Before computing similarity scores and matching, we normally need to check the count the encodings to see if it is consistent with blocking file.
Currently we either load the whole JSON just to get the number of encodings or use ijson to iteratively count the number of encodings.
It would be better to store the count in the metadata and just read this metadata when needing the count.
Before computing similarity scores and matching, we normally need to check the count the encodings to see if it is consistent with blocking file.
Currently we either load the whole JSON just to get the number of encodings or use ijson to iteratively count the number of encodings.
It would be better to store the count in the metadata and just read this metadata when needing the count.