Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Generate recipe from project tree using pypa/build #541

Draft
wants to merge 16 commits into
base: main
Choose a base branch
from
Draft
Show file tree
Hide file tree
Changes from 12 commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 8 additions & 0 deletions grayskull/base/factory.py
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,11 @@
from souschef.recipe import Recipe

from grayskull.strategy.cran import CranStrategy

try:
from grayskull.strategy.py_build import PyBuild
except ImportError: # requires conda
PyBuild = None
from grayskull.strategy.pypi import PypiStrategy


Expand All @@ -16,6 +21,9 @@ class GrayskullFactory(ABC):
"cran": CranStrategy,
}

if PyBuild:
REGISTERED_STRATEGY["pybuild"] = PyBuild

@staticmethod
def create_recipe(repo_type: str, config, pkg_name=None, sections_populate=None):
if repo_type.lower() not in GrayskullFactory.REGISTERED_STRATEGY:
Expand Down
1 change: 1 addition & 0 deletions grayskull/config.py
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,7 @@ class Configuration:
is_arch: bool = False
repo_github: Optional[str] = None
from_local_sdist: bool = False
from_tree: bool = False
local_sdist: Optional[str] = None
missing_deps: set = field(default_factory=set)
extras_require_test: Optional[str] = None
Expand Down
27 changes: 22 additions & 5 deletions grayskull/license/discovery.py
Original file line number Diff line number Diff line change
Expand Up @@ -62,6 +62,18 @@
]


def _match_scrambled_exact(candidate, licenses) -> str | None:
Fixed Show fixed Hide fixed
"""
Return license with rearranged word order only.

Fancy scorer confuses BSD-3-Clause with DEC-3-Clause.
"""
bag = set(re.findall(r"\w+", candidate.lower()))
for license in licenses:
if bag == set(re.findall(r"\w+", license.lower())):
return license


def match_license(name: str) -> dict:
"""Match if the given license name matches any license present on
spdx.org
Expand All @@ -75,11 +87,16 @@
name = re.sub(r"\s+license\s*", "", name.strip(), flags=re.IGNORECASE)
name = name.strip()

best_matches = process.extract(
name, _get_all_license_choice(all_licenses), scorer=partial_ratio
)
best_matches = process.extract(name, [lc for lc, *_ in best_matches])
spdx_license = best_matches[0]
exact_match = _match_scrambled_exact(name, _get_all_license_choice(all_licenses))
if exact_match:
best_matches = [(exact_match, 100, 0)]
spdx_license = best_matches[0]
else:
best_matches = process.extract(
name, _get_all_license_choice(all_licenses), scorer=partial_ratio
)
best_matches = process.extract(name, [lc for lc, *_ in best_matches])
spdx_license = best_matches[0]

if spdx_license[1] < 100:
# Prefer "-or-later" licenses over the "-only"
Expand Down
15 changes: 13 additions & 2 deletions grayskull/main.py
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,12 @@
from grayskull.cli import CLIConfig
from grayskull.cli.stdout import print_msg
from grayskull.config import Configuration
from grayskull.utils import generate_recipe, origin_is_github, origin_is_local_sdist
from grayskull.utils import (
generate_recipe,
origin_is_github,
origin_is_local_sdist,
origin_is_tree,
)

init(autoreset=True)
logging.basicConfig(format="%(levelname)s:%(message)s")
Expand Down Expand Up @@ -283,10 +288,13 @@ def generate_recipes_from_list(list_pkgs, args):
for pkg_name in list_pkgs:
logging.debug(f"Starting grayskull for pkg: {pkg_name}")
from_local_sdist = origin_is_local_sdist(pkg_name)
from_tree = origin_is_tree(pkg_name)
if origin_is_github(pkg_name):
pypi_label = ""
elif from_local_sdist:
pypi_label = " (local)"
elif from_tree:
pypi_label = " (tree)"
else:
pypi_label = " (pypi)"
print_msg(
Expand All @@ -304,6 +312,7 @@ def generate_recipes_from_list(list_pkgs, args):
url_pypi_metadata=args.url_pypi_metadata,
sections_populate=args.sections_populate,
from_local_sdist=from_local_sdist,
from_tree=from_tree,
extras_require_test=args.extras_require_test,
github_release_tag=args.github_release_tag,
extras_require_include=tuple(args.extras_require_include),
Expand Down Expand Up @@ -333,7 +342,9 @@ def create_python_recipe(pkg_name, sections_populate=None, **kwargs):
config = Configuration(name=pkg_name, **kwargs)
return (
GrayskullFactory.create_recipe(
"pypi", config, sections_populate=sections_populate
"pybuild" if config.from_tree else "pypi",
config,
sections_populate=sections_populate,
),
config,
)
Expand Down
156 changes: 156 additions & 0 deletions grayskull/strategy/py_build.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,156 @@
"""
Use pypa/build to get project metadata from a checkout. Create a recipe suitable
for inlinining into the first-party project source tree.
"""

import logging
import tempfile
from importlib.metadata import PathDistribution
from pathlib import Path

import build
from conda.exceptions import InvalidMatchSpec
from conda.models.match_spec import MatchSpec
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Having to import conda will make grayskull more difficult to install.
Is it worth adding conda as dependency just for that function?

