risingwave_connector/sink/
starrocks.rs

1// Copyright 2023 RisingWave Labs
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7//     http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15use std::collections::{BTreeMap, HashMap};
16use std::num::NonZeroU64;
17use std::sync::Arc;
18
19use anyhow::anyhow;
20use async_trait::async_trait;
21use bytes::Bytes;
22use mysql_async::Opts;
23use mysql_async::prelude::Queryable;
24use risingwave_common::array::{Op, StreamChunk};
25use risingwave_common::catalog::Schema;
26use risingwave_common::types::DataType;
27use serde::{Deserialize, Serialize};
28use serde_json::Value;
29use serde_with::{DisplayFromStr, serde_as};
30use thiserror_ext::AsReport;
31use url::form_urlencoded;
32use uuid::Uuid;
33use with_options::WithOptions;
34
35use super::decouple_checkpoint_log_sink::DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITH_SINK_DECOUPLE;
36use super::doris_starrocks_connector::{
37    HeaderBuilder, InserterInner, STARROCKS_DELETE_SIGN, STARROCKS_SUCCESS_STATUS,
38    StarrocksTxnRequestBuilder,
39};
40use super::encoder::{JsonEncoder, RowEncoder};
41use super::{
42    SINK_TYPE_APPEND_ONLY, SINK_TYPE_OPTION, SINK_TYPE_UPSERT, SinkError, SinkParam,
43    SinkWriterMetrics,
44};
45use crate::enforce_secret::EnforceSecret;
46use crate::sink::decouple_checkpoint_log_sink::DecoupleCheckpointLogSinkerOf;
47use crate::sink::writer::SinkWriter;
48use crate::sink::{Result, Sink, SinkWriterParam};
49
50pub const STARROCKS_SINK: &str = "starrocks";
51const STARROCK_MYSQL_PREFER_SOCKET: &str = "false";
52const STARROCK_MYSQL_MAX_ALLOWED_PACKET: usize = 1024;
53const STARROCK_MYSQL_WAIT_TIMEOUT: usize = 28800;
54
55pub const fn _default_stream_load_http_timeout_ms() -> u64 {
56    30 * 1000
57}
58
59const fn default_use_https() -> bool {
60    false
61}
62
63#[serde_as]
64#[derive(Deserialize, Debug, Clone, WithOptions)]
65pub struct StarrocksCommon {
66    /// The `StarRocks` host address.
67    #[serde(rename = "starrocks.host")]
68    pub host: String,
69    /// The port to the MySQL server of `StarRocks` FE.
70    #[serde(rename = "starrocks.mysqlport", alias = "starrocks.query_port")]
71    pub mysql_port: String,
72    /// The port to the HTTP server of `StarRocks` FE.
73    #[serde(rename = "starrocks.httpport", alias = "starrocks.http_port")]
74    pub http_port: String,
75    /// The user name used to access the `StarRocks` database.
76    #[serde(rename = "starrocks.user")]
77    pub user: String,
78    /// The password associated with the user.
79    #[serde(rename = "starrocks.password")]
80    pub password: String,
81    /// The `StarRocks` database where the target table is located
82    #[serde(rename = "starrocks.database")]
83    pub database: String,
84    /// The `StarRocks` table you want to sink data to.
85    #[serde(rename = "starrocks.table")]
86    pub table: String,
87
88    /// Whether to use https to connect to the `StarRocks` server.
89    #[serde(rename = "starrocks.use_https")]
90    #[serde(default = "default_use_https")]
91    #[serde_as(as = "DisplayFromStr")]
92    pub use_https: bool,
93}
94
95impl EnforceSecret for StarrocksCommon {
96    const ENFORCE_SECRET_PROPERTIES: phf::Set<&'static str> = phf::phf_set! {
97        "starrocks.password", "starrocks.user"
98    };
99}
100
101#[serde_as]
102#[derive(Clone, Debug, Deserialize, WithOptions)]
103pub struct StarrocksConfig {
104    #[serde(flatten)]
105    pub common: StarrocksCommon,
106
107    /// The timeout in milliseconds for stream load http request, defaults to 10 seconds.
