private void initIfNeed() {
if (myInitialized) return;
myInitialized = true;
for (Map.Entry<String, Pair<String, List<String>>> entry : myTree.entrySet()) {
final String group = entry.getKey();
if (CORE.equals(group)) continue;
List<IdSet> idSets = new ArrayList<IdSet>();
StringBuilder description = new StringBuilder();
for (String idDescription : entry.getValue().getSecond()) {
IdSet idSet = new IdSet(idDescription);
String idSetTitle = idSet.getTitle();
if (idSetTitle == null) continue;
idSets.add(idSet);
if (description.length() > 0) {
description.append(", ");
}
description.append(idSetTitle);
}
myGroups.put(group, idSets);
if (description.length() > MAX_DESCR_LENGTH) {
int lastWord = description.lastIndexOf(",", MAX_DESCR_LENGTH);
description.delete(lastWord, description.length()).append("...");
}
description.insert(0, "<html><body><center><i>");
myDescriptions.put(group, description.toString());
}
}
public StringBuilder findLocations(Document doc, StringBuilder sb) {
for (Tag tag : doc.tags) {
StringBuilder nb = new StringBuilder();
for (int i = tag.begin; i < tag.end; i++) {
nb.append(doc.terms.get(i) + " ");
}
if (!(nb.lastIndexOf(" ") == -1)) {
nb.deleteCharAt(nb.lastIndexOf(" "));
}
String name = nb.toString();
sb.append(name);
sb.append(",");
sb.append(doc.name);
sb.append("\n");
}
return sb;
}
/**
* pretty print the warning message supplied
*
* @param logger logger for the message
* @param message the message
*/
private static void prettyPrintWarningMessage(Logger logger, String message) {
StringBuilder builder = new StringBuilder(message);
while (builder.length() > 70) {
int space = builder.lastIndexOf(" ", 70);
if (space <= 0) space = 70;
logger.warn(String.format("* %s", builder.substring(0, space)));
builder.delete(0, space + 1);
}
logger.warn(String.format("* %s", builder));
}
@NotNull
private static String stripPath(@NotNull String path) {
String[] endingsToStrip = {"/", "!", ".jar"};
StringBuilder buffer = new StringBuilder(path);
for (String ending : endingsToStrip) {
if (buffer.lastIndexOf(ending) == buffer.length() - ending.length()) {
buffer.setLength(buffer.length() - ending.length());
}
}
return buffer.toString();
}
/** Return the next element of the json document */
private String getNextNode() {
final int popState;
switch (states.peek()) {
case ARRAY:
switch (rand.nextInt(9)) {
case 0: // String case
final String val = "stepha" + this.getWhitespace() + "n" + this.getWhitespace() + "e";
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add(val);
types.add(ExtendedJsonTokenizer.getTokenTypes()[LITERAL]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_STRING);
return "\"" + val + "\"" + this.getWhitespace() + ",";
case 1: // DOUBLE case
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add("34.560e-9");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NUMBER]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_DOUBLE);
return "34.560e-9" + this.getWhitespace() + ",";
case 2: // LONG case
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add("34560e-9");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NUMBER]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_LONG);
return "34560e-9" + this.getWhitespace() + ",";
case 3: // true case
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add("true");
types.add(ExtendedJsonTokenizer.getTokenTypes()[TRUE]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_BOOLEAN);
return "true" + this.getWhitespace() + ",";
case 4: // false case
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add("false");
types.add(ExtendedJsonTokenizer.getTokenTypes()[FALSE]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_BOOLEAN);
return "false" + this.getWhitespace() + ",";
case 5: // null case
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
images.add("null");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NULL]);
incr.add(1);
datatypes.add(XSDDatatype.XSD_STRING);
return "null" + this.getWhitespace() + ",";
case 6: // nested array case
if (states.size() <= MAX_DEPTH) {
this.addToLastNode(1);
this.incrNodeObjectPath();
states.add(ARRAY);
return "[";
}
return "";
case 7: // nested object case
if (states.size() <= MAX_DEPTH) {
this.addToLastNode(1);
this.incrNodeObjectPath();
states.add(ARRAY_OBJECT);
return "{";
}
return "";
case 8: // closing array case
this.decrNodeObjectPath();
popState = states.pop();
if (popState != ARRAY) {
shouldFail = true;
}
// Remove previous comma, this is not allowed
final int comma = sb.lastIndexOf(",");
if (comma != -1 && sb.substring(comma + 1).matches("\\s*")) {
sb.deleteCharAt(comma);
}
return "],";
}
case ARRAY_OBJECT:
case OBJECT_ATT:
switch (rand.nextInt(3)) {
case 0: // new object field
types.add(ExtendedJsonTokenizer.getTokenTypes()[LITERAL]);
images.add("ste ph ane");
