chore: fix some meta typo

.typos.toml

@@ -1,39 +1,58 @@
 # If there are words that you think should not be treated as typo.
 # Please list here along with a comment.
+
+[default]
+extend-ignore-re = [
+    "(?Rm)^.*(#|//)\\s*typos:disable-line$",
+    "(#|//)\\s*typos:ignore-next-line\\n.*",
+    "(?s)(#|//)\\s*typos:off.*?\\n\\s*(#|//)\\s*typos:on",
+]
+
 [default.extend-words]
 # Preserved for NdJson
-"Nd" = "Nd"
-# Microsoft K8s Server
-"AKS" = "AKS"
+Nd = "Nd"
+typ = "typ"
+aranges = "aranges"
+
+[default.extend-identifiers]
+
 # Wrong cases
-"ba" = "ba"
-"ue" = "ue"
-"INOUT" = "INOUT"
-"ser" = "ser"
-"Ser" = "Ser"
-"flate" = "flate"
-"Tke" = "Tke"
-"typ" = "typ"
-"dows" = "dows"
-# created_ons & updated_ons
-"ons" = "ons"
-
-# Intentional typos used in SQL error suggestion test cases
-"tabl" = "tabl"
-"optmize" = "optmize" 
-"vacum" = "vacum"
-"creat" = "creat"
-"crate" = "crate"
+Tke = "Tke"
+ba = "ba"
+
+# sql function name: diff_dows, between_dows
+diff_dows = "diff_dows"
+between_dows = "between_dows"
+
+into_flate2 = "into_flate2"
 
 [files]
 extend-exclude = [
     "**/Cargo.toml",
-    "scripts/benchmark/",
     "benchmark/",
-    "tests/",
-    "src/meta",
-    # masked secrets
-    "src/query/config/src/mask.rs",
-    # Forked from upstream
-    "src/common/parquet2"
+    "tests/sqllogictests",
+    "tests/nox/java_client/testng.xml",
+    "tests/data/ddl",
 ]
+
+# todo: fix me
+[type.py]
+check-file = false
+
+# todo: fix me
+[type.sh]
+check-file = false
+
+[type.txt]
+check-file = false
+
+[type.sql]
+check-file = false
+
+[type.csr]
+extend-glob = ["*.csr"]
+check-file = false
+
+[type.result]
+extend-glob = ["*.result"]
+check-file = false

src/common/hashtable/src/utils.rs

@@ -234,7 +234,7 @@ pub struct MergeIntoBlockInfoIndex {
 pub type Interval = (u32, u32);
 
 /// the segment blocks are not sequential,because we do parallel hashtable build.
-/// the block lay out in chunks could be like belows:
+/// the block lay out in chunks could be like bellows:
 /// segment0_block1 |
 /// segment1_block0 |  chunk0
 /// segment0_block0 |

src/common/io/tests/it/main.rs

@@ -24,4 +24,4 @@ mod borsh_serialization;
 mod cursor_ext;
 mod escape;
 mod interval;
-mod serialization_format_compatability;
+mod serialization_format_compatibility;

src/meta/api/src/garbage_collection_api.rs

@@ -129,7 +129,7 @@ async fn remove_copied_files_for_dropped_table(
     // Loop until:
     // - all cleaned
     // - or table is removed from meta-service
-    // - or is no longer in `droppped` state.
+    // - or is no longer in `dropped` state.
     for i in 0..usize::MAX {
         let mut txn = TxnRequest::default();
 

src/meta/api/src/kv_pb_api/mod.rs

@@ -559,7 +559,7 @@ mod tests {
         ) -> Result<KVStream<Self::Error>, Self::Error> {
             let mut res = Vec::with_capacity(keys.len());
             for (i, key) in keys.iter().enumerate() {
-                // For tesing early return stream.
+                // For testing early return stream.
                 if let Some(early_return) = self.early_return {
                     if i >= early_return {
                         break;

