org.jooq.meta.postgres.PostgresDatabase Maven / Gradle / Ivy
/*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* https://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
* Other licenses:
* -----------------------------------------------------------------------------
* Commercial licenses for this work are available. These replace the above
* Apache-2.0 license and offer limited warranties, support, maintenance, and
* commercial database integrations.
*
* For more information, please visit: https://www.jooq.org/legal/licensing
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*/
package org.jooq.meta.postgres;
import static java.util.Arrays.asList;
import static java.util.stream.Collectors.mapping;
import static java.util.stream.Collectors.toList;
import static org.jooq.Records.intoList;
import static org.jooq.Records.mapping;
import static org.jooq.Rows.toRowArray;
import static org.jooq.SQLDialect.POSTGRES;
// ...
// ...
// ...
// ...
import static org.jooq.impl.DSL.array;
import static org.jooq.impl.DSL.case_;
import static org.jooq.impl.DSL.cast;
import static org.jooq.impl.DSL.coalesce;
import static org.jooq.impl.DSL.condition;
import static org.jooq.impl.DSL.count;
import static org.jooq.impl.DSL.currentCatalog;
import static org.jooq.impl.DSL.decode;
import static org.jooq.impl.DSL.falseCondition;
import static org.jooq.impl.DSL.field;
import static org.jooq.impl.DSL.greatest;
import static org.jooq.impl.DSL.inline;
import static org.jooq.impl.DSL.max;
import static org.jooq.impl.DSL.name;
import static org.jooq.impl.DSL.noCondition;
import static org.jooq.impl.DSL.not;
import static org.jooq.impl.DSL.nullif;
import static org.jooq.impl.DSL.one;
import static org.jooq.impl.DSL.partitionBy;
import static org.jooq.impl.DSL.power;
import static org.jooq.impl.DSL.replace;
import static org.jooq.impl.DSL.row;
import static org.jooq.impl.DSL.rowNumber;
import static org.jooq.impl.DSL.select;
import static org.jooq.impl.DSL.selectFrom;
import static org.jooq.impl.DSL.sql;
import static org.jooq.impl.DSL.substring;
import static org.jooq.impl.DSL.table;
import static org.jooq.impl.DSL.values;
import static org.jooq.impl.DSL.when;
import static org.jooq.impl.SQLDataType.BIGINT;
import static org.jooq.impl.SQLDataType.BOOLEAN;
import static org.jooq.impl.SQLDataType.DECIMAL_INTEGER;
import static org.jooq.impl.SQLDataType.INTEGER;
import static org.jooq.impl.SQLDataType.NUMERIC;
import static org.jooq.impl.SQLDataType.VARCHAR;
import static org.jooq.meta.postgres.information_schema.Tables.ATTRIBUTES;
import static org.jooq.meta.postgres.information_schema.Tables.CHECK_CONSTRAINTS;
import static org.jooq.meta.postgres.information_schema.Tables.COLUMNS;
import static org.jooq.meta.postgres.information_schema.Tables.DOMAINS;
import static org.jooq.meta.postgres.information_schema.Tables.KEY_COLUMN_USAGE;
import static org.jooq.meta.postgres.information_schema.Tables.PARAMETERS;
import static org.jooq.meta.postgres.information_schema.Tables.ROUTINES;
import static org.jooq.meta.postgres.information_schema.Tables.SEQUENCES;
import static org.jooq.meta.postgres.information_schema.Tables.TABLES;
import static org.jooq.meta.postgres.information_schema.Tables.TRIGGERS;
import static org.jooq.meta.postgres.information_schema.Tables.VIEWS;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_CLASS;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_CONSTRAINT;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_DEPEND;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_DESCRIPTION;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_ENUM;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_INDEX;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_INHERITS;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_NAMESPACE;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_PROC;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_SEQUENCE;
import static org.jooq.meta.postgres.pg_catalog.Tables.PG_TYPE;
import static org.jooq.tools.StringUtils.defaultIfNull;
import static org.jooq.util.postgres.PostgresDSL.arrayAppend;
import java.math.BigDecimal;
import java.math.BigInteger;
import java.sql.SQLException;
import java.util.ArrayList;
import java.util.EnumSet;
import java.util.HashMap;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.stream.Collectors;
import org.jooq.CommonTableExpression;
import org.jooq.Condition;
import org.jooq.DSLContext;
import org.jooq.Field;
import org.jooq.Name;
// ...
import org.jooq.Record;
import org.jooq.Record1;
import org.jooq.Record12;
import org.jooq.Record14;
import org.jooq.Record4;
import org.jooq.Record5;
import org.jooq.Record6;
import org.jooq.Record7;
import org.jooq.Result;
import org.jooq.ResultQuery;
import org.jooq.SQLDialect;
import org.jooq.Select;
import org.jooq.SortOrder;
import org.jooq.Table;
import org.jooq.TableField;
import org.jooq.TableOptions.TableType;
// ...
// ...
// ...
import org.jooq.conf.ParseUnknownFunctions;
import org.jooq.exception.DataAccessException;
import org.jooq.impl.DSL;
import org.jooq.impl.ParserException;
import org.jooq.impl.SQLDataType;
import org.jooq.meta.AbstractDatabase;
import org.jooq.meta.AbstractIndexDefinition;
import org.jooq.meta.ArrayDefinition;
import org.jooq.meta.CatalogDefinition;
import org.jooq.meta.ColumnDefinition;
import org.jooq.meta.DataTypeDefinition;
import org.jooq.meta.DefaultCheckConstraintDefinition;
import org.jooq.meta.DefaultDataTypeDefinition;
import org.jooq.meta.DefaultDomainDefinition;
import org.jooq.meta.DefaultEnumDefinition;
import org.jooq.meta.DefaultIndexColumnDefinition;
import org.jooq.meta.DefaultRelations;
import org.jooq.meta.DefaultSequenceDefinition;
// ...
import org.jooq.meta.DomainDefinition;
import org.jooq.meta.EnumDefinition;
import org.jooq.meta.IndexColumnDefinition;
import org.jooq.meta.IndexDefinition;
import org.jooq.meta.PackageDefinition;
import org.jooq.meta.ResultQueryDatabase;
import org.jooq.meta.RoutineDefinition;
import org.jooq.meta.SchemaDefinition;
import org.jooq.meta.SequenceDefinition;
import org.jooq.meta.TableDefinition;
// ...
