Invalid sign exception due to B* two-digit exponent misaligning TLE columns

[KrailH]

Description:

When parsing Starlink TLEs from Celestrak (https://celestrak.org/NORAD/elements/gp.php?GROUP=starlink&FORMAT=tle), SGP.NET throws:

SGPdotNET.Exception.TleException: Invalid sign

Example TLE that triggers the issue:

STARLINK-4553
1 53577U 22101BC 25345.55693763 -.00000288 00000+0 87000-10 0 9990
2 53577 53.2164 89.5151 0001372 89.9326 270.1823 15.08845301183964

Root cause:

The B* drag term (87000-10) has a two-digit exponent (-10).
In a fixed-width TLE, this shifts the preceding space and misaligns columns.
SGP.NET expects B* to be strictly 8 characters: 5-digit mantissa + 1-digit exponent.
The misalignment causes the parser to see an invalid sign and throw an exception.

Expected behavior:

SGP.NET should either:

1. Support B* values with two-digit exponents while keeping proper column alignment, or
2. Provide a way to preprocess/normalize such TLEs to compatible single-digit exponent format.

20251212095028.txt

[parzivail]

Thanks for making this issue, however, it's pretty clearly AI generated so I'm going to be a bit blunt.

I tested the TLEs from your provided link, and was able to parse all 9,119 of them without issue. In fact, I was able to parse all 13,673 active elements from Celestrak.

As far as I understand, the format has a fixed number of characters for all terms including B*, which is 8 characters.

If you have a reputable source of a TLE that contains an abnormal format, I'd be happy to take a look, but the example you gave is invalid for STARLINK-4553 anyway.

[KrailH]

Thanks for making this issue, however, it's pretty clearly AI generated so I'm going to be a bit blunt.

I tested the TLEs from your provided link, and was able to parse all 9,119 of them without issue. In fact, I was able to parse all 13,673 active elements from Celestrak.

As far as I understand, the format has a fixed number of characters for all terms including B*, which is 8 characters.

If you have a reputable source of a TLE that contains an abnormal format, I'd be happy to take a look, but the example you gave is invalid for STARLINK-4553 anyway.

The ephemeris file was not generated by AI. The file 20251212095028.txt was downloaded directly from
https://celestrak.org/NORAD/elements/gp.php?GROUP=starlink&FORMAT=tle.
This is the first time I have encountered this issue. It occurred in the morning of December 12, 2025. When I checked the ephemeris again from the same webpage in the afternoon, this issue no longer appeared.
I am not familiar with the detailed format of this type of file, so I am not sure whether this situation is normal. My English is not very good, so the English content was written with the help of AI.

[parzivail]

No problem, thanks for clarifying. The format is still invalid, so I suspect this might be an issue on Celestrak's or NORAD's end that was resolved. I think we might still benefit from allowing a file to be partially parsed instead of dying for a single invalid entry, so there are improvements that can be made nonetheless. Anyway, I might caution you against using a LLM for translation tasks since it introduces implications or tone that you might not have intended.

…

On Sun, Dec 14, 2025, 4:53 AM KrailH ***@***.***> wrote: *KrailH* left a comment (parzivail/SGP.NET#34) <#34 (comment)> Thanks for making this issue, however, it's pretty clearly AI generated so I'm going to be a bit blunt. I tested the TLEs from your provided link, and was able to parse all 9,119 of them without issue. In fact, I was able to parse all 13,673 active elements from Celestrak. As far as I understand, the format has a fixed number of characters for all terms including B*, which is 8 characters. If you have a reputable source of a TLE that contains an abnormal format, I'd be happy to take a look, but the example you gave is invalid for STARLINK-4553 anyway. The ephemeris file was not generated by AI. The file 20251212095028.txt was downloaded directly from https://celestrak.org/NORAD/elements/gp.php?GROUP=starlink&FORMAT=tle. This is the first time I have encountered this issue. It occurred in the morning of December 12, 2025. When I checked the ephemeris again from the same webpage in the afternoon, this issue no longer appeared. I am not familiar with the detailed format of this type of file, so I am not sure whether this situation is normal. My English is not very good, so the English content was written with the help of AI. — Reply to this email directly, view it on GitHub <#34 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACCTMD3H25IG6XR4FA32C534BUXQVAVCNFSM6AAAAACOZWEIEWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTMNJQGY2DIMJZG4> . You are receiving this because you commented.Message ID: <parzivail/SGP. ***@***.***>