src/meta/api/src/schema_api_test_suite.rs

@@ -2714,7 +2714,7 @@ impl SchemaApiTestSuite {
                 // Change the comment, this modification should not be committed
                 req.new_table_meta.comment = "some new comment".to_string();
                 // Expects no KV api level error, and no app level error.
-                // For the convenience of reviewing, using explict type signature
+                // For the convenience of reviewing, using explicit type signature
                 let _r: databend_common_meta_app::schema::UpdateTableMetaReply = mt
                     .update_multi_table_meta_with_sender(
                         UpdateMultiTableMetaReq {
@@ -5526,7 +5526,7 @@ impl SchemaApiTestSuite {
             let handle = runtime.spawn(async move {
                 let resp = arc_mt.create_table(create_table_req.clone()).await;
 
-                // assert that when create table concurrently with corret params return error,
+                // assert that when create table concurrently with correct params return error,
                 // the error MUST be TxnRetryMaxTimes
                 if resp.is_err() {
                     assert!(matches!(
@@ -5547,7 +5547,7 @@ impl SchemaApiTestSuite {
                 };
                 let resp = arc_mt.commit_table_meta(commit_table_req).await;
 
-                // assert that when commit_table_meta concurrently with corret params return error,
+                // assert that when commit_table_meta concurrently with correct params return error,
                 // the error MUST be TxnRetryMaxTimes or CommitTableMetaError(prev table id has been changed)
                 if resp.is_err() {
                     assert!(

src/meta/app/src/principal/procedure.rs

@@ -71,7 +71,7 @@ impl Display for ProcedureMeta {
     fn fmt(&self, f: &mut Formatter) -> fmt::Result {
         write!(
             f,
-            "Lanuage: {:?}, args {} return_type: {}, CreatedOn: {:?}, Script: {}, Comment: {:?}",
+            "Language: {:?}, args {} return_type: {}, CreatedOn: {:?}, Script: {}, Comment: {:?}",
             self.procedure_language,
             self.arg_names.display_n(1000),
             self.return_types.display_n(1000),

src/meta/app/src/schema/table/mod.rs

@@ -137,7 +137,7 @@ pub struct TableStatistics {
     pub number_of_blocks: Option<u64>,
 }
 
-/// Iceberg table parition
+/// Iceberg table partition
 #[derive(serde::Serialize, serde::Deserialize, Clone, Debug, Eq, PartialEq)]
 pub enum TablePartition {
     Identity { columns: Vec<String> },

src/meta/client/src/grpc_client.rs

@@ -440,7 +440,7 @@ impl MetaGrpcClient {
         resp_err: Option<&(dyn std::error::Error + 'static)>,
     ) {
         // TODO: this current endpoint is not stable, may be modified by other thread.
-        //       The caller should pasing the in use endpoint.
+        //       The caller should passing the in use endpoint.
         let current_endpoint = self.get_current_endpoint();
 
         let Some(endpoint) = current_endpoint else {

src/meta/proto-conv/tests/it/common.rs

@@ -32,7 +32,7 @@ where MT: FromToProto + PartialEq + Debug {
     prost::Message::encode(&p, &mut buf)?;
 
     let var_name = n.split("::").last().unwrap();
-    // The encoded data should be saved for compatability test.
+    // The encoded data should be saved for compatibility test.
     println!("// Encoded data of version {} of {}:", VER, n);
     println!("// It is generated with common::test_pb_from_to().");
     println!(

src/meta/proto-conv/tests/it/v137_add_grant_object_connection.rs

@@ -59,7 +59,7 @@ fn test_decode_v137_grant_object() -> anyhow::Result<()> {
                     mt::principal::GrantObject::Connection("c1".to_string()),
                     make_bitflags!(UserPrivilegeType::{AccessConnection}),
                 ),
-                // test new global privilege CreateConneciton, AccessConnection
+                // test new global privilege CreateConnection, AccessConnection
                 mt::principal::GrantEntry::new(
                     mt::principal::GrantObject::Global,
                     make_bitflags!(UserPrivilegeType::{Create | Select | Insert | Update | Delete | Drop | Alter | Super | CreateUser | DropUser | CreateRole | DropRole | Grant | CreateStage | Set | CreateDataMask | Ownership | Read | Write | CreateWarehouse | CreateConnection | AccessConnection }),

