// TODO also substract place number in out
// TODO also contains (or more comparison)
// TODO also contains (or more)
private List<CandidateField> getFromOutByPathAndValue(
HashMap<String, CandidateField> foundOutFields, CandidateField inQueryField) {
ArrayList<CandidateField> res = new ArrayList<CandidateField>();
CandidateField found = foundOutFields.get(inQueryField.getPath());
if (found != null && inQueryField.getValue().equals(found.getValue())) {
res.add(found);
}
return res;
}
/**
* XML differs from JSON in : - typing, which is like "byPath", in a perfect way when it's on
* concrete (business) types, and in a less accurate way when it matches only on extended
* (including primitive) types (ex. similar to a getBySubpath, or even exactly a getByValue if
* primitive type). They can be handled in a good way by prefixing by ns. NB. without typing & ns
* (or empty ns everywhere) it is similar to JSON & "byPath". - attributes, which are typed fields
* like others, or field-typed fields if no typing. They can be handled in a good way by prefixing
* by ns and "@".
*
* @param xmlInFields
* @param xmlOutFields
*/
private void correlateXML(
HashMap<String, CandidateField> xmlInFields, HashMap<String, CandidateField> xmlOutFields) {
// looking for correlations :
int level = 16;
// if in content, value and path correlations
ArrayList<Object[]> correlations = new ArrayList<Object[]>();
for (CandidateField inContentField : xmlInFields.values()) {
List<CandidateField> foundByTypeAndValueFields =
getFromOutByTypeAndValue(xmlOutFields, inContentField);
if (foundByTypeAndValueFields.size() != 0) {
for (CandidateField field : foundByTypeAndValueFields) {
// TODO also contains (or more comparison)
correlations.add(new Object[] {level, inContentField, field, "byTypeAndValue"});
}
level = level / 4;
}
List<CandidateField> foundByPathAndValueFields =
getFromOutByPathAndValue(xmlOutFields, inContentField);
if (foundByPathAndValueFields.size() != 0) {
for (CandidateField field : foundByPathAndValueFields) {
// TODO also contains (or more comparison)
correlations.add(
new Object[] {
level, inContentField, field, "byPathAndValue"
}); // not really interesting since already found by type ; else containing path ??
}
level = level / 4;
}
List<CandidateField> foundByNameAndValueFields =
getFromOutByNameAndValue(xmlOutFields, inContentField);
foundByNameAndValueFields.removeAll(
foundByPathAndValueFields); // removing fields already found by path
if (foundByNameAndValueFields.size() != 0) {
for (CandidateField field : foundByNameAndValueFields) {
// TODO also contains (or more comparison)
// substract depth to level
correlations.add(
new Object[] {
level - field.getPath().split("/").length * 2,
inContentField,
field,
"byNameAndValue"
});
}
// level = level / 2;
}
}
System.out.println("Found correlations :");
for (Object[] correlation : correlations) {
System.out.println(
correlation[0] + "\t" + correlation[1] + "\t" + correlation[2] + "\t" + correlation[3]);
}
}
/**
* First correlation algorithm : tries to find correlation - if in content, value and path
* correlations - else (otherwise lowered) if in query, value and (path or name) correlations -
* else (otherwise lowered) for all path elements, value correlations
*
* @param jsonExchange
* @param inPathFields
* @param inQueryFields
* @param inContentFields
* @param foundOutFields
*/
private void correlate(
ExchangeRecord jsonExchange,
HashMap<String, CandidateField> inPathFields,
HashMap<String, CandidateField> inQueryFields,
HashMap<String, CandidateField> inContentFields,
HashMap<String, CandidateField> foundOutFields) {
// looking for correlations :
int level = 16;
// if in content, value and path correlations
// ArrayList<ReqResFieldCorrelation> correlations = new ArrayList<ReqResFieldCorrelation>();
HashMap<Integer, CorrelationLevel> correlationLevels = new HashMap<Integer, CorrelationLevel>();
if (jsonExchange.getInMessage().getPostData() != null) {