108    #[serde(
109        rename = "starrocks.stream_load.http.timeout.ms",
110        default = "_default_stream_load_http_timeout_ms"
111    )]
112    #[serde_as(as = "DisplayFromStr")]
113    #[with_option(allow_alter_on_fly)]
114    pub stream_load_http_timeout_ms: u64,
115
116    /// Set this option to a positive integer n, RisingWave will try to commit data
117    /// to Starrocks at every n checkpoints by leveraging the
118    /// [StreamLoad Transaction API](https://docs.starrocks.io/docs/loading/Stream_Load_transaction_interface/),
119    /// also, in this time, the `sink_decouple` option should be enabled as well.
120    /// Defaults to 10 if `commit_checkpoint_interval` <= 0
121    #[serde(default = "default_commit_checkpoint_interval")]
122    #[serde_as(as = "DisplayFromStr")]
123    #[with_option(allow_alter_on_fly)]
124    pub commit_checkpoint_interval: u64,
125
126    /// Enable partial update
127    #[serde(rename = "starrocks.partial_update")]
128    pub partial_update: Option<String>,
129
130    pub r#type: String, // accept "append-only" or "upsert"
131}
132
133impl EnforceSecret for StarrocksConfig {
134    fn enforce_one(prop: &str) -> crate::error::ConnectorResult<()> {
135        StarrocksCommon::enforce_one(prop)
136    }
137}
138
139fn default_commit_checkpoint_interval() -> u64 {
140    DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITH_SINK_DECOUPLE
141}
142
143impl StarrocksConfig {
144    pub fn from_btreemap(properties: BTreeMap<String, String>) -> Result<Self> {
145        let config =
146            serde_json::from_value::<StarrocksConfig>(serde_json::to_value(properties).unwrap())
147                .map_err(|e| SinkError::Config(anyhow!(e)))?;
148        if config.r#type != SINK_TYPE_APPEND_ONLY && config.r#type != SINK_TYPE_UPSERT {
149            return Err(SinkError::Config(anyhow!(
150                "`{}` must be {}, or {}",
151                SINK_TYPE_OPTION,
152                SINK_TYPE_APPEND_ONLY,
153                SINK_TYPE_UPSERT
154            )));
155        }
156        if config.commit_checkpoint_interval == 0 {
157            return Err(SinkError::Config(anyhow!(
158                "`commit_checkpoint_interval` must be greater than 0"
159            )));
160        }
161        Ok(config)
162    }
163}
164
165#[derive(Debug)]
166pub struct StarrocksSink {
167    pub config: StarrocksConfig,
168    schema: Schema,
169    pk_indices: Vec<usize>,
170    is_append_only: bool,
171}
172
173impl EnforceSecret for StarrocksSink {
174    fn enforce_secret<'a>(
175        prop_iter: impl Iterator<Item = &'a str>,
176    ) -> crate::error::ConnectorResult<()> {
177        for prop in prop_iter {
178            StarrocksConfig::enforce_one(prop)?;
179        }
180        Ok(())
181    }
182}
183
184impl StarrocksSink {
185    pub fn new(param: SinkParam, config: StarrocksConfig, schema: Schema) -> Result<Self> {
186        let pk_indices = param.downstream_pk_or_empty();
187        let is_append_only = param.sink_type.is_append_only();
188        Ok(Self {
189            config,
190            schema,
191            pk_indices,
192            is_append_only,
193        })
194    }
195}
196
197impl StarrocksSink {
198    fn check_column_name_and_type(
199        &self,
200        starrocks_columns_desc: HashMap<String, String>,
201    ) -> Result<()> {
202        let rw_fields_name = self.schema.fields();
203        if rw_fields_name.len() > starrocks_columns_desc.len() {
204            return Err(SinkError::Starrocks("The columns of the sink must be equal to or a superset of the target table's columns.".to_owned()));
205        }
206
207        for i in rw_fields_name {
208            let value = starrocks_columns_desc.get(&i.name).ok_or_else(|| {
209                SinkError::Starrocks(format!(
210                    "Column name don't find in starrocks, risingwave is {:?} ",
211                    i.name
212                ))
213            })?;
214            if !Self::check_and_correct_column_type(&i.data_type, value)? {
215                return Err(SinkError::Starrocks(format!(
