s434730/RandomSec
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

RdfTripleImporter handles row dependencies.

Browse Source 
git-svn-id: http://google-refine.googlecode.com/svn/trunk@818 7d457c2a-affb-35e4-300a-418c747d4874
This commit is contained in:
Iain Sproat 2010-05-18 21:08:37 +00:00
parent cfd0f2219e
commit 0e4682f453
3 changed files with 151 additions and 106 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
src/main/java/com/metaweb/gridworks/importers/RdfTripleImporter.java
View File

@ -38,50 +38,53 @@ public class RdfTripleImporter implements Importer{
String baseUrl = options.getProperty("base-url");
Graph graph = JrdfFactory.getNewGraph();
//System.out.println("--------------------------------");
//System.out.println("initial number of triples before parsing is : " + graph.getNumberOfTriples());
LineHandler lineHandler = nTriplesParserFactory.createParser(graph, newMapFactory);
GraphLineParser parser = new GraphLineParser(graph, lineHandler);
parser.parse(reader, baseUrl); //fills JRDF graph
//System.out.println("number of triples parsed is : " + graph.getNumberOfTriples());
//first column is subject
project.columnModel.columns.add(0, new Column(0, "subject"));
project.columnModel.setKeyColumnIndex(0); //the subject will be the key column
project.columnModel.update();
ClosableIterable<Triple> triples = graph.find(ANY_SUBJECT_NODE, ANY_PREDICATE_NODE, ANY_OBJECT_NODE);
try {
for (Triple triple : triples) {
//System.out.println("Triple : " + triple);
String subject = triple.getSubject().toString();
String predicate = triple.getPredicate().toString();
String object = triple.getObject().toString();
//System.out.println("subject : " + subject);
//System.out.println("predicate : " + predicate);
//System.out.println("object : " + object);
//System.out.println("predicate relates to column : " + project.columnModel.getColumnByName(predicate));
int candidateMergeRowIndex = -1;
//creates new column for every predicate
int columnIndex = project.columnModel.getColumnIndexByName(predicate);
if(columnIndex == -1){
candidateMergeRowIndex = AddNewColumn(project, predicate, subject);
AddNewColumn(project, predicate, subject);
}
columnIndex = project.columnModel.getColumnIndexByName(predicate);
//now find row to match with
int candidateMergeRowIndex = -1;
for(int i = 0; i < project.rows.size(); i++){
//check to see if the subjects are the same (merge if they are)
Cell cell = project.rows.get(i).cells.get(0);
if(cell != null){
if(project.rows.get(i).cells.get(0).value == subject){
candidateMergeRowIndex = i;
}
}
}
columnIndex = project.columnModel.getColumnIndexByName(predicate);
if(candidateMergeRowIndex > -1){
if(project.rows.get(candidateMergeRowIndex).cells.get(columnIndex) == null){
Cell cell = project.rows.get(candidateMergeRowIndex).cells.get(columnIndex);
if(cell == null){
//empty, so merge in this value
MergeWithRow(project, candidateMergeRowIndex, columnIndex, object);
}else{
//can't overwrite existing, so add new row
AddNewRow(project, subject, predicate, object); //TODO group to original row.
//can't overwrite existing, so add new dependent row
AddNewDependentRow(project, subject, candidateMergeRowIndex, columnIndex, object); //TODO group to original row.
