Add support for data_files python-poetry#890

docs/docs/pyproject.md

@@ -133,6 +133,59 @@ packages = [
     Poetry is clever enough to detect Python subpackages.
 
     Thus, you only have to specify the directory where your root package resides.
+
+## data_files
+
+A list of files to be installed using the [data_files](https://docs.python.org/2/distutils/setupscript.html#installing-additional-files)
+installation mechanism.  Data files are particularly useful when shipping non-code artifacts that may need to be used
+by other packages at a well-known location.  Example uses include distribution of Protobuf proto definition files and 
+Avro avsc schemas.
+
+```toml
+[tool.poetry.data_files]
+my_package_name = ["a.txt", "subdir/subsubdir/b.txt"]
+"my_target_dir.has.dots" = ["**/*.txt"]
+"my_package_name/with/subdirectories" = ["subdir/subsubdir/b.txt"]
+```
+
+### Effect on produced wheel archives
+
+The above TOML snippet will result in the addition of a [.data](https://www.python.org/dev/peps/pep-0427/#the-data-directory) 
+directory to your wheel.  For example, given my package, `my_package_name`, and a version of `2.3.4`, the wheel will now
+contain the following:
+
+```
+my_package_name-2.3.4.data/my_package_name/a.txt
+my_package_name-2.3.4.data/my_package_name/b.txt
+my_package_name-2.3.4.data/my_target_dir.has.dots/a.txt
+my_package_name-2.3.4.data/my_target_dir.has.dots/b.txt
+my_package_name-2.3.4.data/my_packge_name/with/subdirectories/b.txt
+```
+
+All of the entries added to the wheel are also added to the RECORD with their appropriate secure hashes.
+
+### Effect on produced sdist archives
+
+The above TOML snippet will result in the addition of the following `data_files` element to the generated setup.py:
+
+```python
+# [...]
+data_files = \
+[('my_package_name', ['a.txt', 'subdir/subsubdir/b.txt']),
+ ('my_target_dir.has.dots', ['a.txt', 'subdir/subsubdir/b.txt']),
+ ('my_package_name/with/subdirectories', ['subdir/subsubdir/b.txt'])]
+
+setup_kwargs = {
+    # [...]
+    'data_files': data_files,
+}
+```
+
+
+!!!note
+
+    The path information in the files or globs is discarded during installation.  If you need your files to be placed
+    in a nested directory, it must be a part of the "name" of the `data_files` element.
 
 ## include and exclude
 

poetry/masonry/builders/builder.py

@@ -38,6 +38,7 @@ def __init__(self, poetry, env, io):
             self._path.as_posix(),
             packages=self._package.packages,
             includes=self._package.include,
+            data_files=self._package.data_files,
         )
         self._meta = Metadata.from_package(self._package)
 

poetry/masonry/builders/sdist.py

@@ -15,6 +15,7 @@
 from poetry.utils._compat import to_str
 
 from ..utils.helpers import normalize_file_permissions
+from ..utils.data_file_include import DataFileInclude
 from ..utils.package_include import PackageInclude
 
 from .builder import Builder
@@ -119,6 +120,7 @@ def build_setup(self):  # type: () -> bytes
         modules = []
         packages = []
         package_data = {}
+        data_files = {}
         for include in self._module.includes:
             if isinstance(include, PackageInclude):
                 if include.is_package():
@@ -137,6 +139,10 @@ def build_setup(self):  # type: () -> bytes
 
                     if module not in modules:
                         modules.append(module)
+            elif isinstance(include, DataFileInclude):
+                data_files.setdefault(include.data_file_path_prefix, []).extend(
+                    str(element.relative_to(self._path)) for element in include.elements
+                )
             else:
                 pass
 
@@ -153,8 +159,16 @@ def build_setup(self):  # type: () -> bytes
             extra.append("'package_data': package_data,")
 
         if modules:
-            before.append("modules = \\\n{}".format(pformat(modules)))
-            extra.append("'py_modules': modules,".format())
+            before.append("modules = \\\n{}\n".format(pformat(modules)))
+            extra.append("'py_modules': modules,")
+
+        if data_files:
+            before.append(
+                "data_files = \\\n{}\n".format(
+                    pformat([(k, v) for k, v in data_files.items()])
+                )
+            )
+            extra.append("'data_files': data_files,")
 
         dependencies, extras = self.convert_dependencies(
             self._package, self._package.requires

poetry/masonry/builders/wheel.py

@@ -17,13 +17,15 @@
 from poetry.semver import parse_constraint
 
 from ..utils.helpers import normalize_file_permissions
+from ..utils.data_file_include import DataFileInclude
 from ..utils.package_include import PackageInclude
 from ..utils.tags import get_abbr_impl
 from ..utils.tags import get_abi_tag
 from ..utils.tags import get_impl_ver
 from ..utils.tags import get_platform
 from .builder import Builder
 
