refactor(cubesql): Use LazyLock instead of lazy_static, move more reg…

…exps to statics (#8675)

* Replace lazy_static with LazyLock in many places
* Replace lazy_static with Once for testing logger init
* Remove unused testing logging static in config
* Move more static regexps to LazyLocks

rust/cubesql/cubesql/Cargo.toml

@@ -16,7 +16,6 @@ thiserror = "1.0.50"
 cubeclient = { path = "../cubeclient" }
 pg-srv = { path = "../pg-srv" }
 sqlparser = { git = 'https://github.com/cube-js/sqlparser-rs.git', rev = "6a54d27d3b75a04b9f9cbe309a83078aa54b32fd" }
-lazy_static = "1.4.0"
 base64 = "0.13.0"
 tokio = { version = "^1.35", features = ["full", "rt", "tracing"] }
 serde = { version = "^1.0", features = ["derive"] }

rust/cubesql/cubesql/src/compile/engine/df/wrapper.rs

@@ -32,8 +32,16 @@ use itertools::Itertools;
 use regex::{Captures, Regex};
 use serde::{Deserialize, Serialize};
 use std::{
-    any::Any, cmp::min, collections::HashMap, convert::TryInto, fmt, future::Future, iter,
-    pin::Pin, result, sync::Arc,
+    any::Any,
+    cmp::min,
+    collections::HashMap,
+    convert::TryInto,
+    fmt,
+    future::Future,
+    iter,
+    pin::Pin,
+    result,
+    sync::{Arc, LazyLock},
 };
 
 #[derive(Debug, Clone, Deserialize)]
@@ -160,12 +168,12 @@ impl SqlQuery {
     }
 
     pub fn finalize_query(&mut self, sql_templates: Arc<SqlTemplates>) -> Result<()> {
+        static REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\$(\d+)\$").unwrap());
+
         let mut params = Vec::new();
         let mut rendered_params = HashMap::new();
-        let regex = Regex::new(r"\$(\d+)\$")
-            .map_err(|e| DataFusionError::Execution(format!("Can't parse regex: {}", e)))?;
         let mut res = Ok(());
-        let replaced_sql = regex.replace_all(self.sql.as_str(), |c: &Captures<'_>| {
+        let replaced_sql = REGEX.replace_all(self.sql.as_str(), |c: &Captures<'_>| {
             let param = c.get(1).map(|x| x.as_str());
             match self.render_param(sql_templates.clone(), param, &rendered_params, params.len()) {
                 Ok((param_index, param, push_param)) => {
@@ -260,9 +268,7 @@ pub struct SqlGenerationResult {
     pub request: TransportLoadRequestQuery,
 }
 
-lazy_static! {
-    static ref DATE_PART_REGEX: Regex = Regex::new("^[A-Za-z_ ]+$").unwrap();
-}
+static DATE_PART_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new("^[A-Za-z_ ]+$").unwrap());
 
 macro_rules! generate_sql_for_timestamp {
     (@generic $value:ident, $value_block:expr, $sql_generator:expr, $sql_query:expr) => {
@@ -950,8 +956,9 @@ impl CubeScanWrapperNode {
         ungrouped_scan_node: Option<Arc<CubeScanNode>>,
         subqueries: Arc<HashMap<String, String>>,
     ) -> result::Result<(Vec<AliasedColumn>, SqlQuery), CubeError> {
-        let non_id_regex = Regex::new(r"[^a-zA-Z0-9_]")
-            .map_err(|e| CubeError::internal(format!("Can't parse regex: {}", e)))?;
+        static NON_ID_REGEX: LazyLock<Regex> =
+            LazyLock::new(|| Regex::new(r"[^a-zA-Z0-9_]").unwrap());
+
         let mut aliased_columns = Vec::new();
         for original_expr in exprs {
             let expr = if let Some(column_remapping) = column_remapping.as_ref() {
@@ -1001,7 +1008,7 @@ impl CubeScanWrapperNode {
 
             let alias = if can_rename_columns {
                 let alias = expr_name(&expr, &schema)?;
-                let mut truncated_alias = non_id_regex
+                let mut truncated_alias = NON_ID_REGEX
                     .replace_all(&alias, "_")
                     .trim_start_matches("_")
                     .to_lowercase();

