)
.unwrap()
.remove(0);
assert_eq!(&query, stmt.query.as_ref());
assert_eq!(
[("format", "parquet")]
.into_iter()
.collect::<HashMap<_, _>>(),
stmt.arg.with.to_str_map()
);
assert_eq!(
[("foo", "Bar"), ("one", "two")]
.into_iter()
.collect::<HashMap<_, _>>(),
stmt.arg.connection.to_str_map()
);
}
#[test]
fn test_invalid_copy_query_to() {
{
let sql = "COPY SELECT * FROM tbl WHERE ts > 10 TO 'tbl_file.parquet' WITH (FORMAT = 'parquet') CONNECTION (FOO='Bar', ONE='two')";
assert!(ParserContext::create_with_dialect(
sql,
&GreptimeDbDialect {},
ParseOptions::default()
)
.is_err())
}
{
let sql = "COPY SELECT * FROM tbl WHERE ts > 10) TO 'tbl_file.parquet' WITH (FORMAT = 'parquet') CONNECTION (FOO='Bar', ONE='two')";
assert!(ParserContext::create_with_dialect(
sql,
&GreptimeDbDialect {},
ParseOptions::default()
)
.is_err())
}
{
let sql = "COPY (SELECT * FROM tbl WHERE ts > 10 TO 'tbl_file.parquet' WITH (FORMAT = 'parquet') CONNECTION (FOO='Bar', ONE='two')";
assert!(ParserContext::create_with_dialect(
sql,
&GreptimeDbDialect {},
ParseOptions::default()
)
.is_err())
}
}
}
+56
View File
@@ -18,12 +18,14 @@ use serde::Serialize;
use sqlparser::ast::ObjectName;
use sqlparser_derive::{Visit, VisitMut};
use crate::statements::statement::Statement;
use crate::statements::OptionMap;
#[derive(Debug, Clone, PartialEq, Eq, Visit, VisitMut, Serialize)]
pub enum Copy {
CopyTable(CopyTable),
CopyDatabase(CopyDatabase),
CopyQueryTo(CopyQueryTo),
}
impl Display for Copy {
@@ -31,10 +33,32 @@ impl Display for Copy {
match self {
Copy::CopyTable(s) => s.fmt(f),
Copy::CopyDatabase(s) => s.fmt(f),
Copy::CopyQueryTo(s) => s.fmt(f),
}
}
}
#[derive(Debug, Clone, PartialEq, Eq, Visit, VisitMut, Serialize)]
pub struct CopyQueryTo {
pub query: Box<Statement>,
pub arg: CopyQueryToArgument,
}
impl Display for CopyQueryTo {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
write!(f, "COPY ({}) TO {}", &self.query, &self.arg.location)?;
if !self.arg.with.is_empty() {
let options = self.arg.with.kv_pairs();
write!(f, " WITH ({})", options.join(", "))?;
}
if !self.arg.connection.is_empty() {
let options = self.arg.connection.kv_pairs();
write!(f, " CONNECTION ({})", options.join(", "))?;
}
Ok(())
}
}
#[derive(Debug, Clone, PartialEq, Eq, Visit, VisitMut, Serialize)]
pub enum CopyTable {
To(CopyTableArgument),
@@ -115,6 +139,13 @@ pub struct CopyTableArgument {
pub limit: Option<u64>,
}
#[derive(Debug, Clone, PartialEq, Eq, Visit, VisitMut, Serialize)]
pub struct CopyQueryToArgument {
pub with: OptionMap,
pub connection: OptionMap,
pub location: String,
}
#[cfg(test)]
impl CopyTableArgument {
pub fn format(&self) -> Option<String> {
@@ -238,4 +269,29 @@ mod tests {
}
}
}
#[test]
fn test_display_copy_query_to() {
let sql = r"copy (SELECT * FROM tbl WHERE ts > 10) to 's3://my-bucket/data.parquet'
with (format = 'parquet', pattern = '.*parquet.*')
connection(region = 'us-west-2', secret_access_key = '12345678');";
let stmts =
ParserContext::create_with_dialect(sql, &GreptimeDbDialect {}, ParseOptions::default())
.unwrap();
assert_eq!(1, stmts.len());
assert_matches!(&stmts[0], Statement::Copy { .. });
match &stmts[0] {
Statement::Copy(copy) => {
let new_sql = format!("{}", copy);
assert_eq!(
r#"COPY (SELECT * FROM tbl WHERE ts > 10) TO s3://my-bucket/data.parquet WITH (format = 'parquet', pattern = '.*parquet.*') CONNECTION (region = 'us-west-2', secret_access_key = '******')"#,
&new_sql
);
}
_ => {
unreachable!();
}
}
}
}