public static UpdateContainer createSemanticTypesAndSVGAlignmentUpdates(
String worksheetId, Workspace workspace, Alignment alignment) {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
UpdateContainer c = new UpdateContainer();
c.add(new SemanticTypesUpdate(worksheet, worksheetId, alignment));
c.add(new AlignmentSVGVisualizationUpdate(worksheetId, alignment));
return c;
}
public static void detectSelectionStatusChange(
String worksheetId, Workspace workspace, Command command) {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
for (Selection sel : worksheet.getSelectionManager().getAllDefinedSelection()) {
Set<String> inputColumns = new HashSet<String>(sel.getInputColumns());
inputColumns.retainAll(command.getOutputColumns());
if (inputColumns.size() > 0) sel.invalidateSelection();
}
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
UpdateContainer uc = new UpdateContainer();
SuperSelection sel = this.getSuperSelection(workspace);
for (int i = 0; i < updates.length(); i++) {
String update = updates.getString(i);
switch (update) {
case "headers":
uc.add(new WorksheetHeadersUpdate(worksheetId));
break;
case "list":
uc.add(new WorksheetListUpdate());
break;
case "data":
uc.add(new WorksheetDataUpdate(worksheetId, sel));
break;
case "alignment":
{
Alignment alignment =
AlignmentManager.Instance()
.getAlignmentOrCreateIt(
workspace.getId(), worksheetId, workspace.getOntologyManager());
uc.add(new AlignmentSVGVisualizationUpdate(worksheetId, alignment));
break;
}
case "semanticTypes":
{
Alignment alignment =
AlignmentManager.Instance()
.getAlignmentOrCreateIt(
workspace.getId(), worksheetId, workspace.getOntologyManager());
uc.add(
new SemanticTypesUpdate(
workspace.getWorksheet(worksheetId), worksheetId, alignment));
break;
}
case "regenerate":
uc.add(new RegenerateWorksheetUpdate(worksheetId));
break;
case "all":
uc = WorksheetUpdateFactory.createRegenerateWorksheetUpdates(worksheetId, sel);
break;
case "cleaning":
uc.add(new WorksheetCleaningUpdate(worksheetId, false, sel));
break;
}
}
return uc;
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
RepFactory factory = workspace.getFactory();
SuperSelection superSel = getSuperSelection(worksheet);
HTable hTable = factory.getHTable(factory.getHNode(hNodeId).getHTableId());
Selection currentSel = superSel.getSelection(hTable.getId());
if (currentSel != null) {
currentSel.updateSelection();
}
CommandHistory history = workspace.getCommandHistory();
List<Command> tmp =
gatherAllOperateSelectionCommands(
history.getCommandsFromWorksheetId(worksheetId), workspace);
if (tmp.size() > 0) {
JSONArray inputJSON = new JSONArray();
inputJSON.put(
CommandInputJSONUtil.createJsonObject(
"worksheetId", worksheetId, ParameterType.worksheetId));
inputJSON.put(
CommandInputJSONUtil.createJsonObject("hNodeId", hNodeId, ParameterType.hNodeId));
inputJSON.put(
CommandInputJSONUtil.createJsonObject(
"operation", Operation.Intersect.name(), ParameterType.other));
inputJSON.put(
CommandInputJSONUtil.createJsonObject(
"pythonCode", SelectionManager.defaultCode, ParameterType.other));
inputJSON.put(CommandInputJSONUtil.createJsonObject("onError", "false", ParameterType.other));
inputJSON.put(
CommandInputJSONUtil.createJsonObject(
"selectionName", superSel.getName(), ParameterType.other));
Command t = null;
try {
t = new OperateSelectionCommandFactory().createCommand(inputJSON, workspace);
} catch (Exception e) {
}
if (t != null) history._getHistory().add(t);
history._getHistory().addAll(tmp);
}
UpdateContainer uc =
WorksheetUpdateFactory.createWorksheetHierarchicalAndCleaningResultsUpdates(
worksheetId, superSel);
uc.add(new HistoryUpdate(history));
return uc;
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet wk = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(wk);
String Msg =
String.format("begin, Time,%d, Worksheet,%s", System.currentTimeMillis(), worksheetId);
logger.info(Msg);
