Create purls.yml path without leading segment #660

    * leading segment will be the repo itself, not a directory within the repo

Signed-off-by: Jono Yang <[email protected]>

Author: JonoYang

minecode_pipelines/miners/__init__.py

@@ -13,6 +13,7 @@
 from pathlib import Path
 
 from aboutcode import hashid
+from packageurl import PackageURL
 from scanpipe.pipes import federatedcode
 
 
@@ -31,7 +32,9 @@ def write_data_to_file(path, data):
 
 def write_purls_to_repo(repo, package, packages, push_commit=False):
     # save purls to yaml
-    ppath = hashid.get_package_purls_yml_file_path(package)
+    path_elements = hashid.package_path_elements(package)
+    _, core_path, _, _ = path_elements
+    ppath = core_path / hashid.PURLS_FILENAME
     purls = [p.purl for p in packages]
     federatedcode.write_data_as_yaml(
         base_path=repo.working_dir,

minecode_pipelines/pipes/maven.py

@@ -692,15 +692,15 @@ def get_packages(self, content=None, increments=False):
 def collect_packages_from_maven(commits_per_push=10, logger=None):
     # download and iterate through maven nexus index
     maven_nexus_collector = MavenNexusCollector()
-    prev_package = None
+    prev_purl = None
     current_packages = []
-    for i, (current_package, package) in enumerate(maven_nexus_collector.get_packages(), start=1):
-        if not prev_package:
-            prev_package = current_package
-        elif prev_package != current_package:
+    for i, (current_purl, package) in enumerate(maven_nexus_collector.get_packages(), start=1):
+        if not prev_purl:
+            prev_purl = current_purl
+        elif prev_purl != current_purl:
             # check out repo
             repo_url, _ = federatedcode.get_package_repository(
-                project_purl=prev_package,
+                project_purl=prev_purl,
                 logger=logger
             )
             repo = federatedcode.clone_repository(
@@ -712,7 +712,7 @@ def collect_packages_from_maven(commits_per_push=10, logger=None):
             # save purls to yaml
             write_purls_to_repo(
                 repo=repo,
-                package=prev_package,
+                package=prev_purl,
                 packages=current_packages,
                 push_commit=push_commit
             )
@@ -721,5 +721,5 @@ def collect_packages_from_maven(commits_per_push=10, logger=None):
             federatedcode.delete_local_clone(repo)
 
             current_packages = []
-            prev_package = current_package
+            prev_purl = current_purl
         current_packages.append(package)