risingwave_connector/source/cdc/external/
mysql.rs

1// Copyright 2025 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::HashMap;
16
17use anyhow::{Context, anyhow};
18use chrono::{DateTime, NaiveDateTime};
19use futures::stream::BoxStream;
20use futures::{StreamExt, pin_mut, stream};
21use futures_async_stream::try_stream;
22use itertools::Itertools;
23use mysql_async::prelude::*;
24use mysql_common::params::Params;
25use mysql_common::value::Value;
26use risingwave_common::bail;
27use risingwave_common::catalog::{CDC_OFFSET_COLUMN_NAME, ColumnDesc, ColumnId, Field, Schema};
28use risingwave_common::row::OwnedRow;
29use risingwave_common::types::{DataType, Datum, Decimal, F32, ScalarImpl};
30use risingwave_common::util::iter_util::ZipEqFast;
31use sea_schema::mysql::def::{ColumnDefault, ColumnKey, ColumnType};
32use sea_schema::mysql::discovery::SchemaDiscovery;
33use sea_schema::mysql::query::SchemaQueryBuilder;
34use sea_schema::sea_query::{Alias, IntoIden};
35use serde::{Deserialize, Serialize};
36use sqlx::MySqlPool;
37use sqlx::mysql::MySqlConnectOptions;
38use thiserror_ext::AsReport;
39
40use crate::error::{ConnectorError, ConnectorResult};
41use crate::source::CdcTableSnapshotSplit;
42use crate::source::cdc::external::{
43    CdcOffset, CdcOffsetParseFunc, CdcTableSnapshotSplitOption, DebeziumOffset,
44    ExternalTableConfig, ExternalTableReader, SchemaTableName, SslMode, mysql_row_to_owned_row,
45};
46
47#[derive(Debug, Clone, Default, PartialEq, PartialOrd, Serialize, Deserialize)]
48pub struct MySqlOffset {
49    pub filename: String,
50    pub position: u64,
51}
52
53impl MySqlOffset {
54    pub fn new(filename: String, position: u64) -> Self {
55        Self { filename, position }
56    }
57}
58
59impl MySqlOffset {
60    pub fn parse_debezium_offset(offset: &str) -> ConnectorResult<Self> {
61        let dbz_offset: DebeziumOffset = serde_json::from_str(offset)
62            .with_context(|| format!("invalid upstream offset: {}", offset))?;
63
64        Ok(Self {
65            filename: dbz_offset
66                .source_offset
67                .file
68                .context("binlog file not found in offset")?,
69            position: dbz_offset
70                .source_offset
71                .pos
72                .context("binlog position not found in offset")?,
73        })
74    }
75}
76
77pub struct MySqlExternalTable {
78    column_descs: Vec<ColumnDesc>,
79    pk_names: Vec<String>,
80}
81
82impl MySqlExternalTable {
83    pub async fn connect(config: ExternalTableConfig) -> ConnectorResult<Self> {
84        tracing::debug!("connect to mysql");
85        let options = MySqlConnectOptions::new()
86            .username(&config.username)
87            .password(&config.password)
88            .host(&config.host)
89            .port(config.port.parse::<u16>().unwrap())
90            .database(&config.database)
91            .ssl_mode(match config.ssl_mode {
92                SslMode::Disabled => sqlx::mysql::MySqlSslMode::Disabled,
93                SslMode::Preferred => sqlx::mysql::MySqlSslMode::Preferred,
94                SslMode::Required => sqlx::mysql::MySqlSslMode::Required,
95                _ => {
96                    return Err(anyhow!("unsupported SSL mode").into());
97                }
98            });
99
100        let connection = MySqlPool::connect_with(options).await?;
101        let mut schema_discovery = SchemaDiscovery::new(connection, config.database.as_str());
102
103        // discover system version first
104        let system_info = schema_discovery.discover_system().await?;
105        schema_discovery.query = SchemaQueryBuilder::new(system_info.clone());
106        let schema = Alias::new(config.database.as_str()).into_iden();
107        let table = Alias::new(config.table.as_str()).into_iden();
108        let columns = schema_discovery
109            .discover_columns(schema, table, &system_info)
110            .await?;
111        let mut column_descs = vec![];
112        let mut pk_names = vec![];
113        for col in columns {
114            let data_type = mysql_type_to_rw_type(&col.col_type)?;
115            // column name in mysql is case-insensitive, convert to lowercase
116            let col_name = col.name.to_lowercase();
117            let column_desc = if let Some(default) = col.default {
118                let snapshot_value = derive_default_value(default.clone(), &data_type)
119                    .unwrap_or_else(|e| {
120                        tracing::warn!(