// Get the HNode
HashMap<String, HashMap<String, String>> rows = new HashMap<String, HashMap<String, String>>();
HNodePath selectedPath = null;
List<HNodePath> columnPaths = wk.getHeaders().getAllPaths();
for (HNodePath path : columnPaths) {
if (path.getLeaf().getId().equals(hNodeId)) {
selectedPath = path;
}
}
// random nodes
Collection<Node> nodes = new ArrayList<Node>();
wk.getDataTable().collectNodes(selectedPath, nodes, selection);
HashSet<Integer> indSet = this.obtainIndexs(nodes.size());
int index = 0;
for (Iterator<Node> iterator = nodes.iterator(); iterator.hasNext(); ) {
Node node = iterator.next();
if (indSet.contains(index)) {
String id = node.getId();
String originalVal = node.getValue().asString();
HashMap<String, String> x = new HashMap<String, String>();
x.put("Org", originalVal);
x.put("Tar", originalVal);
x.put("Orgdis", originalVal);
x.put("Tardis", originalVal);
rows.put(id, x);
}
index++;
}
Msg = String.format("end, Time,%d, Worksheet,%s", System.currentTimeMillis(), worksheetId);
logger.info(Msg);
return new UpdateContainer(new FetchResultUpdate(hNodeId, rows));
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
CSVFileExport csvFileExport = new CSVFileExport(worksheet);
try {
final String fileName = csvFileExport.publishCSV();
if (fileName == null)
return new UpdateContainer(
new ErrorUpdate("No data to export! Have you aligned the worksheet?"));
return new UpdateContainer(
new AbstractUpdate() {
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
JSONObject outputObject = new JSONObject();
try {
outputObject.put(JsonKeys.updateType.name(), "PublishCSVUpdate");
outputObject.put(JsonKeys.fileUrl.name(), fileName);
outputObject.put(JsonKeys.worksheetId.name(), worksheetId);
pw.println(outputObject.toString(4));
} catch (JSONException e) {
logger.error("Error occured while generating JSON!");
}
}
});
} catch (FileNotFoundException e) {
logger.error("CSV folder not found!", e);
return new UpdateContainer(new ErrorUpdate("Error occurred while exporting CSV file!"));
} catch (Exception e) {
logger.error("CSV Export Error", e);
return new UpdateContainer(new ErrorUpdate("Error occurred while exporting CSV file!"));
}
}
/*
* Pedro
*
* We should have an abstraction for Commands that operate on worksheets,
* and this method should go there.
*/
protected String formatWorsheetId(Workspace workspace, String worksheetId) {
return worksheetId + " (" + workspace.getWorksheet(worksheetId).getTitle() + ")";
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet wk = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(wk);
String msg =
String.format(
"Gen rule start,Time,%d, Worksheet,%s", System.currentTimeMillis(), worksheetId);
logger.info(msg);
// Get the HNode
HashMap<String, String> rows = new HashMap<String, String>();
HashMap<String, Integer> amb = new HashMap<String, Integer>();
HNodePath selectedPath = null;
List<HNodePath> columnPaths = wk.getHeaders().getAllPaths();
for (HNodePath path : columnPaths) {
if (path.getLeaf().getId().equals(hNodeId)) {
selectedPath = path;
}
}
Collection<Node> nodes = new ArrayList<Node>();
wk.getDataTable().collectNodes(selectedPath, nodes, selection);
for (Node node : nodes) {
String id = node.getId();
if (!this.nodeIds.contains(id)) continue;
String originalVal = node.getValue().asString();
rows.put(id, originalVal);
this.compResultString += originalVal + "\n";
calAmbScore(id, originalVal, amb);
}
RamblerValueCollection vc = new RamblerValueCollection(rows);
HashMap<String, Vector<String[]>> expFeData = new HashMap<String, Vector<String[]>>();
inputs = new RamblerTransformationInputs(examples, vc);
// generate the program
boolean results = false;
int iterNum = 0;
RamblerTransformationOutput rtf = null;
// initialize the vocabulary
Iterator<String> iterx = inputs.getInputValues().getValues().iterator();
Vector<String> v = new Vector<String>();
int vb_cnt = 0;