}
}else{
AddNewRow(project, subject, predicate, object);
AddNewRow(project, subject, columnIndex, object);
}
}
@ -90,46 +93,42 @@ public class RdfTripleImporter implements Importer{
}
}
protected int AddNewColumn(Project project, String predicate, String subject){
//System.out.println("adding new column");
protected void AddNewColumn(Project project, String predicate, String subject){
int numberOfColumns = project.columnModel.columns.size();
project.columnModel.columns.add(numberOfColumns, new Column(numberOfColumns, predicate));
project.columnModel.setMaxCellIndex(numberOfColumns);
project.columnModel.update();
int candidateMergeRowIndex = -1;
//update existing rows with new column
for(int i = 0; i < project.rows.size(); i++){
project.rows.get(i).cells.add(numberOfColumns, null);
if(project.rows.get(i).cells.get(0).value == subject){
candidateMergeRowIndex = i;
}
}
//numberOfColumns = project.columnModel.columns.size();
//System.out.println("New total number of columns : " + numberOfColumns);
return candidateMergeRowIndex;
}
protected void MergeWithRow(Project project, int candidateMergeRowIndex, int columnIndex, String object){
project.rows.get(candidateMergeRowIndex).setCell(columnIndex, new Cell(object, null));
}
protected void AddNewRow(Project project, String subject, String predicate, String object){
int numberOfColumns = project.columnModel.columns.size();
protected void AddNewDependentRow(Project project, String subject, int candidateMergeRowIndex, int columnIndex, String object){
Row row = AddNewRow(project, subject, columnIndex, object);
Project.setRowDependency(project, row, columnIndex, candidateMergeRowIndex, project.columnModel.getKeyColumnIndex());
row.cells.set(project.columnModel.getKeyColumnIndex(), null); //the subject can now be null, as the dependencies are set
}
protected Row AddNewRow(Project project, String subject, int columnIndex, String object){
int numberOfColumns = project.columnModel.columns.size();
//add subject
Row row = new Row(numberOfColumns);
row.setCell(0, new Cell(subject, null));
//add object to a row
int columnIndex = project.columnModel.getColumnIndexByName(predicate);
//System.out.println("predicate relates to columnIndex : " + columnIndex);
row.setCell(columnIndex, new Cell(object, null));
//System.out.println("Number of cells in new row : " + row.cells.size());
project.rows.add(row);
//System.out.println("New total number of rows : " + project.rows.size());
return row;
}
@Override

133
src/main/java/com/metaweb/gridworks/model/Project.java
View File

@ -32,80 +32,80 @@ import com.metaweb.gridworks.util.Pool;
public class Project {
final public long id;
final public ColumnModel columnModel = new ColumnModel();
final public List<Row> rows = new ArrayList<Row>();
final public History history;
public Protograph protograph;
transient public ProcessManager processManager = new ProcessManager();
transient public Date lastSave = new Date();
final static Logger logger = LoggerFactory.getLogger("project");
static public long generateID() {
return System.currentTimeMillis() + Math.round(Math.random() * 1000000000000L);
}
public Project() {
id = generateID();
history = new History(this);
}
protected Project(long id) {
this.id = id;
this.history = new History(this);
}
public ProjectMetadata getMetadata() {
return ProjectManager.singleton.getProjectMetadata(id);
}
synchronized public void save() {
synchronized (this) {
File dir = ProjectManager.singleton.getProjectDir(id);
File tempFile = new File(dir, "data.temp.zip");
try {
saveToFile(tempFile);
} catch (Exception e) {
e.printStackTrace();
logger.warn("Failed to save project {}", id);
