Merge pull request aboutcode-org#3682 from nexB/update-package-handlers

Update package handlers

src/packagedcode/__init__.py

@@ -28,7 +28,6 @@
 from packagedcode import godeps
 from packagedcode import golang
 from packagedcode import haxe
-from packagedcode import jar_manifest
 from packagedcode import maven
 from packagedcode import misc
 from packagedcode import npm
@@ -84,6 +83,7 @@
     cran.CranDescriptionFileHandler,
 
     debian_copyright.DebianCopyrightFileInPackageHandler,
+    debian_copyright.StandaloneDebianCopyrightFileHandler,
     debian.DebianDscFileHandler,
 
     debian.DebianControlFileInExtractedDebHandler,
@@ -216,13 +216,7 @@
 
     debian.DebianInstalledFilelistHandler,
     debian.DebianInstalledMd5sumFilelistHandler,
-    debian.DebianInstalledStatusDatabaseHandler,
-    debian.DebianControlFileInSourceHandler,
-    debian.DebianDscFileHandler,
-    debian.DebianSourcePackageTarballHandler,
-    debian.DebianSourcePackageMetadataTarballHandler,
-    debian.DebianDebPackageHandler,
-    debian_copyright.StandaloneDebianCopyrightFileHandler
+    debian.DebianInstalledStatusDatabaseHandler
 ]
 
 if on_linux:

src/packagedcode/build.py

@@ -47,7 +47,7 @@ def logger_debug(*args):
         )
 
 
-class AutotoolsConfigureHandler(models.DatafileHandler):
+class AutotoolsConfigureHandler(models.NonAssemblableDatafileHandler):
     datasource_id = 'autotools_configure'
     path_patterns = ('*/configure', '*/configure.ac',)
     default_package_type = 'autotools'
@@ -74,14 +74,6 @@ def parse(cls, location):
             version=version,
         )
 
-    @classmethod
-    def assign_package_to_resources(cls, package, resource, codebase, package_adder):
-        models.DatafileHandler.assign_package_to_parent_tree(
-            package=package,
-            resource=resource,
-            codebase=codebase,
-            package_adder=package_adder,
-        )
 
 
 def check_rule_name_ending(rule_name, starlark_rule_types=('binary', 'library')):

src/packagedcode/debian.py

@@ -582,7 +582,10 @@ def parse_debian_files_list(location, datasource_id, package_type):
         name, _, arch = filename.partition(':')
         qualifiers['arch'] = arch
     else:
-        name = filename
+        name = None
+        # For DebianMd5sumFilelistInPackageHandler we cannot infer name
+        if not name == "md5sums":
+            name = filename
 
     file_references = []
     with open(location) as info_file:
@@ -643,12 +646,14 @@ def build_package_data(debian_data, datasource_id, package_type='deb', distro=No
 
     maintainer = debian_data.get('maintainer')
     if maintainer:
-        party = models.Party(role='maintainer', name=maintainer)
+        maintainer_name, maintainer_email = parse_debian_maintainers(maintainer)
+        party = models.Party(role='maintainer', name=maintainer_name, email=maintainer_email)
         parties.append(party)
 
     orig_maintainer = debian_data.get('original_maintainer')
     if orig_maintainer:
-        party = models.Party(role='original_maintainer', name=orig_maintainer)
+        maintainer_name, maintainer_email = parse_debian_maintainers(orig_maintainer)
+        party = models.Party(role='maintainer', name=maintainer_name, email=maintainer_email)
         parties.append(party)
 
     keywords = []
@@ -708,6 +713,26 @@ def build_package_data(debian_data, datasource_id, package_type='deb', distro=No
     )
 
 
+def parse_debian_maintainers(maintainer):
+    """
+    Get name and email values from a debian maintainer string.
+
+    Example string:
+    Debian systemd Maintainers <[email protected]>
+    """
+    email_wrappers = ["<", ">"]
+    has_email = "@" in maintainer and all([
+        True 
+        for char in email_wrappers
+        if char in maintainer
+    ])
+    if not has_email:
+        return maintainer, None
+
+    name, _, email = maintainer.rpartition("<")
+    return name.rstrip(" "), email.rstrip(">")
+
+
 def populate_debian_namespace(packages):
     """
     For an iterable of debian `packages`, populate the