import org.jooq.meta.UDTDefinition;
import org.jooq.meta.UniqueKeyDefinition;
import org.jooq.meta.XMLSchemaCollectionDefinition;
import org.jooq.meta.hsqldb.HSQLDBDatabase;
import org.jooq.meta.postgres.information_schema.tables.CheckConstraints;
import org.jooq.meta.postgres.information_schema.tables.KeyColumnUsage;
import org.jooq.meta.postgres.information_schema.tables.Routines;
import org.jooq.meta.postgres.information_schema.tables.Triggers;
import org.jooq.meta.postgres.pg_catalog.tables.PgClass;
import org.jooq.meta.postgres.pg_catalog.tables.PgConstraint;
import org.jooq.meta.postgres.pg_catalog.tables.PgIndex;
import org.jooq.meta.postgres.pg_catalog.tables.PgInherits;
import org.jooq.meta.postgres.pg_catalog.tables.PgType;
import org.jooq.tools.JooqLogger;
import org.jetbrains.annotations.Nullable;
/**
* Postgres uses the ANSI default INFORMATION_SCHEMA, but unfortunately ships
* with a non-capitalised version of it: information_schema. Hence
* the {@link HSQLDBDatabase} is not used here.
*
* @author Lukas Eder
*/
public class PostgresDatabase extends AbstractDatabase implements ResultQueryDatabase {
private static final JooqLogger log = JooqLogger.getLogger(PostgresDatabase.class);
private Boolean is84;
private Boolean is94;
private Boolean is10;
private Boolean is11;
private Boolean is12;
private Boolean canUseRoutines;
private Boolean canCastToEnumType;
private Boolean canCombineArrays;
private Boolean canUseTupleInPredicates;
@Override
protected List getIndexes0() throws SQLException {
List result = new ArrayList<>();
PgIndex i = PG_INDEX.as("i");
PgConstraint c = PG_CONSTRAINT.as("c");
indexLoop:
for (Record6 record : create()
.select(
i.tableClass().pgNamespace().NSPNAME,
i.tableClass().RELNAME,
i.indexClass().RELNAME,
i.INDISUNIQUE,
array(
select(field("pg_get_indexdef({0}, k + 1, true)", String.class, i.INDEXRELID))
.from("generate_subscripts({0}, 1) as k", i.INDKEY)
.orderBy(field("k"))
).as("columns"),
field("{0}::int[]", Integer[].class, i.INDOPTION).as("asc_or_desc")
)
.from(i)
.where(i.tableClass().pgNamespace().NSPNAME.in(getInputSchemata()))
.and(getIncludeSystemIndexes()
? noCondition()
: row(i.tableClass().pgNamespace().NSPNAME, i.indexClass().RELNAME).notIn(
select(c.pgNamespace().NSPNAME, c.CONNAME).from(c)
))
.orderBy(1, 2, 3)) {
final SchemaDefinition tableSchema = getSchema(record.get(i.tableClass().pgNamespace().NSPNAME));
if (tableSchema == null)
continue indexLoop;
final String indexName = record.get(i.indexClass().RELNAME);
final String tableName = record.get(i.tableClass().RELNAME);
final String[] columns = record.value5();
final Integer[] options = record.value6();
final TableDefinition table = getTable(tableSchema, tableName);
if (table == null)
continue indexLoop;
final boolean unique = record.get(i.INDISUNIQUE);
for (int k = 0; k < columns.length; k++)
// [#6310] [#6620] Function-based indexes are not yet supported
// [#11047] Even without supporting function-based indexes, we might have to parse
// the column expression, because it might be quoted
// [#16237] Alternatively, the column could be hidden or excluded
if (table.getColumn(columns[k]) == null && table.getColumn(columns[k] = tryParseColumnName(columns[k])) == null)
continue indexLoop;
// [#10466] Neither are INCLUDE columns, which are reported as
// columns without options
else if (k >= options.length)
continue indexLoop;
result.add(new AbstractIndexDefinition(tableSchema, indexName, table, unique) {
List indexColumns = new ArrayList<>();
{
for (int ordinal = 0; ordinal < columns.length; ordinal++) {
ColumnDefinition column = table.getColumn(columns[ordinal]);
// [#6307] Some background info on this bitwise operation here:
// https://stackoverflow.com/a/18128104/521799
SortOrder order = (options[ordinal] & 1) == 1 ? SortOrder.DESC : SortOrder.ASC;
indexColumns.add(new DefaultIndexColumnDefinition(
this,
column,
order,
ordinal + 1
));
}
}
@Override
protected List getIndexColumns0() {
return indexColumns;
}
});
}
return result;
}
private String tryParseColumnName(String string) {
try {
return create()
.configuration()
.deriveSettings(s -> s.withParseUnknownFunctions(ParseUnknownFunctions.IGNORE))
.dsl()
.parser().parseField(string).getName();
}
catch (ParserException e) {
log.info("Parse error", "Error when parsing column name : " + string, e);
return string;
}
}
@Override
protected void loadPrimaryKeys(DefaultRelations relations) throws SQLException {
for (Record record : primaryKeys(getInputSchemata())) {
SchemaDefinition schema = getSchema(record.get(KEY_COLUMN_USAGE.TABLE_SCHEMA));
String key = record.get(KEY_COLUMN_USAGE.CONSTRAINT_NAME);
String tableName = record.get(KEY_COLUMN_USAGE.TABLE_NAME);
