Update tests for copyright_tallies.py

    * Remove copyright_summary.py
    * Update summarizer help text

Reference: #2842
Signed-off-by: Jono Yang <jyang@nexb.com>

src/summarycode/plugin_consolidate.py

@@ -22,7 +22,7 @@
 from packagedcode.utils import combine_expressions
 from plugincode.post_scan import PostScanPlugin
 from plugincode.post_scan import post_scan_impl
-from summarycode import copyright_summary
+from summarycode import copyright_tallies
 
 # Tracing flags
 TRACE = False
@@ -289,10 +289,10 @@ def get_consolidated_packages(codebase):
             c = Consolidation(
                 core_license_expression=package_license_expression,
                 # Sort holders by holder key
-                core_holders=[h for h, _ in sorted(copyright_summary.cluster(package_holders), key=lambda t: t[0].key)],
+                core_holders=[h for h, _ in sorted(copyright_tallies.cluster(package_holders), key=lambda t: t[0].key)],
                 other_license_expression=simplified_discovered_license_expression,
                 # Sort holders by holder key
-                other_holders=[h for h, _ in sorted(copyright_summary.cluster(discovered_holders), key=lambda t: t[0].key)],
+                other_holders=[h for h, _ in sorted(copyright_tallies.cluster(discovered_holders), key=lambda t: t[0].key)],
                 files_count=len([package_resource for package_resource in package_resources if package_resource.is_file]),
                 resources=package_resources,
             )
@@ -310,12 +310,12 @@ def get_consolidated_packages(codebase):
 
 
 def process_holders(holders):
-    holders = [copyright_summary.Text(key=holder, original=holder) for holder in holders]
+    holders = [copyright_tallies.Text(key=holder, original=holder) for holder in holders]
 
     for holder in holders:
         holder.normalize()
 
-    holders = list(copyright_summary.filter_junk(holders))
+    holders = list(copyright_tallies.filter_junk(holders))
 
     for holder in holders:
         holder.normalize()

src/summarycode/summarizer.py

@@ -59,8 +59,11 @@ class ScanSummary(PostScanPlugin):
 
     options = [
         PluggableCommandLineOption(('--summary',),
-            is_flag=True, default=False,
-            help='Summarize license, copyright and other scans at the codebase level.',
+            is_flag=True,
+            default=False,
+            help='Summarize scans by providing declared origin '
+                 'information and other detected origin info at the '
+                 'codebase attribute level.',
             help_group=POST_SCAN_GROUP,
             required_options=['classify']
         )

tests/cluecode/cluecode_test_utils.py

@@ -60,9 +60,9 @@ class CopyrightTest(object):
     holders = attr.ib(default=attr.Factory(list))
     authors = attr.ib(default=attr.Factory(list))
 
-    holders_summary = attr.ib(default=attr.Factory(list))
-    copyrights_summary = attr.ib(default=attr.Factory(list))
-    authors_summary = attr.ib(default=attr.Factory(list))
+    holders_tallies = attr.ib(default=attr.Factory(list))
+    copyrights_tallies = attr.ib(default=attr.Factory(list))
+    authors_tallies = attr.ib(default=attr.Factory(list))
 
     expected_failures = attr.ib(default=attr.Factory(list))
     notes = attr.ib(default=None)
@@ -83,14 +83,14 @@ def __attrs_post_init__(self, *args, **kwargs):
                 raise Exception(msg)
 
         # fix counts to be ints: saneyaml loads everything as string
-        for holders_sum in self.holders_summary:
-            holders_sum['count'] = int(holders_sum['count'])
+        for holders_tally in self.holders_tallies:
+            holders_tally['count'] = int(holders_tally['count'])
 
-        for copyrs_sum in self.copyrights_summary:
-            copyrs_sum['count'] = int(copyrs_sum['count'])
+        for copyrs_tally in self.copyrights_tallies:
+            copyrs_tally['count'] = int(copyrs_tally['count'])
 
-        for auths_sum in self.authors_summary:
-            auths_sum['count'] = int(auths_sum['count'])
+        for auths_tally in self.authors_tallies:
+            auths_tally['count'] = int(auths_tally['count'])
 
     def to_dict(self):
         """
@@ -175,32 +175,32 @@ def make_copyright_test_functions(
     name. Create only a single function for multiple tests (e.g. copyrights and
     holders together).
     """
-    from summarycode.copyright_summary import summarize_copyrights
-    from summarycode.copyright_summary import summarize_persons
+    from summarycode.copyright_tallies import tally_copyrights
+    from summarycode.copyright_tallies import tally_persons
 
     def closure_test_function(*args, **kwargs):
         detections = detect_copyrights(test_file)
         copyrights, holders, authors = Detection.split_values(detections)
 
-        holders_summary = []
-        if 'holders_summary' in test.what:
-            holders_summary = as_sorted_mapping(summarize_persons(holders))
+        holders_tallies = []
+        if 'holders_tallies' in test.what:
+            holders_tallies = as_sorted_mapping(tally_persons(holders))
 
-        copyrights_summary = []
-        if 'copyrights_summary' in test.what:
-            copyrights_summary = as_sorted_mapping(summarize_copyrights(copyrights))
+        copyrights_tallies = []
+        if 'copyrights_tallies' in test.what:
+            copyrights_tallies = as_sorted_mapping(tally_copyrights(copyrights))
 
