#253 Recognize Python .egg files as packages

Signed-off-by: Philippe Ombredanne <[email protected]>

Author: rakesh balusa

src/packagedcode/pypi.py

@@ -46,6 +46,45 @@
 # logger.setLevel(logging.DEBUG)
 
 
+PKG_INFO_ATTRIBUTES = [
+    'Name',
+    'Version',
+    'Author',
+    'Author-email',
+    'License',
+    'Description',
+    'Platform',
+    'Home-page',
+    'Summary'
+]
+
+
+def parse_pkg_info(location):
+    """
+    parse a 'PKG-INFO' file at 'location' and return a PythonPackage object.
+    """
+    infos = {}
+    with open(location, 'rb') as txt_file:
+        lines = []
+        for line in txt_file:
+            lines.append(line)
+        lines = ''.join(lines)
+        for attribute in PKG_INFO_ATTRIBUTES:
+            infos[attribute] = re.findall('^'+attribute+'[\s:]*.*', lines, flags=re.MULTILINE)[0]
+            infos[attribute] = re.sub('^'+attribute+'[\s:]*', '', infos[attribute], flags=re.MULTILINE)
+            if infos[attribute] == 'UNKNOWN':
+                infos[attribute] = None
+        package = PythonPackage(
+            name=infos.get('Name'),
+            version=infos.get('Version'),
+            summary=infos.get('Summary'),
+            homepage_url=infos.get('Home-page'),
+            asserted_licenses=[AssertedLicense(license=infos.get('License'))],
+            authors=[infos.get('Author')],
+        )
+    return package
+
+
 def get_attribute(setup_location, attribute):
     """
     Return the value specified for a given 'attribute' mentioned in a 'setup.py'
@@ -115,3 +154,5 @@ def parse(location):
         return package
     if file_name == 'metadata.json':
         parse_metadata(location)
+    if file_name == 'PKG-INFO':
+        parse_pkg_info(location)

tests/packagedcode/data/pypi/PKG-INFO

@@ -0,0 +1,10 @@
+Metadata-Version: 1.0
+Name: TicketImport
+Version: 0.7a
+Summary: Import CSV and Excel files
+Home-page: http://nexb.com
+Author: Francois Granade
+Author-email: [email protected]
+License: BSD
+Description: UNKNOWN
+Platform: UNKNOWN

tests/packagedcode/test_pypi.py

@@ -59,3 +59,13 @@ def test_parse_metadata(self):
         assert 'MIT' == package.asserted_licenses[0].license
         assert ['Benjamin Peterson'] == package.authors
         assert 'http://pypi.python.org/pypi/six/' == package.homepage_url
+
+    def test_parse_pkg_info(self):
+        test_file = self.get_test_loc('pypi/PKG-INFO')
+        package = pypi.parse_pkg_info(test_file)
+        assert 'TicketImport' == package.name
+        assert '0.7a' == package.version
+        assert 'Import CSV and Excel files' == package.summary
+        assert 'BSD' == package.asserted_licenses[0].license
+        assert 'http://nexb.com' == package.homepage_url
+        assert ['Francois Granade'] == package.authors