String columnName = record.get(KEY_COLUMN_USAGE.COLUMN_NAME);
TableDefinition table = getTable(schema, tableName);
if (table != null)
relations.addPrimaryKey(key, table, table.getColumn(columnName));
}
}
@Override
protected void loadUniqueKeys(DefaultRelations relations) throws SQLException {
for (Record record : uniqueKeys(getInputSchemata())) {
SchemaDefinition schema = getSchema(record.get(KEY_COLUMN_USAGE.TABLE_SCHEMA));
String key = record.get(KEY_COLUMN_USAGE.CONSTRAINT_NAME);
String tableName = record.get(KEY_COLUMN_USAGE.TABLE_NAME);
String columnName = record.get(KEY_COLUMN_USAGE.COLUMN_NAME);
TableDefinition table = getTable(schema, tableName);
if (table != null)
relations.addUniqueKey(key, table, table.getColumn(columnName));
}
}
@Override
public ResultQuery> primaryKeys(List schemas) {
return keys(schemas, inline("p"));
}
@Override
public ResultQuery> uniqueKeys(List schemas) {
return keys(schemas, inline("u"));
}
private ResultQuery> keys(List schemas, Field constraintType) {
KeyColumnUsage k = KEY_COLUMN_USAGE.as("k");
PgConstraint c = PG_CONSTRAINT.as("c");
return create()
.select(
k.TABLE_CATALOG,
k.TABLE_SCHEMA,
k.TABLE_NAME,
k.CONSTRAINT_NAME,
k.COLUMN_NAME,
k.ORDINAL_POSITION)
.from(c)
.join(k)
.on(k.TABLE_SCHEMA.eq(c.pgClass().pgNamespace().NSPNAME))
.and(k.TABLE_NAME.eq(c.pgClass().RELNAME))
.and(k.CONSTRAINT_SCHEMA.eq(c.pgNamespace().NSPNAME))
.and(k.CONSTRAINT_NAME.eq(c.CONNAME))
.where(c.CONTYPE.eq(constraintType))
.and(c.pgNamespace().NSPNAME.in(schemas))
.orderBy(
k.TABLE_SCHEMA.asc(),
k.TABLE_NAME.asc(),
k.CONSTRAINT_NAME.asc(),
k.ORDINAL_POSITION.asc());
}
@Override
protected void loadForeignKeys(DefaultRelations relations) throws SQLException {
// [#3520] PostgreSQL INFORMATION_SCHEMA.REFERENTIAL_CONSTRAINTS contains incomplete information about foreign keys
// The (CONSTRAINT_CATALOG, CONSTRAINT_SCHEMA, CONSTRAINT_NAME) tuple is non-unique, in case two tables share the
// same CONSTRAINT_NAME.
// The JDBC driver implements this correctly through the pg_catalog, although the sorting and column name casing is wrong, too.
Result result = create()
.fetch(getConnection().getMetaData().getExportedKeys(null, null, null))
.sortAsc("key_seq")
.sortAsc("fk_name")
.sortAsc("fktable_name")
.sortAsc("fktable_schem");
resultLoop:
for (Record record : result) {
SchemaDefinition foreignKeySchema = getSchema(record.get("fktable_schem", String.class));
SchemaDefinition uniqueKeySchema = getSchema(record.get("pktable_schem", String.class));
if (foreignKeySchema == null || uniqueKeySchema == null)
continue resultLoop;
String foreignKey = record.get("fk_name", String.class);
String foreignKeyTableName = record.get("fktable_name", String.class);
String foreignKeyColumn = record.get("fkcolumn_name", String.class);
String uniqueKey = record.get("pk_name", String.class);
String uniqueKeyTableName = record.get("pktable_name", String.class);
String uniqueKeyColumn = record.get("pkcolumn_name", String.class);
TableDefinition foreignKeyTable = getTable(foreignKeySchema, foreignKeyTableName);
TableDefinition uniqueKeyTable = getTable(uniqueKeySchema, uniqueKeyTableName);
if (foreignKeyTable != null && uniqueKeyTable != null) {
relations.addForeignKey(
foreignKey,
foreignKeyTable,
foreignKeyTable.getColumn(foreignKeyColumn),
uniqueKey,
uniqueKeyTable,
uniqueKeyTable.getColumn(uniqueKeyColumn),
true
);
for (IndexDefinition index : getIndexes(uniqueKeyTable)) {
if (index.getName().equals(uniqueKey)) {
for (UniqueKeyDefinition uk : uniqueKeyTable.getKeys()) {
if (uk.getKeyColumns().equals(index.getIndexColumns().stream().map(i -> i.getColumn()).collect(toList()))) {
relations.addForeignKey(
foreignKey,
foreignKeyTable,
foreignKeyTable.getColumn(foreignKeyColumn),
uk.getName(),
uniqueKeyTable,
uniqueKeyTable.getColumn(uniqueKeyColumn),
true
);
}
}
}
}
}
}
}
@Override
protected void loadCheckConstraints(DefaultRelations relations) throws SQLException {
CheckConstraints cc = CHECK_CONSTRAINTS.as("cc");
// [#10940] [#10992] Workaround for issue caused by re-using implicit join paths
PgConstraint pc = PG_CONSTRAINT.as("pc");
for (Record record : create()
.select(
pc.pgClass().pgNamespace().NSPNAME,
pc.pgClass().RELNAME,
pc.CONNAME.as(cc.CONSTRAINT_NAME),
replace(field("pg_get_constraintdef({0}.oid)", VARCHAR, pc), inline("CHECK "), inline("")).as(cc.CHECK_CLAUSE))
.from(pc)
.where(pc.pgClass().pgNamespace().NSPNAME.in(getInputSchemata()))
.and(pc.CONTYPE.eq(inline("c")))
.unionAll(
getIncludeSystemCheckConstraints()
? select(
pc.pgClass().pgNamespace().NSPNAME,
pc.pgClass().RELNAME,
cc.CONSTRAINT_NAME,
cc.CHECK_CLAUSE
)
.from(pc)
.join(cc)
.on(pc.CONNAME.eq(cc.CONSTRAINT_NAME))