return;
}
File file = new File(dir, "data.zip");
File oldFile = new File(dir, "data.old.zip");
if (file.exists()) {
file.renameTo(oldFile);
}
tempFile.renameTo(file);
if (oldFile.exists()) {
oldFile.delete();
}
lastSave = new Date();
logger.info("Saved project '{}'",id);
}
}
protected void saveToFile(File file) throws Exception {
ZipOutputStream out = new ZipOutputStream(new FileOutputStream(file));
try {
Pool pool = new Pool();
out.putNextEntry(new ZipEntry("data.txt"));
try {
saveToOutputStream(out, pool);
} finally {
out.closeEntry();
}
out.putNextEntry(new ZipEntry("pool.txt"));
try {
pool.save(out);
@ -116,35 +116,35 @@ public class Project {
out.close();
}
}
protected void saveToOutputStream(OutputStream out, Pool pool) throws IOException {
Writer writer = new OutputStreamWriter(out);
try {
Properties options = new Properties();
options.setProperty("mode", "save");
options.put("pool", pool);
saveToWriter(writer, options);
} finally {
writer.flush();
}
}
protected void saveToWriter(Writer writer, Properties options) throws IOException {
writer.write(Gridworks.getVersion()); writer.write('\n');
writer.write("columnModel=\n"); columnModel.save(writer, options);
writer.write("history=\n"); history.save(writer, options);
if (protograph != null) {
writer.write("protograph="); protograph.save(writer, options); writer.write('\n');
}
writer.write("rowCount="); writer.write(Integer.toString(rows.size())); writer.write('\n');
for (Row row : rows) {
row.save(writer, options); writer.write('\n');
}
}
static public Project load(File dir, long id) {
try {
File file = new File(dir, "data.zip");
@ -154,7 +154,7 @@ public class Project {
} catch (Exception e) {
e.printStackTrace();
}
try {
File file = new File(dir, "data.temp.zip");
if (file.exists()) {
@ -163,7 +163,7 @@ public class Project {
} catch (Exception e) {
e.printStackTrace();
}
try {
File file = new File(dir, "data.old.zip");
if (file.exists()) {
@ -172,12 +172,12 @@ public class Project {
} catch (Exception e) {
e.printStackTrace();
}
return null;
}
static protected Project loadFromFile(
File file,
File file,
long id
) throws Exception {
ZipFile zipFile = new ZipFile(file);
@ -188,7 +188,7 @@ public class Project {
pool.load(new InputStreamReader(
zipFile.getInputStream(poolEntry)));
} // else, it's a legacy project file
return loadFromReader(
new LineNumberReader(
new InputStreamReader(
@ -201,25 +201,25 @@ public class Project {
zipFile.close();
}
}
static protected Project loadFromReader(
LineNumberReader reader,
LineNumberReader reader,
long id,
Pool pool
) throws Exception {
long start = System.currentTimeMillis();
/* String version = */ reader.readLine();
Project project = new Project(id);
int maxCellCount = 0;
String line;
while ((line = reader.readLine()) != null) {
int equal = line.indexOf('=');
CharSequence field = line.subSequence(0, equal);
String value = line.substring(equal + 1);
if ("columnModel".equals(field)) {
project.columnModel.load(reader);
} else if ("history".equals(field)) {
@ -228,7 +228,7 @@ public class Project {
project.protograph = Protograph.load(project, value);
} else if ("rowCount".equals(field)) {
int count = Integer.parseInt(value);
for (int i = 0; i < count; i++) {
line = reader.readLine();
if (line != null) {
@ -239,35 +239,35 @@ public class Project {
}
}
}
project.columnModel.setMaxCellIndex(maxCellCount - 1);
logger.info(
"Loaded project {} from disk in {} sec(s)",id,Long.toString((System.currentTimeMillis() - start) / 1000)
);
project.recomputeRowContextDependencies();
return project;
}
static protected class Group {
int[] cellIndices;
int keyCellIndex;