addCorrelationsFromOutByPathOrNameAndValue(
correlationLevels, level, foundOutFields, inContentFields);
level = level / 4;
}
// else (otherwise lowered) if in query, value and path correlations
if (jsonExchange.getInMessage().getQueryString() != null) {
addCorrelationsFromOutByPathOrNameAndValue(
correlationLevels, level, foundOutFields, inQueryFields);
level = level / 4;
}
// else (otherwise lowered) for all path elements, value correlations
for (CandidateField inPathField : inPathFields.values()) {
List<CandidateField> foundByValueFields = getFromOutByValue(foundOutFields, inPathField);
for (CandidateField field : foundByValueFields) {
// exact only
// correlations.add(new ReqResFieldCorrelation(level - field.getPath().split("/").length*2,
// inPathField, field, "byValue"));
// correlationLevels.putCorrelation(new ReqResFieldCorrelation(level -
// field.getPath().split("/").length*2, inPathField, field, "byValue"));
putCorrelation(
correlationLevels,
new ReqResFieldCorrelation(
level - field.getPath().split("/").length * 2, inPathField, field, "byValue"));
}
}
// printCorrelations(correlations);
int inFieldNb = inPathFields.size() + inQueryFields.size() + inContentFields.size();
int outFieldNb = foundOutFields.size();
printCorrelations(correlationLevels, inFieldNb, outFieldNb);
}
/**
* Second correlation algorithm : tries to find correlation - if in content, value and (sub)path
* correlations - else (otherwise lowered) if in query, value and (sub)path correlations - else
* (otherwise lowered) for all path elements, value correlations
*
* @param jsonExchange
* @param inPathFields
* @param inQueryFields
* @param inContentFields
* @param foundOutFields
*/
private void correlateWithSubpath(
ExchangeRecord jsonExchange,
HashMap<String, CandidateField> inPathFields,
HashMap<String, CandidateField> inQueryFields,
HashMap<String, CandidateField> inContentFields,
HashMap<String, CandidateField> foundOutFields) {
// looking for correlations :
int level = 16;
// if in content, value and path correlations
ArrayList<Object[]> correlations = new ArrayList<Object[]>();
if (jsonExchange.getInMessage().getPostData() != null) {
addCorrelationsFromOutBySubpathAndValue(correlations, level, foundOutFields, inContentFields);
level = level / 4;
}
// else (otherwise lowered) if in query, value and path correlations
if (jsonExchange.getInMessage().getQueryString() != null) {
addCorrelationsFromOutBySubpathAndValue(correlations, level, foundOutFields, inQueryFields);
level = level / 4;
}
// else (otherwise lowered) for all path elements, value correlations
for (CandidateField inPathField : inPathFields.values()) {
List<CandidateField> foundByValueFields = getFromOutByValue(foundOutFields, inPathField);
for (CandidateField field : foundByValueFields) {
// exact only
correlations.add(
new Object[] {
level - field.getPath().split("/").length * 2, inPathField, field, "byValue"
});
}
}
System.out.println("Found correlations with subpath :");
for (Object[] correlation : correlations) {
System.out.println(
correlation[0] + "\t" + correlation[1] + "\t" + correlation[2] + "\t" + correlation[3]);
}
System.out.println();
}
private void addCorrelationsFromOutByPathOrNameAndValue(
HashMap<Integer, CorrelationLevel> correlationLevels,
int level,
HashMap<String, CandidateField> foundOutFields,
HashMap<String, CandidateField> inFields) {
for (CandidateField inField : inFields.values()) {
int queryLevel = level;
List<CandidateField> foundByPathAndValueFields =
getFromOutByPathAndValue(foundOutFields, inField);
if (foundByPathAndValueFields.size() != 0) {
for (CandidateField field : foundByPathAndValueFields) {
// TODO also contains (or more comparison)
ReqResFieldCorrelation reqResFieldCorrelation =
new ReqResFieldCorrelation(queryLevel, inField, field, "byPathAndValue");
putCorrelation(correlationLevels, reqResFieldCorrelation);
}
queryLevel = queryLevel / 4;
}
List<CandidateField> foundByNameAndValueFields =
