/**
* Returns true if the given field is a member of any primary key on the given class, for the
* given source.
*
* @param model the Model in which to find ClassDescriptors
* @param clazz the Class in which to look
* @param fieldName the name of the field to check
* @param source the Source that the keys belong to
* @return true if the field is a primary key
*/
public static boolean fieldIsPrimaryKey(
Model model, Class<?> clazz, String fieldName, Source source) {
Map<PrimaryKeyCacheKey, Set<String>> cache = primaryKeyCache.get();
PrimaryKeyCacheKey key = new PrimaryKeyCacheKey(model, clazz, source);
Set<String> fields = cache.get(key);
if (fields == null) {
fields = new HashSet<String>();
for (ClassDescriptor cld : model.getClassDescriptorsForClass(clazz)) {
for (PrimaryKey pk : getPrimaryKeys(cld, source, null)) {
fields.addAll(pk.getFieldNames());
}
}
cache.put(key, fields);
}
return fields.contains(fieldName);
}
/**
* Returns whether this primary key can be fetched now.
*
* @param pk the PrimaryKey
* @param cld the ClassDescriptor that the PrimaryKey is in
* @param pksNotDone a Map of pks not yet fetched
* @return a boolean
*/
protected boolean canDoPkNow(
PrimaryKey pk, ClassDescriptor cld, Map<PrimaryKey, ClassDescriptor> pksNotDone) {
boolean canDoPkNow = true;
Iterator<String> fieldNameIter = pk.getFieldNames().iterator();
while (fieldNameIter.hasNext() && canDoPkNow) {
String fieldName = fieldNameIter.next();
FieldDescriptor fd = cld.getFieldDescriptorByName(fieldName);
if (fd.isReference()) {
Iterator<ClassDescriptor> otherCldIter = pksNotDone.values().iterator();
while (otherCldIter.hasNext() && canDoPkNow) {
ClassDescriptor otherCld = otherCldIter.next();
Class<? extends FastPathObject> fieldClass =
((ReferenceDescriptor) fd).getReferencedClassDescriptor().getType();
if (otherCld.getType().isAssignableFrom(fieldClass)
|| fieldClass.isAssignableFrom(otherCld.getType())) {
canDoPkNow = false;
}
}
}
}
return canDoPkNow;
}
/**
* Fetches equivalent objects for a particular primary key.
*
* @param pk the PrimaryKey
* @param cld the ClassDescriptor of the PrimaryKey
* @param results a Map to hold results that are to be added to the cache
* @param objectsForCld a List of objects relevant to this PrimaryKey
* @param fetchedObjectIds a Set to hold ids of objects that are fetched, to prefetch from the
* data tracker later
* @throws ObjectStoreException if something goes wrong
*/
protected void doPk(
PrimaryKey pk,
ClassDescriptor cld,
Map<InterMineObject, Set<InterMineObject>> results,
List<InterMineObject> objectsForCld,
Set<Integer> fetchedObjectIds)
throws ObjectStoreException {
Iterator<InterMineObject> objectsForCldIter = objectsForCld.iterator();
while (objectsForCldIter.hasNext()) {
int objCount = 0;
int origObjCount = 0;
Query q = new Query();
QueryClass qc = new QueryClass(cld.getType());
q.addFrom(qc);
q.addToSelect(qc);
ConstraintSet cs = new ConstraintSet(ConstraintOp.AND);
q.setConstraint(cs);
Map<String, Set<Object>> fieldNameToValues = new HashMap<String, Set<Object>>();
for (String fieldName : pk.getFieldNames()) {
try {
QueryField qf = new QueryField(qc, fieldName);
q.addToSelect(qf);
Set<Object> values = new HashSet<Object>();
fieldNameToValues.put(fieldName, values);
cs.addConstraint(new BagConstraint(qf, ConstraintOp.IN, values));
} catch (IllegalArgumentException e) {
QueryForeignKey qf = new QueryForeignKey(qc, fieldName);
q.addToSelect(qf);
Set<Object> values = new HashSet<Object>();
fieldNameToValues.put(fieldName, values);
cs.addConstraint(new BagConstraint(qf, ConstraintOp.IN, values));
}
}
// Now make a map from the primary key values to source objects
Map<List<Object>, InterMineObject> keysToSourceObjects =
new HashMap<List<Object>, InterMineObject>();
while (objectsForCldIter.hasNext() && (objCount < 500)) {