216                    "Column type don't match, column name is {:?}. starrocks type is {:?} risingwave type is {:?} ",
217                    i.name, value, i.data_type
218                )));
219            }
220        }
221        Ok(())
222    }
223
224    fn check_and_correct_column_type(
225        rw_data_type: &DataType,
226        starrocks_data_type: &String,
227    ) -> Result<bool> {
228        match rw_data_type {
229            risingwave_common::types::DataType::Boolean => {
230                Ok(starrocks_data_type.contains("tinyint") | starrocks_data_type.contains("boolean"))
231            }
232            risingwave_common::types::DataType::Int16 => {
233                Ok(starrocks_data_type.contains("smallint"))
234            }
235            risingwave_common::types::DataType::Int32 => Ok(starrocks_data_type.contains("int")),
236            risingwave_common::types::DataType::Int64 => Ok(starrocks_data_type.contains("bigint")),
237            risingwave_common::types::DataType::Float32 => {
238                Ok(starrocks_data_type.contains("float"))
239            }
240            risingwave_common::types::DataType::Float64 => {
241                Ok(starrocks_data_type.contains("double"))
242            }
243            risingwave_common::types::DataType::Decimal => {
244                Ok(starrocks_data_type.contains("decimal"))
245            }
246            risingwave_common::types::DataType::Date => Ok(starrocks_data_type.contains("date")),
247            risingwave_common::types::DataType::Varchar => {
248                Ok(starrocks_data_type.contains("varchar"))
249            }
250            risingwave_common::types::DataType::Time => Err(SinkError::Starrocks(
251                "TIME is not supported for Starrocks sink. Please convert to VARCHAR or other supported types.".to_owned(),
252            )),
253            risingwave_common::types::DataType::Timestamp => {
254                Ok(starrocks_data_type.contains("datetime"))
255            }
256            risingwave_common::types::DataType::Timestamptz => Err(SinkError::Starrocks(
257                "TIMESTAMP WITH TIMEZONE is not supported for Starrocks sink as Starrocks doesn't store time values with timezone information. Please convert to TIMESTAMP first.".to_owned(),
258            )),
259            risingwave_common::types::DataType::Interval => Err(SinkError::Starrocks(
260                "INTERVAL is not supported for Starrocks sink. Please convert to VARCHAR or other supported types.".to_owned(),
261            )),
262            risingwave_common::types::DataType::Struct(_) => Err(SinkError::Starrocks(
263                "STRUCT is not supported for Starrocks sink.".to_owned(),
264            )),
265            risingwave_common::types::DataType::List(list) => {
266                // For compatibility with older versions starrocks
267                if starrocks_data_type.contains("unknown") {
268                    return Ok(true);
269                }
270                let check_result = Self::check_and_correct_column_type(list.elem(), starrocks_data_type)?;
271                Ok(check_result && starrocks_data_type.contains("array"))
272            }
273            risingwave_common::types::DataType::Bytea => Err(SinkError::Starrocks(
274                "BYTEA is not supported for Starrocks sink. Please convert to VARCHAR or other supported types.".to_owned(),
275            )),
276            risingwave_common::types::DataType::Jsonb => Ok(starrocks_data_type.contains("json")),
277            risingwave_common::types::DataType::Serial => {
278                Ok(starrocks_data_type.contains("bigint"))
279            }
280            risingwave_common::types::DataType::Int256 => Err(SinkError::Starrocks(
281                "INT256 is not supported for Starrocks sink.".to_owned(),
282            )),
283            risingwave_common::types::DataType::Map(_) => Err(SinkError::Starrocks(
284                "MAP is not supported for Starrocks sink.".to_owned(),
285            )),
286            DataType::Vector(_) => Err(SinkError::Starrocks(
287                "VECTOR is not supported for Starrocks sink.".to_owned(),