121                            column = col_name,
122                            ?default,
123                            %data_type,
124                            error = %e.as_report(),
125                            "failed to derive column default value, fallback to `NULL`",
126                        );
127                        None
128                    });
129
130                ColumnDesc::named_with_default_value(
131                    col_name.clone(),
132                    ColumnId::placeholder(),
133                    data_type.clone(),
134                    snapshot_value,
135                )
136            } else {
137                ColumnDesc::named(col_name.clone(), ColumnId::placeholder(), data_type)
138            };
139
140            column_descs.push(column_desc);
141            if matches!(col.key, ColumnKey::Primary) {
142                pk_names.push(col_name);
143            }
144        }
145
146        if pk_names.is_empty() {
147            return Err(anyhow!("MySQL table doesn't define the primary key").into());
148        }
149        Ok(Self {
150            column_descs,
151            pk_names,
152        })
153    }
154
155    pub fn column_descs(&self) -> &Vec<ColumnDesc> {
156        &self.column_descs
157    }
158
159    pub fn pk_names(&self) -> &Vec<String> {
160        &self.pk_names
161    }
162}
163
164fn derive_default_value(default: ColumnDefault, data_type: &DataType) -> ConnectorResult<Datum> {
165    let datum = match default {
166        ColumnDefault::Null => None,
167        ColumnDefault::Int(val) => match data_type {
168            DataType::Int16 => Some(ScalarImpl::Int16(val as _)),
169            DataType::Int32 => Some(ScalarImpl::Int32(val as _)),
170            DataType::Int64 => Some(ScalarImpl::Int64(val)),
171            DataType::Varchar => {
172                // should be the Enum type which is mapped to Varchar
173                Some(ScalarImpl::from(val.to_string()))
174            }
175            _ => bail!("unexpected default value type for integer"),
176        },
177        ColumnDefault::Real(val) => match data_type {
178            DataType::Float32 => Some(ScalarImpl::Float32(F32::from(val as f32))),
179            DataType::Float64 => Some(ScalarImpl::Float64(val.into())),
180            DataType::Decimal => Some(ScalarImpl::Decimal(
181                Decimal::try_from(val).context("failed to convert default value to decimal")?,
182            )),
183            _ => bail!("unexpected default value type for real"),
184        },
185        ColumnDefault::String(mut val) => {
186            // mysql timestamp is mapped to timestamptz, we use UTC timezone to
187            // interpret its value
188            if data_type == &DataType::Timestamptz {
189                val = timestamp_val_to_timestamptz(val.as_str())?;
190            }
191            Some(ScalarImpl::from_text(val.as_str(), data_type).map_err(|e| anyhow!(e)).context(
192                "failed to parse mysql default value expression, only constant is supported",
193            )?)
194        }
195        ColumnDefault::CurrentTimestamp | ColumnDefault::CustomExpr(_) => {
196            bail!("MySQL CURRENT_TIMESTAMP and custom expression default value not supported")
197        }
198    };
199    Ok(datum)
200}
201
202pub fn timestamp_val_to_timestamptz(value_text: &str) -> ConnectorResult<String> {
203    let format = "%Y-%m-%d %H:%M:%S";
204    let naive_datetime = NaiveDateTime::parse_from_str(value_text, format)
205        .map_err(|err| anyhow!("failed to parse mysql timestamp value").context(err))?;
206    let postgres_timestamptz: DateTime<chrono::Utc> =
207        DateTime::<chrono::Utc>::from_naive_utc_and_offset(naive_datetime, chrono::Utc);
208    Ok(postgres_timestamptz
209        .format("%Y-%m-%d %H:%M:%S%:z")
210        .to_string())
211}
212
213pub fn type_name_to_mysql_type(ty_name: &str) -> Option<ColumnType> {
214    macro_rules! column_type {
215        ($($name:literal => $variant:ident),* $(,)?) => {
216            match ty_name.to_lowercase().as_str() {
217                $(
218                    $name => Some(ColumnType::$variant(Default::default())),
219                )*
220                "json" => Some(ColumnType::Json),
221                "date" => Some(ColumnType::Date),
222                "bool" => Some(ColumnType::Bool),
223                "tinyblob" => Some(ColumnType::TinyBlob),
224                "mediumblob" => Some(ColumnType::MediumBlob),
225                "longblob" => Some(ColumnType::LongBlob),
