Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
Description
The logic used to generate the "str:" prefix for string values comes from here: https://github.com/apache/sling-org-apache-sling-feature-cpconverter/blob/a98317ba836d9acedb334518a438f4bcc8ce721d/src/main/java/org/apache/sling/feature/cpconverter/index/IndexDefinitionsJsonWriter.java#L108
For prefixes, we should probably use the logic used in Oak. AFAIK this is implemented in https://github.com/apache/jackrabbit-oak/blob/838780a7aaf9775ad0bf7b6be65c1a1619e65eb7/oak-store-spi/src/main/java/org/apache/jackrabbit/oak/json/JsonSerializer.java#L269-L291:
public void serialize(PropertyState property, Type<?> type, int index) {
if (type == BOOLEAN) {
json.value(property.getValue(BOOLEAN, index));
} else if (type == LONG) {
json.value(property.getValue(LONG, index));
} else if (type == DOUBLE) {
Double value = property.getValue(DOUBLE, index);
if (value.isNaN() || value.isInfinite()) {
json.value(TypeCodes.encode(type.tag(), value.toString()));
} else {
json.encodedValue(value.toString());
}
} else if (type == BINARY) {
Blob blob = property.getValue(BINARY, index);
json.value(TypeCodes.encode(type.tag(), blobs.serialize(blob)));
} else {
String value = property.getValue(STRING, index);
if (type != STRING || TypeCodes.split(value) != -1) {
value = TypeCodes.encode(type.tag(), value);
}
json.value(value);
}
}
public static int split(String jsonString) {
if (jsonString.startsWith(":blobId:")) { // See OAK-428
return 7;
}
else if (jsonString.length() >= 4 && jsonString.charAt(3) == ':') {
return 3;
}
else {
return -1;
}
}
So basically, the logic for String types is: if there is a ':' at position 4, then the "str:" prefix is needed. Otherwise, not. Note that we can't reuse the code since it uses the NodeState/PropertyState API.
Attachments
Issue Links
- links to