Run the extract_archives step in ScanSinglePackage #1081 (#1083)

Signed-off-by: tdruez <tdruez@nexb.com>

Author: tdruez

CHANGELOG.rst

@@ -38,6 +38,9 @@ v33.2.0 (unreleased)
   unit paths when available from elfs.
   https://github.com/nexB/purldb/issues/260
 
+- Extract all archives recursively in the `scan_single_package` pipeline.
+  https://github.com/nexB/scancode.io/issues/1081
+
 v33.1.0 (2024-02-02)
 --------------------
 

scanpipe/pipelines/__init__.py

@@ -248,6 +248,18 @@ def flag_ignored_resources(self):
         if ignored_patterns := self.env.get("ignored_patterns"):
             flag.flag_ignored_patterns(self.project, patterns=ignored_patterns)
 
+    def extract_archives(self):
+        """Extract archives located in the codebase/ directory with extractcode."""
+        from scanpipe.pipes import scancode
+
+        extract_errors = scancode.extract_archives(
+            location=self.project.codebase_path,
+            recurse=self.env.get("extract_recursively", True),
+        )
+
+        if extract_errors:
+            self.add_error("\n".join(extract_errors))
+
 
 def is_pipeline(obj):
     """

scanpipe/pipelines/deploy_to_develop.py

@@ -57,7 +57,7 @@ def steps(cls):
         return (
             cls.get_inputs,
             cls.extract_inputs_to_codebase_directory,
-            cls.extract_archives_in_place,
+            cls.extract_archives,
             cls.collect_and_create_codebase_resources,
             cls.fingerprint_codebase_directories,
             cls.flag_empty_files,
@@ -133,16 +133,6 @@ def extract_inputs_to_codebase_directory(self):
         if errors:
             self.add_error("\n".join(errors))
 
-    def extract_archives_in_place(self):
-        """Extract recursively from* and to* archives in place with extractcode."""
-        extract_errors = scancode.extract_archives(
-            self.project.codebase_path,
-            recurse=self.env.get("extract_recursively", True),
-        )
-
-        if extract_errors:
-            self.add_error("\n".join(extract_errors))
-
     def collect_and_create_codebase_resources(self):
         """Collect and create codebase resources."""
         pipes.collect_and_create_codebase_resources(self.project)

scanpipe/pipelines/scan_codebase.py

@@ -54,16 +54,6 @@ def copy_inputs_to_codebase_directory(self):
         """
         copy_inputs(self.project.inputs(), self.project.codebase_path)
 
-    def extract_archives(self):
-        """Extract archives with extractcode."""
-        extract_errors = scancode.extract_archives(
-            location=self.project.codebase_path,
-            recurse=self.env.get("extract_recursively", True),
-        )
-
-        if extract_errors:
-            self.add_error("\n".join(extract_errors))
-
     def collect_and_create_codebase_resources(self):
         """Collect and create codebase resources."""
         pipes.collect_and_create_codebase_resources(self.project)

scanpipe/pipelines/scan_single_package.py

@@ -31,7 +31,6 @@
 from scanpipe.pipes import scancode
 from scanpipe.pipes.input import copy_input
 from scanpipe.pipes.input import is_archive
-from scanpipe.pipes.scancode import extract_archive
 
 
 class ScanSinglePackage(Pipeline):
@@ -50,6 +49,7 @@ def steps(cls):
             cls.get_package_input,
             cls.collect_input_information,
             cls.extract_input_to_codebase_directory,
+            cls.extract_archives,
             cls.run_scan,
             cls.load_inventory_from_toolkit_scan,
             cls.make_summary_from_scan_results,
@@ -93,7 +93,11 @@ def extract_input_to_codebase_directory(self):
             copy_input(self.input_path, self.project.codebase_path)
             return
 
-        extract_errors = extract_archive(self.input_path, self.project.codebase_path)
+        extract_errors = scancode.extract_archive(
+            location=self.input_path,
+            target=self.project.codebase_path,
+        )
+
         if extract_errors:
             self.add_error("\n".join(extract_errors))