|
| 1 | +# Custom JDBC Operations |
| 2 | + |
| 3 | +Doobie aims to provide a nice API for typical database operations, |
| 4 | +but sometimes you need more - such as calling specific JDBC methods. |
| 5 | +In this page we list ways you can customize your JDBC query, from high level to lower level. |
| 6 | + |
| 7 | +## Low-level JDBC API (`doobie.free.*` / `doobie.F*`) |
| 8 | + |
| 9 | +In the low-level API, Doobie encodes the JDBC API as a [free monad](https://typelevel.org/cats/datatypes/freemonad.html) |
| 10 | +(hence `free` in the package name). |
| 11 | +Each standard JDBC type has its own free monad counterpart covering its operations |
| 12 | + |
| 13 | +- `doobie.free.ConnectionIO`/`doobie.FC` represents operations over `java.sql.Connection` |
| 14 | +- `doobie.free.PreparedStatement`/`doobie.FPS` represents operations over `java.sql.PreparedStatement` |
| 15 | +- `doobie.free.ResultSetIO`/`doobie.FRS` represents operations over `java.sql.ResultSet` |
| 16 | +- ...and similarly for other JDBC types |
| 17 | + |
| 18 | +For example, in `doobie.FC`/`doobie.free.connection` module you will find equivalents to the methods on `java.sql.Connection` class. |
| 19 | +(FC is an acronym where F == Free and C == Connection). Similarly, `doobie.FPS`/`doobie.free.preparedstatement` module contains equivalents to the methods on `java.sql.PreparedStatement` class. |
| 20 | + |
| 21 | +Just like we need to be careful to close Connections and PreparedStatements with JDBC, **care must be taken when using |
| 22 | +doobie's low-level API to ensure resources like Connections and PreparedStatements are closed.** |
| 23 | + |
| 24 | +### Useful low-level operations |
| 25 | + |
| 26 | +#### `raw`: Access to the underlying JDBC object |
| 27 | + |
| 28 | +Provides access to the underlying JDBC object. |
| 29 | + |
| 30 | +```scala mdoc |
| 31 | +import doobie.FRS // alias for doobie.free.resultset |
| 32 | + |
| 33 | +FRS.raw { resultSet: java.sql.ResultSet => |
| 34 | + // ... do something with the ResultSet |
| 35 | + resultSet.getInt(1) |
| 36 | +} |
| 37 | + |
| 38 | +``` |
| 39 | + |
| 40 | +#### `embed`: Convert between various `*IO`s |
| 41 | + |
| 42 | +The `embed` method allows you to compose a "lower-level" JDBC operation |
| 43 | +chain into a "higher-level" one. |
| 44 | + |
| 45 | +In the example below, we embed a `ResultSetIO` into a `PreparedStatementIO`, which is itself then embedded into a `ConnectionIO` |
| 46 | + |
| 47 | +```scala mdoc:silent |
| 48 | +import doobie.{FPS, FC} |
| 49 | +import doobie.{ResultSetIO, ConnectionIO} |
| 50 | +import cats.effect.syntax.all._ |
| 51 | +import doobie.syntax.all._ |
| 52 | + |
| 53 | +val readFirstRowCol1: ResultSetIO[String] = FRS.next.flatMap(_ => FRS.getString(1)) |
| 54 | + |
| 55 | +// Use bracket to ensure resources are closed |
| 56 | +val embedExample: ConnectionIO[String] = FC.prepareStatement("SELECT 1").bracket { preparedStatement => |
| 57 | + FC.embed( |
| 58 | + preparedStatement, |
| 59 | + FPS.executeQuery.bracket { resultSet => |
| 60 | + FPS.embed(resultSet, readFirstRowCol1) |
| 61 | + }(resultSet => FPS.embed(resultSet, FRS.close))) |
| 62 | +}(ps => FC.embed(ps, FPS.close)) |
| 63 | +``` |
| 64 | + |
| 65 | +#### Other operations |
| 66 | + |
| 67 | +You will find many familiar methods from `cats.effect.IO` |
| 68 | +with the same semantics, such as: |
| 69 | + |
| 70 | +- `pure`: Lift a pure value |
| 71 | +- `raiseError`: Raise an error |
| 72 | +- `delay`: Suspend a computation |