288            )),
289        }
290    }
291}
292
293impl Sink for StarrocksSink {
294    type LogSinker = DecoupleCheckpointLogSinkerOf<StarrocksSinkWriter>;
295
296    const SINK_NAME: &'static str = STARROCKS_SINK;
297
298    async fn validate(&self) -> Result<()> {
299        if !self.is_append_only && self.pk_indices.is_empty() {
300            return Err(SinkError::Config(anyhow!(
301                "Primary key not defined for upsert starrocks sink (please define in `primary_key` field)"
302            )));
303        }
304        // check reachability
305        let mut client = StarrocksSchemaClient::new(
306            self.config.common.host.clone(),
307            self.config.common.mysql_port.clone(),
308            self.config.common.table.clone(),
309            self.config.common.database.clone(),
310            self.config.common.user.clone(),
311            self.config.common.password.clone(),
312        )
313        .await?;
314        let (read_model, pks) = client.get_pk_from_starrocks().await?;
315
316        if !self.is_append_only && read_model.ne("PRIMARY_KEYS") {
317            return Err(SinkError::Config(anyhow!(
318                "If you want to use upsert, please set the keysType of starrocks to PRIMARY_KEY"
319            )));
320        }
321
322        for (index, filed) in self.schema.fields().iter().enumerate() {
323            if self.pk_indices.contains(&index) && !pks.contains(&filed.name) {
324                return Err(SinkError::Starrocks(format!(
325                    "Can't find pk {:?} in starrocks",
326                    filed.name
327                )));
328            }
329        }
330
331        let starrocks_columns_desc = client.get_columns_from_starrocks().await?;
332
333        self.check_column_name_and_type(starrocks_columns_desc)?;
334        Ok(())
335    }
336
337    fn validate_alter_config(config: &BTreeMap<String, String>) -> Result<()> {
338        StarrocksConfig::from_btreemap(config.clone())?;
339        Ok(())
340    }
341
342    async fn new_log_sinker(&self, writer_param: SinkWriterParam) -> Result<Self::LogSinker> {
343        let commit_checkpoint_interval =
344            NonZeroU64::new(self.config.commit_checkpoint_interval).expect(
345                "commit_checkpoint_interval should be greater than 0, and it should be checked in config validation",
346            );
347
348        let writer = StarrocksSinkWriter::new(
349            self.config.clone(),
350            self.schema.clone(),
351            self.pk_indices.clone(),
352            self.is_append_only,
353        )?;
354
355        let metrics = SinkWriterMetrics::new(&writer_param);
356
357        Ok(DecoupleCheckpointLogSinkerOf::new(
358            writer,
359            metrics,
360            commit_checkpoint_interval,
361        ))
362    }
363}
364
365pub struct StarrocksSinkWriter {
366    pub config: StarrocksConfig,
367    #[expect(dead_code)]
368    schema: Schema,
369    #[expect(dead_code)]
370    pk_indices: Vec<usize>,
371    is_append_only: bool,
372    client: Option<StarrocksClient>,
373    txn_client: Arc<StarrocksTxnClient>,
374    row_encoder: JsonEncoder,
375    curr_txn_label: Option<String>,
376}
377
378impl TryFrom<SinkParam> for StarrocksSink {
379    type Error = SinkError;
380
381    fn try_from(param: SinkParam) -> std::result::Result<Self, Self::Error> {
382        let schema = param.schema();
383        let config = StarrocksConfig::from_btreemap(param.properties.clone())?;
384        StarrocksSink::new(param, config, schema)
385    }
386}
387
388impl StarrocksSinkWriter {
389    pub fn new(
390        config: StarrocksConfig,
391        schema: Schema,
392        pk_indices: Vec<usize>,
393        is_append_only: bool,
394    ) -> Result<Self> {
395        let mut field_names = schema.names_str();
396        if !is_append_only {
397            field_names.push(STARROCKS_DELETE_SIGN);
398        };
399        // we should quote field names in `MySQL` style to prevent `StarRocks` from rejecting the request due to
400        // a field name being a reserved word. For example, `order`, 'from`, etc.
