import java.util.List;
import java.util.Set;
import java.util.function.Consumer;
+import java.util.function.IntFunction;
import javax.annotation.CheckForNull;
import javax.annotation.Nullable;
import org.sonar.api.utils.log.Logger;
*/
private static final String[] TABLE_TYPE = {"TABLE"};
+ protected DatabaseUtils() {
+ throw new IllegalStateException("Utility class");
+ }
+
public static void closeQuietly(@Nullable Connection connection) {
if (connection != null) {
try {
* and with MsSQL when there's more than 2000 parameters in a query
*/
public static <OUTPUT, INPUT extends Comparable<INPUT>> List<OUTPUT> executeLargeInputs(Collection<INPUT> input, Function<List<INPUT>, List<OUTPUT>> function) {
- return executeLargeInputs(input, function, size -> size == 0 ? Collections.emptyList() : new ArrayList<>(size));
+ return executeLargeInputs(input, function, i -> i);
}
- public static <OUTPUT, INPUT extends Comparable<INPUT>> Set<OUTPUT> executeLargeInputsIntoSet(Collection<INPUT> input, Function<List<INPUT>, Set<OUTPUT>> function) {
- return executeLargeInputs(input, function, size -> size == 0 ? Collections.emptySet() : new HashSet<>(size));
+ /**
+ * Partition by 1000 elements a list of input and execute a function on each part.
+ *
+ * The goal is to prevent issue with ORACLE when there's more than 1000 elements in a 'in ('X', 'Y', ...)'
+ * and with MsSQL when there's more than 2000 parameters in a query
+ */
+ public static <OUTPUT, INPUT extends Comparable<INPUT>> List<OUTPUT> executeLargeInputs(Collection<INPUT> input, Function<List<INPUT>, List<OUTPUT>> function,
+ IntFunction<Integer> partitionSizeManipulations) {
+ return executeLargeInputs(input, function, size -> size == 0 ? Collections.emptyList() : new ArrayList<>(size), partitionSizeManipulations);
+ }
+
+ public static <OUTPUT, INPUT extends Comparable<INPUT>> Set<OUTPUT> executeLargeInputsIntoSet(Collection<INPUT> input, Function<List<INPUT>, Set<OUTPUT>> function,
+ IntFunction<Integer> partitionSizeManipulations) {
+ return executeLargeInputs(input, function, size -> size == 0 ? Collections.emptySet() : new HashSet<>(size), partitionSizeManipulations);
}
private static <OUTPUT, INPUT extends Comparable<INPUT>, RESULT extends Collection<OUTPUT>> RESULT executeLargeInputs(Collection<INPUT> input,
- Function<List<INPUT>, RESULT> function, java.util.function.Function<Integer, RESULT> outputInitializer) {
+ Function<List<INPUT>, RESULT> function, java.util.function.Function<Integer, RESULT> outputInitializer, IntFunction<Integer> partitionSizeManipulations) {
if (input.isEmpty()) {
return outputInitializer.apply(0);
}
RESULT results = outputInitializer.apply(input.size());
- for (List<INPUT> partition : toUniqueAndSortedPartitions(input)) {
+ for (List<INPUT> partition : toUniqueAndSortedPartitions(input, partitionSizeManipulations)) {
RESULT subResults = function.apply(partition);
if (subResults != null) {
results.addAll(subResults);
* and with MsSQL when there's more than 2000 parameters in a query
*/
public static <INPUT extends Comparable<INPUT>> void executeLargeUpdates(Collection<INPUT> inputs, Consumer<List<INPUT>> consumer) {
- Iterable<List<INPUT>> partitions = toUniqueAndSortedPartitions(inputs);
+ executeLargeUpdates(inputs, consumer, i -> i);
+ }
+
+ /**
+ * Partition by 1000 elements a list of input and execute a consumer on each part.
+ *
+ * The goal is to prevent issue with ORACLE when there's more than 1000 elements in a 'in ('X', 'Y', ...)'
+ * and with MsSQL when there's more than 2000 parameters in a query
+ *
+ * @param inputs the whole list of elements to be partitioned
+ * @param consumer the mapper method to be executed, for example {@code mapper(dbSession)::selectByUuids}
+ * @param partitionSizeManipulations the function that computes the number of usages of a partition, for example
+ * {@code partitionSize -> partitionSize / 2} when the partition of elements
+ * in used twice in the SQL request.
+ */
+ public static <INPUT extends Comparable<INPUT>> void executeLargeUpdates(Collection<INPUT> inputs, Consumer<List<INPUT>> consumer,
+ IntFunction<Integer> partitionSizeManipulations) {
+ Iterable<List<INPUT>> partitions = toUniqueAndSortedPartitions(inputs, partitionSizeManipulations);
for (List<INPUT> partition : partitions) {
consumer.accept(partition);
}
* The goal is to prevent issue with ORACLE when there's more than 1000 elements in a 'in ('X', 'Y', ...)'