InterMineObject object = objectsForCldIter.next();
origObjCount++;
try {
if (DataLoaderHelper.objectPrimaryKeyNotNull(model, object, cld, pk, source, idMap)) {
List<Collection<Object>> values = new ArrayList<Collection<Object>>();
boolean skipObject = false;
Map<String, Set<Object>> fieldsValues = new HashMap<String, Set<Object>>();
for (String fieldName : pk.getFieldNames()) {
try {
Object value = object.getFieldProxy(fieldName);
Set<Object> fieldValues;
if (value instanceof InterMineObject) {
Integer id = idMap.get(((InterMineObject) value).getId());
if (id == null) {
Set<InterMineObject> eqs = results.get(value);
if (eqs == null) {
value = object.getFieldValue(fieldName);
eqs = queryEquivalentObjects((InterMineObject) value, source);
}
fieldValues = new HashSet<Object>();
for (InterMineObject obj : eqs) {
fieldValues.add(obj.getId());
}
} else {
fieldValues = Collections.singleton((Object) id);
}
} else {
fieldValues = Collections.singleton(value);
}
values.add(fieldValues);
fieldsValues.put(fieldName, fieldValues);
for (Object fieldValue : fieldValues) {
long time = System.currentTimeMillis();
boolean pkQueryFruitless =
hints.pkQueryFruitless(cld.getType(), fieldName, fieldValue);
String summaryName = Util.getFriendlyName(cld.getType()) + "." + fieldName;
if (!savedTimes.containsKey(summaryName)) {
savedTimes.put(summaryName, new Long(System.currentTimeMillis() - time));
savedCounts.put(summaryName, new Integer(0));
}
if (pkQueryFruitless) {
skipObject = true;
}
}
} catch (IllegalAccessException e) {
throw new RuntimeException(e);
}
}
if (!skipObject) {
objCount++;
for (String fieldName : pk.getFieldNames()) {
fieldNameToValues.get(fieldName).addAll(fieldsValues.get(fieldName));
}
for (List<Object> valueSet : CollectionUtil.fanOutCombinations(values)) {
if (keysToSourceObjects.containsKey(valueSet)) {
throw new ObjectStoreException(
"Duplicate objects found for pk "
+ cld.getName()
+ "."
+ pk.getName()
+ ": "
+ object);
}
keysToSourceObjects.put(valueSet, object);
}
}
}
} catch (MetaDataException e) {
throw new ObjectStoreException(e);
}
}
// Prune BagConstraints using the hints system.
// boolean emptyQuery = false;
// Iterator<String> fieldNameIter = pk.getFieldNames().iterator();
// while (fieldNameIter.hasNext() && (!emptyQuery)) {
// String fieldName = fieldNameIter.next();
// Set values = fieldNameToValues.get(fieldName);
// Iterator valueIter = values.iterator();
// while (valueIter.hasNext()) {
// if (hints.pkQueryFruitless(cld.getType(), fieldName, valueIter.next())) {
// valueIter.remove();
// }
// }
// if (values.isEmpty()) {
// emptyQuery = true;
// }
// }
if (objCount > 0) {
// Iterate through query, and add objects to results
// long time = System.currentTimeMillis();
int matches = 0;
Results res = lookupOs.execute(q, 2000, false, false, false);
@SuppressWarnings("unchecked")
List<ResultsRow<Object>> tmpRes = (List) res;
for (ResultsRow<Object> row : tmpRes) {
List<Object> values = new ArrayList<Object>();
for (int i = 1; i <= pk.getFieldNames().size(); i++) {
values.add(row.get(i));
}
Set<InterMineObject> set = results.get(keysToSourceObjects.get(values));
if (set != null) {
set.add((InterMineObject) row.get(0));
matches++;
}
fetchedObjectIds.add(((InterMineObject) row.get(0)).getId());
}
// LOG.info("Fetched " + res.size() + " equivalent objects for " + objCount
// + " objects in " + (System.currentTimeMillis() - time) + " ms for "
// + cld.getName() + "." + pk.getName());
}
}
}
/**
* Look a the values of the given primary key in the object and return true if and only if some
* part of the primary key is null. If the primary key contains a reference it is sufficient for
* any of the primary keys of the referenced object to be non-null (ie objectPrimaryKeyIsNull()
* returning true).