src/meta/protos/proto/virtual_schema.proto

@@ -59,7 +59,7 @@ message VirtualDataSchema {
 
   // Virtual Fields in the table
   repeated VirtualDataField fields = 1;
-  // Other optional virutal columns info
+  // Other optional virtual columns info
   map<string, string> metadata = 2;
   // start from 3_000_000_001, Avoid conflicts with the column ids of normal columns
   uint32 next_column_id = 3;

src/meta/raft-store/src/applier/mod.rs

@@ -315,7 +315,7 @@ where SM: StateMachineApi<SysData> + 'static
         debug!("apply txn cmd {}: {}", self.cmd_ctx, req);
 
         // 1. Evaluate conditional operations one by one.
-        //    Once one of them is successful, execute the corresponding operations and retrun.
+        //    Once one of them is successful, execute the corresponding operations and return.
         //    Otherwise, try next.
         for (i, conditional) in req.operations.iter().enumerate() {
             let success = if let Some(predicate) = &conditional.predicate {

src/meta/raft-store/src/sm_v003/writer_v003.rs

@@ -133,7 +133,7 @@ impl WriterV003 {
     /// It returns a sender to send entries and a handle to wait for the thread to finish.
     /// Internally it calls tokio::spawn_blocking.
     ///
-    /// When a [`WritenEntry::Finish`] is received, the thread will flush the data to disk and return
+    /// When a [`WriteEntry::Finish`] is received, the thread will flush the data to disk and return
     /// a [`TempSnapshotDataV003`] and a [`SnapshotStat`].
     ///
     /// [`TempSnapshotDataV003`] is a temporary snapshot data that will be renamed to the final path by the caller.

src/meta/raft-store/src/snapshot_config.rs

@@ -84,7 +84,7 @@ impl SnapshotConfig {
         (self.snapshot_dir(), temp_snapshot_id)
     }
 
-    // TODO: remvoe this
+    // TODO: remove this
     pub fn snapshot_temp_path(&self) -> String {
         let temp_snapshot_id = self.temp_snapshot_id();
         format!("{}/{}", self.snapshot_dir(), temp_snapshot_id)

src/meta/raft-store/src/state/raft_state_kv.rs

@@ -31,6 +31,7 @@ pub enum RaftStateKey {
     /// The node id.
     Id,
 
+    // typos:ignore-next-line
     /// Hard state of the raft log, including `current_term` and `voted_vor`.
     HardState,
 

src/meta/raft-store/src/utils.rs

@@ -118,7 +118,7 @@ mod tests {
     #[test]
     fn test_prefix_right_bound_basic() {
         // Basic functionality test
-        assert_eq!(prefix_right_bound("abc"), Some(s("abd")));
+        assert_eq!(prefix_right_bound("abc"), Some(s("abd"))); // typos:disable-line
     }
 
     #[test]

src/meta/semaphore/src/errors/mod.rs

@@ -17,11 +17,11 @@ mod acquirer_closed;
 mod connection_closed;
 mod early_removed;
 mod either;
-mod processer_error;
+mod processor_error;
 
 pub use acquire_error::AcquireError;
 pub use acquirer_closed::AcquirerClosed;
 pub use connection_closed::ConnectionClosed;
 pub use early_removed::EarlyRemoved;
 pub use either::Either;
-pub use processer_error::ProcessorError;
+pub use processor_error::ProcessorError;

src/meta/service/src/meta_node/meta_node.rs

@@ -183,7 +183,7 @@ impl MetaNode {
             max_in_snapshot_log_to_keep: config.max_applied_log_to_keep,
             snapshot_max_chunk_size: config.snapshot_chunk_size,
             // Allow Leader to reset replication if a follower clears its log.
-            // Usefull in a testing environment.
+            // Useful in a testing environment.
             allow_log_reversion: Some(true),
             ..Default::default()
         }