}
synchronized public void recomputeRowContextDependencies() {
List<Group> keyedGroups = new ArrayList<Group>();
addRootKeyedGroup(keyedGroups);
for (ColumnGroup group : columnModel.columnGroups) {
if (group.keyColumnIndex >= 0) {
Group keyedGroup = new Group();
keyedGroup.keyCellIndex = columnModel.columns.get(group.keyColumnIndex).getCellIndex();
keyedGroup.cellIndices = new int[group.columnSpan - 1];
int c = 0;
for (int i = 0; i < group.columnSpan; i++) {
int columnIndex = group.startColumnIndex + i;
@ -276,44 +276,45 @@ public class Project {
keyedGroup.cellIndices[c++] = cellIndex;
}
}
keyedGroups.add(keyedGroup);
}
}
Collections.sort(keyedGroups, new Comparator<Group>() {
public int compare(Group o1, Group o2) {
return o2.cellIndices.length - o1.cellIndices.length; // larger groups first
}
});
int[] lastNonBlankRowsByGroup = new int[keyedGroups.size()];
for (int i = 0; i < lastNonBlankRowsByGroup.length; i++) {
lastNonBlankRowsByGroup[i] = -1;
}
int rowCount = rows.size();
int groupCount = keyedGroups.size();
int recordIndex = 0;
for (int r = 0; r < rowCount; r++) {
Row row = rows.get(r);
row.contextRows = null;
row.contextRowSlots = null;
row.contextCellSlots = null;
for (int g = 0; g < groupCount; g++) {
Group group = keyedGroups.get(g);
if (!ExpressionUtils.isNonBlankData(row.getCellValue(group.keyCellIndex))) {
int contextRowIndex = lastNonBlankRowsByGroup[g];
if (contextRowIndex >= 0) {
for (int dependentCellIndex : group.cellIndices) {
if (ExpressionUtils.isNonBlankData(row.getCellValue(dependentCellIndex))) {
setRowDependency(
row,
dependentCellIndex,
contextRowIndex,
this,
row,
dependentCellIndex,
contextRowIndex,
group.keyCellIndex
);
}
@ -323,7 +324,7 @@ public class Project {
lastNonBlankRowsByGroup[g] = r;
}
}
if (row.contextRowSlots != null && row.contextRowSlots.length > 0) {
row.recordIndex = -1;
row.contextRows = new ArrayList<Integer>();
@ -333,19 +334,19 @@ public class Project {
}
}
Collections.sort(row.contextRows);
columnModel._hasDependentRows = true;
} else {
row.recordIndex = recordIndex++;
}
}
}
protected void addRootKeyedGroup(List<Group> keyedGroups) {
int count = columnModel.getMaxCellIndex() + 1;
if (count > 0 && columnModel.getKeyColumnIndex() < columnModel.columns.size()) {
Group rootKeyedGroup = new Group();
rootKeyedGroup.cellIndices = new int[count - 1];
rootKeyedGroup.keyCellIndex = columnModel.columns.get(columnModel.getKeyColumnIndex()).getCellIndex();
@ -359,19 +360,19 @@ public class Project {
keyedGroups.add(rootKeyedGroup);
}
}
protected void setRowDependency(Row row, int cellIndex, int contextRowIndex, int contextCellIndex) {
int count = columnModel.getMaxCellIndex() + 1;
public static void setRowDependency(Project project, Row row, int cellIndex, int contextRowIndex, int contextCellIndex) {
int count = project.columnModel.getMaxCellIndex() + 1;
if (row.contextRowSlots == null || row.contextCellSlots == null) {
row.contextRowSlots = new int[count];
row.contextCellSlots = new int[count];
for (int i = 0; i < count; i++) {
row.contextRowSlots[i] = -1;
row.contextCellSlots[i] = -1;
}
}
row.contextRowSlots[cellIndex] = contextRowIndex;
row.contextCellSlots[cellIndex] = contextCellIndex;
}

51
tests/java/src/com/metaweb/gridworks/tests/importers/RdfTripleImporterTests.java
View File

@ -63,14 +63,31 @@ public class RdfTripleImporterTests {
Assert.fail();
}
//columns
Assert.assertEquals(project.columnModel.columns.size(), 2);
Assert.assertEquals(project.columnModel.columns.get(0).getName(), "subject");
Assert.assertEquals(project.columnModel.columns.get(1).getName(), "http://rdf.freebase.com/ns/music.artist.album");