.and(pc.pgNamespace().NSPNAME.eq(cc.CONSTRAINT_NAME))
.where(pc.pgNamespace().NSPNAME.in(getInputSchemata()))
.and(row(pc.pgClass().pgNamespace().NSPNAME, pc.pgClass().RELNAME, cc.CONSTRAINT_NAME).notIn(
select(
pc.pgClass().pgNamespace().NSPNAME,
pc.pgClass().RELNAME,
pc.CONNAME)
.from(pc)
.where(pc.CONTYPE.eq(inline("c")))
))
: select(inline(""), inline(""), inline(""), inline("")).where(falseCondition()))
.orderBy(1, 2, 3)
) {
SchemaDefinition schema = getSchema(record.get(pc.pgClass().pgNamespace().NSPNAME));
TableDefinition table = getTable(schema, record.get(pc.pgClass().RELNAME));
if (table != null) {
relations.addCheckConstraint(table, new DefaultCheckConstraintDefinition(
schema,
table,
record.get(cc.CONSTRAINT_NAME),
record.get(cc.CHECK_CLAUSE)
));
}
}
}
@Override
protected List getTables0() throws SQLException {
List result = new ArrayList<>();
Map map = new HashMap<>();
Select> empty =
select(inline(""), inline(""), inline(""), inline(""), inline(""), inline(""), inline(""))
.where(falseCondition());
for (Record record : create()
.select()
.from(
select(
TABLES.TABLE_SCHEMA,
TABLES.TABLE_NAME,
TABLES.TABLE_NAME.as("specific_name"),
PG_DESCRIPTION.DESCRIPTION,
when(TABLES.TABLE_TYPE.eq(inline("VIEW")), inline(TableType.VIEW.name()))
.else_(inline(TableType.TABLE.name())).as("table_type"),
inline("").as(ROUTINES.TYPE_UDT_SCHEMA),
inline("").as(ROUTINES.TYPE_UDT_NAME))
.from(TABLES)
.join(PG_CLASS)
.on(PG_CLASS.RELNAME.eq(TABLES.TABLE_NAME))
.and(PG_CLASS.pgNamespace().NSPNAME.eq(TABLES.TABLE_SCHEMA))
.leftJoin(PG_DESCRIPTION)
.on(PG_DESCRIPTION.OBJOID.eq(PG_CLASS.OID))
.and(PG_DESCRIPTION.CLASSOID.eq(field("'pg_class'::regclass", BIGINT)))
.and(PG_DESCRIPTION.OBJSUBID.eq(inline(0)))
.leftJoin(VIEWS)
.on(TABLES.TABLE_SCHEMA.eq(VIEWS.TABLE_SCHEMA))
.and(TABLES.TABLE_NAME.eq(VIEWS.TABLE_NAME))
.where(TABLES.TABLE_SCHEMA.in(getInputSchemata()))
// To stay on the safe side, if the INFORMATION_SCHEMA ever
// includes materialised views, let's exclude them from here
.and(canUseTupleInPredicates()
? row(TABLES.TABLE_SCHEMA, TABLES.TABLE_NAME).notIn(
select(
PG_CLASS.pgNamespace().NSPNAME,
PG_CLASS.RELNAME)
.from(PG_CLASS)
.where(PG_CLASS.RELKIND.eq(inline("m"))))
: noCondition()
)
// [#3254] Materialised views are reported only in PG_CLASS, not
// in INFORMATION_SCHEMA.TABLES
// [#8478] CockroachDB cannot compare "sql_identifier" types (varchar)
// from information_schema with "name" types from pg_catalog
.unionAll(
select(
field("{0}::varchar", PG_CLASS.pgNamespace().NSPNAME.getDataType(), PG_CLASS.pgNamespace().NSPNAME),
field("{0}::varchar", PG_CLASS.RELNAME.getDataType(), PG_CLASS.RELNAME),
field("{0}::varchar", PG_CLASS.RELNAME.getDataType(), PG_CLASS.RELNAME),
PG_DESCRIPTION.DESCRIPTION,
inline(TableType.MATERIALIZED_VIEW.name()).as("table_type"),
inline(""),
inline(""))
.from(PG_CLASS)
.leftOuterJoin(PG_DESCRIPTION)
.on(PG_DESCRIPTION.OBJOID.eq(PG_CLASS.OID))
.and(PG_DESCRIPTION.CLASSOID.eq(field("'pg_class'::regclass", BIGINT)))
.and(PG_DESCRIPTION.OBJSUBID.eq(inline(0)))
.where(PG_CLASS.pgNamespace().NSPNAME.in(getInputSchemata()))
.and(PG_CLASS.RELKIND.eq(inline("m"))))
// [#3375] [#3376] Include table-valued functions in the set of tables
.unionAll(
tableValuedFunctions()
? select(
ROUTINES.ROUTINE_SCHEMA,
ROUTINES.ROUTINE_NAME,
ROUTINES.SPECIFIC_NAME,
inline(""),
inline(TableType.FUNCTION.name()).as("table_type"),
ROUTINES.TYPE_UDT_SCHEMA,
ROUTINES.TYPE_UDT_NAME)
.from(ROUTINES)
.join(PG_PROC).on(PG_PROC.pgNamespace().NSPNAME.eq(ROUTINES.SPECIFIC_SCHEMA))
.and(PG_PROC.PRONAME.concat("_").concat(PG_PROC.OID).eq(ROUTINES.SPECIFIC_NAME))
.where(ROUTINES.ROUTINE_SCHEMA.in(getInputSchemata()))
.and(PG_PROC.PRORETSET)
: empty)
.asTable("tables"))
.orderBy(1, 2)
.fetch()) {
SchemaDefinition schema = getSchema(record.get(TABLES.TABLE_SCHEMA));
String name = record.get(TABLES.TABLE_NAME);
String comment = record.get(PG_DESCRIPTION.DESCRIPTION, String.class);
TableType tableType = record.get("table_type", TableType.class);
switch (tableType) {
case FUNCTION: {
result.add(new PostgresTableValuedFunction(
schema, name,
record.get(ROUTINES.SPECIFIC_NAME),
comment, null,
getSchema(record.get(ROUTINES.TYPE_UDT_SCHEMA)),
record.get(ROUTINES.TYPE_UDT_NAME)
));
break;
}
case MATERIALIZED_VIEW: {
result.add(new PostgresMaterializedViewDefinition(schema, name, comment));
break;
}
default: {
PostgresTableDefinition t = new PostgresTableDefinition(schema, name, comment, tableType, null);
result.add(t);
map.put(name(schema.getName(), name), t);
break;
}
}
}
PgClass ct = PG_CLASS.as("ct");
PgInherits i = PG_INHERITS.as("i");
PgClass pt = PG_CLASS.as("pt");