while (iterx.hasNext() && vb_cnt < 30) {
String eString = iterx.next();
v.add(eString);
vb_cnt++;
}
Vector<String> vob = UtilTools.buildDict(v);
inputs.setVocab(vob.toArray(new String[vob.size()]));
while (iterNum < 1 && !results) // try to find an program within iterNum
{
rtf = new RamblerTransformationOutput(inputs);
if (rtf.getTransformations().keySet().size() > 0) {
results = true;
}
iterNum++;
}
Iterator<String> iter = rtf.getTransformations().keySet().iterator();
// id:{org: tar: orgdis: tardis: }
HashMap<String, HashMap<String, String>> resdata =
new HashMap<String, HashMap<String, String>>();
HashSet<String> keys = new HashSet<String>();
while (iter.hasNext()) {
String tpid = iter.next();
ValueCollection rvco = rtf.getTransformedValues_debug(tpid);
if (rvco == null) continue;
// constructing displaying data
HashMap<String, String[]> xyzHashMap = new HashMap<String, String[]>();
for (String key : rvco.getNodeIDs()) {
HashMap<String, String> dict = new HashMap<String, String>();
// add to the example selection
boolean isExp = false;
String org = vc.getValue(key);
String classLabel = rvco.getClass(key);
String pretar = rvco.getValue(key);
String dummyValue = pretar;
if (pretar.indexOf("_FATAL_ERROR_") != -1) {
dummyValue = org;
// dummyValue = "#ERROR";
}
try {
UtilTools.StringColorCode(org, dummyValue, dict);
} catch (Exception ex) {
logger.info(String.format("ColorCoding Exception%s, %s", org, dummyValue));
// set dict
dict.put("Org", org);
dict.put("Tar", "ERROR");
dict.put("Orgdis", org);
dict.put("Tardis", "ERROR");
}
for (TransformationExample exp : examples) {
if (exp.getNodeId().compareTo(key) == 0) {
if (!expFeData.containsKey(classLabel)) {
Vector<String[]> vstr = new Vector<String[]>();
String[] texp = {dict.get("Org"), pretar};
vstr.add(texp);
expFeData.put(classLabel, vstr);
} else {
String[] texp = {dict.get("Org"), pretar};
expFeData.get(classLabel).add(texp);
}
isExp = true;
}
}
if (!isExp) {
String[] pair = {dict.get("Org"), dict.get("Tar"), pretar, classLabel};
xyzHashMap.put(key, pair);
}
resdata.put(key, dict);
}
if (!rtf.nullRule) keys.add(getBestExample(xyzHashMap, expFeData));
}
// find the best row
String vars = "";
String expstr = "";
String recmd = "";
for (TransformationExample x : examples) {
expstr += String.format("%s|%s", x.getBefore(), x.getAfter());
}
expstr += "|";
if (rtf.nullRule) {
keys.clear();
// keys.add("-2"); // "-2 indicates null rule"
}
if (!resdata.isEmpty() && !rtf.nullRule) {
recmd = resdata.get(keys.iterator().next()).get("Org");
} else {
recmd = "";
}
msg =
String.format(
"Gen rule end, Time,%d, Worksheet,%s,Examples:%s,Recmd:%s",
System.currentTimeMillis(), worksheetId, expstr, recmd);
logger.info(msg);
return new UpdateContainer(new CleaningResultUpdate(hNodeId, resdata, vars, keys));
}
@Override
public UpdateContainer doIt(Workspace workspace) {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(worksheet);
String worksheetName = worksheet.getTitle();
try {
// preparing model file name
final String modelFileName =
workspace.getCommandPreferencesId() + worksheetId + "-" + worksheetName + "-model.ttl";
final String modelFileLocalPath =
ServletContextParameterMap.getParameterValue(ContextParameter.R2RML_PUBLISH_DIR)
+ modelFileName;
File f = new File(modelFileLocalPath);
// preparing the graphUri where the model is published in the triple store
String graphName =
worksheet
.getMetadataContainer()
.getWorksheetProperties()
.getPropertyValue(Property.graphName);
if (graphName == null || graphName.isEmpty()) {
SimpleDateFormat sdf = new SimpleDateFormat("dd-MMM-yyyy-kkmmssS");
String ts = sdf.format(Calendar.getInstance().getTime());
graphName =
"http://localhost/"
+ workspace.getCommandPreferencesId()
+ "/"
+ worksheetId
+ "/model/"
+ ts;
worksheet
.getMetadataContainer()
.getWorksheetProperties()
.setPropertyValue(Property.graphName, graphName);
}
// If the model is not published, publish it!
