/**
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You 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
*
* http://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.
*/
package org.apache.mahout.cf.taste.hadoop.item;
import java.io.IOException;
import java.io.InputStream;
import java.util.HashMap;
import java.util.Map;
import java.util.regex.Pattern;
import com.google.common.collect.Maps;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.mahout.cf.taste.impl.common.FastIDSet;
import org.apache.mahout.common.HadoopUtil;
import org.apache.mahout.common.iterator.FileLineIterable;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.google.common.io.Closeables;
/**
* Reads user ids and item ids from files specified in usersFile, itemsFile or userItemFile options in item-based
* recommender. Composes a list of users and a list of items which can be used by
* {@link org.apache.mahout.cf.taste.hadoop.item.UserVectorSplitterMapper} and
* {@link org.apache.mahout.cf.taste.hadoop.item.AggregateAndRecommendReducer}.
*/
public class IDReader {
static final String USER_ITEM_FILE = "userItemFile";
private static final Logger log = LoggerFactory.getLogger(IDReader.class);
private static final Pattern SEPARATOR = Pattern.compile("[\t,]");
private Configuration conf;
private String usersFile;
private String itemsFile;
private String userItemFile;
private FastIDSet userIds;
private FastIDSet itemIds;
private FastIDSet emptySet;
/* Key - user id, value - a set of item ids to include into recommendations for this user */
private Map<Long, FastIDSet> userItemFilter;
/**
* Creates a new IDReader
*
* @param conf Job configuration
*/
public IDReader(Configuration conf) {
this.conf = conf;
emptySet = new FastIDSet();
usersFile = conf.get(UserVectorSplitterMapper.USERS_FILE);
itemsFile = conf.get(AggregateAndRecommendReducer.ITEMS_FILE);
userItemFile = conf.get(USER_ITEM_FILE);
}
/**
* Reads user ids and item ids from files specified in a job configuration
*
* @throws IOException if an error occurs during file read operation
*
* @throws IllegalStateException if userItemFile option is specified together with usersFile or itemsFile
*/
public void readIDs() throws IOException, IllegalStateException {
if (isUserItemFileSpecified()) {
readUserItemFilterIfNeeded();
}
if (isUsersFileSpecified() || isUserItemFilterSpecified()) {
readUserIds();
}
if (isItemsFileSpecified() || isUserItemFilterSpecified()) {
readItemIds();
}
}
/**
* Gets a collection of items which should be recommended for a user
*
* @param userId ID of a user we are interested in
* @return if a userItemFile option is specified, and that file contains at least one item ID for the user,
* then this method returns a {@link FastIDSet} object populated with item IDs. Otherwise, this
* method returns an empty set.
*/
public FastIDSet getItemsToRecommendForUser(Long userId) {
if (isUserItemFilterSpecified() && userItemFilter.containsKey(userId)) {
return userItemFilter.get(userId);
} else {
return emptySet;
}
}
private void readUserIds() throws IOException, IllegalStateException {
if (isUsersFileSpecified() && !isUserItemFileSpecified()) {
userIds = readIDList(usersFile);
} else if (isUserItemFileSpecified() && !isUsersFileSpecified()) {
readUserItemFilterIfNeeded();
userIds = extractAllUserIdsFromUserItemFilter(userItemFilter);
} else if (!isUsersFileSpecified() && !isUserItemFileSpecified()) {
throw new IllegalStateException("Neither usersFile nor userItemFile options are specified");
} else {
throw new IllegalStateException("usersFile and userItemFile options cannot be used simultaneously");
}
}
private void readItemIds() throws IOException, IllegalStateException {
if (isItemsFileSpecified() && !isUserItemFileSpecified()) {
itemIds = readIDList(itemsFile);
} else if (isUserItemFileSpecified() && !isItemsFileSpecified()) {
readUserItemFilterIfNeeded();
itemIds = extractAllItemIdsFromUserItemFilter(userItemFilter);
} else if (!isItemsFileSpecified() && !isUserItemFileSpecified()) {
throw new IllegalStateException("Neither itemsFile nor userItemFile options are specified");
} else {
throw new IllegalStateException("itemsFile and userItemFile options cannot be specified simultaneously");
}
}
private void readUserItemFilterIfNeeded() throws IOException {
if (!isUserItemFilterSpecified() && isUserItemFileSpecified()) {
userItemFilter = readUserItemFilter(userItemFile);
}
}
private Map<Long, FastIDSet> readUserItemFilter(String pathString) throws IOException {
Map<Long, FastIDSet> result = Maps.newHashMap();
InputStream in = openFile(pathString);
try {
for (String line : new FileLineIterable(in)) {
try {
String[] tokens = SEPARATOR.split(line.toString());
Long userId = Long.parseLong(tokens[0]);
Long itemId = Long.parseLong(tokens[1]);
addUserAndItemIdToUserItemFilter(result, userId, itemId);
} catch (NumberFormatException nfe) {
log.warn("userItemFile line ignored: {}", line);
}
}
} finally {
Closeables.close(in, true);
}
return result;
}
void addUserAndItemIdToUserItemFilter(Map<Long, FastIDSet> filter, Long userId, Long itemId) {
FastIDSet itemIds;
if (filter.containsKey(userId)) {
itemIds = filter.get(userId);
} else {
itemIds = new FastIDSet();
filter.put(userId, itemIds);
}
itemIds.add(itemId);
}
static FastIDSet extractAllUserIdsFromUserItemFilter(Map<Long, FastIDSet> filter) {
FastIDSet result = new FastIDSet();
for (Long userId : filter.keySet()) {
result.add(userId);
}
return result;
}
private FastIDSet extractAllItemIdsFromUserItemFilter(Map<Long, FastIDSet> filter) {
FastIDSet result = new FastIDSet();
for (FastIDSet itemIds : filter.values()) {
result.addAll(itemIds);
}
return result;
}
private FastIDSet readIDList(String pathString) throws IOException {
FastIDSet result = null;
if (pathString != null) {
result = new FastIDSet();
InputStream in = openFile(pathString);
try {
for (String line : new FileLineIterable(in)) {
try {
result.add(Long.parseLong(line));
} catch (NumberFormatException nfe) {
log.warn("line ignored: {}", line);
}
}
} finally {
Closeables.close(in, true);
}
}
return result;
}
private InputStream openFile(String pathString) throws IOException {
return HadoopUtil.openStream(new Path(pathString), conf);
}
public boolean isUsersFileSpecified () {
return usersFile != null;
}
public boolean isItemsFileSpecified () {
return itemsFile != null;
}
public boolean isUserItemFileSpecified () {
return userItemFile != null;
}
public boolean isUserItemFilterSpecified() {
return userItemFilter != null;
}
public FastIDSet getUserIds() {
return userIds;
}
public FastIDSet getItemIds() {
return itemIds;
}
}