rust/cubesql/cubesql/src/compile/engine/udf/common.rs

@@ -1,4 +1,8 @@
-use std::{any::type_name, sync::Arc, thread};
+use std::{
+    any::type_name,
+    sync::{Arc, LazyLock},
+    thread,
+};
 
 use chrono::{Datelike, Days, Duration, Months, NaiveDate, NaiveDateTime, NaiveTime};
 use datafusion::{
@@ -3329,17 +3333,18 @@ pub fn create_current_setting_udf() -> ScalarUDF {
 }
 
 pub fn create_quote_ident_udf() -> ScalarUDF {
+    static RE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^[a-z_][a-z0-9_]*$").unwrap());
+
     let fun = make_scalar_function(move |args: &[ArrayRef]| {
         assert!(args.len() == 1);
 
         let idents = downcast_string_arg!(args[0], "str", i32);
 
-        let re = Regex::new(r"^[a-z_][a-z0-9_]*$").unwrap();
         let result = idents
             .iter()
             .map(|ident| {
                 ident.map(|ident| {
-                    if re.is_match(ident) {
+                    if RE.is_match(ident) {
                         return ident.to_string();
                     }
                     format!("\"{}\"", ident.replace("\"", "\"\""))

rust/cubesql/cubesql/src/compile/parser.rs

@@ -1,4 +1,4 @@
-use std::collections::HashMap;
+use std::{collections::HashMap, sync::LazyLock};
 
 use regex::Regex;
 use sqlparser::{
@@ -36,9 +36,9 @@ impl Dialect for MySqlDialectWithBackTicks {
     }
 }
 
-lazy_static! {
-    static ref SIGMA_WORKAROUND: Regex = Regex::new(r#"(?s)^\s*with\s+nsp\sas\s\(.*nspname\s=\s.*\),\s+tbl\sas\s\(.*relname\s=\s.*\).*select\s+attname.*from\spg_attribute.*$"#).unwrap();
-}
+static SIGMA_WORKAROUND: LazyLock<Regex> = LazyLock::new(|| {
+    Regex::new(r#"(?s)^\s*with\s+nsp\sas\s\(.*nspname\s=\s.*\),\s+tbl\sas\s\(.*relname\s=\s.*\).*select\s+attname.*from\spg_attribute.*$"#).unwrap()
+});
 
 pub fn parse_sql_to_statements(
     query: &String,
@@ -118,13 +118,18 @@ pub fn parse_sql_to_statements(
     // Sigma Computing WITH query workaround
     // TODO: remove workaround when subquery is supported in JOIN ON conditions
     let query = if SIGMA_WORKAROUND.is_match(&query) {
-        let relnamespace_re = Regex::new(r#"(?s)from\spg_catalog\.pg_class\s+where\s+relname\s=\s(?P<relname>'(?:[^']|'')+'|\$\d+)\s+and\s+relnamespace\s=\s\(select\soid\sfrom\snsp\)"#).unwrap();
-        let relnamespace_replaced = relnamespace_re.replace(
+        static RELNAMESPACE_RE: LazyLock<Regex> = LazyLock::new(|| {
+            Regex::new(r#"(?s)from\spg_catalog\.pg_class\s+where\s+relname\s=\s(?P<relname>'(?:[^']|'')+'|\$\d+)\s+and\s+relnamespace\s=\s\(select\soid\sfrom\snsp\)"#).unwrap()
+        });
+        static ATTRELID_RE: LazyLock<Regex> = LazyLock::new(|| {
+            Regex::new(r#"(?s)left\sjoin\spg_description\son\s+attrelid\s=\sobjoid\sand\s+attnum\s=\sobjsubid\s+where\s+attnum\s>\s0\s+and\s+attrelid\s=\s\(select\soid\sfrom\stbl\)"#).unwrap()
+        });
+
+        let relnamespace_replaced = RELNAMESPACE_RE.replace(
             &query,
             "from pg_catalog.pg_class join nsp on relnamespace = nsp.oid where relname = $relname",
         );
-        let attrelid_re = Regex::new(r#"(?s)left\sjoin\spg_description\son\s+attrelid\s=\sobjoid\sand\s+attnum\s=\sobjsubid\s+where\s+attnum\s>\s0\s+and\s+attrelid\s=\s\(select\soid\sfrom\stbl\)"#).unwrap();
-        let attrelid_replaced = attrelid_re.replace(&relnamespace_replaced, "left join pg_description on attrelid = objoid and attnum = objsubid join tbl on attrelid = tbl.oid where attnum > 0");
+        let attrelid_replaced = ATTRELID_RE.replace(&relnamespace_replaced, "left join pg_description on attrelid = objoid and attnum = objsubid join tbl on attrelid = tbl.oid where attnum > 0");
         attrelid_replaced.to_string()
     } else {
         query