*
* @param model the Model in which to find ClassDescriptors
* @param obj the Object to check
* @param cld one of the classes that obj is. Only primary keys for this classes will be checked
* @param pk the primary key to check
* @param source the Source database
* @param idMap an IntToIntMap from source IDs to destination IDs
* @return true if the the given primary key is non-null for the given object
* @throws MetaDataException if anything goes wrong
*/
public static boolean objectPrimaryKeyNotNull(
Model model,
InterMineObject obj,
ClassDescriptor cld,
PrimaryKey pk,
Source source,
IntToIntMap idMap)
throws MetaDataException {
for (String fieldName : pk.getFieldNames()) {
FieldDescriptor fd = cld.getFieldDescriptorByName(fieldName);
if (fd instanceof AttributeDescriptor) {
Object value;
try {
value = obj.getFieldValue(fieldName);
} catch (IllegalAccessException e) {
throw new MetaDataException(
"Failed to get field " + fieldName + " for key " + pk + " from " + obj, e);
}
if (value == null) {
return false;
}
} else if (fd instanceof CollectionDescriptor) {
throw new MetaDataException(
"Primary key "
+ pk.getName()
+ " for class "
+ cld.getName()
+ " cannot contain collection "
+ fd.getName()
+ ": collections cannot be part of a primary key. Please edit"
+ model.getName()
+ "_keyDefs.properties");
} else if (fd instanceof ReferenceDescriptor) {
InterMineObject refObj;
try {
refObj = (InterMineObject) obj.getFieldProxy(fieldName);
} catch (IllegalAccessException e) {
throw new MetaDataException(
"Failed to get field " + fieldName + " for key " + pk + " from " + obj, e);
}
if (refObj == null) {
return false;
}
if ((refObj.getId() != null) && (idMap.get(refObj.getId()) != null)) {
// We have previously loaded the object in this reference.
continue;
}
if (refObj instanceof ProxyReference) {
refObj = ((ProxyReference) refObj).getObject();
}
boolean foundNonNullKey = false;
boolean foundKey = false;
Set<ClassDescriptor> classDescriptors =
model.getClassDescriptorsForClass(refObj.getClass());
CLDS:
for (ClassDescriptor refCld : classDescriptors) {
Set<PrimaryKey> primaryKeys;
if (source == null) {
primaryKeys =
new LinkedHashSet<PrimaryKey>(PrimaryKeyUtil.getPrimaryKeys(refCld).values());
} else {
primaryKeys = DataLoaderHelper.getPrimaryKeys(refCld, source, null);
}
for (PrimaryKey refPK : primaryKeys) {
foundKey = true;
if (objectPrimaryKeyNotNull(model, refObj, refCld, refPK, source, idMap)) {
foundNonNullKey = true;
break CLDS;
}
}
}
if (foundKey && (!foundNonNullKey)) {
return false;
}
}
}
return true;
}
/**
* Return a Set of PrimaryKeys relevant to a given Source for a ClassDescriptor. The Set contains
* all the primary keys that exist on a particular class that are used by the source, without
* performing any recursion. The Model.getClassDescriptorsForClass() method is recommended if you
* wish for all the primary keys of the class' parents as well.