src/meta/types/src/sys_data.rs

@@ -216,7 +216,7 @@ mod tests {
         assert_eq!(want, serialized);
     }
 
-    /// Test serialze will all fields set.
+    /// Test serialize will all fields set.
     #[test]
     fn test_sys_data_serialize() {
         let sys_data = SysData {

src/query/ast/src/parser/error_suggestion.rs

@@ -242,11 +242,11 @@ mod tests {
     fn test_typo_corrections() {
         // Typo corrections - may return multiple suggestions if scores are close
         assert_eq!(
-            suggest_correction("show tabl"),
+            suggest_correction("show tabl"), // typos:disable-line
             Some("Did you mean `SHOW TABLE FUNCTIONS` or `SHOW TABLES`?".to_string())
         );
         assert_eq!(
-            suggest_correction("vacum drop table"),
+            suggest_correction("vacum drop table"), // typos:disable-line
             Some("Did you mean `VACUUM DROP TABLE`?".to_string())
         );
         assert_eq!(
@@ -316,7 +316,7 @@ mod tests {
     fn test_edit_distance() {
         assert_eq!(edit_distance("show", "show"), 0);
         assert_eq!(edit_distance("tempare", "temporary"), 3);
-        assert_eq!(edit_distance("tabl", "tables"), 2);
+        assert_eq!(edit_distance("tabl", "tables"), 2); // typos:disable-line
     }
 
     #[test]

src/query/ast/src/parser/statement.rs

@@ -3298,7 +3298,7 @@ pub fn column_def(i: Input) -> IResult<ColumnDefinition> {
         value(ColumnConstraint::Nullable(true), rule! { NULL }),
         value(ColumnConstraint::Nullable(false), rule! { NOT ~ ^NULL }),
     ));
-    let identity_parmas = alt((
+    let identity_params = alt((
         map(
             rule! {
                 "(" ~ ^#literal_u64 ~ ^"," ~ ^#literal_i64 ~ ^")"
@@ -3341,7 +3341,7 @@ pub fn column_def(i: Input) -> IResult<ColumnDefinition> {
         map(
             rule! {
                 (AUTOINCREMENT | IDENTITY)
-                ~ #identity_parmas?
+                ~ #identity_params?
                 ~ (ORDER | NOORDER)?
             },
             |(_, params, order_token)| {

src/query/ast/tests/it/parser.rs

@@ -302,7 +302,7 @@ SELECT * from s;"#,
         r#"ALTER TABLE t CLUSTER BY(c1);"#,
         r#"ALTER TABLE t1 swap with t2;"#,
         r#"ALTER TABLE t refresh cache;"#,
-        r#"ALTER TABLE t COMMENT='t1-commnet';"#,
+        r#"ALTER TABLE t COMMENT='t1-commnet';"#, // typos:disable-line
         r#"ALTER TABLE t DROP CLUSTER KEY;"#,
         r#"ALTER TABLE t RECLUSTER FINAL WHERE c1 > 0 LIMIT 10;"#,
         r#"ALTER TABLE t ADD c int null;"#,
@@ -1049,7 +1049,7 @@ fn test_statement_error() {
         r#"GRANT select ON UDF a TO 'test-grant';"#,
         r#"REVOKE SELECT, CREATE, ALL PRIVILEGES ON * FROM 'test-grant';"#,
         r#"REVOKE SELECT, CREATE ON * TO 'test-grant';"#,
-        r#"COPY INTO mytable FROM 's3://bucket' CONECTION= ();"#,
+        r#"COPY INTO mytable FROM 's3://bucket' CONECTION= ();"#, // typos:disable-line
         r#"COPY INTO mytable FROM @mystage CONNECTION = ();"#,
         r#"CALL system$test"#,
         r#"CALL system$test(a"#,
@@ -1097,6 +1097,7 @@ fn test_statement_error() {
         r#"GRANT OWNERSHIP ON *.* TO ROLE 'd20_0015_owner';"#,
         r#"CREATE FUNCTION IF NOT EXISTS isnotempty AS(p) -> not(is_null(p)"#,
         r#"CREATE FUNCTION my_agg (INT) STATE { s STRING } RETURNS BOOLEAN LANGUAGE javascript HANDLER = 'my_agg' ADDRESS = 'http://0.0.0.0:8815';"#,
+        // typos:ignore-next-line
         r#"CREATE FUNCTION my_agg (INT) STATE { s STRIN } RETURNS BOOLEAN LANGUAGE javascript ADDRESS = 'http://0.0.0.0:8815';"#,
         r#"drop table :a"#,
         r#"drop table IDENTIFIER(a)"#,

