Merge inishchith:fix_summary_by_facet_for_packages branch

This is for PR #1864 and  #1759

Signed-off-by: Philippe Ombredanne <[email protected]>

Author: pombredanne

src/summarycode/summarizer.py

@@ -19,7 +19,6 @@
 from summarycode.utils import get_resource_summary
 from summarycode.utils import set_resource_summary
 
-
 # Tracing flags
 TRACE = False
 TRACE_LIGHT = False
@@ -279,6 +278,7 @@ def summarize_values(values, attribute):
         holders=summarize_holders,
         authors=summarize_holders,
         programming_language=summarize_languages,
+        packages=summarize_packages,
     )
     return value_summarizers_by_attr[attribute](values)
 
@@ -328,7 +328,7 @@ def summarize_codebase_key_files(codebase, **kwargs):
         'authors',
         'programming_language',
         # 'packages',
-    ])  
+    ])
     summarizable_attributes = [k for k in summarizable_attributes
         if k in really_summarizable_attributes]
 
@@ -480,11 +480,28 @@ def package_summarizer(resource, children, keep_details=False):
         logger_debug('package_summarizer: for:', resource,
                      'packages are:', packs)
 
-    # Collect direct children packages summary
-    for child in children:
-        child_summaries = get_resource_summary(child, key='packages', as_attribute=False) or []
-        packages.extend(child_summaries)
+    package_urls = []
+    for package in packages:
+        purl = package.get('purl')
+        if purl:
+            package_urls.append(purl)
 
     # summarize proper
-    set_resource_summary(resource, key='packages', value=packages, as_attribute=False)
-    return packages
+    packages_counter = summarize_packages(package_urls)
+    summarized = sorted_counter(packages_counter)
+    set_resource_summary(
+        resource=resource,
+        key='packages',
+        value=summarized,
+        as_attribute=keep_details,
+    )
+
+    return summarized
+
+
+def summarize_packages(package_urls):
+    """
+    Given a list of package urls, return a mapping of {expression: count
+    of occurences}
+    """
+    return Counter(package_urls)

tests/summarycode/data/full_summary/summary_by_facet.expected.json

@@ -5,6 +5,7 @@
       "options": {
         "input": "<path>",
         "--copyright": true,
+        "--email": true,
         "--facet": [
           "dev=*.java",
           "dev=*.cs",

tests/summarycode/test_summarizer.py

@@ -99,7 +99,7 @@ def test_full_summary_by_facet(self):
         result_file = self.get_temp_file('json')
         expected_file = self.get_test_loc('full_summary/summary_by_facet.expected.json')
         run_scan_click([
-            '-cli',
+            '-clpieu',
             '--facet', 'dev=*.java',
             '--facet', 'dev=*.cs',
             '--facet', 'dev=*ada*',