if (!f.exists() || !f.isFile()) {
GenerateR2RMLModelCommandFactory factory = new GenerateR2RMLModelCommandFactory();
GenerateR2RMLModelCommand cmd =
(GenerateR2RMLModelCommand)
factory.createCommand(
workspace,
worksheetId,
TripleStoreUtil.defaultModelsRepoUrl,
graphName,
selection.getName());
cmd.doIt(workspace);
} else {
// if the model was published 30 min ago, publish it again, just to be sure
long diff = Calendar.getInstance().getTimeInMillis() - f.lastModified();
if ((diff / 1000L / 60L) > 30) {
f.delete();
GenerateR2RMLModelCommandFactory factory = new GenerateR2RMLModelCommandFactory();
GenerateR2RMLModelCommand cmd =
(GenerateR2RMLModelCommand)
factory.createCommand(
workspace,
worksheetId,
TripleStoreUtil.defaultModelsRepoUrl,
graphName,
selection.getName());
cmd.doIt(workspace);
}
}
// TripleStoreUtil tUtil = new TripleStoreUtil();
StringBuffer query =
new StringBuffer(
"prefix rr: <http://www.w3.org/ns/r2rml#> prefix km-dev: <http://isi.edu/integration/karma/dev#> ");
/* ****** this is the query for the list of columns.
PREFIX km-dev: <http://isi.edu/integration/karma/dev#>
PREFIX rr: <http://www.w3.org/ns/r2rml#>
select distinct ?class where {
{
?x1 rr:subjectMap/km-dev:alignmentNodeId "------- The full url of the column/class --------".
?x1 rr:predicateObjectMap/rr:objectMap/rr:column ?column .
?x1 rr:subjectMap/rr:predicate ?class .
}
UNION
{
?x1 rr:subjectMap/km-dev:alignmentNodeId "------- The full url of the column/class --------".
?x1 (rr:predicateObjectMap/rr:objectMap/rr:parentTriplesMap)* ?x2 .
?x2 rr:predicateObjectMap/rr:objectMap/rr:column ?column .
?x2 rr:predicateObjectMap/rr:predicate ?class .
}
}
* */
query.append("select distinct ?class ?column where { ");
if (graphName != null && !graphName.trim().isEmpty()) {
query.append(" graph <" + graphName + "> { ");
}
query
.append("{ ?x1 rr:subjectMap/km-dev:alignmentNodeId \"")
.append(this.nodeId)
.append(
"\" . ?x1 rr:predicateObjectMap/rr:objectMap/rr:column ?column . ?x1 rr:subjectMap/rr:predicate ?class .")
.append(" } UNION { ")
.append("?x1 rr:subjectMap/km-dev:alignmentNodeId \"")
.append(this.nodeId)
.append("\" . ?x1 (rr:predicateObjectMap/rr:objectMap/rr:parentTriplesMap)* ?x2 .")
.append(" ?x2 rr:predicateObjectMap ?x3 . ")
.append(" ?x3 rr:objectMap/rr:column ?column . ?x3 rr:predicate ?class .")