I haven't checked how requirements are parsed in the current code.

from packaging.requirements import Requirement
from souschef.recipe import Recipe

from grayskull.config import Configuration
from grayskull.strategy.abstract_strategy import AbstractStrategy
from grayskull.strategy.pypi import compose_test_section

log = logging.getLogger(__name__)


class PyBuild(AbstractStrategy):
@staticmethod
def fetch_data(recipe: Recipe, config: Configuration, sections=None):
project = build.ProjectBuilder(config.name)

recipe["source"]["path"] = "../"
# XXX relative to output. "git_url: ../" is also a good choice.

with tempfile.TemporaryDirectory(prefix="grayskull") as output:
build_system_requires = project.build_system_requires
requires_for_build = project.get_requires_for_build("wheel")
# If those are already installed, we can get the extras requirements
# without invoking pip e.g. setuptools_scm[toml]
print("Requires for build:", build_system_requires, requires_for_build)

# Example of finding extra dependencies for a distribution 'scm' (is
# there a dict API?) Subtract "has non-extra marker" dependencies from
# this set.
#
# for e in scm.metadata.get_all('provides-extra'):
# print (e, [x for x in r if x.marker and x.marker.evaluate({'extra':e})])
#
# docs [<Requirement('entangled-cli[rich]; extra == "docs"')>, <Requirement('mkdocs; extra == "docs"')>, <Requirement('mkdocs-entangled-plugin; extra == "docs"')>, <Requirement('mkdocs-material; extra == "docs"')>, <Requirement('mkdocstrings[python]; extra == "docs"')>, <Requirement('pygments; extra == "docs"')>]
# rich [<Requirement('rich; extra == "rich"')>]
# test [<Requirement('build; extra == "test"')>, <Requirement('pytest; extra == "test"')>, <Requirement('rich; extra == "test"')>, <Requirement('wheel; extra == "test"')>]
# toml []

# build the project's metadata "dist-info" directory
metadata_path = Path(project.metadata_path(output_directory=output))

distribution = PathDistribution(metadata_path)

# real distribution name not pathname
config.name = distribution.name # see also _normalized_name
config.version = distribution.version

# grayskull thought the name was the path. correct that.
if recipe[0] == '#% set name = "." %}': # XXX fragile
# recipe[0] = x does not work
recipe._yaml._yaml_get_pre_comment()[0].value = (
f'#% set name = "{config.name}" %}}\n'
f'#% set version = "{config.version}" %}}\n'
)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Instead of that, you can use

set_global_jinja_var(recipe, "name", config.name)
set_global_jinja_var(recipe, "version", config.version)


recipe["package"]["version"] = "{{ version }}"
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This puts quotes around version: '{{ version }}'; how do we control the YAML more closely with conda-souschef's Recipe?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

conda-souschef is just a wrapper around ruamel yaml, you can still manipulate the ruamel yaml object directly.
you can do something like:

recipe["package"]["version"].value = "{{ version }}"

elif config.name not in str(recipe[0]):
log.warning("Package name not found in first line of recipe")

metadata = distribution.metadata

requires_python = metadata["requires-python"]
if requires_python:
requires_python = f"python { requires_python }"
else:
requires_python = "python"

recipe["requirements"]["host"] = [requires_python] + sorted(
(*build_system_requires, *requires_for_build)
)

requirements = [Requirement(r) for r in distribution.requires or []]
active_requirements = [
str(r).rsplit(";", 1)[0]
for r in requirements
if not r.marker or r.marker.evaluate()
]
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How do we include the marker as a YAML comment?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These are evaluated in the environment grayskull runs in, and might not be correct for the end user.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

there is a property called .comment that you can use for that
a recipe/section object uses this mixin
https://github.com/marcelotrevisani/souschef/blob/92cee9159fb65fa035593f1a8b60a931413ab581/src/souschef/mixins.py#L14

# XXX to normalize space between name and version, MatchSpec(r).spec
normalized_requirements = []
for requirement in active_requirements:
try:
normalized_requirements.append(
# MatchSpec uses a metaclass hiding its constructor from
# the type checker
MatchSpec(requirement).spec # type: ignore
)
except InvalidMatchSpec:
log.warning("%s is not a valid MatchSpec", requirement)
normalized_requirements.append(requirement)