226                _ => None,
227            }
228        };
229    }
230
231    column_type! {
232        "bit" => Bit,
233        "tinyint" => TinyInt,
234        "smallint" => SmallInt,
235        "mediumint" => MediumInt,
236        "int" => Int,
237        "bigint" => BigInt,
238        "decimal" => Decimal,
239        "float" => Float,
240        "double" => Double,
241        "time" => Time,
242        "datetime" => DateTime,
243        "timestamp" => Timestamp,
244        "char" => Char,
245        "nchar" => NChar,
246        "varchar" => Varchar,
247        "nvarchar" => NVarchar,
248        "binary" => Binary,
249        "varbinary" => Varbinary,
250        "text" => Text,
251        "tinytext" => TinyText,
252        "mediumtext" => MediumText,
253        "longtext" => LongText,
254        "blob" => Blob,
255        "enum" => Enum,
256        "set" => Set,
257        "geometry" => Geometry,
258        "point" => Point,
259        "linestring" => LineString,
260        "polygon" => Polygon,
261        "multipoint" => MultiPoint,
262        "multilinestring" => MultiLineString,
263        "multipolygon" => MultiPolygon,
264        "geometrycollection" => GeometryCollection,
265    }
266}
267
268pub fn mysql_type_to_rw_type(col_type: &ColumnType) -> ConnectorResult<DataType> {
269    let dtype = match col_type {
270        ColumnType::Serial => DataType::Int32,
271        ColumnType::Bit(attr) => {
272            if let Some(1) = attr.maximum {
273                DataType::Boolean
274            } else {
275                return Err(
276                    anyhow!("BIT({}) type not supported", attr.maximum.unwrap_or(0)).into(),
277                );
278            }
279        }
280        ColumnType::TinyInt(_) | ColumnType::SmallInt(_) => DataType::Int16,
281        ColumnType::Bool => DataType::Boolean,
282        ColumnType::MediumInt(_) => DataType::Int32,
283        ColumnType::Int(_) => DataType::Int32,
284        ColumnType::BigInt(_) => DataType::Int64,
285        ColumnType::Decimal(_) => DataType::Decimal,
286        ColumnType::Float(_) => DataType::Float32,
287        ColumnType::Double(_) => DataType::Float64,
288        ColumnType::Date => DataType::Date,
289        ColumnType::Time(_) => DataType::Time,
290        ColumnType::DateTime(_) => DataType::Timestamp,
291        ColumnType::Timestamp(_) => DataType::Timestamptz,
292        ColumnType::Year => DataType::Int32,
293        ColumnType::Char(_)
294        | ColumnType::NChar(_)
295        | ColumnType::Varchar(_)
296        | ColumnType::NVarchar(_) => DataType::Varchar,
297        ColumnType::Binary(_) | ColumnType::Varbinary(_) => DataType::Bytea,
298        ColumnType::Text(_)
299        | ColumnType::TinyText(_)
300        | ColumnType::MediumText(_)
301        | ColumnType::LongText(_) => DataType::Varchar,
302        ColumnType::Blob(_)
303        | ColumnType::TinyBlob
304        | ColumnType::MediumBlob
305        | ColumnType::LongBlob => DataType::Bytea,
306        ColumnType::Enum(_) => DataType::Varchar,
307        ColumnType::Json => DataType::Jsonb,
308        ColumnType::Set(_) => {
309            return Err(anyhow!("SET type not supported").into());
310        }
311        ColumnType::Geometry(_) => {
312            return Err(anyhow!("GEOMETRY type not supported").into());
313        }
314        ColumnType::Point(_) => {
315            return Err(anyhow!("POINT type not supported").into());
316        }
317        ColumnType::LineString(_) => {
318            return Err(anyhow!("LINE string type not supported").into());
319        }
320        ColumnType::Polygon(_) => {
321            return Err(anyhow!("POLYGON type not supported").into());
322        }
323        ColumnType::MultiPoint(_) => {
324            return Err(anyhow!("MULTI POINT type not supported").into());
325        }
326        ColumnType::MultiLineString(_) => {
327            return Err(anyhow!("MULTI LINE STRING type not supported").into());
328        }
329        ColumnType::MultiPolygon(_) => {
330            return Err(anyhow!("MULTI POLYGON type not supported").into());
331        }
332        ColumnType::GeometryCollection(_) => {
333            return Err(anyhow!("GEOMETRY COLLECTION type not supported").into());
334        }
335        ColumnType::Unknown(_) => {
336            return Err(anyhow!("Unknown MySQL data type").into());
337        }
338    };
339
340    Ok(dtype)
341}
342
343pub struct MySqlExternalTableReader {
344    rw_schema: Schema,
345    field_names: String,