rust/cubesql/cubesql/src/compile/rewrite/converter.rs

@@ -59,7 +59,7 @@ use std::{
     collections::{HashMap, HashSet},
     env,
     ops::Index,
-    sync::Arc,
+    sync::{Arc, LazyLock},
 };
 
 pub use super::rewriter::CubeRunner;
@@ -170,8 +170,8 @@ macro_rules! add_plan_list_node {
     }};
 }
 
-lazy_static! {
-    static ref EXCLUDED_PARAM_VALUES: HashSet<ScalarValue> = vec![
+static EXCLUDED_PARAM_VALUES: LazyLock<HashSet<ScalarValue>> = LazyLock::new(|| {
+    vec![
         ScalarValue::Utf8(Some("second".to_string())),
         ScalarValue::Utf8(Some("minute".to_string())),
         ScalarValue::Utf8(Some("hour".to_string())),
@@ -182,8 +182,8 @@ lazy_static! {
     ]
     .into_iter()
     .chain((0..50).map(|i| ScalarValue::Int64(Some(i))))
-    .collect();
-}
+    .collect()
+});
 
 pub struct LogicalPlanToLanguageConverter {
     graph: EGraph<LogicalPlanLanguage, LogicalPlanAnalysis>,

rust/cubesql/cubesql/src/compile/rewrite/rules/members.rs

@@ -46,7 +46,7 @@ use std::{
     collections::{HashMap, HashSet},
     fmt::Display,
     ops::{Index, IndexMut},
-    sync::Arc,
+    sync::{Arc, LazyLock},
 };
 
 pub struct MemberRules {
@@ -2857,27 +2857,30 @@ pub fn add_member_error(
     ]))
 }
 
-lazy_static! {
-    static ref STANDARD_GRANULARITIES_PARENTS: HashMap<&'static str, Vec<&'static str>> = [
-        (
-            "year",
-            vec!["year", "quarter", "month", "day", "hour", "minute", "second"]
-        ),
-        (
-            "quarter",
-            vec!["quarter", "month", "day", "hour", "minute", "second"]
-        ),
-        ("month", vec!["month", "day", "hour", "minute", "second"]),
-        ("week", vec!["week", "day", "hour", "minute", "second"]),
-        ("day", vec!["day", "hour", "minute", "second"]),
-        ("hour", vec!["hour", "minute", "second"]),
-        ("minute", vec!["minute", "second"]),
-        ("second", vec!["second"]),
-    ]
-    .iter()
-    .cloned()
-    .collect();
-}
+static STANDARD_GRANULARITIES_PARENTS: LazyLock<HashMap<&'static str, Vec<&'static str>>> =
+    LazyLock::new(|| {
+        [
+            (
+                "year",
+                vec![
+                    "year", "quarter", "month", "day", "hour", "minute", "second",
+                ],
+            ),
+            (
+                "quarter",
+                vec!["quarter", "month", "day", "hour", "minute", "second"],
+            ),
+            ("month", vec!["month", "day", "hour", "minute", "second"]),
+            ("week", vec!["week", "day", "hour", "minute", "second"]),
+            ("day", vec!["day", "hour", "minute", "second"]),
+            ("hour", vec!["hour", "minute", "second"]),
+            ("minute", vec!["minute", "second"]),
+            ("second", vec!["second"]),
+        ]
+        .iter()
+        .cloned()
+        .collect()
+    });
 
 pub fn min_granularity(granularity_a: &String, granularity_b: &String) -> Option<String> {
     let granularity_a = granularity_a.to_lowercase();

