The skip_docs parameter is passed to the index_stream:

https://github.com/wetneb/opentapioca/blob/1a26df5328aeff18752e65a2d389e2cbd007c038/opentapioca/cli.py#L117-L119

The index_stream method will skip lines:

https://github.com/wetneb/opentapioca/blob/1a26df5328aeff18752e65a2d389e2cbd007c038/opentapioca/taggerfactory.py#L74-L75

But the dumpreader has already used time to load the json, even for skipped lines:

https://github.com/wetneb/opentapioca/blob/1a26df5328aeff18752e65a2d389e2cbd007c038/opentapioca/readers/dumpreader.py#L26-L33

While working on a fix for this, IMO a good workaround is to tail the lines before sending them to the cli, e.g.:

pbzip2 -c -d -p8 latest-all.json.bz2 | tail -n +879322

(Notice here I am using a multi-threaded bzip2 implementation)

Also I suggest switching to orjson since it faster than the system json.