346    pool: mysql_async::Pool,
347    mysql_version: (u8, u8),
348}
349
350impl ExternalTableReader for MySqlExternalTableReader {
351    async fn current_cdc_offset(&self) -> ConnectorResult<CdcOffset> {
352        let mut conn = self.pool.get_conn().await?;
353
354        // Choose SQL command based on MySQL version
355        let sql = if self.is_mysql_8_4_or_later() {
356            "SHOW BINARY LOG STATUS"
357        } else {
358            "SHOW MASTER STATUS"
359        };
360
361        tracing::debug!(
362            "Using SQL command: {} for MySQL version {}.{}",
363            sql,
364            self.mysql_version.0,
365            self.mysql_version.1
366        );
367        let mut rs = conn.query::<mysql_async::Row, _>(sql).await?;
368        let row = rs
369            .iter_mut()
370            .exactly_one()
371            .ok()
372            .context("expect exactly one row when reading binlog offset")?;
373        drop(conn);
374        Ok(CdcOffset::MySql(MySqlOffset {
375            filename: row.take("File").unwrap(),
376            position: row.take("Position").unwrap(),
377        }))
378    }
379
380    fn snapshot_read(
381        &self,
382        table_name: SchemaTableName,
383        start_pk: Option<OwnedRow>,
384        primary_keys: Vec<String>,
385        limit: u32,
386    ) -> BoxStream<'_, ConnectorResult<OwnedRow>> {
387        self.snapshot_read_inner(table_name, start_pk, primary_keys, limit)
388    }
389
390    async fn disconnect(self) -> ConnectorResult<()> {
391        self.pool.disconnect().await.map_err(|e| e.into())
392    }
393
394    fn get_parallel_cdc_splits(
395        &self,
396        _options: CdcTableSnapshotSplitOption,
397    ) -> BoxStream<'_, ConnectorResult<CdcTableSnapshotSplit>> {
398        // TODO(zw): feat: impl
399        stream::empty::<ConnectorResult<CdcTableSnapshotSplit>>().boxed()
400    }
401
402    fn split_snapshot_read(
403        &self,
404        _table_name: SchemaTableName,
405        _left: OwnedRow,
406        _right: OwnedRow,
407        _split_columns: Vec<Field>,
408    ) -> BoxStream<'_, ConnectorResult<OwnedRow>> {
409        todo!("implement MySQL CDC parallelized backfill")
410    }
411}
412
413impl MySqlExternalTableReader {
414    /// Get MySQL version from the connection
415    async fn get_mysql_version(pool: &mysql_async::Pool) -> ConnectorResult<(u8, u8)> {
416        let mut conn = pool.get_conn().await?;
417        let result: Option<String> = conn.query_first("SELECT VERSION()").await?;
418
419        if let Some(version_str) = result {
420            let parts: Vec<&str> = version_str.split('.').collect();
421            if parts.len() >= 2 {
422                let major_version = parts[0]
423                    .parse::<u8>()
424                    .context("Failed to parse major version")?;
425                let minor_version = parts[1]
426                    .parse::<u8>()
427                    .context("Failed to parse minor version")?;
428                return Ok((major_version, minor_version));
429            }
430        }
431        Err(anyhow!("Failed to get MySQL version").into())
432    }
433
434    /// Check if MySQL version is 8.4 or later
435    fn is_mysql_8_4_or_later(&self) -> bool {
436        let (major, minor) = self.mysql_version;
437        major > 8 || (major == 8 && minor >= 4)
438    }
439
440    pub async fn new(config: ExternalTableConfig, rw_schema: Schema) -> ConnectorResult<Self> {
441        let mut opts_builder = mysql_async::OptsBuilder::default()
442            .user(Some(config.username))
443            .pass(Some(config.password))
444            .ip_or_hostname(config.host)
445            .tcp_port(config.port.parse::<u16>().unwrap())
446            .db_name(Some(config.database));
447
448        opts_builder = match config.ssl_mode {
449            SslMode::Disabled | SslMode::Preferred => opts_builder.ssl_opts(None),
450            // verify-ca and verify-full are same as required for mysql now
451            SslMode::Required | SslMode::VerifyCa | SslMode::VerifyFull => {
452                let ssl_without_verify = mysql_async::SslOpts::default()
453                    .with_danger_accept_invalid_certs(true)
454                    .with_danger_skip_domain_validation(true);
455                opts_builder.ssl_opts(Some(ssl_without_verify))
456            }
457        };
458        let pool = mysql_async::Pool::new(opts_builder);
459
460        let field_names = rw_schema
461            .fields