.append(" } }");
if (graphName != null && !graphName.trim().isEmpty()) {
query.append(" } ");
}
logger.info("Query: " + query.toString());
String sData =
TripleStoreUtil.invokeSparqlQuery(
query.toString(), TripleStoreUtil.defaultModelsRepoUrl, "application/json", null);
if (sData == null | sData.isEmpty()) {
logger.error("Empty response object from query : " + query);
}
HashMap<String, String> cols = new HashMap<String, String>();
try {
JSONObject obj1 = new JSONObject(sData);
JSONArray arr = obj1.getJSONObject("results").getJSONArray("bindings");
for (int i = 0; i < arr.length(); i++) {
String colName = arr.getJSONObject(i).getJSONObject("column").getString("value");
String colValue = arr.getJSONObject(i).getJSONObject("class").getString("value");
if (cols.containsKey(colName)) {
logger.error("Duplicate Column <-> property mapping. " + colName + " <=> " + colValue);
} else {
cols.put(colName, colValue);
}
}
} catch (Exception e2) {
logger.error("Error in parsing json response", e2);
}
logger.info("Total Columns fetched : " + cols.size());
final HashMap<String, String> columns = cols;
return new UpdateContainer(
new AbstractUpdate() {
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
JSONObject obj = new JSONObject();
try {
Iterator<String> itr = columns.keySet().iterator();
JSONArray colList = new JSONArray();
while (itr.hasNext()) {
JSONObject o = new JSONObject();
String k = itr.next();
o.put("name", k);
o.put("url", columns.get(k));
colList.put(o);
}
obj.put("updateType", "FetchColumnUpdate");
obj.put("columns", colList);
obj.put("rootId", nodeId);
pw.println(obj.toString());
} catch (JSONException e) {
logger.error("Error occurred while fetching worksheet properties!", e);
}
}
});
} catch (Exception e) {
String msg = "Error occured while fetching columns!";
logger.error(msg, e);
return new UpdateContainer(new ErrorUpdate(msg));
}
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(worksheet);
HNodePath selectedPath = null;
List<HNodePath> columnPaths = worksheet.getHeaders().getAllPaths();
for (HNodePath path : columnPaths) {
if (path.getLeaf().getId().equals(hNodeId)) {
selectedPath = path;
}
}
Collection<Node> nodes = new ArrayList<Node>();
workspace
.getFactory()
.getWorksheet(worksheetId)
.getDataTable()
.collectNodes(selectedPath, nodes, selection);
try {
JSONArray requestJsonArray = new JSONArray();
for (Node node : nodes) {
String id = node.getId();
String originalVal = node.getValue().asString();
JSONObject jsonRecord = new JSONObject();
jsonRecord.put("id", id);
originalVal = originalVal == null ? "" : originalVal;
jsonRecord.put("value", originalVal);
requestJsonArray.put(jsonRecord);
}
String jsonString = null;
jsonString = requestJsonArray.toString();
// String url =
// "http://localhost:8080/cleaningService/IdentifyData";
// String url = "http://localhost:8070/myWS/IdentifyData";
String url =
ServletContextParameterMap.getParameterValue(ContextParameter.CLEANING_SERVICE_URL);
HttpClient httpclient = new DefaultHttpClient();
HttpPost httppost = null;
HttpResponse response = null;
HttpEntity entity;
StringBuffer out = new StringBuffer();
URI u = null;
u = new URI(url);
List<NameValuePair> formparams = new ArrayList<NameValuePair>();
formparams.add(new BasicNameValuePair("json", jsonString));
httppost = new HttpPost(u);
httppost.setEntity(new UrlEncodedFormEntity(formparams, "UTF-8"));
response = httpclient.execute(httppost);
entity = response.getEntity();
if (entity != null) {
BufferedReader buf = new BufferedReader(new InputStreamReader(entity.getContent()));
String line = buf.readLine();
while (line != null) {
out.append(line);
line = buf.readLine();
}
}
// logger.trace(out.toString());
// logger.info("Connnection success : " + url + " Successful.");
final JSONObject data1 = new JSONObject(out.toString());
// logger.trace("Data--->" + data1);
return new UpdateContainer(
new AbstractUpdate() {
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
JSONObject response = new JSONObject();
// logger.trace("Reached here");
try {
response.put("updateType", "CleaningServiceOutput");
response.put("chartData", data1);
response.put("hNodeId", hNodeId);
// logger.trace(response.toString(4));
} catch (JSONException e) {
pw.print("Error");
}
pw.print(response.toString());
}
});
} catch (Exception e) {
e.printStackTrace();
return new UpdateContainer(new ErrorUpdate("Error!"));
}
}
