Read long log lines from file storage correctly

[hg]

Log lines may contain up to 1 MB of text, but with default settings bufio.Scanner only reads lines up to 64k in length and fails on anything larger, rendering the Scanner unusable.

As a result, the user won't see anything after the first very long line, including the line itself.

when:
  event: push

steps:
  logs:
    image: ubuntu:24.04
    commands:
      - seq 1 3
      - tr -dc '[:alnum:]' </dev/urandom | head -c70k
      - echo
      - seq 4 6

The output should be:

1
2
3
zkwmLrpYmk6XcgJZiYqzNO3hsIV26HJfVQfyvSo1ccrMqP ... very long line of random text
4
5
6

Instead, Woodpecker cuts everything after 3:

1
2
3
(end)

I tested this on jobs with up to 20 MBs of output (simply update 70k with something like 20M).

[6543]

please dont forcepush if you can! it makes reviews harder and we squashmerge anyway ...

[woodpecker-bot]

Deployment of preview was torn down

[6543]

https://ci.woodpecker-ci.org/repos/3780/pipeline/19786/24 ... well

[6543]

thanks for not running at the dedicated pull :/

[hg]

Since this issue crops up in other pull requests too, it might be easier on you to mention it in the docs or in a pull request template (example).

[6543]

If you d'like to create a pull? 🙏

[hg]

Sure, I'll have to think about proper messaging for a moment.