normalized input types handling for issue 86

[christakahashi]

changes

• reworked _bytearray to handle input types more uniformly, and added some value checking. _bytearray still defaults to a python bytearray for c_exp<=8. But it now chooses the smallest possible array.array type. This improves performance a little and halves the memory use for the array in cases where 8<e_exp<=16.
• Also removed the global definition of _bytearray and initialization of globals before the call to init_tables(). The former was only needed for find_prime_polys() when python 2.5 and earlier were used, which are not supported versions of python. (Note: global _bytearray is still needed, its just not initialized until init_tables() is called.)
• rs_encode_message now only accepts array-like objects of integral types. Its impossible to know the user's intent when given a string.
• Added a byte packing mode for RSCodec.encode() when pack=True, c_exp=16, and data is given as bytes, or a string with encoding the encoder will pack two bytes per codeword. Note odd numbers of bytes or strings that encode into an odd number of bytes cannot be packed as there is no way to pad the message out and then remove the padding without either making assumptions about the input data or including additional metadata which is not possible within a RS codeword.
• If given an encoding, RSCodec.decode() will attempt to decode the given bytes back into a string.
• Added tests for string modes

[lrq3000]

Thank you very much Chris, the PR looks very good. Unfortunately, I unluckily and coincidentally getting a huge workload suddenly for the whole week, so I cannot even read the code changes. I will review next week as soon as I can. Thank you once again a lot for your contribution and your patience. (Ps if you want to contribute other PRs meanwhile, you can consider your PR merged, I will likely merge it as is or with minor changes). 10 mai 2025 02:03:44 Chris Takahashi ***@***.***>:

…

*changes* * reworked *_bytearray* to handle input types more uniformly, and added some value checking. *_bytearray* still defaults to a python *bytearray* for *c_exp<=8*. But it now chooses the smallest possible array.array type. This improves performance a little and halves the memory use for the array in cases where *8<e_exp<=16*. * Also removed the global definition of *_bytearray* and initialization of globals before the call to *init_tables()*. The former was only needed for *find_prime_polys()* when python 2.5 and earlier were used, which are not supported versions of python. (Note: global *_bytearray* is still needed, its just not initialized until *init_tables()* is called.) * *rs_encode_message* now only accepts array-like objects of integral types. Its impossible to know the user's intent when given a string. * Added a byte packing mode for *RSCodec.encode()* when *pack=True*, *c_exp=16*, and data is given as bytes, or a string with encoding the encoder will pack two bytes per codeword. Note odd numbers of bytes or strings that encode into an odd number of bytes cannot be packed as there is no way to pad the message out and then remove the padding without either making assumptions about the input data or including additional metadata which is not possible within a RS codeword. * If given an encoding, *RSCodec.decode()* will attempt to decode the given bytes back into a string. * Added tests for string modes ---------------------------------------- *You can view, comment on, or merge this pull request online at:*   #88 *Commit Summary* * cb7c24a[cb7c24a] normalized input types handling *File Changes* (2 files[https://github.com/tomerfiliba-org/reedsolomon/pull/88/files]) * *M* src/reedsolo/reedsolo.py[https://github.com/tomerfiliba-org/reedsolomon/pull/88/files#diff-c682d3f91783d3e4d68337fa5f6613ced4674c7e06542dd9710a77d3a936592c] (155) * *M* tests/test_reedsolo.py[https://github.com/tomerfiliba-org/reedsolomon/pull/88/files#diff-8f0b24ed03fc356250f754f5287aacf686d097d2c509de6dd6146137fa7e84d4] (29) *Patch Links:* * https://github.com/tomerfiliba-org/reedsolomon/pull/88.patch * https://github.com/tomerfiliba-org/reedsolomon/pull/88.diff — Reply to this email directly, view it on GitHub[#88], or unsubscribe[https://github.com/notifications/unsubscribe-auth/AAIRFXQAVR5M46IIIT6QPR325U657AVCNFSM6AAAAAB42FLRJOVHI2DSMVQWIX3LMV43ASLTON2WKOZTGA2TGNBSGMYTMNY]. You are receiving this because you are subscribed to this thread. [Image de pistage][https://github.com/notifications/beacon/AAIRFXQP5ZRLL7HTZSQDK6325U657A5CNFSM6AAAAAB42FLRJOWGG33NNVSW45C7OR4XAZNFJFZXG5LFVJRW63LNMVXHIX3JMTHLL74KH4.gif]

[christakahashi]

I found a small bug in the new _bytearray that I've corrected in the PR.