src/query/config/src/mask.rs

@@ -338,7 +338,7 @@ mod tests {
 
         let masked_config = config.mask_display();
         assert_eq!(masked_config.oss_access_key_id, "*************y_id");
-        assert_eq!(masked_config.oss_access_key_secret, "****************ecret");
+        assert_eq!(masked_config.oss_access_key_secret, "****************ecret"); // typos:disable-line
         assert_eq!(
             masked_config.oss_server_side_encryption_key_id,
             "**************************_key_id"

src/query/ee/tests/it/aggregating_index/index_scan.rs

@@ -702,13 +702,13 @@ async fn fuzz(ctx: Arc<QueryContext>, params: FuzzParams) -> Result<()> {
     let actual = DataBlock::concat(&actual)?;
     let actual = DataBlock::sort(&actual, &get_sort_col_descs(actual.num_columns()), None)?;
 
-    let formated_expect = pretty_format_blocks(&[expect])?;
-    let formated_actual = pretty_format_blocks(&[actual])?;
+    let formatted_expect = pretty_format_blocks(&[expect])?;
+    let formatted_actual = pretty_format_blocks(&[actual])?;
 
     assert_eq!(
-        formated_expect, formated_actual,
+        formatted_expect, formatted_actual,
         "Test params:{}\nExpected data block:\n{}\nActual data block:\n{}\n",
-        fuzz_info, formated_expect, formated_actual
+        fuzz_info, formatted_expect, formatted_actual
     );
 
     // Clear

src/query/ee/tests/it/inverted_index/pruning.rs

@@ -336,7 +336,7 @@ async fn test_block_pruner() -> Result<()> {
             "Spill the beans".to_string(),
         ]),
         StringType::from_data(vec![
-            "Put a projet on hold".to_string(),
+            "Put a projet on hold".to_string(), // typos:disable-line
             "To spoil something".to_string(),
             "Saving money for later".to_string(),
             "Reliability is more important than speed".to_string(),

src/query/ee/tests/it/storages/fuse/operations/vacuum.rs

@@ -729,7 +729,7 @@ async fn test_vacuum_dropped_table_clean_ownership() -> Result<()> {
     let v = meta.get_pb(&table_ownership_key).await?;
     assert!(v.is_none());
 
-    // 8. Chek that DbIdTableName mapping is cleaned up
+    // 8. Check that DbIdTableName mapping is cleaned up
     let (seq, _) = get_u64_value(&meta, &db_id_table_name).await?;
     assert_eq!(seq, 0);
 

src/query/expression/tests/it/meta_scalar.rs

@@ -94,7 +94,7 @@ pub fn test_simple_converts() -> databend_common_exception::Result<()> {
                 assert_eq!(scalar, convert_back_scalar);
             }
 
-            // TODO: comment these when we swith string in scalar
+            // TODO: comment these when we switch string in scalar
             // let data = rmp_serde::to_vec(&scalars).unwrap();
             // let new_scalars: Vec<IndexScalar> = rmp_serde::from_slice(&data).unwrap();
             // assert_eq!(simple_scalars, new_scalars);