462            .iter()
463            .filter(|f| f.name != CDC_OFFSET_COLUMN_NAME)
464            .map(|f| Self::quote_column(f.name.as_str()))
465            .join(",");
466
467        // Get MySQL version
468        let mysql_version = Self::get_mysql_version(&pool).await?;
469        tracing::info!(
470            "MySQL version detected: {}.{}",
471            mysql_version.0,
472            mysql_version.1
473        );
474
475        Ok(Self {
476            rw_schema,
477            field_names,
478            pool,
479            mysql_version,
480        })
481    }
482
483    pub fn get_normalized_table_name(table_name: &SchemaTableName) -> String {
484        // schema name is the database name in mysql
485        format!("`{}`.`{}`", table_name.schema_name, table_name.table_name)
486    }
487
488    pub fn get_cdc_offset_parser() -> CdcOffsetParseFunc {
489        Box::new(move |offset| {
490            Ok(CdcOffset::MySql(MySqlOffset::parse_debezium_offset(
491                offset,
492            )?))
493        })
494    }
495
496    #[try_stream(boxed, ok = OwnedRow, error = ConnectorError)]
497    async fn snapshot_read_inner(
498        &self,
499        table_name: SchemaTableName,
500        start_pk_row: Option<OwnedRow>,
501        primary_keys: Vec<String>,
502        limit: u32,
503    ) {
504        let order_key = primary_keys
505            .iter()
506            .map(|col| Self::quote_column(col))
507            .join(",");
508        let sql = if start_pk_row.is_none() {
509            format!(
510                "SELECT {} FROM {} ORDER BY {} LIMIT {limit}",
511                self.field_names,
512                Self::get_normalized_table_name(&table_name),
513                order_key,
514            )
515        } else {
516            let filter_expr = Self::filter_expression(&primary_keys);
517            format!(
518                "SELECT {} FROM {} WHERE {} ORDER BY {} LIMIT {limit}",
519                self.field_names,
520                Self::get_normalized_table_name(&table_name),
521                filter_expr,
522                order_key,
523            )
524        };
525        let mut conn = self.pool.get_conn().await?;
526        // Set session timezone to UTC
527        conn.exec_drop("SET time_zone = \"+00:00\"", ()).await?;
528
529        if let Some(start_pk_row) = start_pk_row {
530            let field_map = self
531                .rw_schema
532                .fields
533                .iter()
534                .map(|f| (f.name.as_str(), f.data_type.clone()))
535                .collect::<HashMap<_, _>>();
536
537            // fill in start primary key params
538            let params: Vec<_> = primary_keys
539                .iter()
540                .zip_eq_fast(start_pk_row.into_iter())
541                .map(|(pk, datum)| {
542                    if let Some(value) = datum {
543                        let ty = field_map.get(pk.as_str()).unwrap();
544                        let val = match ty {
545                            DataType::Boolean => Value::from(value.into_bool()),
546                            DataType::Int16 => Value::from(value.into_int16()),
547                            DataType::Int32 => Value::from(value.into_int32()),
548                            DataType::Int64 => Value::from(value.into_int64()),
549                            DataType::Float32 => Value::from(value.into_float32().into_inner()),
550                            DataType::Float64 => Value::from(value.into_float64().into_inner()),
551                            DataType::Varchar => Value::from(String::from(value.into_utf8())),
552                            DataType::Date => Value::from(value.into_date().0),
553                            DataType::Time => Value::from(value.into_time().0),
554                            DataType::Timestamp => Value::from(value.into_timestamp().0),
555                            DataType::Decimal => Value::from(value.into_decimal().to_string()),
556                            DataType::Timestamptz => {
557                                // Convert timestamptz to NaiveDateTime for MySQL TIMESTAMP comparison
558                                // MySQL expects NaiveDateTime for TIMESTAMP parameters
559                                let ts = value.into_timestamptz();
560                                let datetime_utc = ts.to_datetime_utc();
561                                let naive_datetime = datetime_utc.naive_utc();
562                                Value::from(naive_datetime)
