Refactor inspect subcommand summary

agustingroh · web-flow · commit 116a3b1e4c8e · 2025-06-23T09:34:04.000-03:00
* chore:SP-2777 Refactor on inspect license and component summary

* Upgrades version to v1.26.1

* chore:Updates CHANGELOG.md file
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 - Upcoming changes...
 
+## [1.26.1] - 2025-06-23
+
+### Added
+- Added component count to inspect license summary
+### Changed
+- Modified summaries for inspect subcommand
+
 ## [1.26.0] - 2025-06-20
 ### Added
 - New `inspect license-summary` subcommand to generate license summaries from scan results
@@ -552,4 +559,5 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 [1.25.0]: https://github.com/scanoss/scanoss.py/compare/v1.24.0...v1.25.0
 [1.25.1]: https://github.com/scanoss/scanoss.py/compare/v1.25.0...v1.25.1
 [1.25.2]: https://github.com/scanoss/scanoss.py/compare/v1.25.1...v1.25.2
-[1.26.0]: https://github.com/scanoss/scanoss.py/compare/v1.25.2...v1.26.0
+[1.26.0]: https://github.com/scanoss/scanoss.py/compare/v1.25.2...v1.26.0
+[1.26.1]: https://github.com/scanoss/scanoss.py/compare/v1.26.0...v1.26.1
diff --git a/src/scanoss/__init__.py b/src/scanoss/__init__.py
@@ -22,4 +22,4 @@
   THE SOFTWARE.
 """
 
-__version__ = '1.26.0'
+__version__ = '1.26.1'
diff --git a/src/scanoss/inspection/component_summary.py b/src/scanoss/inspection/component_summary.py
@@ -35,12 +35,18 @@ def _get_component_summary_from_components(self, scan_components: list)-> dict:
         :param components: List of all components
         :return: Dict with license summary information
         """
+        # A component is considered unique by its combination of PURL (Package URL) and license
+        component_licenses = self._group_components_by_license(scan_components)
+        total_components = len(component_licenses)
+        # Get undeclared components
+        undeclared_components = len([c for c in component_licenses if c['status'] == 'pending'])
+
         components: list = []
-        undeclared_components = 0
-        total_components = 0
+        total_undeclared_files = 0
+        total_files_detected = 0
         for component in scan_components:
-            total_components += component['count']
-            undeclared_components += component['undeclared']
+            total_files_detected += component['count']
+            total_undeclared_files += component['undeclared']
             components.append({
                 'purl': component['purl'],
                 'version': component['version'],
@@ -50,10 +56,13 @@ def _get_component_summary_from_components(self, scan_components: list)-> dict:
             })
         ## End for loop components
         return {
-            'components': components,
-            'total': total_components,
-            'undeclared': undeclared_components,
-            'declared': total_components - undeclared_components,
+            "components": component_licenses,
+            'totalComponents': total_components,
+            'undeclaredComponents': undeclared_components,
+            'declaredComponents': total_components - undeclared_components,
+            'totalFilesDetected': total_files_detected,
+            'totalFilesUndeclared': total_undeclared_files,
+            'totalFilesDeclared': total_files_detected - total_undeclared_files,
         }
 
     def _get_components(self):
diff --git a/src/scanoss/inspection/copyleft.py b/src/scanoss/inspection/copyleft.py
@@ -78,12 +78,14 @@ def _json(self, components: list) -> Dict[str, Any]:
         :param components: List of components with copyleft licenses
         :return: Dictionary with formatted JSON details and summary
         """
+        # A component is considered unique by its combination of PURL (Package URL) and license
+        component_licenses = self._group_components_by_license(components)
         details = {}
         if len(components) > 0:
             details = {'components': components}
         return {
             'details': f'{json.dumps(details, indent=2)}\n',
-            'summary': f'{len(components)} component(s) with copyleft licenses were found.\n',
+            'summary': f'{len(component_licenses)} component(s) with copyleft licenses were found.\n',
         }
 
     def _markdown(self, components: list) -> Dict[str, Any]:
@@ -93,24 +95,24 @@ def _markdown(self, components: list) -> Dict[str, Any]:
         :param components: List of components with copyleft licenses
         :return: Dictionary with formatted Markdown details and summary
         """
-        headers = ['Component', 'Version', 'License', 'URL', 'Copyleft']
+        # A component is considered unique by its combination of PURL (Package URL) and license
+        component_licenses = self._group_components_by_license(components)
+        headers = ['Component', 'License', 'URL', 'Copyleft']
         centered_columns = [1, 4]
         rows: [[]] = []
-        for component in components:
-            for lic in component['licenses']:
+        for comp_lic_item in component_licenses:
                 row = [
-                    component['purl'],
-                    component['version'],
-                    lic['spdxid'],
-                    lic['url'],
-                    'YES' if lic['copyleft'] else 'NO',
+                    comp_lic_item['purl'],
+                    comp_lic_item['spdxid'],
+                    comp_lic_item['url'],
+                    'YES' if comp_lic_item['copyleft'] else 'NO',
                 ]
                 rows.append(row)
             # End license loop
         # End component loop
         return {
             'details': f'### Copyleft licenses\n{self.generate_table(headers, rows, centered_columns)}\n',
-            'summary': f'{len(components)} component(s) with copyleft licenses were found.\n',
+            'summary': f'{len(component_licenses)} component(s) with copyleft licenses were found.\n',
         }
 