+from poetry.utils._compat import Path
 
 wheel_file_template = """\
 Wheel-Version: 1.0
@@ -136,6 +138,14 @@ def _copy_module(self, wheel):
 
                 if isinstance(include, PackageInclude) and include.source:
                     rel_file = file.relative_to(include.base)
+                elif isinstance(include, DataFileInclude):
+                    rel_file = Path(
+                        self.wheel_meta_dir_name(
+                            self._package.name, self._meta.version, "data"
+                        ),
+                        include.data_file_path_prefix,
+                        file.name,
+                    )
                 else:
                     rel_file = file.relative_to(self._path)
 
@@ -192,7 +202,7 @@ def find_excluded_files(self):  # type: () -> Set
 
     @property
     def dist_info(self):  # type: () -> str
-        return self.dist_info_name(self._package.name, self._meta.version)
+        return self.wheel_meta_dir_name(self._package.name, self._meta.version)
 
     @property
     def wheel_filename(self):  # type: () -> str
@@ -207,11 +217,13 @@ def supports_python2(self):
             parse_constraint(">=2.0.0 <3.0.0")
         )
 
-    def dist_info_name(self, distribution, version):  # type: (...) -> str
+    def wheel_meta_dir_name(
+        self, distribution, version, suffix="dist-info"
+    ):  # type: (...) -> str
         escaped_name = re.sub(r"[^\w\d.]+", "_", distribution, flags=re.UNICODE)
         escaped_version = re.sub(r"[^\w\d.]+", "_", version, flags=re.UNICODE)
 
-        return "{}-{}.dist-info".format(escaped_name, escaped_version)
+        return "{}-{}.{}".format(escaped_name, escaped_version, suffix)
 
     @property
     def tag(self):

poetry/masonry/utils/data_file_include.py

@@ -0,0 +1,16 @@
+from .include import Include
+
+# noinspection PyProtectedMember
+from poetry.utils._compat import Path
+
+
+class DataFileInclude(Include):
+    def __init__(
+        self, base, include, data_file_path_prefix
+    ):  # type: (Path, str, str) -> None
+        super(DataFileInclude, self).__init__(base, include)
+        self._data_file_path_prefix = data_file_path_prefix
+
+    @property
+    def data_file_path_prefix(self):  # type: () -> str
+        return self._data_file_path_prefix

poetry/masonry/utils/include.py

@@ -21,7 +21,7 @@ def __init__(self, base, include):  # type: (Path, str) -> None
         self._base = base
         self._include = str(include)
 
-        self._elements = sorted(list(self._base.glob(str(self._include))))
+        self._elements = sorted(list(self._base.glob(self._include)))
 
     @property
     def base(self):  # type: () -> Path

poetry/masonry/utils/module.py

@@ -2,6 +2,7 @@
 from poetry.utils.helpers import module_name
 
 from .include import Include
+from .data_file_include import DataFileInclude
 from .package_include import PackageInclude
 
 
@@ -11,14 +12,17 @@ class ModuleOrPackageNotFound(ValueError):
 
 
 class Module:
-    def __init__(self, name, directory=".", packages=None, includes=None):
+    def __init__(
+        self, name, directory=".", packages=None, includes=None, data_files=None
+    ):
         self._name = module_name(name)
         self._in_src = False
         self._is_package = False
         self._path = Path(directory)
         self._includes = []
         packages = packages or []
         includes = includes or []
+        data_files = data_files or {}
 
         if not packages:
             # It must exist either as a .py file or a directory, but not both
@@ -62,6 +66,10 @@ def __init__(self, name, directory=".", packages=None, includes=None):
                 PackageInclude(self._path, package["include"], package.get("from"))
             )
 
+        for base, globs in data_files.items():
+            for glob in globs:
+                self._includes.append(DataFileInclude(self._path, glob, base))
+
         for include in includes:
             self._includes.append(Include(self._path, include))
 

poetry/masonry/utils/package_include.py

@@ -26,7 +26,7 @@ def source(self):  # type: () -> str
     def is_package(self):  # type: () -> bool
         return self._is_package
 