incr.add(1);
this.addToLastNode(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
datatypes.add(JSONDatatype.JSON_FIELD);
states.push(OBJECT_VAL);
return "\"ste ph ane\"" + this.getWhitespace() + ":";
case 1: // close object
if (states.peek() == OBJECT_ATT && nestedObjs > 0) {
this.decrNodeObjectPath();
nestedObjs--;
}
this.decrNodeObjectPath();
popState = states.pop();
if (popState != OBJECT_ATT && popState != ARRAY_OBJECT) {
shouldFail = true;
}
// Remove previous comma, this is not allowed
final int comma = sb.lastIndexOf(",");
if (comma != -1 && sb.substring(comma + 1).matches("\\s*")) {
sb.deleteCharAt(comma);
}
return states.empty() ? "}" : "},";
case 2: // Datatype
if (getLastNode() >= 0) {
// this nested object cannot be a datatype object because other things have been added
// to it
return "";
}
final String field;
if (states.isEmpty()) {
// datatype object at the root are not possible
shouldFail = true;
field = "{";
} else if (states.peek() == OBJECT_ATT) {
// field name
this.addToLastNode(1);
field = "\"field\":{";
types.add(ExtendedJsonTokenizer.getTokenTypes()[LITERAL]);
images.add("field");
incr.add(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
datatypes.add(JSONDatatype.JSON_FIELD);
// value
this.incrNodeObjectPath();
this.setLastNode(0);
} else if (states.peek() == ARRAY) {
this.addToLastNode(1);
field = "{";
} else if (states.peek() == ARRAY_OBJECT) {
this.decrNodeObjectPath();
field = "";
} else {
// should not happen
throw new IllegalStateException("Received unknown state=" + states.peek());
}
types.add(ExtendedJsonTokenizer.getTokenTypes()[LITERAL]);
images.add("Luke Skywalker");
incr.add(1);
nodes.add(IntsRef.deepCopyOf(curNodePath));
datatypes.add("jedi");
// close datatype object
if (states.peek() == ARRAY_OBJECT) {
popState = states.pop();
} else {
this.decrNodeObjectPath();
}
return field
+ this.getWhitespace()
+ "\""
+ ExtendedJsonTokenizer.DATATYPE_LABEL
+ "\":"
+ this.getWhitespace()
+ "\"jedi\","
+ "\""
+ ExtendedJsonTokenizer.DATATYPE_VALUES
+ "\":"
+ this.getWhitespace()
+ "\"Luke Skywalker\""
+ this.getWhitespace()
+ "},";
}
case OBJECT_VAL:
switch (rand.nextInt(8)) {
case 0: // String
return this.doValString(
"stepha" + this.getWhitespace() + "n" + this.getWhitespace() + "e");
case 1: // DOUBLE case
images.add("34.560e-9");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NUMBER]);
incr.add(1);
this.incrNodeObjectPath();
this.setLastNode(0);
nodes.add(IntsRef.deepCopyOf(curNodePath));
this.decrNodeObjectPath();
datatypes.add(XSDDatatype.XSD_DOUBLE);
states.pop(); // remove OBJECT_VAL state
return "34.560e-9" + this.getWhitespace() + ",";
case 2: // LONG case
images.add("34560e-9");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NUMBER]);
incr.add(1);
this.incrNodeObjectPath();
this.setLastNode(0);
nodes.add(IntsRef.deepCopyOf(curNodePath));
this.decrNodeObjectPath();
datatypes.add(XSDDatatype.XSD_LONG);
states.pop(); // remove OBJECT_VAL state
return "34560e-9" + this.getWhitespace() + ",";
case 3: // True
images.add("true");
types.add(ExtendedJsonTokenizer.getTokenTypes()[TRUE]);
incr.add(1);
this.incrNodeObjectPath();
this.setLastNode(0);
nodes.add(IntsRef.deepCopyOf(curNodePath));
this.decrNodeObjectPath();
datatypes.add(XSDDatatype.XSD_BOOLEAN);
states.pop(); // remove OBJECT_VAL state
return "true" + this.getWhitespace() + ",";
case 4: // False
images.add("false");
types.add(ExtendedJsonTokenizer.getTokenTypes()[FALSE]);
incr.add(1);
this.incrNodeObjectPath();
this.setLastNode(0);
nodes.add(IntsRef.deepCopyOf(curNodePath));
this.decrNodeObjectPath();
datatypes.add(XSDDatatype.XSD_BOOLEAN);
states.pop(); // remove OBJECT_VAL state
return "false" + this.getWhitespace() + ",";
case 5: // NULL
images.add("null");
types.add(ExtendedJsonTokenizer.getTokenTypes()[NULL]);
incr.add(1);
this.incrNodeObjectPath();
this.setLastNode(0);
nodes.add(IntsRef.deepCopyOf(curNodePath));
this.decrNodeObjectPath();
datatypes.add(XSDDatatype.XSD_STRING);
states.pop(); // remove OBJECT_VAL state
return "null" + this.getWhitespace() + ",";
case 6: // New array
if (states.size() <= MAX_DEPTH) {
states.pop(); // remove OBJECT_VAL state
this.incrNodeObjectPath();
states.add(ARRAY);
return "[";
}
return this.doValString("");
case 7: // new Object
if (states.size() <= MAX_DEPTH) {
states.pop(); // remove OBJECT_VAL state
// Two incrementations, because the object introduce a "blank" node
nestedObjs++;
this.incrNodeObjectPath();
this.setLastNode(0);
this.incrNodeObjectPath();
states.add(OBJECT_ATT);
return "{";
}
return this.doValString("");
}
default:
throw new IllegalStateException("Got unknown lexical state: " + states.peek());
}
}
@Override
protected void appendNullText(StringBuilder buffer) {
buffer.setLength(buffer.lastIndexOf("\n") + 1);
}