401        let field_names = field_names
402            .into_iter()
403            .map(|name| format!("`{}`", name))
404            .collect::<Vec<String>>();
405        let field_names_str = field_names
406            .iter()
407            .map(|name| name.as_str())
408            .collect::<Vec<&str>>();
409
410        let header = HeaderBuilder::new()
411            .add_common_header()
412            .set_user_password(config.common.user.clone(), config.common.password.clone())
413            .add_json_format()
414            .set_partial_update(config.partial_update.clone())
415            .set_columns_name(field_names_str)
416            .set_db(config.common.database.clone())
417            .set_table(config.common.table.clone())
418            .build();
419
420        let url = if config.common.use_https {
421            format!("https://{}:{}", config.common.host, config.common.http_port)
422        } else {
423            format!("http://{}:{}", config.common.host, config.common.http_port)
424        };
425        let txn_request_builder =
426            StarrocksTxnRequestBuilder::new(url, header, config.stream_load_http_timeout_ms)?;
427
428        Ok(Self {
429            config,
430            schema: schema.clone(),
431            pk_indices,
432            is_append_only,
433            client: None,
434            txn_client: Arc::new(StarrocksTxnClient::new(txn_request_builder)),
435            row_encoder: JsonEncoder::new_with_starrocks(schema, None),
436            curr_txn_label: None,
437        })
438    }
439
440    async fn append_only(&mut self, chunk: StreamChunk) -> Result<()> {
441        for (op, row) in chunk.rows() {
442            if op != Op::Insert {
443                continue;
444            }
445            let row_json_string = Value::Object(self.row_encoder.encode(row)?).to_string();
446            self.client
447                .as_mut()
448                .ok_or_else(|| SinkError::Starrocks("Can't find starrocks sink insert".to_owned()))?
449                .write(row_json_string.into())
450                .await?;
451        }
452        Ok(())
453    }
454
455    async fn upsert(&mut self, chunk: StreamChunk) -> Result<()> {
456        for (op, row) in chunk.rows() {
457            match op {
458                Op::Insert => {
459                    let mut row_json_value = self.row_encoder.encode(row)?;
460                    row_json_value.insert(
461                        STARROCKS_DELETE_SIGN.to_owned(),
462                        Value::String("0".to_owned()),
463                    );
464                    let row_json_string = serde_json::to_string(&row_json_value).map_err(|e| {
465                        SinkError::Starrocks(format!("Json derialize error: {}", e.as_report()))
466                    })?;
467                    self.client
468                        .as_mut()
469                        .ok_or_else(|| {
470                            SinkError::Starrocks("Can't find starrocks sink insert".to_owned())
471                        })?
472                        .write(row_json_string.into())
473                        .await?;
474                }
475                Op::Delete => {
476                    let mut row_json_value = self.row_encoder.encode(row)?;
477                    row_json_value.insert(
478                        STARROCKS_DELETE_SIGN.to_owned(),
479                        Value::String("1".to_owned()),
480                    );
481                    let row_json_string = serde_json::to_string(&row_json_value).map_err(|e| {
482                        SinkError::Starrocks(format!("Json derialize error: {}", e.as_report()))
483                    })?;
484                    self.client
485                        .as_mut()
486                        .ok_or_else(|| {
487                            SinkError::Starrocks("Can't find starrocks sink insert".to_owned())
488                        })?
489                        .write(row_json_string.into())
490                        .await?;
491                }
492                Op::UpdateDelete => {}
493                Op::UpdateInsert => {
494                    let mut row_json_value = self.row_encoder.encode(row)?;
495                    row_json_value.insert(
496                        STARROCKS_DELETE_SIGN.to_owned(),
497                        Value::String("0".to_owned()),
498                    );
499                    let row_json_string = serde_json::to_string(&row_json_value).map_err(|e| {
500                        SinkError::Starrocks(format!("Json derialize error: {}", e.as_report()))
501                    })?;
502                    self.client
503                        .as_mut()
504                        .ok_or_else(|| {
505                            SinkError::Starrocks("Can't find starrocks sink insert".to_owned())
506                        })?
507                        .write(row_json_string.into())
508                        .await?;
509                }
510            }
511        }
512        Ok(())
513    }
514
515    /// Generating a new transaction label, should be unique across all `SinkWriters` even under rewinding.