     def _jira_markdown(self, components: list) -> Dict[str, Any]:
@@ -120,24 +122,24 @@ def _jira_markdown(self, components: list) -> Dict[str, Any]:
         :param components: List of components with copyleft licenses
         :return: Dictionary with formatted Markdown details and summary
         """
-        headers = ['Component', 'Version', 'License', 'URL', 'Copyleft']
+        # A component is considered unique by its combination of PURL (Package URL) and license
+        component_licenses = self._group_components_by_license(components)
+        headers = ['Component', 'License', 'URL', 'Copyleft']
         centered_columns = [1, 4]
         rows: [[]] = []
-        for component in components:
-            for lic in component['licenses']:
+        for comp_lic_item in component_licenses:
                 row = [
-                    component['purl'],
-                    component['version'],
-                    lic['spdxid'],
-                    lic['url'],
-                    'YES' if lic['copyleft'] else 'NO',
+                    comp_lic_item['purl'],
+                    comp_lic_item['spdxid'],
+                    comp_lic_item['url'],
+                    'YES' if comp_lic_item['copyleft'] else 'NO',
                 ]
                 rows.append(row)
             # End license loop
         # End component loop
         return {
             'details': f'{self.generate_jira_table(headers, rows, centered_columns)}',
-            'summary': f'{len(components)} component(s) with copyleft licenses were found.\n',
+            'summary': f'{len(component_licenses)} component(s) with copyleft licenses were found.\n',
         }
 
     def _filter_components_with_copyleft_licenses(self, components: list) -> list:
@@ -161,7 +163,6 @@ def _filter_components_with_copyleft_licenses(self, components: list) -> list:
                     lic.pop('count', None)  # None is default value if key doesn't exist
 
                 filtered_component['licenses'] = copyleft_licenses
-                del filtered_component['status']
                 filtered_components.append(filtered_component)
         # End component loop
         self.print_debug(f'Copyleft components: {filtered_components}')
diff --git a/src/scanoss/inspection/inspect_base.py b/src/scanoss/inspection/inspect_base.py
@@ -372,6 +372,40 @@ def _get_licenses_order_by_source_priority(self,licenses_data):
         self.print_debug("No priority sources found, returning all licenses as list")
         return licenses_data
 
+    def _group_components_by_license(self,components):
+        """
+        Groups components by their unique component-license pairs.
+
+        This method processes a list of components and creates unique entries for each
+        component-license combination. If a component has multiple licenses, it will create
+        separate entries for each license.
+
+        Args:
+            components: A list of component dictionaries. Each component should have:
+                - purl: Package URL identifying the component
+                - licenses: List of license dictionaries, each containing:
+                    - spdxid: SPDX identifier for the license (optional)
+
+        Returns:
+            list: A list of dictionaries, each containing:
+                - purl: The component's package URL
+                - license: The SPDX identifier of the license (or 'Unknown' if not provided)
+        """
+        component_licenses: dict = {}
+        for component in components:
+            for lic in component['licenses']:
+                spdxid = lic.get('spdxid', 'Unknown')
+                if spdxid not in component_licenses:
+                    key = f'{component["purl"]}-{spdxid}'
+                    component_licenses[key] = {
+                        'purl': component['purl'],
+                        'spdxid': spdxid,
+                        'status': component['status'],
+                        'copyleft': lic['copyleft'],
+                        'url': lic['url'],
+                    }
+        return list(component_licenses.values())
+
 
 #
 # End of PolicyCheck Class
diff --git a/src/scanoss/inspection/license_summary.py b/src/scanoss/inspection/license_summary.py
@@ -23,7 +23,6 @@
 """
 
 import json
-from typing import Any, Dict
 
 from .inspect_base import InspectBase
 
@@ -73,63 +72,42 @@ def __init__( # noqa: PLR0913
         self.exclude = exclude
         self.explicit = explicit
 
-    def _validate_license(self, license_data: Dict[str, Any]) -> bool:
-        """
-        Validate that a license has all required fields.
-
-        :param license_data: Dictionary containing license information
-        :return: True if license is valid, False otherwise
-        """
-        for field in self.REQUIRED_LICENSE_FIELDS:
-            value = license_data.get(field)
-            if value is None:
-                self.print_debug(f'WARNING: {field} is empty in license: {license_data}')
-                return False
-        return True
-
-    def _append_license(self, licenses: dict, new_license) -> None:
-        """Add or update a license in the licenses' dictionary."""
-        spdxid = new_license.get("spdxid")
-        url = new_license.get("url")
-        copyleft = new_license.get("copyleft")
-        if spdxid not in licenses:
-            licenses[spdxid] = {
-                'spdxid': spdxid,
-                'url': url,
-                'copyleft':copyleft,
-                'count': new_license.get("count"),
-            }
-        else:
-            licenses[spdxid]['count'] += new_license.get("count")
-
     def _get_licenses_summary_from_components(self, components: list)-> dict:
         """
         Get a license summary from detected components.
 
