yeah, tricky, it comes from the pypi json payload
Grayskull also tries to fetch info from there, for some reason is putting a breakline after the end of the URL
this is part of the json response from pypi for that package:

"project_url": "https://pypi.org/project/pyospackage/",
    "project_urls": {
      "Documentation": "https://github.com/pyopensci/pyospackage#readme",
      "Homepage": "https://pypi.org/project/pyospackage/",
      "Issues": "https://github.com/pyopensci/pyospackage/issues",
      "Source": "https://github.com/pyopensci/pyospackage"
    },

can you see that #readme at the end of Documentation? it is coming from there. I hope that helps, if you need more help with that please let me know :)

Woah, thanks a lot @marcelotrevisani for the advice and quick response! Not sure when I'll be able to make a second pass at this. Looks like maybe something needs to be quoted/escaped?

Woah, thanks a lot @marcelotrevisani for the advice and quick response! Not sure when I'll be able to make a second pass at this. Looks like maybe something needs to be quoted/escaped?

Yeah, probably it does need some escaping. I can try to take a look later to give you a hint about that

@maresb keep in mind that a lot of the code in play here is part of souschef

@msarahan should we submit a bug report to souchef repository as well? i'm just watching this. it would be nice to fix this because otherwise users will get that linting error when they submit their recipe and need to figure out how to fix things.

I think that there are two stacked bugs here.

One bug has to do with the metadata getting passed around without being accidentally dropped. The other has to do with the # in the URL being interpreted as a comment.

I actually think that I fixed the bug with the metadata being dropped, but unmasking the bug due to the # in @lwasser's URL.

@lwasser, as a temporary workaround could you try removing the #... part? I'm thinking that it may be easier to proceed with one bug at a time.

Restarting the CI due to the connection error...

It has seemed like @maresb has this covered. Ben, if you need my help, please ping me. Otherwise, I'm leaving it in your capable hands and cheering you on.

ok great! thanks @msarahan !!
@maresb can you remind me - what link are you referring to with the commented out element? i'm happy to test against this branch but i'm unclear about what you have identified that i can modify. many thanks! ❇️

@lwasser it's your docs link: https://github.com/pyopensci/pyospackage#readme. I think the current PR would fix the recipe if your docs link were instead simply https://github.com/pyopensci/pyospackage.

@lwasser it's your docs link: https://github.com/pyopensci/pyospackage#readme. I think the current PR would fix the recipe if your docs link were instead simply https://github.com/pyopensci/pyospackage.

Thank you. i see that now. @maresb in testing this can i run grayskull against a sdist locally so i can skip releasing another version to pypi? i keep digging through the docs but don't see that option (may be user error)?

You can, but you'll need to adjust the source section.

From README.md:

You can also generate a recipe from a local sdist archive:

grayskull pypi ./pytest-5.3.5.tar.gz

Note that such a recipe isn't really portable as it will depend on the local path of the sdist file. It can be useful if you want to automatically generate a conda package.

Thank you everyone @msarahan i missed that in the readme. i was checking the docs. As a side note would folks here be open to a documentation pr from me that had the get started info in the readme? it was confusing to me as i saw docs and assumed that info would be there rather than in the readme! I suppose the alternative coming with less maintenance would be to just include the readme in the docs! just a thought :)

@maresb Ok i was not understanding the issue and now i see it. There is no reason to have a # in a url as you point out. when i remove that my metadata render perfectly (from your branch) with grayskull running in editable mode on my computer.

The metadata / recipe now look like this:

{% set name = "pyospackage" %}
{% set version = "0.1.10" %}

package:
  name: {{ name|lower }}
  version: {{ version }}

source:
  url: file:///Users/leahawasser/Documents/GitHub/pyos/pyospackage/dist/pyospackage-0.1.10.tar.gz
  sha256: 1db1ceb1d769a081c2945d235016ddb6dfe1b555c68014704c978b198b841ded

build:
  noarch: python
  script: {{ PYTHON }} -m pip install . -vv --no-deps --no-build-isolation
  number: 0

requirements:
  host:
    - python >=3.9
    - hatchling
    - pip
  run:
    - python >=3.9

test:
  imports:
    - pyospackage
  commands:
    - pip check
  requires:
    - pip

about:
  home: https://pypi.org/project/pyospackage/
  summary: A package that adds numbers together
  license: MIT
  license_file: LICENSE

extra:
  recipe-maintainers:
    - lwasser

i wonder if this pr could be merged as i am not seeing anything else broken here in the output. thank you all for walking me through what was going on. i think i misunderstood above what "breaking things" related to.

Thank you everyone @msarahan i missed that in the readme. i was checking the docs. As a side note would folks here be open to a documentation pr from me that had the get started info in the readme? it was confusing to me as i saw docs and assumed that info would be there rather than in the readme! I suppose the alternative coming with less maintenance would be to just include the readme in the docs! just a thought :)

@maresb Ok i was not understanding the issue and now i see it. There is no reason to have a # in a url as you point out. when i remove that my metadata render perfectly (from your branch) with grayskull running in editable mode on my computer.

The metadata / recipe now look like this:

{% set name = "pyospackage" %}
{% set version = "0.1.10" %}

package:
  name: {{ name|lower }}
  version: {{ version }}

source:
  url: file:///Users/leahawasser/Documents/GitHub/pyos/pyospackage/dist/pyospackage-0.1.10.tar.gz
  sha256: 1db1ceb1d769a081c2945d235016ddb6dfe1b555c68014704c978b198b841ded

build:
  noarch: python
  script: {{ PYTHON }} -m pip install . -vv --no-deps --no-build-isolation
  number: 0

requirements:
  host:
    - python >=3.9
    - hatchling
    - pip
  run:
    - python >=3.9

test:
  imports:
    - pyospackage
  commands:
    - pip check
  requires:
    - pip

about:
  home: https://pypi.org/project/pyospackage/
  summary: A package that adds numbers together
  license: MIT
  license_file: LICENSE

extra:
  recipe-maintainers:
    - lwasser

i wonder if this pr could be merged as i am not seeing anything else broken here in the output. thank you all for walking me through what was going on. i think i misunderstood above what "breaking things" related to.

I'm glad that this is sorted out, but that is still a bug that I can fix it later, because it should be able to handle the # fine. It seems something related to the parsing and ruamel.yaml
I have been quite busy these days with my day work and personal life that I didn't have much time to look into it, but I hope to fix this soon :)

Thank you everybody for putting efforts into this.

@maresb , would you like to change the status of this PR to "ready for review"? or are you going to work more in this front?