getFromOutByNameAndValue(foundOutFields, inField);
foundByNameAndValueFields.removeAll(
foundByPathAndValueFields); // removing fields already found by path
if (foundByNameAndValueFields.size() != 0) {
for (CandidateField field : foundByNameAndValueFields) {
// TODO also contains (or more comparison)
// substract depth to level
ReqResFieldCorrelation reqResFieldCorrelation =
new ReqResFieldCorrelation(
queryLevel - field.getPath().split("/").length * 2,
inField,
field,
"byNameAndValue");
putCorrelation(correlationLevels, reqResFieldCorrelation);
}
// queryLevel = queryLevel / 2;
}
}
}
/**
* @param obj
* @param foundFields
* @param pathStack
*/
private void putCandidateFields(
Object obj, HashMap<String, CandidateField> foundFields, Stack<Object> pathStack) {
CandidateField candidateField = new CandidateField(toPath(pathStack), String.valueOf(obj));
foundFields.put(candidateField.getPath(), candidateField);
}
/**
* TODO also contains (or more comparison) TODO merge fullpath with subpath, once
*
* @param correlations
* @param level
* @param foundOutFields
* @param inField
* @return
*/
private int addCorrelationsFromOutBySubpathAndValue(
ArrayList<Object[]> correlations,
int level,
HashMap<String, CandidateField> foundOutFields,
CandidateField inField) {
// ArrayList<CandidateField> res = new ArrayList<CandidateField>();
// full path :
CandidateField fullPathOutField = foundOutFields.get(inField.getPath());
if (fullPathOutField != null) {
if (inField.getValue().equals(fullPathOutField.getValue())) {
// res.add(found);
// TODO also more comparison
correlations.add(new Object[] {level, inField, fullPathOutField, "byPathAndValue"});
// either hit hard on level or even stop there
level = level - 6; // 4
// return level;
} else {
// path only, trying out ?!?
correlations.add(new Object[] {level / 4, inField, fullPathOutField, "byPath"}); // ?!?
}
}
int matchingInFieldSubPathDepth = inField.getPath().split("/").length;
// full path partially :
boolean foundAtLeastOne = false;
for (CandidateField outField : foundOutFields.values()) {
if (outField.getPath().endsWith('/' + inField.getPath())
&& inField.getValue().equals(outField.getValue())) {
// res.add(outField);
// TODO also more comparison
// outDepth is bad (at least 1), more than 1 matchingInFieldSubPathDepth is good
int outDepth = outField.getPath().split("/").length - matchingInFieldSubPathDepth;
correlations.add(
new Object[] {
level - outDepth * 2 + 2 * (matchingInFieldSubPathDepth - 1),
inField,
outField,
"byPartialPathAndValue"
});
foundAtLeastOne = true;
}
}
if (foundAtLeastOne) {
// either hit hard on level or even stop there
level = level - 6; // 4
// return level;
}
int inDepth = 1; // = nonMatchingInFieldSubpathDepth
int indexOfSlashPlusOne = 0;
while ((indexOfSlashPlusOne = inField.getPath().indexOf('/', indexOfSlashPlusOne) + 1) != 0) {
String subpath = inField.getPath().substring(indexOfSlashPlusOne);
matchingInFieldSubPathDepth--;
inDepth++;
boolean foundAtLeastOneInSubpath = false;
for (CandidateField outField : foundOutFields.values()) {
if (outField.getPath().equals(subpath)
|| outField.getPath().endsWith('/' + subpath)
&& inField.getValue().equals(outField.getValue())) {
// res.add(outField);
// TODO also more comparison
// outDepth is bad, inDepth worse (but already >= 1), more than 1
// matchingInFieldSubPathDepth is good
int outDepth = outField.getPath().split("/").length - matchingInFieldSubPathDepth;
correlations.add(
new Object[] {
level - outDepth * 2 - inDepth * 2 + 2 * (matchingInFieldSubPathDepth - 1),
inField,
outField,
"bySubpathAndValue:" + subpath
});
foundAtLeastOneInSubpath = true;
}
}
if (foundAtLeastOneInSubpath) {
// either hit hard on level or even stop there
level = level - 6; // 4
// return level;
}
}
return level;
}
private void putField(HashMap<String, CandidateField> fields, CandidateField candidateField) {
fields.put(candidateField.getPath(), candidateField);
}