// [#2916] If window functions are not supported (prior to PostgreSQL 8.4), then
// don't execute the following query:
if (is84()) {
for (Record5 inheritance : create()
.select(
ct.pgNamespace().NSPNAME,
ct.RELNAME,
pt.pgNamespace().NSPNAME,
pt.RELNAME,
max(i.INHSEQNO).over().partitionBy(i.INHRELID).as("m")
)
.from(ct)
.join(i).on(i.INHRELID.eq(ct.OID))
.join(pt).on(i.INHPARENT.eq(pt.OID))
.where(ct.pgNamespace().NSPNAME.in(getInputSchemata()))
.and(pt.pgNamespace().NSPNAME.in(getInputSchemata()))
) {
Name child = name(inheritance.value1(), inheritance.value2());
Name parent = name(inheritance.value3(), inheritance.value4());
if (inheritance.value5() > 1) {
log.info("Multiple inheritance",
"Multiple inheritance is not supported by jOOQ: " +
child +
" inherits from " +
parent);
}
else {
PostgresTableDefinition childTable = map.get(child);
PostgresTableDefinition parentTable = map.get(parent);
if (childTable != null && parentTable != null) {
childTable.setParentTable(parentTable);
parentTable.getChildTables().add(childTable);
}
}
}
}
return result;
}
@Override
protected List getCatalogs0() throws SQLException {
List result = new ArrayList<>();
result.add(new CatalogDefinition(this, "", ""));
return result;
}
@Override
protected List getSchemata0() throws SQLException {
// [#1409] Shouldn't select from INFORMATION_SCHEMA.SCHEMATA, as that
// would only return schemata of which CURRENT_USER is the owner
return
create().select(PG_NAMESPACE.NSPNAME)
.from(PG_NAMESPACE)
.orderBy(PG_NAMESPACE.NSPNAME)
.collect(mapping(r -> new SchemaDefinition(this, r.value1(), ""), toList()));
}
@Override
public ResultQuery> sources(List schemas) {
// [#9483] Some dialects include materialized views in the INFORMATION_SCHEMA.VIEWS view
PgClass c = PG_CLASS.as("c");
return create()
.select(
currentCatalog(),
c.pgNamespace().NSPNAME,
c.RELNAME,
when(c.RELKIND.eq(inline("m")), inline("create materialized view \""))
.else_(inline("create view \""))
.concat(c.RELNAME)
.concat(inline("\" as "))
.concat(field("pg_get_viewdef({0})", VARCHAR, c.OID)))
.from(c)
.where(c.RELKIND.in(inline("v"), inline("m")))
.and(c.pgNamespace().NSPNAME.in(schemas))
.orderBy(1, 2, 3);
}
@Override
public ResultQuery> comments(List schemas) {
return null;
}
@Override
public ResultQuery> sequences(List schemas) {
CommonTableExpression> s = name("schemas").fields("schema").as(selectFrom(values(schemas.stream().collect(toRowArray(DSL::val)))));
return create()
.with(s)
.select(
inline(null, VARCHAR).as("catalog"),
SEQUENCES.SEQUENCE_SCHEMA,
SEQUENCES.SEQUENCE_NAME,
SEQUENCES.DATA_TYPE,
SEQUENCES.NUMERIC_PRECISION,
SEQUENCES.NUMERIC_SCALE,
nullif(SEQUENCES.START_VALUE.cast(BIGINT), inline(1L)).as(SEQUENCES.START_VALUE),
nullif(SEQUENCES.INCREMENT.cast(BIGINT), inline(1L)).as(SEQUENCES.INCREMENT),
nullif(SEQUENCES.MINIMUM_VALUE.cast(BIGINT), inline(1L)).coerce(NUMERIC).as(SEQUENCES.MINIMUM_VALUE),
nullif(SEQUENCES.MAXIMUM_VALUE.cast(DECIMAL_INTEGER),
power(cast(inline(2L), DECIMAL_INTEGER), cast(SEQUENCES.NUMERIC_PRECISION.minus(inline(1L)), DECIMAL_INTEGER)).minus(inline(1L))).coerce(NUMERIC).as(SEQUENCES.MAXIMUM_VALUE),
SEQUENCES.CYCLE_OPTION.cast(BOOLEAN).as(SEQUENCES.CYCLE_OPTION),
inline(null, BIGINT).as("cache"))
.from(SEQUENCES)
.where(SEQUENCES.SEQUENCE_SCHEMA.in(selectFrom(s)))
.and(!getIncludeSystemSequences()
? row(SEQUENCES.SEQUENCE_SCHEMA, SEQUENCES.SEQUENCE_NAME).notIn(
select(COLUMNS.TABLE_SCHEMA, COLUMNS.TABLE_NAME.concat(inline("_")).concat(COLUMNS.COLUMN_NAME).concat(inline("_seq")))
.from(COLUMNS)
.where(COLUMNS.COLUMN_DEFAULT.eq(
inline("nextval('").concat(COLUMNS.TABLE_NAME.concat(inline("_")).concat(COLUMNS.COLUMN_NAME)).concat(inline("_seq'::regclass)"))
))
.or(COLUMNS.COLUMN_DEFAULT.eq(
inline("nextval('").concat(COLUMNS.TABLE_SCHEMA.concat(inline(".")).concat(COLUMNS.TABLE_NAME).concat(inline("_")).concat(COLUMNS.COLUMN_NAME)).concat(inline("_seq'::regclass)"))
))
)
: noCondition())
.unionAll(is10() && getIncludeSystemSequences()
? select(
inline(null, VARCHAR),
PG_SEQUENCE.pgClass().pgNamespace().NSPNAME,
PG_SEQUENCE.pgClass().RELNAME,
PG_SEQUENCE.pgType().TYPNAME,
// See https://github.com/postgres/postgres/blob/master/src/backend/catalog/information_schema.sql
field("information_schema._pg_numeric_precision({0}, {1})", INTEGER, PG_SEQUENCE.pgType().TYPBASETYPE, PG_SEQUENCE.pgType().TYPTYPMOD),
inline(0),
PG_SEQUENCE.SEQSTART,
PG_SEQUENCE.SEQINCREMENT,
PG_SEQUENCE.SEQMIN.coerce(NUMERIC),
PG_SEQUENCE.SEQMAX.coerce(NUMERIC),
PG_SEQUENCE.SEQCYCLE,
inline(null, BIGINT).as("cache"))
.from(PG_SEQUENCE)
.where(PG_SEQUENCE.pgClass().pgNamespace().NSPNAME.in(selectFrom(s)))
.and(PG_SEQUENCE.pgClass().OID.in(
select(PG_DEPEND.OBJID)
.from(PG_DEPEND)