-    def is_module(self):  # type: ()
+    def is_module(self):  # type: () -> bool
         return self._is_module
 
     def refresh(self):  # type: () -> PackageInclude

poetry/packages/project_package.py

@@ -14,6 +14,7 @@ def __init__(self, name, version, pretty_version=None):
         self.packages = []
         self.include = []
         self.exclude = []
+        self.data_files = {}
 
         if self._python_versions == "*":
             self._python_constraint = parse_constraint("~2.7 || >=3.4")

poetry/poetry.py

@@ -184,6 +184,9 @@ def create(cls, cwd):  # type: (Path) -> Poetry
         if "packages" in local_config:
             package.packages = local_config["packages"]
 
+        if "data_files" in local_config:
+            package.data_files = local_config["data_files"]
+
         # Moving lock if necessary (pyproject.lock -> poetry.lock)
         lock = poetry_file.parent / "poetry.lock"
         if not lock.exists():

tests/masonry/builders/fixtures/data_files/pyproject.toml

@@ -0,0 +1,10 @@
+[tool.poetry]
+name = "data_files_example"
+version = "0.1.0"
+description = "An example TOML file describing a package with data_files"
+authors = ["delphyne <delphyne@gmail.com>"]
+
+[tool.poetry.data_files]
+easy = ["a.txt", "**/c.csv"]
+"a.little.more.difficult" = ["**/*.txt"]
+"nested/directories" = ["subdir/subsubdir/b.txt"]

tests/masonry/builders/test_sdist.py

@@ -510,3 +510,38 @@ def test_proper_python_requires_if_three_digits_precision_version_specified():
     parsed = p.parsestr(to_str(pkg_info))
 
     assert parsed["Requires-Python"] == "==2.7.15"
+
+
+def test_with_data_files():
+    poetry = Poetry.create(project("data_files"))
+    builder = SdistBuilder(poetry, NullEnv(), NullIO())
+
+    # Check setup.py
+    setup = builder.build_setup()
+    setup_ast = ast.parse(setup)
+
+    setup_ast.body = [n for n in setup_ast.body if isinstance(n, ast.Assign)]
+    ns = {}
+    exec(compile(setup_ast, filename="setup.py", mode="exec"), ns)
+
+    assert [
+        ("easy", ["a.txt", "subdir/subsubdir/c.csv"]),
+        ("a.little.more.difficult", ["a.txt", "subdir/subsubdir/b.txt"]),
+        ("nested/directories", ["subdir/subsubdir/b.txt"]),
+    ] == ns.get("data_files")
+
+    assert ns.get("setup_kwargs", {}).get("data_files") == ns.get("data_files")
+
+    builder.build()
+
+    sdist = fixtures_dir / "data_files" / "dist" / "data_files_example-0.1.0.tar.gz"
+
+    assert sdist.exists()
+
+    with tarfile.open(str(sdist), "r") as tar:
+        names = tar.getnames()
+        assert "data_files_example-0.1.0/data_files_example/__init__.py" in names
+        assert "data_files_example-0.1.0/a.txt" in names
+        assert "data_files_example-0.1.0/subdir/subsubdir/b.txt" in names
+        assert "data_files_example-0.1.0/subdir/subsubdir/c.csv" in names
+        assert "data_files_example-0.1.0/setup.py" in names

tests/masonry/builders/test_wheel.py

@@ -144,3 +144,28 @@ def test_write_metadata_file_license_homepage_default(mocker):
     # Assertion
     mocked_file.write.assert_any_call("Home-page: UNKNOWN\n")
     mocked_file.write.assert_any_call("License: UNKNOWN\n")
+
+
+def test_with_data_files():
+    module_path = fixtures_dir / "data_files"
+    p = Poetry.create(str(module_path))
+    WheelBuilder.make(p, NullEnv(), NullIO())
+    whl = module_path / "dist" / "data_files_example-0.1.0-py2.py3-none-any.whl"
+    assert whl.exists()
+
+    with zipfile.ZipFile(str(whl)) as z:
+        names = z.namelist()
+        with z.open("data_files_example-0.1.0.dist-info/RECORD") as record_file:
+            record = record_file.readlines()
+
+            def validate(path):
+                assert path in names
+                assert (
+                    len([r for r in record if r.decode("utf-8").startswith(path)]) == 1
+                )
+
+            validate("data_files_example-0.1.0.data/easy/a.txt")
+            validate("data_files_example-0.1.0.data/easy/c.csv")
+            validate("data_files_example-0.1.0.data/a.little.more.difficult/a.txt")
+            validate("data_files_example-0.1.0.data/a.little.more.difficult/b.txt")
+            validate("data_files_example-0.1.0.data/nested/directories/b.txt")