//rows
Assert.assertEquals(project.rows.size(), 3);
//row0
Assert.assertEquals(project.rows.get(0).cells.size(), 2);
Assert.assertEquals(project.rows.get(0).cells.get(0).value, "http://rdf.freebase.com/ns/en.bob_dylan");
Assert.assertEquals(project.rows.get(0).cells.get(1).value, "http://rdf.freebase.com/ns/en.blood_on_the_tracks");
//row1
Assert.assertEquals(project.rows.get(2).cells.size(), 2);
Assert.assertNull(project.rows.get(1).cells.get(0));
Assert.assertEquals(project.rows.get(1).contextRowSlots[1], 0);
Assert.assertEquals(project.rows.get(1).contextCellSlots[1], 0);
Assert.assertEquals(project.rows.get(1).cells.get(1).value, "http://rdf.freebase.com/ns/en.bringing_it_all_back_home"); //NB triples aren't created in order they were input
//row2
Assert.assertEquals(project.rows.get(2).cells.size(), 2);
Assert.assertEquals(project.rows.get(2).contextRowSlots[1], 0);
Assert.assertEquals(project.rows.get(2).contextCellSlots[1], 0);
Assert.assertNull(project.rows.get(2).cells.get(0));
Assert.assertEquals(project.rows.get(2).cells.get(1).value, "http://rdf.freebase.com/ns/en.under_the_red_sky"); //NB triples aren't created in order they were input
}
@ -87,18 +104,46 @@ public class RdfTripleImporterTests {
Assert.fail();
}
//columns
Assert.assertEquals(project.columnModel.columns.size(), 3);
Assert.assertEquals(project.columnModel.columns.get(0).getName(), "subject");
Assert.assertEquals(project.columnModel.columns.get(1).getName(), "http://rdf.freebase.com/ns/music.artist.album");
Assert.assertEquals(project.columnModel.columns.get(2).getName(), "http://rdf.freebase.com/ns/music.artist.genre");
//rows
Assert.assertEquals(project.rows.size(), 2);
//row0
Assert.assertEquals(project.rows.get(0).cells.size(), 3);
Assert.assertEquals(project.rows.get(0).cells.get(0).value, "http://rdf.freebase.com/ns/en.bob_dylan");
Assert.assertEquals(project.rows.get(0).cells.get(1).value, "http://rdf.freebase.com/ns/en.blood_on_the_tracks");
Assert.assertNull(project.rows.get(0).cells.get(2));
Assert.assertEquals(project.rows.get(0).cells.get(2).value, "http://rdf.freebase.com/ns/en.folk_rock");
//row1
Assert.assertEquals(project.rows.get(1).cells.size(), 3);
Assert.assertEquals(project.rows.get(1).cells.get(0).value, "http://rdf.freebase.com/ns/en.bob_dylan");
Assert.assertEquals(project.rows.get(1).contextRowSlots[1], 0);
Assert.assertEquals(project.rows.get(1).contextCellSlots[1], 0);
Assert.assertNull(project.rows.get(1).cells.get(0));
Assert.assertEquals(project.rows.get(1).cells.get(1).value, "http://rdf.freebase.com/ns/en.bringing_it_all_back_home");
Assert.assertEquals(project.rows.get(1).cells.get(2).value, "http://rdf.freebase.com/ns/en.folk_rock");
Assert.assertNull(project.rows.get(1).cells.get(2));
}
@Test
public void CanParseTripleWithValue(){
String sampleRdf = "<http://rdf.freebase.com/ns/en.bob_dylan> <http://rdf.freebase.com/ns/common.topic.alias> \"Robert Zimmerman\"@en.";
StringReader reader = new StringReader(sampleRdf);
try {
SUT.read(reader, project, options);
} catch (Exception e) {
Assert.fail();
}
Assert.assertEquals(project.columnModel.columns.size(), 2);
Assert.assertEquals(project.columnModel.columns.get(0).getName(), "subject");
Assert.assertEquals(project.columnModel.columns.get(1).getName(), "http://rdf.freebase.com/ns/common.topic.alias");
Assert.assertEquals(project.rows.size(), 1);
Assert.assertEquals(project.rows.get(0).cells.size(), 2);
Assert.assertEquals(project.rows.get(0).cells.get(0).value, "http://rdf.freebase.com/ns/en.bob_dylan");
Assert.assertEquals(project.rows.get(0).cells.get(1).value, "\"Robert Zimmerman\"@en");
}
}