-        authors_summary = []
-        if 'authors_summary' in test.what:
-            authors_summary = as_sorted_mapping(summarize_persons(authors))
+        authors_tallies = []
+        if 'authors_tallies' in test.what:
+            authors_tallies = as_sorted_mapping(tally_persons(authors))
 
         results = dict(
             copyrights=copyrights,
             authors=authors,
             holders=holders,
-            holders_summary=holders_summary,
-            copyrights_summary=copyrights_summary,
-            authors_summary=authors_summary,
+            holders_tallies=holders_tallies,
+            copyrights_tallies=copyrights_tallies,
+            authors_tallies=authors_tallies,
         )
 
         expected_yaml = test.dumps()

tests/cluecode/data/authors/author-config.rpath.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Gordon Matzigkeit <gord@gnu.ai.mit.edu>
-authors_summary:
+authors_tallies:
   - value: Gordon Matzigkeit <gord@gnu.ai.mit.edu>
     count: 1

tests/cluecode/data/authors/author_addr_c-addr_c.c.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - John Doe
-authors_summary:
+authors_tallies:
   - value: John Doe
     count: 1

tests/cluecode/data/authors/author_avinash-BitVector_py.py.yml

@@ -1,9 +1,9 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Avinash Kak (kak@purdue.edu)
   - Avinash Kak (kak@purdue.edu)
-authors_summary:
+authors_tallies:
   - value: Avinash Kak (kak@purdue.edu)
     count: 2

tests/cluecode/data/authors/author_avinash_kak-BitVector_py.py.yml

@@ -1,9 +1,9 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Avinash Kak (kak@purdue.edu)
   - Avinash Kak (kak@purdue.edu)
-authors_summary:
+authors_tallies:
   - value: Avinash Kak (kak@purdue.edu)
     count: 2

tests/cluecode/data/authors/author_complex_author-strtol_c.c.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - the University of California, Berkeley and its contributors
-authors_summary:
+authors_tallies:
   - value: the University of California, Berkeley and its contributors
     count: 1

tests/cluecode/data/authors/author_correct-detail_9_html.html.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_do_not_detect_authorize_as_author.csv.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_does_not_report_incorrect_junk.txt.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_does_not_report_trailing_coma.txt.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Gordon Matzigkeit
-authors_summary:
+authors_tallies:
   - value: Gordon Matzigkeit
     count: 1

tests/cluecode/data/authors/author_does_not_report_trailing_junk_and_incorrect_authors.txt.yml

@@ -1,10 +1,10 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Mohit Agarwal
   - Mitsubishi and NTT.
-authors_summary:
+authors_tallies:
   - value: Mitsubishi and NTT.
     count: 1
   - value: Mohit Agarwal

tests/cluecode/data/authors/author_expat-expat_h.h.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_gary-ProjectInfo_java.java.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Gary O'Neall
-authors_summary:
+authors_tallies:
   - value: Gary O'Neall
     count: 1

tests/cluecode/data/authors/author_gthomas_c-c.c.yml

@@ -1,10 +1,10 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - gthomas, sorin@netappi.com
   - gthomas, sorin@netappi.com, andrew.lunn@ascom.ch
-authors_summary:
+authors_tallies:
   - value: gthomas, sorin@netappi.com
     count: 1
   - value: gthomas, sorin@netappi.com, andrew.lunn@ascom.ch

tests/cluecode/data/authors/author_in_java-MergeSort_java.java.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Scott Violet
-authors_summary:
+authors_tallies:
   - value: Scott Violet
     count: 1

tests/cluecode/data/authors/author_in_java_tag-java.java.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Apple Banana Car
-authors_summary:
+authors_tallies:
   - value: Apple Banana Car
     count: 1

tests/cluecode/data/authors/author_in_postcript-9__ps.ps.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_in_visio_doc-Glitch_ERD_vsd.vsd.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_iproute2.c.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Patrick McHardy <kaber@trash.net>
-authors_summary:
+authors_tallies:
   - value: Patrick McHardy <kaber@trash.net>
     count: 1

tests/cluecode/data/authors/author_nathan-KEYS.yml

@@ -1,6 +1,6 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 notes: |
   name +email is not enough to create an author
   hence this is not detected correctly

tests/cluecode/data/authors/author_no_author-c.c.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none-wrong.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none_c-c.c.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none_fp-false_positives_c.c.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none_js-editor_beta_de_js.js.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none_license-LICENSE.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_none_sample_java-java.java.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - not attributable
-authors_summary:
+authors_tallies:
   - value: not attributable
     count: 1

tests/cluecode/data/authors/author_russ_c-c.c.yml

@@ -1,5 +1,5 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 notes: these are detected as copyrights, not authors Russ Dill <Russ.Dill@asu.edu> Vladimir
   Oleynik <dzo@simtreas.ru>

tests/cluecode/data/authors/author_sample-c.c.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_samplepy-py.py.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_snippet.yml

@@ -1,3 +1,3 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies

tests/cluecode/data/authors/author_stacktrace_cpp-stacktrace_cpp.cpp.yml

@@ -1,8 +1,8 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - faith@dict.org
-authors_summary:
+authors_tallies:
   - value: faith@dict.org
     count: 1

tests/cluecode/data/authors/author_treetablemodeladapter_java-TreeTableModelAdapter_java.java.yml

@@ -1,10 +1,10 @@
 what:
   - authors
-  - authors_summary
+  - authors_tallies
 authors:
   - Philip Milne
   - Scott Violet
-authors_summary:
+authors_tallies:
   - value: Philip Milne
     count: 1
   - value: Scott Violet