# conda does support ~=3.0.0 "compatibility release" matches
recipe["requirements"]["run"] = [requires_python] + normalized_requirements
# includes extras as markers e.g. ; extra == 'testing'. Evaluate
# using Marker().

recipe["build"]["entry_points"] = [
f"{ep.name} = {ep.value}"
for ep in distribution.entry_points
if ep.group == "console_scripts"
]

recipe["build"]["noarch"] = "python"
recipe["build"][
"script"
] = "{{ PYTHON }} -m pip install . -vv --no-deps --no-build-isolation"
# XXX also --no-index?
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would be nice to be careful about not fetching pip requirements, even the build-system requirements (poetry, flit-core, hatchling, ...)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sorry, I think I didn't follow, what do you mean?


# distribution.metadata.keys() for grayskull is
# Metadata-Version
# Name
# Version
# Summary
# Author-email
# License
# Project-URL
# Keywords
# Requires-Python
# Description-Content-Type
# License-File
# License-File
# Requires-Dist (many times)
# Provides-Extra (several times)
# Description or distribution.metadata.get_payload()

about = {
"summary": metadata["summary"],
"license": metadata["license"],
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

On an example I tested, license was set to the full license text content. The pyproject.toml was using license = {file = "LICENSE"}. In the recipe, we want the SPDX identifier.

# there are two license-file in grayskull e.g.
"license_file": metadata["license-file"],
}
recipe["about"] = about

metadata_dict = dict(
metadata
) # XXX not what compose_test_section expects at all
metadata_dict["name"] = config.name
metadata_dict["entry_points"] = [
f"{ep.name} = {ep.value}"
for ep in distribution.entry_points
if ep.group == "console_scripts"
]
recipe["test"] = compose_test_section(metadata_dict, [])
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe this sort of "copy all fields from one format to another" code is destined to be messy...


# raise NotImplementedError()
8 changes: 7 additions & 1 deletion grayskull/utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -108,6 +108,12 @@ def origin_is_local_sdist(name: str) -> bool:
)


def origin_is_tree(name: str) -> bool:
"""Return True if it is a directory"""
path = Path(name)
return path.is_dir()
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We could look for pyproject.toml or setup.py; it should work with either.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

that can work, but you also need to look for setup.cfg

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

With the pypa/build method, pyproject.toml is the only necessary file to get the metadata.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I agree, but not all projects uses it :/

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pypa/build works with setup.py or pyproject.toml. It asks the underlying build system to create a wheel with package.dist-info/METADATA. Then we read METADATA and not pyproject.toml. So with this technique we don't touch setup.py, setup.cfg or pyproject.toml; pypa/build writes METADATA and we read that.



def sha256_checksum(filename, block_size=65536):
sha256 = hashlib.sha256()
with open(filename, "rb") as f:
Expand Down Expand Up @@ -214,7 +220,7 @@ def generate_recipe(
recipe_dir = Path(folder_path) / pkg_name
logging.debug(f"Generating recipe on: {recipe_dir}")
if not recipe_dir.is_dir():
recipe_dir.mkdir()
recipe_dir.mkdir(parents=True)
recipe_path = recipe_dir / "meta.yaml"
recipe_folder = recipe_dir
add_new_lines_after_section(recipe.yaml)
Expand Down
10 changes: 6 additions & 4 deletions pyproject.toml
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@ dynamic = ["version"]
requires-python = ">=3.8"
dependencies = [
"beautifulsoup4",
"build", # conda install python-build
"colorama",
"conda-souschef >=2.2.3",
"packaging >=21.3",
Expand Down Expand Up @@ -50,11 +51,12 @@ docs = [
"mdit-py-plugins>=0.3.0",
]

# both spellings of gray/grey
[project.scripts]
grayskull = "grayskull.main:main"
greyskull = "grayskull.main:main"
conda-grayskull = "grayskull.main:main"
conda-greyskull = "grayskull.main:main"
grayskull = "grayskull.__main__:main"
greyskull = "grayskull.__main__:main"
conda-grayskull = "grayskull.__main__:main"
conda-greyskull = "grayskull.__main__:main"

[project.urls]
Source = "https://github.com/conda/grayskull"
Expand Down
1 change: 1 addition & 0 deletions tests/cli/test_cli_cmds.py
Original file line number Diff line number Diff line change
Expand Up @@ -79,6 +79,7 @@ def test_change_pypi_url(mocker):
url_pypi_metadata="http://url_pypi.com/abc",
sections_populate=None,
from_local_sdist=False,
from_tree=False,
extras_require_test=None,
github_release_tag=None,
extras_require_include=tuple(),
Expand Down
Loading