How to ignore module imports

[nrbnlulu]

currently generating for one module like this

    context = Context(directory=root_dir)
    loader = PythonLoader(search_path=['gqlauth'], modules=['decorators'])
    renderer = MarkdownRenderer(render_module_header=False,)

    loader.init(context)
    renderer.init(context)
    header = """
# Decorates

---

    """
    modules = loader.load()
    with open(root_dir / 'docs' / 'decorators.md', 'w+') as f:
        f.write(header + renderer.render_to_string(modules))

this is the result:

I want only the functions or classes how do i remove the import statements?

[lucacorbucci]

Hi @nrbnlulu have you found a solution? I have the same problem

[nrbnlulu]

Supreme! maybe create a PR for this?

[lucacorbucci]

Yes, I'll do more tests and then open a PR.

[nrbnlulu]

I moved this to an issue to be referred in the PR... Thank U!

[lucacorbucci]

Hi, I made some more tests and I noticed that probably we don't need to open a PR. Using this code I managed to generate the markdown ignoring the imports.

This is for MarkdownRenderer:

from pydoc_markdown import PydocMarkdown
from pydoc_markdown.contrib.processors.crossref import CrossrefProcessor
from pydoc_markdown.contrib.processors.filter import FilterProcessor
from pydoc_markdown.contrib.processors.smart import SmartProcessor
from pydoc_markdown.contrib.renderers.markdown import MarkdownRenderer

config = PydocMarkdown(
    loaders=[PythonLoader(search_path=["src/Models"])],
    processors=[FilterProcessor(skip_empty_modules=True), CrossrefProcessor(), SmartProcessor()],
    renderer=MarkdownRenderer(
        render_module_header=False,
    ),
)

modules = config.load_modules()
config.process(modules)
config.render(modules)

This is for DocusaurusRenderer:

from pydoc_markdown.contrib.loaders.python import PythonLoader
from pydoc_markdown import PydocMarkdown
from pydoc_markdown.contrib.processors.crossref import CrossrefProcessor
from pydoc_markdown.contrib.processors.filter import FilterProcessor
from pydoc_markdown.contrib.processors.smart import SmartProcessor
from pydoc_markdown.contrib.renderers.docusaurus import DocusaurusRenderer

docs_path = "./docs/docs"
config = PydocMarkdown(
    loaders=[PythonLoader(search_path=["src/Models"])],
    processors=[FilterProcessor(skip_empty_modules=True), CrossrefProcessor(), SmartProcessor()],
    renderer=DocusaurusRenderer(
        docs_base_path="./docs/docs",
        sidebar_top_level_label="Models",
        relative_sidebar_path="Models.json",
        relative_output_path="Models",
    ),
)

modules = config.load_modules()
config.process(modules)
config.render(modules)

I found this code in this test: https://github.com/NiklasRosenstein/pydoc-markdown/blob/develop/test/renderers/test_docusaurus.py

[nrbnlulu]

Unfortunately I Just want to generate string... renderer.render_to_string(modules)
This is not working for me yet.. even with render_module_header=False,

[NiklasRosenstein]

Sorry for the late reply.

Since you're using the API directly rather than the command line, it appears you are missing what the filter that is usually applied by default.

modules = loader.load()

processor = FilterProcessor()
processor.process(modules)

# ...

The default configuration of the processor will remove API objects that have no docstrings and no child elements (so your class that has no docstring but has a method with a docstring will be kept).

[nrbnlulu]

Thanks I needed to cast this modules = loader.load() to list for some reason.

[ahopkins]

FWIW, I tackled this by subclassing FilterProcessor and extending the functionality to fit my needs:

class MayimFilter(FilterProcessor):
    def _match(self, obj: docspec.ApiObject) -> bool:
        result = super()._match(obj)
        if result:
            result = self._additional(obj)
        return result

    def _additional(self, obj: docspec.ApiObject) -> bool:
        if isinstance(obj, docspec.Indirection) and not obj.target.startswith(
            "."
        ):
            return False
        elif isinstance(obj, docspec.Variable) and obj.name.isupper():
            return False
        elif obj.name.startswith("_") and obj.name not in WHITELIST:
            return False
        return True

...
MayimFilter(
    documented_only=False,
    do_not_filter_modules=False,
    skip_empty_modules=True,
).process(modules, resolver)

https://github.com/ahopkins/mayim/blob/main/build_api_docs.py#L207