rust/cubesql/cubesql/src/compile/test/mod.rs

@@ -915,14 +915,10 @@ impl TestContext {
     }
 }
 
-lazy_static! {
-    pub static ref TEST_LOGGING_INITIALIZED: std::sync::RwLock<bool> =
-        std::sync::RwLock::new(false);
-}
+static TEST_LOGGING_INITIALIZED: std::sync::Once = std::sync::Once::new();
 
 pub fn init_testing_logger() {
-    let mut initialized = TEST_LOGGING_INITIALIZED.write().unwrap();
-    if !*initialized {
+    TEST_LOGGING_INITIALIZED.call_once(|| {
         let log_level = log::Level::Trace;
         let logger = simple_logger::SimpleLogger::new()
             .with_level(log::Level::Error.to_level_filter())
@@ -933,8 +929,7 @@ pub fn init_testing_logger() {
 
         log::set_boxed_logger(Box::new(logger)).unwrap();
         log::set_max_level(log_level.to_level_filter());
-        *initialized = true;
-    }
+    });
 }
 
 pub async fn convert_select_to_query_plan_customized(

rust/cubesql/cubesql/src/config/mod.rs

@@ -237,11 +237,6 @@ impl ConfigObj for ConfigObjImpl {
     }
 }
 
-lazy_static! {
-    pub static ref TEST_LOGGING_INITIALIZED: tokio::sync::RwLock<bool> =
-        tokio::sync::RwLock::new(false);
-}
-
 impl Config {
     pub fn default() -> Config {
         Config {

rust/cubesql/cubesql/src/lib.rs

@@ -15,8 +15,6 @@
 
 // trace_macros!(false);
 
-#[macro_use]
-extern crate lazy_static;
 extern crate core;
 
 pub mod compile;

rust/cubesql/cubesql/src/sql/session.rs

@@ -3,7 +3,7 @@ use log::trace;
 use rand::Rng;
 use std::{
     collections::HashMap,
-    sync::{Arc, RwLock as RwLockSync, Weak},
+    sync::{Arc, LazyLock, RwLock as RwLockSync, Weak},
     time::{Duration, SystemTime},
 };
 use tokio_util::sync::CancellationToken;
@@ -23,8 +23,6 @@ use crate::{
     RWLockAsync,
 };
 
-extern crate lazy_static;
-
 #[derive(Debug, Clone)]
 pub struct SessionProperties {
     user: Option<String>,
@@ -37,10 +35,10 @@ impl SessionProperties {
     }
 }
 
-lazy_static! {
-    static ref POSTGRES_DEFAULT_VARIABLES: DatabaseVariables = postgres_default_session_variables();
-    static ref MYSQL_DEFAULT_VARIABLES: DatabaseVariables = mysql_default_session_variables();
-}
+static POSTGRES_DEFAULT_VARIABLES: LazyLock<DatabaseVariables> =
+    LazyLock::new(postgres_default_session_variables);
+static MYSQL_DEFAULT_VARIABLES: LazyLock<DatabaseVariables> =
+    LazyLock::new(mysql_default_session_variables);
 
 #[derive(Debug)]
 pub enum TransactionState {

rust/cubesql/cubesql/src/telemetry/mod.rs

@@ -1,12 +1,14 @@
 use crate::{compile::DatabaseProtocolDetails, sql::SessionState, CubeError};
 use arc_swap::ArcSwap;
 use log::{Level, LevelFilter};
-use std::{collections::HashMap, fmt::Debug, sync::Arc};
-
-lazy_static! {
-    static ref REPORTER: ArcSwap<Box<dyn LogReporter>> =
-        ArcSwap::from_pointee(Box::new(LocalReporter::new()));
-}
+use std::{
+    collections::HashMap,
+    fmt::Debug,
+    sync::{Arc, LazyLock},
+};
+
+static REPORTER: LazyLock<ArcSwap<Box<dyn LogReporter>>> =
+    LazyLock::new(|| ArcSwap::from_pointee(Box::new(LocalReporter::new())));
 
 pub trait LogReporter: Send + Sync + Debug {
     fn log(&self, event: String, properties: HashMap<String, String>, level: Level);