The [java.sql](https://docs.oracle.com/javase/8/docs/api/java/sql/package-summary.html) API is horrible, [ResultSet.wasNull()](https://docs.oracle.com/javase/8/docs/api/java/sql/ResultSet.html#wasNull--) ?, enough said. Hibernate is black magic that generates some truly
All decisions as to which ResultSet method(s) to call are based on the Java type being mapped to, because we have no
knowledge of any database schema. These mappings rarely if ever need changed, they can be overridden with QueryMapper
but not currently at compile-time with JdbcMapper.
If you are thinking 'shut up and show me the code already' refer to [ResultSetUtil.java](https://github.com/moparisthebest/JdbcMapper/blob/master/common/src/main/java/com/moparisthebest/jdbc/util/ResultSetUtil.java)
which contains the implementations actually called.
For the purposes of this mapping, consider 'rs' an instance of ResultSet, and 'index' an int index of a ResultSet column.
### numeric primitives
if the SQL value is NULL, 0 is returned for these, and no exception is thrown
##### byte
```java
return rs.getByte(index);
```
##### short
```java
return rs.getShort(index);
```
##### int
```java
return rs.getInt(index);
```
##### long
```java
return rs.getLong(index);
```
##### float
```java
return rs.getFloat(index);
```
##### double
```java
return rs.getDouble(index);
```
### numeric objects
these wrapper types are retrieved using the same function returning their primitives above, except null is returned
if the SQL value is NULL instead of 0
```java
x ret = rs.getX(index);
return rs.wasNull() ? null : ret;
```
##### java.math.BigDecimal
```java
return rs.getBigDecimal(index);
```
### boolean
in all cases of SQL NULL being returned, if primitive boolean is requested an SQLException is thrown, if Object Boolean
is requested then null is returned.
boolean has special handling due to many popular databases not actually having a boolean type (hi Oracle!),
forcing application level workarounds.
0/1 numeric types convert to boolean using the standard ResultSet API, but many systems use char/varchar of Y/N or T/F,
which we default to Y/N but can be set via system properties:
ResultSetUtil.TRUE=Y
ResultSetUtil.FALSE=N
First the standard ResultSet API is attempted:
```java
return rs.getBoolean(index);
```
If this does not throw an SQLException, it is returned directly
If SQLException is thrown, then we try to compare as a String:
```java
String bool = rs.getString(index);
boolean ret = ResultSetUtil.TRUE.equals(bool);
if (!ret && !ResultSetUtil.FALSE.equals(bool))
throw new SQLException(String.format("Implicit conversion of database string to boolean failed on column '%d'. Returned string needs to be '%s' or '%s' and was instead '%s'.", index, ResultSetUtil.TRUE, ResultSetUtil.FALSE, bool));
return ret;
```
The returned string MUST be either TRUE or FALSE (or null, for Object Boolean) or an exception will be thrown
### Misc Objects
For all of these, when SQL NULL is returned, it maps to null
##### String
```java
return rs.getString(index);
```
##### byte[]
```java
return rs.getBytes(index);
```
##### java.sql.Ref
```java
return rs.getRef(index);
```
##### java.sql.Blob
```java
return rs.getBlob(index);
```
##### java.sql.Clob
```java
return rs.getClob(index);
```
##### java.sql.Array
```java
return rs.getArray(index);
```
##### java.sql.Struct
```java
return rs.getObject(index);
```
##### *
If nothing else fits, we call getObject and cross our fingers with QueryMapper at runtime, this is a compile-time error
with JdbcMapper. todo: is this actually a compile-time error? it *should* be, check...
Write an interface or abstract class with methods that make sense for accessing your database, annotate the methods with
SQL, and on compilation an annotation processor will generate the required java.sql API code to execute your query and
return what you wanted. This code is guaranteed to be the fastest code possible because hand written code would look
the same, just more error prone and harder to maintain. The annotation processor also checks that the SQL queries are
valid, have all the right bind parameters, and can bind the result columns to all the correct fields on the result object.
If anything is wrong it's a compile error pointing you to the exact problem.
Example:
```
@JdbcMapper.Mapper(jndiName = "java:/comp/env/jdbc/testPool") // omit jndiName and you must send in a java.sql.Connection
public interface PersonDAO extends Closeable { // Closeable is optional but must have a 'void close()' method to use cachePreparedStatements or jndiName
@JdbcMapper.SQL("INSERT INTO person (person_no, birth_date, last_name, first_name) VALUES ({personNo}, {birthDate}, {firstName}, {lastName})")
int insertPerson(long personNo, Date birthDate, String firstName, String lastName);
@JdbcMapper.SQL("UPDATE person SET first_name = {firstName} WHERE person_no = {personNo}")
int setFirstName(String firstName, long personNo); // returning int will return number of rows modified, can also return void
@JdbcMapper.SQL("SELECT first_name FROM person WHERE person_no = {personNo}")
String getFirstName(long personNo) throws SQLException; // can map directly to simple types
@JdbcMapper.SQL("SELECT person_no, first_name, last_name, birth_date FROM person WHERE person_no = {personNo}")
Person getPerson(long personNo) throws SQLException; // or multiple fields, set methods, or constructor parameters on a POJO
@JdbcMapper.SQL("SELECT person_no, first_name, last_name, birth_date FROM person WHERE last_name = {lastName}")
List<Person> getPeople(String lastName) throws SQLException; // all rows in any Collection<T> (like Set<T>, LinkedList<T> etc), T[], ResultSetIterable<T> or Stream<T> (java8+) works too
System.out.println(qm.executeUpdate("INSERT INTO person (person_no, birth_date, last_name, first_name) VALUES (?, ?, ?, ?)", 0, null, "First", "Person")); // 1
System.out.println(qm.executeUpdate("INSERT INTO person (person_no, birth_date, last_name, first_name) VALUES (?, ?, ?, ?)", 1, null, "First", "Person")); // 1
System.out.println(qm.executeUpdate("UPDATE person SET first_name = ? WHERE person_no = ?", "Second", 1)); // 1
// can map directly to simple types
System.out.println(qm.toObject("SELECT first_name FROM person WHERE person_no = ?", String.class, 0)); // First
System.out.println(qm.toObject("SELECT first_name FROM person WHERE person_no = ?", String.class, 1)); // Second
// or multiple fields, set methods, or constructor parameters on a POJO
System.out.println(qm.toObject("SELECT person_no, first_name, last_name, birth_date FROM person WHERE person_no = ?", String.class, 0)); // Person{personNo=0,birthDate=null,firstName=First,lastName=Person}
System.out.println(qm.toObject("SELECT person_no, first_name, last_name, birth_date FROM person WHERE person_no = ?", String.class, 1)); // Person{personNo=1,birthDate=null,firstName=Second,lastName=Person}
// instead of toCollection can use toList, toArray, toResultSetIterable, toStream (java8+)
System.out.println(qm.toCollection("SELECT person_no, first_name, last_name, birth_date FROM person WHERE last_name = ?", new ArrayList<String>(), String.class, "Person")); // [Person{personNo=0,birthDate=null,firstName=First,lastName=Person}, Person{personNo=1,birthDate=null,firstName=Second,lastName=Person}]