516    #[inline(always)]
517    fn new_txn_label(&self) -> String {
518        format!(
519            "rw-txn-{}-{}",
520            Uuid::new_v4(),
521            chrono::Utc::now().timestamp_micros()
522        )
523    }
524
525    async fn prepare_and_commit(&self, txn_label: String) -> Result<()> {
526        tracing::debug!(?txn_label, "prepare transaction");
527        let txn_label_res = self.txn_client.prepare(txn_label.clone()).await?;
528        if txn_label != txn_label_res {
529            return Err(SinkError::Starrocks(format!(
530                "label {} returned from prepare transaction {} differs from the current one",
531                txn_label, txn_label_res
532            )));
533        }
534        tracing::debug!(?txn_label, "commit transaction");
535        let txn_label_res = self.txn_client.commit(txn_label.clone()).await?;
536        if txn_label != txn_label_res {
537            return Err(SinkError::Starrocks(format!(
538                "label {} returned from commit transaction {} differs from the current one",
539                txn_label, txn_label_res
540            )));
541        }
542        Ok(())
543    }
544}
545
546impl Drop for StarrocksSinkWriter {
547    fn drop(&mut self) {
548        if let Some(txn_label) = self.curr_txn_label.take() {
549            let txn_client = self.txn_client.clone();
550            tokio::spawn(async move {
551                if let Err(e) = txn_client.rollback(txn_label.clone()).await {
552                    tracing::error!(
553                        "starrocks rollback transaction error: {:?}, txn label: {}",
554                        e.as_report(),
555                        txn_label
556                    );
557                }
558            });
559        }
560    }
561}
562
563#[async_trait]
564impl SinkWriter for StarrocksSinkWriter {
565    async fn begin_epoch(&mut self, _epoch: u64) -> Result<()> {
566        Ok(())
567    }
568
569    async fn write_batch(&mut self, chunk: StreamChunk) -> Result<()> {
570        // We check whether start a new transaction in `write_batch`. Therefore, if no data has been written
571        // within the `commit_checkpoint_interval` period, no meta requests will be made. Otherwise if we request
572        // `prepare` against an empty transaction, the `StarRocks` will report a `hasn't send any data yet` error.
573        if self.curr_txn_label.is_none() {
574            let txn_label = self.new_txn_label();
575            tracing::debug!(?txn_label, "begin transaction");
576            let txn_label_res = self.txn_client.begin(txn_label.clone()).await?;
577            if txn_label != txn_label_res {
578                return Err(SinkError::Starrocks(format!(
579                    "label {} returned from StarRocks {} differs from generated one",
580                    txn_label, txn_label_res
581                )));
582            }
583            self.curr_txn_label = Some(txn_label.clone());
584        }
585        if self.client.is_none() {
586            let txn_label = self.curr_txn_label.clone();
587            self.client = Some(StarrocksClient::new(
588                self.txn_client.load(txn_label.unwrap()).await?,
589            ));
590        }
591        if self.is_append_only {
592            self.append_only(chunk).await
593        } else {
594            self.upsert(chunk).await
595        }
596    }
597
598    async fn barrier(&mut self, is_checkpoint: bool) -> Result<()> {
599        if let Some(client) = self.client.take() {
600            // Here we finish the `/api/transaction/load` request when a barrier is received. Therefore,
601            // one or more load requests should be made within one commit_checkpoint_interval period.
602            // StarRocks will take care of merging those splits into a larger one during prepare transaction.
603            // Thus, only one version will be produced when the transaction is committed. See Stream Load
604            // transaction interface for more information.
605            client.finish().await?;
606        }
607
608        if is_checkpoint
609            && let Some(txn_label) = self.curr_txn_label.take()
610            && let Err(err) = self.prepare_and_commit(txn_label.clone()).await
611        {
612            match self.txn_client.rollback(txn_label.clone()).await {
613                Ok(_) => tracing::warn!(
614                    ?txn_label,
615                    "transaction is successfully rolled back due to commit failure"
616                ),
617                Err(err) => {
618                    tracing::warn!(?txn_label, error = ?err.as_report(), "Couldn't roll back transaction after commit failed")
619                }
620            }
621
622            return Err(err);
623        }
624        Ok(())
625    }
626
627    async fn abort(&mut self) -> Result<()> {
628        if let Some(txn_label) = self.curr_txn_label.take() {
629            tracing::debug!(?txn_label, "rollback transaction");
630            self.txn_client.rollback(txn_label).await?;
631        }
632        Ok(())
633    }
634}
635
636pub struct StarrocksSchemaClient {
637    table: String,
638    db: String,
639    conn: mysql_async::Conn,
640}
641
642impl StarrocksSchemaClient {
643    pub async fn new(
644        host: String,
645        port: String,
646        table: String,
647        db: String,
648        user: String,
649        password: String,
650    ) -> Result<Self> {
651        // username & password may contain special chars, so we need to do URL encoding on them.