563                            }
564                            _ => bail!("unsupported primary key data type: {}", ty),
565                        };
566                        ConnectorResult::Ok((pk.to_lowercase(), val))
567                    } else {
568                        bail!("primary key {} cannot be null", pk);
569                    }
570                })
571                .try_collect::<_, _, ConnectorError>()?;
572
573            tracing::debug!("snapshot read params: {:?}", &params);
574            let rs_stream = sql
575                .with(Params::from(params))
576                .stream::<mysql_async::Row, _>(&mut conn)
577                .await?;
578
579            let row_stream = rs_stream.map(|row| {
580                // convert mysql row into OwnedRow
581                let mut row = row?;
582                Ok::<_, ConnectorError>(mysql_row_to_owned_row(&mut row, &self.rw_schema))
583            });
584            pin_mut!(row_stream);
585            #[for_await]
586            for row in row_stream {
587                let row = row?;
588                yield row;
589            }
590        } else {
591            let rs_stream = sql.stream::<mysql_async::Row, _>(&mut conn).await?;
592            let row_stream = rs_stream.map(|row| {
593                // convert mysql row into OwnedRow
594                let mut row = row?;
595                Ok::<_, ConnectorError>(mysql_row_to_owned_row(&mut row, &self.rw_schema))
596            });
597            pin_mut!(row_stream);
598            #[for_await]
599            for row in row_stream {
600                let row = row?;
601                yield row;
602            }
603        }
604        drop(conn);
605    }
606
607    // mysql cannot leverage the given key to narrow down the range of scan,
608    // we need to rewrite the comparison conditions by our own.
609    // (a, b) > (x, y) => (`a` > x) OR ((`a` = x) AND (`b` > y))
610    fn filter_expression(columns: &[String]) -> String {
611        let mut conditions = vec![];
612        // push the first condition
613        conditions.push(format!(
614            "({} > :{})",
615            Self::quote_column(&columns[0]),
616            columns[0].to_lowercase()
617        ));
618        for i in 2..=columns.len() {
619            // '=' condition
620            let mut condition = String::new();
621            for (j, col) in columns.iter().enumerate().take(i - 1) {
622                if j == 0 {
623                    condition.push_str(&format!(
624                        "({} = :{})",
625                        Self::quote_column(col),
626                        col.to_lowercase()
627                    ));
628                } else {
629                    condition.push_str(&format!(
630                        " AND ({} = :{})",
631                        Self::quote_column(col),
632                        col.to_lowercase()
633                    ));
634                }
635            }
636            // '>' condition
637            condition.push_str(&format!(
638                " AND ({} > :{})",
639                Self::quote_column(&columns[i - 1]),
640                columns[i - 1].to_lowercase()
641            ));
642            conditions.push(format!("({})", condition));
643        }
644        if columns.len() > 1 {
645            conditions.join(" OR ")
646        } else {
647            conditions.join("")
648        }
649    }
650
651    fn quote_column(column: &str) -> String {
652        format!("`{}`", column)
653    }
654}
655
656#[cfg(test)]
657mod tests {
658    use std::collections::HashMap;
659
660    use futures::pin_mut;
661    use futures_async_stream::for_await;
662    use maplit::{convert_args, hashmap};
663    use risingwave_common::catalog::{ColumnDesc, ColumnId, Field, Schema};
664    use risingwave_common::types::DataType;
665
666    use crate::source::cdc::external::mysql::MySqlExternalTable;
667    use crate::source::cdc::external::{
668        CdcOffset, ExternalTableConfig, ExternalTableReader, MySqlExternalTableReader, MySqlOffset,
669        SchemaTableName,
670    };
671
672    #[ignore]
673    #[tokio::test]
674    async fn test_mysql_schema() {
675        let config = ExternalTableConfig {
676            connector: "mysql-cdc".to_owned(),
677            host: "localhost".to_owned(),
678            port: "8306".to_owned(),
679            username: "root".to_owned(),
680            password: "123456".to_owned(),
681            database: "mydb".to_owned(),
