/*
Copyright 2010, Google Inc.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above
copyright notice, this list of conditions and the following disclaimer
in the documentation and/or other materials provided with the
distribution.
* Neither the name of Google Inc. nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
package com.google.refine.model.recon;
import java.io.DataOutputStream;
import java.io.InputStream;
import java.io.StringWriter;
import java.net.URL;
import java.net.URLConnection;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Properties;
import java.util.Set;
import org.apache.commons.lang.StringUtils;
import org.json.JSONArray;
import org.json.JSONException;
import org.json.JSONObject;
import org.json.JSONWriter;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.google.refine.expr.ExpressionUtils;
import com.google.refine.model.Cell;
import com.google.refine.model.Project;
import com.google.refine.model.Recon;
import com.google.refine.model.Recon.Judgment;
import com.google.refine.model.ReconCandidate;
import com.google.refine.model.RecordModel.RowDependency;
import com.google.refine.model.Row;
import com.google.refine.util.ParsingUtilities;
public class StandardReconConfig extends ReconConfig {
final static Logger logger = LoggerFactory.getLogger("refine-standard-recon");
static public class ColumnDetail {
final public String columnName;
final public String propertyName;
final public String propertyID;
public ColumnDetail(String columnName, String propertyName, String propertyID) {
this.columnName = columnName;
this.propertyName = propertyName;
this.propertyID = propertyID;
}
}
static public ReconConfig reconstruct(JSONObject obj) throws Exception {
List<ColumnDetail> columnDetails = null;
if (obj.has("columnDetails")) {
JSONArray columnDetailsA = obj.getJSONArray("columnDetails");
int l = columnDetailsA.length();
columnDetails = new ArrayList<ColumnDetail>(l);
for (int i = 0; i < l; i++) {
JSONObject o = columnDetailsA.getJSONObject(i);
if (o.has("property")) { // legacy
JSONObject p = o.getJSONObject("property");
columnDetails.add(new ColumnDetail(
o.getString("column"),
p.has("name") ? p.getString("name") : null,
p.has("id") ? p.getString("id") : null
));
} else {
columnDetails.add(new ColumnDetail(
o.getString("column"),
o.has("propertyName") ? o.getString("propertyName") : null,
o.has("propertyID") ? o.getString("propertyID") : null
));
}
}
} else {
columnDetails = new ArrayList<ColumnDetail>();
}
JSONObject t = obj.has("type") && !obj.isNull("type") ? obj.getJSONObject("type") : null;
return new StandardReconConfig(
obj.getString("service"),
obj.has("identifierSpace") ? obj.getString("identifierSpace") : null,
obj.has("schemaSpace") ? obj.getString("schemaSpace") : null,
t == null ? null : t.getString("id"),
t == null ? null : (t.has("name") ? t.getString("name") : null),
obj.getBoolean("autoMatch"),
columnDetails
);
}
static protected class StandardReconJob extends ReconJob {
String text;
String code;
@Override
public int getKey() {
return code.hashCode();
}
}
final public String service;
final public String identifierSpace;
final public String schemaSpace;
final public String typeID;
final public String typeName;
final public boolean autoMatch;
final public List<ColumnDetail> columnDetails;
public StandardReconConfig(
String service,
String identifierSpace,
String schemaSpace,
String typeID,
String typeName,
boolean autoMatch,
List<ColumnDetail> columnDetails
) {
this.service = service;
this.identifierSpace = identifierSpace;
this.schemaSpace = schemaSpace;
this.typeID = typeID;
this.typeName = typeName;
this.autoMatch = autoMatch;
this.columnDetails = columnDetails;
}
@Override
public void write(JSONWriter writer, Properties options)
throws JSONException {
writer.object();
writer.key("mode"); writer.value("standard-service");
writer.key("service"); writer.value(service);
writer.key("identifierSpace"); writer.value(identifierSpace);
writer.key("schemaSpace"); writer.value(schemaSpace);
writer.key("type");
if (typeID == null) {
writer.value(null);
} else {
writer.object();
writer.key("id"); writer.value(typeID);
writer.key("name"); writer.value(typeName);
writer.endObject();
}
writer.key("autoMatch"); writer.value(autoMatch);