.where(PG_DEPEND.DEPTYPE.eq(inline("i")))
.and(PG_DEPEND.CLASSID.eq(field("'pg_class'::regclass", PG_DEPEND.CLASSID.getDataType())))
))
// AND NOT EXISTS (SELECT 1 FROM pg_depend WHERE classid = 'pg_class'::regclass AND objid = c.oid AND deptype = 'i')
: select(inline(""), inline(""), inline(""), inline(""), inline(0), inline(0), inline(0L), inline(0L), inline(BigDecimal.ZERO), inline(BigDecimal.ZERO), inline(false), inline(0L))
.where(falseCondition()))
.orderBy(2, 3);
}
@Override
protected List getSequences0() throws SQLException {
List result = new ArrayList<>();
for (Record record : sequences(getInputSchemata())) {
SchemaDefinition schema = getSchema(record.get(SEQUENCES.SEQUENCE_SCHEMA));
DataTypeDefinition type = new DefaultDataTypeDefinition(
this, schema,
record.get(SEQUENCES.DATA_TYPE),
0,
record.get(SEQUENCES.NUMERIC_PRECISION),
record.get(SEQUENCES.NUMERIC_SCALE),
false,
(String) null
);
result.add(new DefaultSequenceDefinition(
schema,
record.get(SEQUENCES.SEQUENCE_NAME),
type,
null,
record.get(SEQUENCES.START_VALUE, BigInteger.class),
record.get(SEQUENCES.INCREMENT, BigInteger.class),
record.get(SEQUENCES.MINIMUM_VALUE, BigInteger.class),
record.get(SEQUENCES.MAXIMUM_VALUE, BigInteger.class),
record.get(SEQUENCES.CYCLE_OPTION, Boolean.class),
null // [#9442] The CACHE flag is not available from SEQUENCES
));
}
return result;
}
@Override
public ResultQuery> enums(List schemas) {
return null;
}
static record Identifier(String schema, String name) {}
@Override
protected List getEnums0() throws SQLException {
List result = new ArrayList<>();
// [#2736] This table is unavailable in Amazon Redshift
if (exists(PG_ENUM)) {
// [#2707] Fetch all enum type names first, in order to be able to
// perform enumlabel::[typname] casts in the subsequent query for
// cross-version compatible enum literal ordering
for (Identifier type : create()
.select(
PG_TYPE.pgNamespace().NSPNAME,
PG_TYPE.TYPNAME)
.from(PG_TYPE)
.where(PG_TYPE.pgNamespace().NSPNAME.in(getInputSchemata()))
.and(PG_TYPE.OID.in(select(PG_ENUM.ENUMTYPID).from(PG_ENUM)))
.orderBy(
PG_TYPE.pgNamespace().NSPNAME,
PG_TYPE.TYPNAME)
.fetch(mapping(Identifier::new))) {
DefaultEnumDefinition definition = null;
for (String label : enumLabels(type.schema, type.name)) {
SchemaDefinition schema = getSchema(type.schema);
String typeName = String.valueOf(type.name);
if (definition == null || !definition.getName().equals(typeName)) {
definition = new DefaultEnumDefinition(schema, typeName, null);
result.add(definition);
}
definition.addLiteral(label);
}
}
}
return result;
}
@Override
protected List getDomains0() throws SQLException {
Map result = new LinkedHashMap<>();
if (existAll(PG_CONSTRAINT, PG_TYPE)) {
PgConstraint c = PG_CONSTRAINT.as("c");
PgType d = PG_TYPE.as("d");
PgType b = PG_TYPE.as("b");
Field src = field(name("domains", "src"), String[].class);
Field constraintDef = field("pg_get_constraintdef({0})", VARCHAR, c.OID);
for (Record record : create()
.withRecursive("domains",
"domain_id",
"base_id",
"typbasetype",
"conname",
"src"
)
.as(
select(
d.OID,
d.OID,
d.TYPBASETYPE,
c.CONNAME,
when(c.OID.isNotNull(), array(constraintDef))
)
.from(d)
.leftJoin(c)
.on(d.OID.eq(c.CONTYPID))
.where(d.TYPTYPE.eq("d"))
.and(d.pgNamespace().NSPNAME.in(getInputSchemata()))
.unionAll(
select(
field(name("domains", "domain_id"), Long.class),
d.OID,
d.TYPBASETYPE,
c.CONNAME,
decode()
.when(c.CONBIN.isNull(), src)
.otherwise(arrayAppend(src, constraintDef))
)
.from(name("domains"))
.join(d)
.on(field(name("domains", d.TYPBASETYPE.getName())).eq(d.OID))
.leftJoin(c)
.on(d.OID.eq(c.CONTYPID))
))
.select(
d.pgNamespace().NSPNAME,
d.TYPNAME,
d.TYPNOTNULL,
d.TYPDEFAULT,
b.TYPNAME,
// See https://github.com/postgres/postgres/blob/master/src/backend/catalog/information_schema.sql
field("information_schema._pg_char_max_length({0}, {1})", INTEGER, d.TYPBASETYPE, d.TYPTYPMOD).as(DOMAINS.CHARACTER_MAXIMUM_LENGTH),
// [#15555]
coalesce(
field("information_schema._pg_datetime_precision({0}, {1})", INTEGER, d.TYPBASETYPE, d.TYPTYPMOD),
field("information_schema._pg_numeric_precision({0}, {1})", INTEGER, d.TYPBASETYPE, d.TYPTYPMOD)
).as(DOMAINS.NUMERIC_PRECISION),
field("information_schema._pg_numeric_scale({0}, {1})", INTEGER, d.TYPBASETYPE, d.TYPTYPMOD).as(DOMAINS.NUMERIC_SCALE),
src)
.from(d)
.join(name("domains"))
.on(field(name("domains", "typbasetype")).eq(0))
.and(field(name("domains", "domain_id")).eq(d.OID))
.join(b)
.on(field(name("domains", "base_id")).eq(b.OID))
.where(d.TYPTYPE.eq("d"))
.and(d.pgNamespace().NSPNAME.in(getInputSchemata()))
.orderBy(d.pgNamespace().NSPNAME, d.TYPNAME, field(name("domains", "conname")))
) {
String schemaName = record.get(d.pgNamespace().NSPNAME);