682            schema: "".to_owned(),
683            table: "part".to_owned(),
684            ssl_mode: Default::default(),
685            ssl_root_cert: None,
686            encrypt: "false".to_owned(),
687        };
688
689        let table = MySqlExternalTable::connect(config).await.unwrap();
690        println!("columns: {:?}", &table.column_descs);
691        println!("primary keys: {:?}", &table.pk_names);
692    }
693
694    #[test]
695    fn test_mysql_filter_expr() {
696        let cols = vec!["id".to_owned()];
697        let expr = MySqlExternalTableReader::filter_expression(&cols);
698        assert_eq!(expr, "(`id` > :id)");
699
700        let cols = vec!["aa".to_owned(), "bb".to_owned(), "cc".to_owned()];
701        let expr = MySqlExternalTableReader::filter_expression(&cols);
702        assert_eq!(
703            expr,
704            "(`aa` > :aa) OR ((`aa` = :aa) AND (`bb` > :bb)) OR ((`aa` = :aa) AND (`bb` = :bb) AND (`cc` > :cc))"
705        );
706    }
707
708    #[test]
709    fn test_mysql_binlog_offset() {
710        let off0_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000001", "pos": 105622, "snapshot": true }, "isHeartbeat": false }"#;
711        let off1_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000007", "pos": 1062363217, "snapshot": true }, "isHeartbeat": false }"#;
712        let off2_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000007", "pos": 659687560, "snapshot": true }, "isHeartbeat": false }"#;
713        let off3_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000008", "pos": 7665875, "snapshot": true }, "isHeartbeat": false }"#;
714        let off4_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000008", "pos": 7665875, "snapshot": true }, "isHeartbeat": false }"#;
715
716        let off0 = CdcOffset::MySql(MySqlOffset::parse_debezium_offset(off0_str).unwrap());
717        let off1 = CdcOffset::MySql(MySqlOffset::parse_debezium_offset(off1_str).unwrap());
718        let off2 = CdcOffset::MySql(MySqlOffset::parse_debezium_offset(off2_str).unwrap());
719        let off3 = CdcOffset::MySql(MySqlOffset::parse_debezium_offset(off3_str).unwrap());
720        let off4 = CdcOffset::MySql(MySqlOffset::parse_debezium_offset(off4_str).unwrap());
721
722        assert!(off0 <= off1);
723        assert!(off1 > off2);
724        assert!(off2 < off3);
725        assert_eq!(off3, off4);
726    }
727
728    // manual test case
729    #[ignore]
730    #[tokio::test]
731    async fn test_mysql_table_reader() {
732        let columns = [
733            ColumnDesc::named("v1", ColumnId::new(1), DataType::Int32),
734            ColumnDesc::named("v2", ColumnId::new(2), DataType::Decimal),
735            ColumnDesc::named("v3", ColumnId::new(3), DataType::Varchar),
736            ColumnDesc::named("v4", ColumnId::new(4), DataType::Date),
737        ];
738        let rw_schema = Schema {
739            fields: columns.iter().map(Field::from).collect(),
740        };
741        let props: HashMap<String, String> = convert_args!(hashmap!(
742                "hostname" => "localhost",
743                "port" => "8306",
744                "username" => "root",
745                "password" => "123456",
746                "database.name" => "mytest",
747                "table.name" => "t1"));
748
749        let config =
750            serde_json::from_value::<ExternalTableConfig>(serde_json::to_value(props).unwrap())
751                .unwrap();
752        let reader = MySqlExternalTableReader::new(config, rw_schema)
753            .await
754            .unwrap();
755        let offset = reader.current_cdc_offset().await.unwrap();
756        println!("BinlogOffset: {:?}", offset);
757
758        let off0_str = r#"{ "sourcePartition": { "server": "test" }, "sourceOffset": { "ts_sec": 1670876905, "file": "binlog.000001", "pos": 105622, "snapshot": true }, "isHeartbeat": false }"#;
759        let parser = MySqlExternalTableReader::get_cdc_offset_parser();
760        println!("parsed offset: {:?}", parser(off0_str).unwrap());
761        let table_name = SchemaTableName {
762            schema_name: "mytest".to_owned(),
763            table_name: "t1".to_owned(),
764        };
765
766        let stream = reader.snapshot_read(table_name, None, vec!["v1".to_owned()], 1000);
767        pin_mut!(stream);
768        #[for_await]
769        for row in stream {
770            println!("OwnedRow: {:?}", row);
771        }
772    }
773}
risingwave_connector/source/cdc/external/mysql.rs

risingwave_connector/source/cdc/external/
mysql.rs
