'utf-8' codec can't decode

[paul-gauthier]

          mypath.../node_modules/faker/lib/locales/it/name/first_name.js: 'utf-8' codec can't

decode byte 0xf9 in position 5170: invalid start byte

Is there anyway to have Aider ignore node_modules?

I'm not sure if this issue is related, but I've already tried upgrading Aider.

It also seems to hang.

du -sh .aider.*
12K .aider.chat.history.md
27M .aider.ident.cache.v1
4.0K .aider.input.history
166M .aider.tags.cache.v1
(AI-CODING-ENV) neo@tulum  ~/Documents/Business/SmallCloud/git/ec-app-frontend   main ±  wc -l .aider.tags.cache.v1
wc: .aider.tags.cache.v1: read: Is a directory
(AI-CODING-ENV) ✘ neo@tulum  ~/Documents/Business/SmallCloud/git/ec-app-frontend   main ±  ls .aider.tags.cache.v1
00 13 26 38 4c 5e 70 84 9a b1 c2 d3 e5 fa
01 14 27 39 4d 5f 71 86 9d b2 c3 d4 e6 fb
02 15 28 3a 4e 60 72 87 9e b3 c4 d6 e7 fc
03 16 29 3b 4f 61 73 88 9f b4 c5 d7 e8 fd
04 17 2b 3d 50 62 74 89 a0 b5 c6 d8 e9 fe
05 18 2c 3e 51 63 75 8b a1 b6 c7 d9 ea
07 19 2d 3f 52 64 76 8c a2 b7 c8 da ec
08 1a 2e 41 53 65 78 8d a4 b8 c9 db ef
09 1b 2f 42 54 66 79 8e a5 b9 ca dc f0
0a 1c 30 43 56 67 7a 8f a7 ba cache.db dd f1
0b 1d 31 44 57 68 7c 91 a8 bb cb de f2
0c 1e 32 46 58 69 7e 94 a9 bc cc df f3
0d 1f 33 47 59 6a 7f 95 aa bd cd e0 f4
0e 21 34 48 5a 6b 80 96 ab be ce e1 f6
0f 22 35 49 5b 6c 81 97 ac bf d0 e2 f7
10 23 36 4a 5c 6e 82 98 ae c0 d1 e3 f8
12 25 37 4b 5d 6f 83 99 af c1 d2 e4 f9

My project is 241M if I include Aiders 166M cache (so far).... it may have been caching things, but i'm not sure. It seems to be hanging....

Originally posted by @funkytaco in #56 (comment)

[paul-gauthier]

Thanks for trying aider and reporting this issue. You seem to have 2 issues going on here:

Aider is scanning all of your node_modules directory.

Aider scans all the files in your git repo once, and then re-scans any files which are modified.

Have you added node_modules and all of its content to your git repo? Normally folks do not do that. If it is in your repo, that would explain why aider trying to scan all of those files.

Aider is encountering lots of non-utf8 files.

Aider prints a warning message if it encounters a file which does not match the encoding it is using. The warnings are harmless as long as those files aren't important to the work you want to do with GPT. Aider will remember that the file can't be decoded and ignore it.

Or, if all your files are in a non-utf8 encoding, you can specify it with the --encoding command line switch.

[funkytaco]

So are you saying to add node_modules to .gitignore?

Anyhow, I tried on an empty project and it's working really well with React. This project is stellar.

[paul-gauthier]

Yes, node_modules is usually added to .gitignore.

But it looks like those files have already been added to your git repo? If so you will also need to remove them. You should see if git rm --cached node_modules would do what you want. But please confirm and understand any risks of using that command before running it.

[paul-gauthier]

I'm going to close this issue for now, but feel free to re-open or file a new issue any time.

[alexturpin]

@paul-gauthier I'm running into this in a repo with lots of committed, non code files (e.g. PDFs). I understand that it's harmless but the warning still causes a big wall of red whenever Aider first tries to help! Just wanted to add my use case out here, maybe there's a world where we can have a flag to silence this? Would be happy to contribute.

[DWDuq]

I'm having this issue and it is a problem for me because I'm adding the pdf as a reference using the /read command. The output isn't very good, so I'm guessing it's because it cannot read the pdf due to this error

[paul-gauthier]

The read command doesn't support pdf at this time.