writer.key("columnDetails");
writer.array();
for (ColumnDetail c : columnDetails) {
writer.object();
writer.key("column"); writer.value(c.columnName);
writer.key("propertyName"); writer.value(c.propertyName);
writer.key("propertyID"); writer.value(c.propertyID);
writer.endObject();
}
writer.endArray();
writer.endObject();
}
@Override
public int getBatchSize() {
return 10;
}
@Override
public String getBriefDescription(Project project, String columnName) {
return "Reconcile cells in column " + columnName + " to type " + typeID;
}
@Override
public ReconJob createJob(Project project, int rowIndex, Row row,
String columnName, Cell cell) {
StandardReconJob job = new StandardReconJob();
try {
StringWriter stringWriter = new StringWriter();
JSONWriter jsonWriter = new JSONWriter(stringWriter);
jsonWriter.object();
jsonWriter.key("query"); jsonWriter.value(cell.value.toString());
if (typeID != null) {
jsonWriter.key("type"); jsonWriter.value(typeID);
jsonWriter.key("type_strict"); jsonWriter.value("should");
}
if (columnDetails.size() > 0) {
jsonWriter.key("properties");
jsonWriter.array();
for (ColumnDetail c : columnDetails) {
int detailCellIndex = project.columnModel.getColumnByName(c.columnName).getCellIndex();
Cell cell2 = row.getCell(detailCellIndex);
if (cell2 == null || !ExpressionUtils.isNonBlankData(cell2.value)) {
int cellIndex = project.columnModel.getColumnByName(columnName).getCellIndex();
RowDependency rd = project.recordModel.getRowDependency(rowIndex);
if (rd != null && rd.cellDependencies != null) {
int contextRowIndex = rd.cellDependencies[cellIndex].rowIndex;
if (contextRowIndex >= 0 && contextRowIndex < project.rows.size()) {
Row row2 = project.rows.get(contextRowIndex);
cell2 = row2.getCell(detailCellIndex);
}
}
}
if (cell2 != null && ExpressionUtils.isNonBlankData(cell2.value)) {
jsonWriter.object();
jsonWriter.key("pid"); jsonWriter.value(c.propertyID);
jsonWriter.key("v");
if (cell2.recon != null && cell2.recon.match != null) {
jsonWriter.object();
jsonWriter.key("id"); jsonWriter.value(cell2.recon.match.id);
jsonWriter.key("name"); jsonWriter.value(cell2.recon.match.name);
jsonWriter.endObject();
} else {
jsonWriter.value(cell2.value.toString());
}
jsonWriter.endObject();
}
}
jsonWriter.endArray();
}
jsonWriter.endObject();
job.text = cell.value.toString();
job.code = stringWriter.toString();
} catch (JSONException e) {
//
}
return job;
}
@Override
public List<Recon> batchRecon(List<ReconJob> jobs, long historyEntryID) {
List<Recon> recons = new ArrayList<Recon>(jobs.size());
StringWriter stringWriter = new StringWriter();
stringWriter.write("{");
for (int i = 0; i < jobs.size(); i++) {
StandardReconJob job = (StandardReconJob) jobs.get(i);
if (i > 0) {
stringWriter.write(",");
}
stringWriter.write("\"q" + i + "\":");
stringWriter.write(job.code);
}
stringWriter.write("}");
String queriesString = stringWriter.toString();
try {
URL url = new URL(service);
URLConnection connection = url.openConnection();
{
connection.setRequestProperty("Content-Type", "application/x-www-form-urlencoded");
connection.setConnectTimeout(30000);
connection.setDoOutput(true);
DataOutputStream dos = new DataOutputStream(connection.getOutputStream());
try {
String body = "queries=" + ParsingUtilities.encode(queriesString);
dos.writeBytes(body);
} finally {
dos.flush();
dos.close();
}
connection.connect();
}
InputStream is = connection.getInputStream();
try {
String s = ParsingUtilities.inputStreamToString(is);
JSONObject o = ParsingUtilities.evaluateJsonStringToObject(s);
for (int i = 0; i < jobs.size(); i++) {
StandardReconJob job = (StandardReconJob) jobs.get(i);
Recon recon = null;
String text = job.text;
String key = "q" + i;
if (o.has(key)) {
JSONObject o2 = o.getJSONObject(key);
if (o2.has("result")) {
JSONArray results = o2.getJSONArray("result");
recon = createReconServiceResults(text, results, historyEntryID);
} else {
logger.warn("Service error for text: " + text + "\n Job code: " + job.code + "\n Response: " + o2.toString());
}
} else {
logger.warn("Service error for text: " + text + "\n Job code: " + job.code);
}
if (recon != null) {
recon.service = service;
}
recons.add(recon);
}
} finally {
is.close();
}
// } catch (IOException e) {
// // TODO: Retry on HTTP 500 errors?
