Backport changes from #3315 to v31.2.6

Signed-off-by: Jono Yang <jyang@nexb.com>

Author: JonoYang

requirements.txt

@@ -9,7 +9,7 @@ chardet==5.0.0
 charset-normalizer==2.1.0
 click==8.1.3
 colorama==0.4.5
-commoncode==31.0.0
+commoncode==31.0.2
 construct==2.10.68
 container-inspector==31.1.0
 cryptography==37.0.4
@@ -48,7 +48,7 @@ pefile==2022.5.30
 pip-requirements-parser==32.0.1
 pkginfo2==30.0.0
 pluggy==1.0.0
-plugincode==31.0.0
+plugincode==32.0.0
 ply==3.11
 publicsuffix2==2.20191221
 pyahocorasick==2.0.0b1

setup.cfg

@@ -68,7 +68,7 @@ install_requires =
     chardet >= 3.0.0
     click >= 6.7, !=7.0
     colorama >= 0.3.9
-    commoncode >= 31.0.0
+    commoncode >= 31.0.2
     container-inspector >= 31.0.0
     debian-inspector >= 31.0.0
     dparse2 >= 0.7.0
@@ -95,7 +95,7 @@ install_requires =
     pkginfo2 >= 30.0.0
     pip-requirements-parser >= 32.0.1
     pluggy >= 1.0.0
-    plugincode >= 31.0.0
+    plugincode >= 32.0.0
     publicsuffix2
     pyahocorasick >= 2.0.0b1
     pygmars >= 0.7.0
@@ -158,7 +158,6 @@ console_scripts =
 scancode_pre_scan =
     ignore = scancode.plugin_ignore:ProcessIgnore
     facet = summarycode.facet:AddFacet
-    classify = summarycode.classify:FileClassifier
 
 
 # scancode_scan is the entry point for scan plugins that run a scan after the
@@ -190,6 +189,7 @@ scancode_post_scan =
     filter-clues = cluecode.plugin_filter_clues:RedundantCluesFilter
     consolidate = summarycode.plugin_consolidate:Consolidator
     licenses-reference = licensedcode.plugin_licenses_reference:LicensesReference
+    classify = summarycode.classify:FileClassifier
 
 
 # scancode_output_filter is the entry point for filter plugins executed after

src/cluecode/plugin_copyright.py

@@ -28,6 +28,7 @@ class CopyrightScanner(ScanPlugin):
         ('authors',attr.ib(default=attr.Factory(list))),
     ])
 
+    run_order = 4
     sort_order = 4
 
     options = [

src/cluecode/plugin_email.py

@@ -25,6 +25,7 @@ class EmailScanner(ScanPlugin):
     """
     resource_attributes = dict(emails=attr.ib(default=attr.Factory(list)))
 
+    run_order = 8
     sort_order = 8
 
     options = [

src/cluecode/plugin_filter_clues.py

@@ -41,6 +41,7 @@ class RedundantCluesFilter(PostScanPlugin):
     Filter redundant clues (copyrights, authors, emails, and urls) that are
     already contained in a matched license text.
     """
+    run_order = 1
     sort_order = 1
 
     options = [

src/cluecode/plugin_url.py

@@ -26,6 +26,7 @@ class UrlScanner(ScanPlugin):
 
     resource_attributes = dict(urls=attr.ib(default=attr.Factory(list)))
 
+    run_order = 10
     sort_order = 10
 
     options = [

src/licensedcode/plugin_license.py

@@ -89,6 +89,7 @@ class LicenseScanner(ScanPlugin):
         ('percentage_of_license_text', attr.ib(default=0)),
     ])
 
+    run_order = 2
     sort_order = 2
 
     options = [

src/licensedcode/plugin_license_policy.py

@@ -28,6 +28,7 @@ class LicensePolicy(PostScanPlugin):
 
     resource_attributes = dict(license_policy=attr.ib(default=attr.Factory(dict)))
 
+    run_order = 9
     sort_order = 9
 
     options = [

src/packagedcode/maven.py

@@ -155,6 +155,21 @@ def assign_package_to_resources(cls, package, resource, codebase, package_adder)
     def compute_normalized_license(cls, package):
         return compute_normalized_license(declared_license=package.declared_license)
 
+    @classmethod
+    def get_top_level_resources(cls, manifest_resource, codebase):
+        """
+        Yield Resources that are top-level based on a JAR's directory structure
+        """
+        if 'META-INF' in manifest_resource.path:
+            path_segments = manifest_resource.path.split('META-INF')
+            leading_segment = path_segments[0].strip()
+            meta_inf_dir_path = f'{leading_segment}/META-INF'
+            meta_inf_resource = codebase.get_resource(meta_inf_dir_path)
+            if meta_inf_resource:
+                yield meta_inf_resource
+                yield from meta_inf_resource.walk(codebase)
+
+
 # TODO: assemble with its pom!!
 class MavenPomPropertiesHandler(models.NonAssemblableDatafileHandler):
     datasource_id = 'maven_pom_properties'

src/packagedcode/models.py

@@ -1201,6 +1201,13 @@ def create_default_package_data(cls, **kwargs):
             **kwargs,
         )
 
+    @classmethod
+    def get_top_level_resources(cls, manifest_resource, codebase):
+        """
+        Yield Resources that are considered top-level for a Package type.
+        """
+        pass
+
 
 class NonAssemblableDatafileHandler(DatafileHandler):
     """