packagedcode: add CondaEnvironmentYmlHandler to parse environment.yml

src/packagedcode/__init__.py

@@ -76,6 +76,7 @@
     conda.CondaMetaJsonHandler,
     conda.CondaMetaYamlHandler,
     conda.CondaYamlHandler,
+    conda.CondaEnvironmentYmlHandler,
 
     conan.ConanFileHandler,
     conan.ConanDataHandler,

src/packagedcode/conda.py

@@ -666,3 +666,79 @@ def get_variables(location):
                 parts = line.split('=')
                 result[parts[0].strip()] = parts[-1].strip().strip('"')
     return result
+
+
+class CondaEnvironmentYmlHandler(BaseDependencyFileHandler):
+    datasource_id = 'conda_environment_yml'
+    path_patterns = ('*environment.yml', '*environment.yaml')
+    default_package_type = 'conda'
+    description = 'Conda environment.yml'
+
+    @classmethod
+    def parse(cls, location, package_only=False):
+        import re
+        with open(location) as fi:
+            conda_data = saneyaml.load(fi.read())
+
+        if not conda_data or not isinstance(conda_data, dict):
+            conda_data = {}
+
+        name = conda_data.get('name')
+        extra_data = {}
+        channels = conda_data.get('channels')
+        if channels:
+            extra_data['channels'] = channels
+
+        dependencies = []
+        deps_data = conda_data.get('dependencies') or []
+        for dep in deps_data:
+            if isinstance(dep, str):
+                match = re.split(r'(>=|<=|==|=|>|<)', dep, maxsplit=1)
+                dep_name = match[0].strip()
+                version = None
+
+                if len(match) > 1:
+                    version = match[2].strip()
+
+                purl = PackageURL(type='conda', name=dep_name)
+
+                dependencies.append(
+                    models.DependentPackage(
+                        purl=purl.to_string(),
+                        extracted_requirement=version,
+                        scope='runtime',
+                        is_runtime=True,
+                        is_optional=False,
+                    )
+                )
+            elif isinstance(dep, dict) and 'pip' in dep:
+                pip_deps = dep.get('pip') or []
+                for pip_dep in pip_deps:
+                    match = re.split(r'(>=|<=|==|=|>|<|~=)', pip_dep, maxsplit=1)
+                    pip_name = match[0].strip()
+                    version = None
+                    if len(match) > 1:
+                        version = match[2].strip()
+
+                    purl = PackageURL(type='pypi', name=pip_name)
+
+                    dependencies.append(
+                        models.DependentPackage(
+                            purl=purl.to_string(),
+                            extracted_requirement=version,
+                            scope='runtime',
+                            is_runtime=True,
+                            is_optional=False,
+                        )
+                    )
+
+        package_data = dict(
+            datasource_id=cls.datasource_id,
+            type=cls.default_package_type,
+            name=name,
+            dependencies=dependencies,
+        )
+        if extra_data:
+            package_data['extra_data'] = extra_data
+
+        yield models.PackageData.from_data(package_data, package_only)

tests/packagedcode/data/conda/environment_yml/multiregex-environment.yml

@@ -0,0 +1,11 @@
+name: myenv
+channels:
+  - conda-forge
+  - defaults
+dependencies:
+  - python=3.10
+  - numpy=1.24.0
+  - pandas
+  - pip:
+    - requests==2.28.0
+    - flask>=2.0.0

tests/packagedcode/data/conda/environment_yml/simple-environment.yml

@@ -0,0 +1,9 @@
+name: testenv
+channels:
+  - conda-forge
+  - defaults
+dependencies:
+  - python=3.10
+  - numpy=1.24.0
+  - pip:
+    - requests==2.28.0

tests/packagedcode/test_conda_environment_yml.py

@@ -0,0 +1,63 @@
+import os
+from packages_test_utils import PackageTester
+from packagedcode import conda
+from scancode_config import REGEN_TEST_FIXTURES
+
+class TestCondaEnvironmentYml(PackageTester):
+    test_data_dir = os.path.join(os.path.dirname(__file__), 'data')
+
+    def test_conda_environment_yml_is_datafile(self):
+        test_file = self.get_test_loc('conda/environment_yml/simple-environment.yml')
+        assert conda.CondaEnvironmentYmlHandler.is_datafile(test_file)
+
+    def test_parse_simple_environment_yml(self):
+        test_file = self.get_test_loc('conda/environment_yml/simple-environment.yml')
+        packages = list(conda.CondaEnvironmentYmlHandler.parse(test_file))
+        assert len(packages) == 1
+        package = packages[0]
+
+        assert package.name == 'testenv'
+        assert package.extra_data == {'channels': ['conda-forge', 'defaults']}
+
+        deps = package.dependencies
+
+        numpy_dep = next((d for d in deps if d.purl == 'pkg:conda/numpy'), None)
+        assert numpy_dep is not None
+        assert numpy_dep.extracted_requirement == '1.24.0'
+
+        requests_dep = next((d for d in deps if d.purl == 'pkg:pypi/requests'), None)
+        assert requests_dep is not None
+        assert requests_dep.extracted_requirement == '2.28.0'
+
+    def test_parse_real_environment_yml(self):
+        test_file = self.get_test_loc('conda/environment_yml/multiregex-environment.yml')
+        packages = list(conda.CondaEnvironmentYmlHandler.parse(test_file))
+        assert len(packages) == 1
+        package = packages[0]
+
+        assert package.name == 'myenv'
+        assert len(package.dependencies) > 0
+        deps = [d.purl for d in package.dependencies]
+        assert 'pkg:conda/pandas' in deps
+
+    def test_parse_empty_dependencies(self):
+        test_file = self.get_temp_file('empty-deps.yml')
+        with open(test_file, 'w') as f:
+            f.write('name: nodeps\nchannels:\n  - defaults\ndependencies:\n')
+
+        packages = list(conda.CondaEnvironmentYmlHandler.parse(test_file))
+        assert len(packages) == 1
+        package = packages[0]
+        assert package.name == 'nodeps'
+        assert package.dependencies == []
+
+    def test_parse_missing_name(self):
+        test_file = self.get_temp_file('noname.yml')
+        with open(test_file, 'w') as f:
+            f.write('channels:\n  - defaults\ndependencies:\n  - python=3.10\n')
+
+        packages = list(conda.CondaEnvironmentYmlHandler.parse(test_file))
+        assert len(packages) == 1
+        package = packages[0]
+        assert package.name is None
+        assert len(package.dependencies) == 1