| 73 | + |
| 74 | +These are eventually interpreted into equivalent `cats.effect.IO` operations. |
| 75 | + |
| 76 | +## High-level API (`doobie.hi.*` / `doobie.H*`) |
| 77 | + |
| 78 | +The `doobie.hi.*` modules provide high-level APIs which handle concerns like |
| 79 | +closing Connections and logging for you. The high-level module builds upont the low-level API (`doobie.free.*`). |
| 80 | + |
| 81 | +In the example below, we use `doobie.HC.executeWithResultSet` |
| 82 | +to execute a query and obtain the results. There is no need explicitly close the Connection, PreparedStatement or ResultSet |
| 83 | +because it's handled by `HC.executeWithResultSet` already. |
| 84 | + |
| 85 | +```scala mdoc:silent |
| 86 | +import cats.effect.IO |
| 87 | +import cats.effect.unsafe.implicits.global // To allow .unsafeRunSync |
| 88 | +import doobie.Transactor |
| 89 | + |
| 90 | +// Create the transactor |
| 91 | +val xa: Transactor[IO] = Transactor.fromDriverManager[IO]( |
| 92 | + driver = "org.postgresql.Driver", |
| 93 | + url = "jdbc:postgresql:world", |
| 94 | + user = "postgres", |
| 95 | + password = "password", |
| 96 | + logHandler = None |
| 97 | +) |
| 98 | +``` |
| 99 | + |
| 100 | +```scala mdoc:silent |
| 101 | +import doobie.HC // High-level API over java.sql.Connection |
| 102 | +import doobie.HRS // High-level API over java.sql.ResultSet |
| 103 | +import doobie.ConnectionIO |
| 104 | +import doobie.util.log.{LoggingInfo, Parameters} |
| 105 | +import cats.effect.unsafe.implicits.global |
| 106 | + |
| 107 | +val sql = "SELECT * FROM (VALUES (1, '1'), (2, '2'))" |
| 108 | +val program: ConnectionIO[List[(Int, String)]] = HC.executeWithResultSet( |
| 109 | + create = FC.prepareStatement(sql), |
| 110 | + prep = FPS.unit, |
| 111 | + exec = FPS.executeQuery, |
| 112 | + process = HRS.list[(Int, String)], |
| 113 | + loggingInfo = LoggingInfo(sql, Parameters.NonBatch(List.empty), label = doobie.util.unlabeled) |
| 114 | +) |
| 115 | +``` |
| 116 | + |
| 117 | +```scala mdoc |
| 118 | +program.transact(xa).unsafeRunSync() |
| 119 | +``` |
| 120 | + |
| 121 | +### Useful high-level APIs in `doobie.hi.connection`/`doobie.HC` |
| 122 | + |
| 123 | +- `executeWithResultSet`: Create and execute a `PreparedStatement` and then process the ResultSet |
| 124 | +- `executeWithoutResultSet`: Create and execute a `PreparedStatement` which immediately returns the result without reading from a `ResultSet` |
| 125 | + (e.g. for updates since the updated row count does not require reading from a `ResultSet`) |
| 126 | +- `stream`: Execute a `PreparedStatement` query and provide rows in chunks, streamed via `fs2.Stream` |
| 127 | + |
| 128 | +## Tweaking `Query`/`Update` execution with `*AlteringExecution` methods |
| 129 | + |
| 130 | +If you just need to do a small "tweak" to your typical `Query`/`Update` execution steps, |
| 131 | +you can use methods like `toAlteringExecution`/`toMapAlteringExecution` to customize the steps. |
| 132 | + |
| 133 | +```scala mdoc |
| 134 | +import cats.syntax.all._ |
| 135 | +import doobie.hi.connection.PreparedExecution |
| 136 | + |
| 137 | +fr"select name from country order by code limit 10" |
| 138 | + .query[String] |
| 139 | + .toAlteringExecution[List, List[String]] { (steps: PreparedExecution[List[String]]) => |
| 140 | + steps.copy( |
| 141 | + process = FRS.setFetchSize(5) *> steps.process |
| 142 | + ) |
| 143 | + } |
| 144 | + .transact(xa) |
| 145 | + .unsafeRunSync() |
| 146 | + |
| 147 | +``` |
| 148 | + |
0 commit comments