652        // Otherwise, Opts::from_url may report a `Parse error`
653        let user = form_urlencoded::byte_serialize(user.as_bytes()).collect::<String>();
654        let password = form_urlencoded::byte_serialize(password.as_bytes()).collect::<String>();
655
656        let conn_uri = format!(
657            "mysql://{}:{}@{}:{}/{}?prefer_socket={}&max_allowed_packet={}&wait_timeout={}",
658            user,
659            password,
660            host,
661            port,
662            db,
663            STARROCK_MYSQL_PREFER_SOCKET,
664            STARROCK_MYSQL_MAX_ALLOWED_PACKET,
665            STARROCK_MYSQL_WAIT_TIMEOUT
666        );
667        let pool = mysql_async::Pool::new(
668            Opts::from_url(&conn_uri)
669                .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?,
670        );
671        let conn = pool
672            .get_conn()
673            .await
674            .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?;
675
676        Ok(Self { table, db, conn })
677    }
678
679    pub async fn get_columns_from_starrocks(&mut self) -> Result<HashMap<String, String>> {
680        let query = format!(
681            "select column_name, column_type from information_schema.columns where table_name = {:?} and table_schema = {:?};",
682            self.table, self.db
683        );
684        let mut query_map: HashMap<String, String> = HashMap::default();
685        self.conn
686            .query_map(query, |(column_name, column_type)| {
687                query_map.insert(column_name, column_type)
688            })
689            .await
690            .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?;
691        Ok(query_map)
692    }
693
694    pub async fn get_pk_from_starrocks(&mut self) -> Result<(String, String)> {
695        let query = format!(
696            "select table_model, primary_key, sort_key from information_schema.tables_config where table_name = {:?} and table_schema = {:?};",
697            self.table, self.db
698        );
699        let table_mode_pk: (String, String) = self
700            .conn
701            .query_map(
702                query,
703                |(table_model, primary_key, sort_key): (String, String, String)| match table_model
704                    .as_str()
705                {
706                    // Get primary key of aggregate table from the sort_key field
707                    // https://docs.starrocks.io/docs/table_design/table_types/table_capabilities/
708                    // https://docs.starrocks.io/docs/sql-reference/information_schema/tables_config/
709                    "AGG_KEYS" => (table_model, sort_key),
710                    _ => (table_model, primary_key),
711                },
712            )
713            .await
714            .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?
715            .first()
716            .ok_or_else(|| {
717                SinkError::Starrocks(format!(
718                    "Can't find schema with table {:?} and database {:?}",
719                    self.table, self.db
720                ))
721            })?