String domainName = record.get(d.TYPNAME);
String[] check = record.get(src);
DefaultDomainDefinition domain = result.computeIfAbsent(name(schemaName, domainName), k -> {
SchemaDefinition schema = getSchema(record.get(d.pgNamespace().NSPNAME));
DataTypeDefinition baseType = new DefaultDataTypeDefinition(
this,
schema,
record.get(b.TYPNAME),
record.get(DOMAINS.CHARACTER_MAXIMUM_LENGTH),
record.get(DOMAINS.NUMERIC_PRECISION),
record.get(DOMAINS.NUMERIC_SCALE),
!record.get(d.TYPNOTNULL, boolean.class),
record.get(d.TYPDEFAULT),
name(
record.get(d.pgNamespace().NSPNAME),
record.get(b.TYPNAME)
)
);
return new DefaultDomainDefinition(
schema,
record.get(d.TYPNAME),
baseType
);
});
domain.addCheckClause(check);
}
}
return new ArrayList<>(result.values());
}
@Override
protected List getXMLSchemaCollections0() throws SQLException {
List result = new ArrayList<>();
return result;
}
@Override
protected List getUDTs0() throws SQLException {
List result = new ArrayList<>();
// [#2736] This table is unavailable in Amazon Redshift
if (exists(ATTRIBUTES)) {
for (Identifier udt : create()
.select(
ATTRIBUTES.UDT_SCHEMA,
ATTRIBUTES.UDT_NAME)
.from(ATTRIBUTES)
.where(ATTRIBUTES.UDT_SCHEMA.in(getInputSchemata()))
// [#14621] Work around https://github.com/yugabyte/yugabyte-db/issues/16081
.groupBy(
ATTRIBUTES.UDT_SCHEMA,
ATTRIBUTES.UDT_NAME)
.orderBy(
ATTRIBUTES.UDT_SCHEMA,
ATTRIBUTES.UDT_NAME)
.fetch(mapping(Identifier::new))) {
SchemaDefinition schema = getSchema(udt.schema);
if (schema != null)
result.add(new PostgresUDTDefinition(schema, udt.name, null));
}
}
return result;
}
@Override
protected List getArrays0() throws SQLException {
List result = new ArrayList<>();
return result;
}
@Override
protected List getRoutines0() throws SQLException {
List result = new ArrayList<>();
if (!canUseRoutines())
return result;
Routines r1 = ROUTINES.as("r1");
PgType retT = PG_TYPE.as("rett");
// [#7785] The pg_proc.proisagg column has been replaced incompatibly in PostgreSQL 11
Field isAgg = (is11()
? PG_PROC.PROKIND.eq(inline("a"))
: field("{0}.proisagg", SQLDataType.BOOLEAN, PG_PROC)
).as("is_agg");
return
create().select(
r1.ROUTINE_SCHEMA,
r1.ROUTINE_NAME,
r1.SPECIFIC_NAME,
r1.ROUTINE_TYPE,
when(r1.DATA_TYPE.eq(inline("USER-DEFINED")).and(r1.TYPE_UDT_NAME.eq(inline("geometry"))), inline("geometry"))
// Ignore the data type when there is at least one out parameter
.when(canCombineArrays()
? condition("{0} && ARRAY['o','b']::\"char\"[]", PG_PROC.PROARGMODES)
: falseCondition(), inline("void"))
.when(r1.DATA_TYPE.eq(inline("ARRAY")), substring(r1.TYPE_UDT_NAME, inline(2)).concat(inline(" ARRAY")))
.else_(r1.DATA_TYPE).as("data_type"),
r1.CHARACTER_MAXIMUM_LENGTH,
// [#12048] TODO: Maintain whether we know the precision or not
when(r1.NUMERIC_PRECISION.isNull().and(r1.DATA_TYPE.in(
inline("time"),
inline("timetz"),
inline("time without time zone"),
inline("time with time zone"),
inline("timestamp"),
inline("timestamptz"),
inline("timestamp without time zone"),
inline("timestamp with time zone"))), inline(6))
.else_(r1.NUMERIC_PRECISION).as(r1.NUMERIC_PRECISION),
r1.NUMERIC_SCALE,
r1.TYPE_UDT_SCHEMA,
when(r1.DATA_TYPE.eq(inline("ARRAY")), substring(r1.TYPE_UDT_NAME, inline(2)))
.else_(r1.TYPE_UDT_NAME).as(r1.TYPE_UDT_NAME),
// Calculate overload index if applicable
when(
count().over(partitionBy(r1.ROUTINE_SCHEMA, r1.ROUTINE_NAME)).gt(one()),
rowNumber().over(partitionBy(r1.ROUTINE_SCHEMA, r1.ROUTINE_NAME).orderBy(
// [#9754] To stabilise overload calculation, we use the type signature
// replace(field("pg_get_function_arguments({0})", VARCHAR, PG_PROC.OID), inline('"'), inline("")),
r1.SPECIFIC_NAME
))
).as("overload"),
isAgg)
.from(r1)
// [#3375] Exclude table-valued functions as they're already generated as tables
.join(PG_PROC)
.on(PG_PROC.pgNamespace().NSPNAME.eq(r1.SPECIFIC_SCHEMA))
.and(nameconcatoid(r1))
.leftJoin(retT).on(PG_PROC.PRORETTYPE.eq(retT.OID))
.where(r1.ROUTINE_SCHEMA.in(getInputSchemata()))
.and(tableValuedFunctions()
? condition(not(PG_PROC.PRORETSET))
: noCondition())
.and(!getIncludeTriggerRoutines()
? r1.DATA_TYPE.isDistinctFrom(inline("trigger"))
: noCondition())
.orderBy(
r1.ROUTINE_SCHEMA.asc(),
r1.ROUTINE_NAME.asc(),
field(name("overload")).asc())
.collect(mapping(r -> new PostgresRoutineDefinition(this, r), Collectors.toList()));
}
protected Condition nameconcatoid(Routines r1) {
return is12()
? condition("nameconcatoid({0}, {1}) = {2}", PG_PROC.PRONAME, PG_PROC.OID, r1.SPECIFIC_NAME)
: PG_PROC.PRONAME.concat("_").concat(PG_PROC.OID).eq(r1.SPECIFIC_NAME);
}
@Override
protected List getPackages0() throws SQLException {
List result = new ArrayList<>();
return result;