         :param components: List of all components
         :return: Dict with license summary information
         """
+        # A component is considered unique by its combination of PURL (Package URL) and license
+        component_licenses = self._group_components_by_license(components)
+        license_component_count = {}
+        # Count license per component
+        for lic in component_licenses:
+            if lic['spdxid'] not in license_component_count:
+                license_component_count[lic['spdxid']] = 1
+            else:
+                license_component_count[lic['spdxid']] += 1
         licenses:dict = {}
-        licenses_with_copyleft = 0
-        total_licenses = 0
-        for component in components:
-            component_licenses =  component.get("licenses", [])
-            for lic in component_licenses:
-                if not self._validate_license(lic):
-                    continue
-                copyleft = lic.get("copyleft")
-                ## Increment counters
-                total_licenses += lic.get("count")
-                if copyleft:
-                    licenses_with_copyleft += lic.get("count")
-                ## Add license
-                self._append_license(licenses, lic)
+        for comp_lic in component_licenses:
+            spdxid = comp_lic.get("spdxid")
+            url = comp_lic.get("url")
+            copyleft = comp_lic.get("copyleft")
+            if spdxid not in licenses:
+                licenses[spdxid] = {
+                    'spdxid': spdxid,
+                    'url': url,
+                    'copyleft': copyleft,
+                    'componentCount': license_component_count.get(spdxid, 0), # Append component count to license
+                }
             ## End for loop licenses
         ## End for loop components
+        detected_licenses = list(licenses.values())
+        licenses_with_copyleft = [lic for lic in detected_licenses if lic['copyleft']]
         return {
-                 'licenses': list(licenses.values()),
-                 'total': total_licenses,
-                 'copyleft': licenses_with_copyleft
+                 'licenses': detected_licenses,
+                 'detectedLicenses': len(detected_licenses), # Count unique licenses. SPDXID is considered unique
+                 'detectedLicensesWithCopyleft': len(licenses_with_copyleft),
                }
 
 
diff --git a/src/scanoss/inspection/undeclared_component.py b/src/scanoss/inspection/undeclared_component.py
@@ -80,7 +80,6 @@ def _get_undeclared_component(self, components: list) -> list or None:
         for component in components:
             if component['status'] == 'pending':
                 # Remove unused keys
-                del component['status']
                 del component['count']
                 del component['declared']
                 del component['undeclared']
@@ -177,7 +176,7 @@ def _markdown(self, components: list) -> Dict[str, Any]:
         # TODO look at using SpdxLite license name lookup method
         component_licenses = self._group_components_by_license(components)
         for component in component_licenses:
-            rows.append([component.get('purl'), component.get('license')])
+            rows.append([component.get('purl'), component.get('spdxid')])
         return {
             'details': f'### Undeclared components\n{self.generate_table(headers, rows)}\n',
             'summary': self._get_summary(component_licenses),
@@ -195,7 +194,7 @@ def _jira_markdown(self, components: list) -> Dict[str, Any]:
         # TODO look at using SpdxLite license name lookup method
         component_licenses = self._group_components_by_license(components)
         for component in component_licenses:
-            rows.append([component.get('purl'), component.get('license')])
+            rows.append([component.get('purl'), component.get('spdxid')])
         return {
             'details': f'{self.generate_jira_table(headers, rows)}',
             'summary': self._get_jira_summary(component_licenses),
@@ -265,36 +264,6 @@ def _get_components(self):
         # Convert to list and process licenses
         return self._convert_components_to_list(components)
 
-    def _group_components_by_license(self,components):
-        """
-        Groups components by their unique component-license pairs.
-
-        This method processes a list of components and creates unique entries for each
-        component-license combination. If a component has multiple licenses, it will create
-        separate entries for each license.
-
-        Args:
-            components: A list of component dictionaries. Each component should have:
-                - purl: Package URL identifying the component
-                - licenses: List of license dictionaries, each containing:
-                    - spdxid: SPDX identifier for the license (optional)
-
-        Returns:
-            list: A list of dictionaries, each containing:
-                - purl: The component's package URL
-                - license: The SPDX identifier of the license (or 'Unknown' if not provided)
-        """
-        component_licenses: dict = {}
-        for component in components:
-            for lic in component['licenses']:
-                spdxid = lic.get('spdxid', 'Unknown')
-                key = f'{component["purl"]}-{spdxid}'
-                component_licenses[key] = {
-                    'purl': component['purl'],
-                    'license': spdxid,
-                }
-        return list(component_licenses.values())
-
     def run(self):
         """
         Run the undeclared component inspection process.
diff --git a/tests/test_policy_inspect.py b/tests/test_policy_inspect.py
