Bug in converter/history2db.js

[Alpini1980]

Hello,

I found a small bug in the converter/history2db.js which prevents some datapoints to be converted.

I converted a history.0 to a sql.0 instance (MSSQL-Server) with your script and most but not all datapoints were converted. So I searched for the reason why some datapoints were missing in SQL. The script said for example:

Ignore ID history.smartmeter.0.1-0~2_8_0__255.value.json: smartmeter.0.1-0~2_8_0__255.value

The datapoint-name in ioBroker and also SQL for this datapoint is

smartmeter.0.1-0:2_8_0__255.value but the filename is:
smartmeter.0.1-0~2_8_0__255.value.json

So the history-instance replaces ":" chars in filenames with "~" chars because ":" aren't allowed in filenames. But the converter script don't take these replaced chars into account.

As a quick and dirty workaround I changed line 198 from

const id = file.substring(8, file.length - 5);

to

const id = file.substring(8, file.length - 5).replace('~', ':');

which worked for me. But I am not sure if other illegal file name chars can cause the same problem so this shouldn't be a general solution.

Regards, Stefan

[Apollon77]

I think this is a problem .....
The full logic is in

ioBroker.history/lib/getHistory.js

Line 66 in f6d4094

const safeId = id.replace(/[\u0000|*,;"'<>?:\/\\]/g, '~');

... so there is a lot more that gets replaced and the problem currently is that we can not reliably map back this because the pure JSON content does not contain any information on the real object ID ... so I have basically no real solution for that tbh