}
@Override
protected DSLContext create0() {
return DSL.using(getConnection(), SQLDialect.POSTGRES);
}
boolean is84() {
if (is84 == null) {
is84 = configuredDialectIsNotFamilyAndSupports(asList(POSTGRES), () -> {
// [#2916] Window functions were introduced with PostgreSQL 9.0
try {
create(true)
.select(count().over())
.fetch();
return true;
}
catch (DataAccessException e) {
return false;
}
});
}
return is84;
}
boolean is94() {
// [#4254] INFORMATION_SCHEMA.PARAMETERS.PARAMETER_DEFAULT was added
// in PostgreSQL 9.4 only
if (is94 == null)
is94 = configuredDialectIsNotFamilyAndSupports(asList(POSTGRES), () -> exists(PARAMETERS.PARAMETER_DEFAULT));
return is94;
}
boolean is10() {
// [#7785] pg_sequence was added in PostgreSQL 10 only
if (is10 == null)
is10 = configuredDialectIsNotFamilyAndSupports(asList(POSTGRES), () -> exists(PG_SEQUENCE.SEQRELID));
return is10;
}
boolean is11() {
// [#7785] pg_proc.prokind was added in PostgreSQL 11 only, and
// pg_proc.proisagg was removed, incompatibly
if (is11 == null)
is11 = configuredDialectIsNotFamilyAndSupports(asList(POSTGRES), () -> exists(PG_PROC.PROKIND));
return is11;
}
boolean is12() {
// [#11325] column_column_usage was added in PostgreSQL 12 only
if (is12 == null)
is12 = configuredDialectIsNotFamilyAndSupports(asList(POSTGRES), () -> exists(table("column_column_usage")));
return is12;
}
@Override
protected boolean exists0(TableField field) {
return exists1(field, COLUMNS, COLUMNS.TABLE_SCHEMA, COLUMNS.TABLE_NAME, COLUMNS.COLUMN_NAME);
}
@Override
protected boolean exists0(Table table) {
return exists1(table, TABLES, TABLES.TABLE_SCHEMA, TABLES.TABLE_NAME);
}
boolean canCombineArrays() {
if (canCombineArrays == null) {
// [#7270] The ARRAY && ARRAY operator is not implemented in all PostgreSQL
// style databases, e.g. CockroachDB
try {
create(true).select(field("array[1, 2] && array[2, 3]")).fetch();
canCombineArrays = true;
}
catch (DataAccessException e) {
canCombineArrays = false;
}
}
return canCombineArrays;
}
boolean canUseTupleInPredicates() {
if (canUseTupleInPredicates == null) {
// [#7270] The tuple in predicate is not implemented in all PostgreSQL
// style databases, e.g. CockroachDB
// [#8072] Some database versions might support in, but not not in
try {
create(true).select(field("(1, 2) in (select 1, 2)")).fetch();
create(true).select(field("(1, 2) not in (select 1, 2)")).fetch();
canUseTupleInPredicates = true;
}
catch (DataAccessException e) {
canUseTupleInPredicates = false;
}
}
return canUseTupleInPredicates;
}
protected boolean canUseRoutines() {
// [#7892] The information_schema.routines table is not available in all PostgreSQL
// style databases, e.g. CockroachDB
if (canUseRoutines == null)
canUseRoutines = existAll(ROUTINES, PG_PROC);
return canUseRoutines;
}
private List enumLabels(String nspname, String typname) {
// [#10821] Avoid the cast if pg_enum.enumsortorder is available (PG 9.1+)
Field orderBy =
exists(PG_ENUM.ENUMSORTORDER)
? PG_ENUM.ENUMSORTORDER
// [#9511] [#9917] Workaround for regression introduced by avoiding quoting names everywhere
: field("{0}::{1}", PG_ENUM.ENUMLABEL, sql(name(nspname, typname) + ""));
if (canCastToEnumType == null) {
// [#2917] Older versions of PostgreSQL don't support the above cast
try {
canCastToEnumType = true;
return enumLabels(nspname, typname, orderBy);
}
catch (DataAccessException e) {
canCastToEnumType = false;
}
}
return canCastToEnumType ? enumLabels(nspname, typname, orderBy) : enumLabels(nspname, typname, PG_ENUM.ENUMLABEL);
}
private List enumLabels(String nspname, String typname, Field orderBy) {
return
create().select(PG_ENUM.ENUMLABEL)
.from(PG_ENUM)
.where(PG_ENUM.pgType().pgNamespace().NSPNAME.eq(nspname))
.and(PG_ENUM.pgType().TYPNAME.eq(typname))
.orderBy(orderBy)
.collect(intoList());
}
/**
* Translate the DATA_TYPE = 'ARRAY' to the UDT_NAME in standard SQL form,
* for multi dimensional arrays.
*/
Field arrayDataType(Field dataType, Field udtName, Field dims) {
return when(dataType.eq(inline("ARRAY")),
substring(udtName, inline(2))
.concat(DSL.repeat(inline(" ARRAY"), greatest(coalesce(dims, inline(0)), inline(1)))))
.else_(dataType);
}
/**
* Translate the DATA_TYPE = 'ARRAY' to the UDT_NAME in standard SQL form,
* for single dimensional arrays.
*/
Field arrayDataType(Field dataType, Field udtName) {
return when(dataType.eq(inline("ARRAY")), substring(udtName, inline(2)).concat(inline(" ARRAY")))
.else_(dataType);
}
/**
* Translate the UDT_NAME to the base type.
*/
Field arrayUdtName(Field dataType, Field udtName) {
return when(dataType.eq(inline("ARRAY")), substring(udtName, inline(2)))
.else_(udtName);
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy