Fix license tallies to reflect license detections

Fix license tallies to be keyed by license expressions for each
license detection, instead of the detected_license_expression
for a file. Also take into account package license detections.

Signed-off-by: Ayan Sinha Mahapatra <[email protected]>

Author: AyanSinhaMahapatra

src/summarycode/tallies.py

@@ -159,25 +159,46 @@ def license_tallies(resource, children, keep_details=False):
     sorted by decreasing count.
     """
     LIC_EXP = 'detected_license_expression'
+    LIC_DET = 'license_detections'
+    LIC_CLUE = 'license_clues'
     license_expressions = []
 
     # Collect current data
-    lic_expression = getattr(resource, LIC_EXP, None)
-    if not lic_expression and resource.is_file:
+    detected_expressions = []
+    for detection in getattr(resource, LIC_DET, []):
+        detected_expressions.append(detection["license_expression"])
+    for match in getattr(resource, LIC_CLUE, []):
+        detected_expressions.append(match["license_expression"])
+
+    package_license_detections = []
+    PACKAGE_DATA = 'package_data'
+    package_data = getattr(resource, PACKAGE_DATA, [])
+    if package_data:
+        package_license_detections.extend(
+            [
+                detection
+                for detection in getattr(package_data, LIC_DET, [])
+                if detection
+            ]
+        )
+
+    for detection in package_license_detections:
+        detected_expressions.append(detection["license_expression"])
+
+    if not detected_expressions and resource.is_file:
         # also count files with no detection
         license_expressions.append(None)
     else:
-        license_expressions.append(lic_expression)
+        license_expressions.extend(detected_expressions)
 
     # Collect direct children expression tallies
     for child in children:
         child_tallies = get_resource_tallies(child, key=LIC_EXP, as_attribute=keep_details) or []
         for child_tally in child_tallies:
             # TODO: review this: this feels rather weird
             child_sum_val = child_tally.get('value')
-            if child_sum_val:
-                values = [child_sum_val] * child_tally['count']
-                license_expressions.extend(values)
+            values = [child_sum_val] * child_tally['count']
+            license_expressions.extend(values)
 
     # summarize proper
     licenses_counter = tally_licenses(license_expressions)

tests/formattedcode/data/common/manifests-expected.yaml

@@ -47,14 +47,12 @@ summary:
   declared_holder:
   primary_language: Python
   other_license_expressions:
-    - value:
-      count: 1
+    - value: lgpl-3.0
+      count: 3
     - value: apache-2.0
-      count: 1
+      count: 2
     - value: cddl-1.0
       count: 1
-    - value: lgpl-3.0
-      count: 1
     - value: mit
       count: 1
   other_holders:

tests/formattedcode/data/yaml/package-and-licenses-expected.yaml

@@ -47,8 +47,6 @@ summary:
   declared_holder: Example Corp.
   primary_language: Python
   other_license_expressions:
-    - value:
-      count: 1
     - value: apache-2.0 AND (apache-2.0 OR mit)
       count: 1
     - value: mit

tests/summarycode/data/summary/end-2-end/bug-1141.expected.json

@@ -15,7 +15,7 @@
     "other_license_expressions": [
       {
         "value": null,
-        "count": 1
+        "count": 2
       },
       {
         "value": "gpl-2.0-plus",

tests/summarycode/data/summary/holders/clear_holder.expected.json

@@ -13,10 +13,6 @@
     "declared_holder": "Example Corp.",
     "primary_language": "Python",
     "other_license_expressions": [
-      {
-        "value": null,
-        "count": 1
-      },
       {
         "value": "apache-2.0",
         "count": 1

tests/summarycode/data/summary/holders/combined_holders.expected.json

@@ -13,10 +13,6 @@
     "declared_holder": "Example Corp., Demo Corp.",
     "primary_language": "Python",
     "other_license_expressions": [
-      {
-        "value": null,
-        "count": 1
-      },
       {
         "value": "apache-2.0",
         "count": 1

tests/summarycode/data/summary/license_ambiguity/unambiguous.expected.json

@@ -13,10 +13,6 @@
     "declared_holder": "Example Corp.",
     "primary_language": null,
     "other_license_expressions": [
-      {
-        "value": null,
-        "count": 1
-      },
       {
         "value": "apache-2.0",
         "count": 1

tests/summarycode/data/summary/multiple_package_data/multiple_package_data.expected.json

@@ -21,10 +21,6 @@
         "value": "mit",
         "count": 2
       },
-      {
-        "value": null,
-        "count": 1
-      },
       {
         "value": "apache-2.0 AND (apache-2.0 OR mit)",
         "count": 1

tests/summarycode/data/summary/single_file/single_file.expected.json

@@ -12,12 +12,7 @@
     },
     "declared_holder": "Mort Bay Consulting Pty. Ltd. (Australia) and others, Sun Microsystems",
     "primary_language": null,
-    "other_license_expressions": [
-      {
-        "value": null,
-        "count": 1
-      }
-    ],
+    "other_license_expressions": [],
     "other_holders": [],
     "other_languages": []
   },

tests/summarycode/data/summary/summary_without_holder/summary-without-holder-pypi.expected.json

@@ -15,7 +15,7 @@
     "other_license_expressions": [
       {
         "value": null,
-        "count": 1
+        "count": 9
       }
     ],
     "other_holders": [],