} catch (Exception e) {
logger.error("Failed to batch recon with load:\n" + queriesString, e);
}
while (recons.size() < jobs.size()) {
Recon recon = new Recon(historyEntryID, identifierSpace, schemaSpace);
recon.service = service;
recon.identifierSpace = identifierSpace;
recon.schemaSpace = schemaSpace;
recons.add(recon);
}
return recons;
}
@Override
public Recon createNewRecon(long historyEntryID) {
Recon recon = new Recon(historyEntryID, identifierSpace, schemaSpace);
recon.service = service;
return recon;
}
protected Recon createReconServiceResults(String text, JSONArray results, long historyEntryID) {
Recon recon = new Recon(historyEntryID, identifierSpace, schemaSpace);
try {
int length = results.length();
int count = 0;
for (int i = 0; i < length && count < 3; i++) {
JSONObject result = results.getJSONObject(i);
if (!result.has("name")) {
continue;
}
JSONArray types = result.getJSONArray("type");
String[] typeIDs = new String[types.length()];
for (int j = 0; j < typeIDs.length; j++) {
Object type = types.get(j);
typeIDs[j] = type instanceof String ? (String) type :
((JSONObject) type).getString("id");
}
double score = result.getDouble("score");
ReconCandidate candidate = new ReconCandidate(
result.getString("id"),
result.getString("name"),
typeIDs,
score
);
if (autoMatch && i == 0 && result.has("match") && result.getBoolean("match")) {
recon.match = candidate;
recon.matchRank = 0;
recon.judgment = Judgment.Matched;
recon.judgmentAction = "auto";
}
recon.addCandidate(candidate);
count++;
}
if (count > 0) {
ReconCandidate candidate = recon.candidates.get(0);
recon.setFeature(Recon.Feature_nameMatch, text.equalsIgnoreCase(candidate.name));
recon.setFeature(Recon.Feature_nameLevenshtein,
StringUtils.getLevenshteinDistance(StringUtils.lowerCase(text), StringUtils.lowerCase(candidate.name)));
recon.setFeature(Recon.Feature_nameWordDistance, wordDistance(text, candidate.name));
recon.setFeature(Recon.Feature_typeMatch, false);
if (this.typeID != null) {
for (String typeID : candidate.types) {
if (this.typeID.equals(typeID)) {
recon.setFeature(Recon.Feature_typeMatch, true);
break;
}
}
}
}
} catch (JSONException e) {
e.printStackTrace();
}
return recon;
}
static protected double wordDistance(String s1, String s2) {
Set<String> words1 = breakWords(s1);
Set<String> words2 = breakWords(s2);
return words1.size() >= words2.size() ? wordDistance(words1, words2) : wordDistance(words2, words1);
}
static protected double wordDistance(Set<String> longWords, Set<String> shortWords) {
if (longWords.size() == 0) {
return 0.0;
}
double common = 0;
for (String word : shortWords) {
if (longWords.contains(word)) {
common++;
}
}
return common / longWords.size();
}
static final protected Set<String> s_stopWords = new HashSet<String>();
static {
s_stopWords.add("the");
s_stopWords.add("a");
s_stopWords.add("and");
s_stopWords.add("of");
s_stopWords.add("on");
s_stopWords.add("in");
s_stopWords.add("at");
s_stopWords.add("by");
}
static protected Set<String> breakWords(String s) {
String[] words = s.toLowerCase().split("\\s+");
Set<String> set = new HashSet<String>(words.length);
for (String word : words) {
if (!s_stopWords.contains(word)) {
set.add(word);
}
}
return set;
}
}