Fixed #395 Add support for package_url field

 * Introduce a PackageUrlField class in model.py
 * Update pacakgeurl_python to version 0.9.0
 * Update SPEC to v3.1.5 (support `package_url`)

Author: chinyeungli

SPECIFICATION.rst

@@ -1,4 +1,4 @@
-ABOUT File Specification v3.1.4
+ABOUT File Specification v3.1.5
 
 
 Purpose
@@ -284,6 +284,8 @@ Optional Information fields
 
 - changelog_file: Changelog file for the component.
 
+- package_url: Package URL for the package.
+
 - notes: Notes and comments about the component.
 
 

docs/CHANGELOG.rst

@@ -1,3 +1,8 @@
+2020-xx-xx
+    Release 5.1.0
+
+    * Add support for `package_url` #396
+
 2020-08-11
     Release 5.0.0
 
@@ -16,7 +21,6 @@
     * Fix the missing `multi_sort` filter for Jinja2
     * Update help text for `--vartext`
 
-
 2019-10-17
     Release 4.0.1
 

docs/UsingAboutCodetoDocumentYourSoftwareAssets.md

@@ -85,6 +85,11 @@ You should start with a software inventory of your codebase in spreadsheet forma
     <td>URL to the homepage for this component</td>
     <td>Optional</td>
   </tr>
+  <tr>
+    <td>package_url</td>
+    <td>Package URL for this component (See https://github.com/package-url/purl-spec for SPEC)</td>
+    <td>Optional</td>
+  </tr>
   <tr>
     <td>notes</td>
     <td>notes text</td>

docs/UsingAboutCodetoDocumentYourSoftwareAssets.pdf

@@ -79,6 +79,7 @@ def read(*names, **kwargs):
 
         'boolean.py >= 3.5, < 4.0',
         'license_expression >= 0.94',
+        'packageurl_python >= 0.9.0',
     ],
     extras_require={
         ":python_version < '3.6'": ['backports.csv'],

setup.py

@@ -74,6 +74,8 @@
 from attributecode.util import ungroup_licenses
 from attributecode.util import unique
 
+from packageurl import PackageURL
+
 genereated_tk_version = "# Generated with AboutCode Toolkit Version %s \n\n" % __version__
 
 class Field(object):
@@ -351,6 +353,32 @@ def __eq__(self, other):
         if sval == oval:
             return True
 
+class PackageUrlField(StringField):
+    """
+    A Package URL field. The validated value is a purl.
+    """
+    def _validate(self, *args, **kwargs):
+        """
+        Check that Package URL is valid. Return a list of errors.
+        """
+        errors = super(PackageUrlField, self)._validate(*args, ** kwargs)
+        name = self.name
+        val = self.value
+        if not self.is_valid_purl(val):
+            msg = (u'Field %(name)s: Invalid Package URL: %(val)s' % locals())
+            errors.append(Error(WARNING, msg))
+        return errors
+
+    @staticmethod
+    def is_valid_purl(purl):
+        """
+        Return True if a Package URL is valid.
+        """
+        try:
+            return bool(PackageURL.from_string(purl))
+        except:
+            return False
+
 class UrlListField(ListField):
     """
     A URL field. The validated value is a list of URLs.
@@ -717,6 +745,7 @@ def set_standard_fields(self):
             ('download_url', UrlField()),
             ('description', StringField()),
             ('homepage_url', UrlField()),
+            ('package_url', PackageUrlField()),
             ('notes', StringField()),
 
             ('license_expression', StringField()),

src/attributecode/model.py

@@ -118,6 +118,7 @@ def test_Field_init(self):
         model.BooleanField()
         model.PathField()
         model.FileTextField()
+        model.PackageUrlField()
 
     def test_empty_Field_has_no_content(self):
         field = model.Field()
@@ -167,6 +168,12 @@ def test_TextField_loads_file(self):
         expected = {'license.LICENSE': 'some license text'}
         assert expected == field.value
 
+    def test_PackageUrlField_is_valid_url(self):
+        assert model.PackageUrlField.is_valid_purl('pkg:pypi/[email protected]')
+
+    def test_PackageUrlField_is_valid_url_no_version(self):
+        assert model.PackageUrlField.is_valid_purl('pkg:pypi/saneyaml')
+
     def test_UrlField_is_valid_url(self):
         assert model.UrlField.is_valid_url('http://www.google.com')