* and with MsSQL when there's more than 2000 parameters in a query
*
+ * @param inputs the whole list of elements to be partitioned
* @param sqlCaller a {@link Function} which calls the SQL update/delete and returns the number of updated/deleted rows.
- *
+ * @param partitionSizeManipulations the function that computes the number of usages of a partition, for example
+ * {@code partitionSize -> partitionSize / 2} when the partition of elements
+ * in used twice in the SQL request.
* @return the total number of updated/deleted rows (computed as the sum of the values returned by {@code sqlCaller}).
*/
- public static <INPUT extends Comparable<INPUT>> int executeLargeUpdates(Collection<INPUT> inputs, Function<List<INPUT>, Integer> sqlCaller) {
- Iterable<List<INPUT>> partitions = toUniqueAndSortedPartitions(inputs);
+ public static <INPUT extends Comparable<INPUT>> int executeLargeUpdates(Collection<INPUT> inputs, Function<List<INPUT>, Integer> sqlCaller,
+ IntFunction<Integer> partitionSizeManipulations) {
+ Iterable<List<INPUT>> partitions = toUniqueAndSortedPartitions(inputs, partitionSizeManipulations);
Integer res = 0;
for (List<INPUT> partition : partitions) {
res += sqlCaller.apply(partition);
* Ensure values {@code inputs} are unique (which avoids useless arguments) and sorted before creating the partition.
*/
public static <INPUT extends Comparable<INPUT>> Iterable<List<INPUT>> toUniqueAndSortedPartitions(Collection<INPUT> inputs) {
- return Iterables.partition(toUniqueAndSortedList(inputs), PARTITION_SIZE_FOR_ORACLE);
+ return toUniqueAndSortedPartitions(inputs, i -> i);
+ }
+
+ /**
+ * Ensure values {@code inputs} are unique (which avoids useless arguments) and sorted before creating the partition.
+ */
+ public static <INPUT extends Comparable<INPUT>> Iterable<List<INPUT>> toUniqueAndSortedPartitions(Collection<INPUT> inputs, IntFunction<Integer> partitionSizeManipulations) {
+ int partitionSize = partitionSizeManipulations.apply(PARTITION_SIZE_FOR_ORACLE);
+ return Iterables.partition(toUniqueAndSortedList(inputs), partitionSize);
}
/**
*/
package org.sonar.db.permission;
+import java.util.Collection;
import java.util.Collections;
+import java.util.List;
import java.util.Random;
import java.util.Set;
+import java.util.stream.Collectors;
import java.util.stream.IntStream;
import org.junit.Before;
import org.junit.Rule;
.isEmpty();
}
+ @Test
+ public void keepAuthorizedProjectIds_should_be_able_to_handle_lots_of_projects() {
+ List<ComponentDto> projects = IntStream.range(0, 2000).mapToObj(i -> db.components().insertPublicProject(organization)).collect(Collectors.toList());
+
+ Collection<Long> ids = projects.stream().map(ComponentDto::getId).collect(Collectors.toSet());
+ assertThat(underTest.keepAuthorizedProjectIds(dbSession, ids, null, UserRole.USER))
+ .containsOnly(ids.toArray(new Long[0]));
+ }
+
+ @Test
+ public void keepAuthorizedProjectUuids_should_be_able_to_handle_lots_of_projects() {
+ List<ComponentDto> projects = IntStream.range(0, 2000).mapToObj(i -> db.components().insertPublicProject(organization)).collect(Collectors.toList());
+
+ Collection<String> uuids = projects.stream().map(ComponentDto::uuid).collect(Collectors.toSet());
+ assertThat(underTest.keepAuthorizedProjectUuids(dbSession, uuids, null, UserRole.USER))
+ .containsOnly(uuids.toArray(new String[0]));
+ }
+
@Test
public void keepAuthorizedUsersForRoleAndProject_returns_empty_if_user_set_is_empty_on_public_project() {
OrganizationDto organization = db.organizations().insert();
newHashSet(100, 101, 102), "user", PROJECT_ID)).isEmpty();
}
+ @Test
+ public void keepAuthorizedUsersForRoleAndProject_should_be_able_to_handle_lots_of_users() {
+ List<UserDto> users = IntStream.range(0, 2000).mapToObj(i -> db.users().insertUser()).collect(Collectors.toList());
+
+ assertThat(underTest.keepAuthorizedUsersForRoleAndProject(dbSession,
+ users.stream().map(UserDto::getId).collect(Collectors.toSet()), "user", PROJECT_ID)).isEmpty();
+ }
+
+
@Test
public void countUsersWithGlobalPermissionExcludingGroupMember() {
// u1 has the direct permission, u2 and u3 have the permission through their group