*
* @param cld the ClassDescriptor
* @param source the Source
* @param os the ObjectStore that these PrimaryKeys are used in, for creating indexes
* @return a Set of PrimaryKeys
*/
public static Set<PrimaryKey> getPrimaryKeys(ClassDescriptor cld, Source source, ObjectStore os) {
GetPrimaryKeyCacheKey key = new GetPrimaryKeyCacheKey(cld, source);
synchronized (getPrimaryKeyCache) {
Set<PrimaryKey> keySet = getPrimaryKeyCache.get(key);
if (keySet == null) {
keySet = new LinkedHashSet<PrimaryKey>();
Properties keys = getKeyProperties(source);
if (keys != null) {
if (!verifiedSources.contains(source)) {
String packageNameWithDot =
cld.getName().substring(0, cld.getName().lastIndexOf('.') + 1);
LOG.info(
"Verifying primary key config for source "
+ source
+ ", packageName = "
+ packageNameWithDot);
for (Map.Entry<Object, Object> entry : keys.entrySet()) {
String cldName = (String) entry.getKey();
String keyList = (String) entry.getValue();
if (!cldName.contains(".")) {
ClassDescriptor iCld =
cld.getModel().getClassDescriptorByName(packageNameWithDot + cldName);
if (iCld != null) {
Map<String, PrimaryKey> map = PrimaryKeyUtil.getPrimaryKeys(iCld);
String[] tokens = keyList.split(",");
for (int i = 0; i < tokens.length; i++) {
String token = tokens[i].trim();
if (map.get(token) == null) {
throw new IllegalArgumentException(
"Primary key "
+ token
+ " for class "
+ cldName
+ " required by datasource "
+ source.getName()
+ " in "
+ source.getName()
+ "_keys.properties is not defined in "
+ cld.getModel().getName()
+ "_keyDefs.properties");
}
}
} else {
LOG.warn(
"Ignoring entry for "
+ cldName
+ " in file "
+ cld.getModel().getName()
+ "_keyDefs.properties - not in model!");
}
}
}
verifiedSources.add(source);
}
Map<String, PrimaryKey> map = PrimaryKeyUtil.getPrimaryKeys(cld);
String cldName = TypeUtil.unqualifiedName(cld.getName());
String keyList = (String) keys.get(cldName);
if (keyList != null) {
String[] tokens = keyList.split(",");
for (int i = 0; i < tokens.length; i++) {
String token = tokens[i].trim();
if (map.get(token) == null) {
throw new IllegalArgumentException(
"Primary key "
+ token
+ " for class "
+ cld.getName()
+ " required by data source "
+ source.getName()
+ " in "
+ source.getName()
+ "_keys.properties is not defined in "
+ cld.getModel().getName()
+ "_keyDefs.properties");
} else {
keySet.add(map.get(token));
}
}
}
for (Map.Entry<Object, Object> entry : keys.entrySet()) {
String propKey = (String) entry.getKey();
String fieldList = (String) entry.getValue();
int posOfDot = propKey.indexOf('.');
if (posOfDot > 0) {
String propCldName = propKey.substring(0, posOfDot);
if (cldName.equals(propCldName)) {
String keyName = propKey.substring(posOfDot + 1);
PrimaryKey pk = new PrimaryKey(keyName, fieldList, cld);
if (!keySet.contains(pk)) {
keySet.add(pk);
if (os instanceof ObjectStoreInterMineImpl) {
ObjectStoreInterMineImpl osimi = (ObjectStoreInterMineImpl) os;
DatabaseSchema schema = osimi.getSchema();
ClassDescriptor tableMaster = schema.getTableMaster(cld);
String tableName = DatabaseUtil.getTableName(tableMaster);
List<String> fields = new ArrayList<String>();
for (String field : pk.getFieldNames()) {
String colName = DatabaseUtil.generateSqlCompatibleName(field);
if (tableMaster.getReferenceDescriptorByName(field, true) != null) {
colName += "id";
}
fields.add(colName);
}
String sql =
"CREATE INDEX "
+ tableName
+ "__"
+ keyName
+ " ON "
+ tableName
+ " ("
+ StringUtil.join(fields, ", ")
+ ")";
System.out.println("Creating index: " + sql);
LOG.info("Creating index: " + sql);
Connection conn = null;
try {
conn = osimi.getConnection();
conn.createStatement().execute(sql);
} catch (SQLException e) {
LOG.warn("Index creation failed", e);
} finally {
if (conn != null) {
osimi.releaseConnection(conn);
}
}
}
}
}
}
}
} else {
throw new IllegalArgumentException(
"Unable to find keys for source "
+ source.getName()
+ " in file "
+ source.getName()
+ "_keys.properties");
}
getPrimaryKeyCache.put(key, keySet);
}
return keySet;
}
}