722            .clone();
723        Ok(table_mode_pk)
724    }
725}
726
727#[derive(Debug, Serialize, Deserialize)]
728pub struct StarrocksInsertResultResponse {
729    #[serde(rename = "TxnId")]
730    pub txn_id: Option<i64>,
731    #[serde(rename = "Seq")]
732    pub seq: Option<i64>,
733    #[serde(rename = "Label")]
734    pub label: Option<String>,
735    #[serde(rename = "Status")]
736    pub status: String,
737    #[serde(rename = "Message")]
738    pub message: String,
739    #[serde(rename = "NumberTotalRows")]
740    pub number_total_rows: Option<i64>,
741    #[serde(rename = "NumberLoadedRows")]
742    pub number_loaded_rows: Option<i64>,
743    #[serde(rename = "NumberFilteredRows")]
744    pub number_filtered_rows: Option<i32>,
745    #[serde(rename = "NumberUnselectedRows")]
746    pub number_unselected_rows: Option<i32>,
747    #[serde(rename = "LoadBytes")]
748    pub load_bytes: Option<i64>,
749    #[serde(rename = "LoadTimeMs")]
750    pub load_time_ms: Option<i32>,
751    #[serde(rename = "BeginTxnTimeMs")]
752    pub begin_txn_time_ms: Option<i32>,
753    #[serde(rename = "ReadDataTimeMs")]
754    pub read_data_time_ms: Option<i32>,
755    #[serde(rename = "WriteDataTimeMs")]
756    pub write_data_time_ms: Option<i32>,
757    #[serde(rename = "CommitAndPublishTimeMs")]
758    pub commit_and_publish_time_ms: Option<i32>,
759    #[serde(rename = "StreamLoadPlanTimeMs")]
760    pub stream_load_plan_time_ms: Option<i32>,
761    #[serde(rename = "ExistingJobStatus")]
762    pub existing_job_status: Option<String>,
763    #[serde(rename = "ErrorURL")]
764    pub error_url: Option<String>,
765}
766
767pub struct StarrocksClient {
768    insert: InserterInner,
769}
770impl StarrocksClient {
771    pub fn new(insert: InserterInner) -> Self {
772        Self { insert }
773    }
774
775    pub async fn write(&mut self, data: Bytes) -> Result<()> {
776        self.insert.write(data).await?;
777        Ok(())
778    }
779
780    pub async fn finish(self) -> Result<StarrocksInsertResultResponse> {
781        let raw = self.insert.finish().await?;
782        let res: StarrocksInsertResultResponse = serde_json::from_slice(&raw)
783            .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?;
784
785        if !STARROCKS_SUCCESS_STATUS.contains(&res.status.as_str()) {
786            return Err(SinkError::DorisStarrocksConnect(anyhow::anyhow!(
787                "Insert error: {}, {}, {:?}",
788                res.status,
789                res.message,
790                res.error_url,
791            )));
792        };
793        Ok(res)
794    }
795}
796
797pub struct StarrocksTxnClient {
798    request_builder: StarrocksTxnRequestBuilder,
799}
800
801impl StarrocksTxnClient {
802    pub fn new(request_builder: StarrocksTxnRequestBuilder) -> Self {
803        Self { request_builder }
804    }
805
806    fn check_response_and_extract_label(&self, res: Bytes) -> Result<String> {
807        let res: StarrocksInsertResultResponse = serde_json::from_slice(&res)
808            .map_err(|err| SinkError::DorisStarrocksConnect(anyhow!(err)))?;
809        if !STARROCKS_SUCCESS_STATUS.contains(&res.status.as_str()) {
810            return Err(SinkError::DorisStarrocksConnect(anyhow::anyhow!(
811                "transaction error: {}, {}, {:?}",
812                res.status,
813                res.message,
814                res.error_url,
815            )));
816        }
817        res.label.ok_or_else(|| {
818            SinkError::DorisStarrocksConnect(anyhow::anyhow!("Can't get label from response"))
819        })
820    }
821
822    pub async fn begin(&self, label: String) -> Result<String> {
823        let res = self
824            .request_builder
825            .build_begin_request_sender(label)?
826            .send()
827            .await?;
828        self.check_response_and_extract_label(res)
829    }
830
831    pub async fn prepare(&self, label: String) -> Result<String> {
832        let res = self
833            .request_builder
834            .build_prepare_request_sender(label)?
835            .send()
836            .await?;
837        self.check_response_and_extract_label(res)
838    }
839
840    pub async fn commit(&self, label: String) -> Result<String> {
841        let res = self
842            .request_builder
843            .build_commit_request_sender(label)?
844            .send()
845            .await?;
846        self.check_response_and_extract_label(res)
847    }
848
849    pub async fn rollback(&self, label: String) -> Result<String> {
850        let res = self
851            .request_builder
852            .build_rollback_request_sender(label)?
853            .send()
854            .await?;
855        self.check_response_and_extract_label(res)
856    }
857
858    pub async fn load(&self, label: String) -> Result<InserterInner> {
859        self.request_builder.build_txn_inserter(label).await
860    }
861}
risingwave_connector/sink/starrocks.rs

risingwave_connector/sink/
starrocks.rs
