Add authors summaries #1043 #1107

Signed-off-by: Philippe Ombredanne <pombredanne@nexb.com>

src/summarycode/plugin_copyright_summary.py

@@ -78,14 +78,15 @@ class CopyrightSummary(PostScanPlugin):
     attributes = OrderedDict([
         ('copyrights_summary', attr.ib(default=attr.Factory(list))),
         ('holders_summary', attr.ib(default=attr.Factory(list))),
+        ('authors_summary', attr.ib(default=attr.Factory(list))),
     ])
 
     sort_order = 12
 
     options = [
         CommandLineOption(('--copyrights-summary',),
             is_flag=True, default=False,
-            help='Summarize copyrights and holders at the file and '
+            help='Summarize copyrights, holders  and authors at the file and '
                  'directory level.',
             help_group=POST_SCAN_GROUP)
     ]
@@ -107,6 +108,10 @@ def process_codebase(self, codebase, copyrights_summary, **kwargs):
                 {"value": "nexB Inc. and others.", "count": 13},
                 {"value": "MyCo Inc. and others.", "count": 13}
             ],
+        "authors_summary": [
+                {"value": "nexB Inc. and others.", "count": 13},
+                {"value": "MyCo Inc. and others.", "count": 13}
+            ],
         """
 
         def _collect_existing_summary_text_objects(_summaries):
@@ -119,16 +124,17 @@ def _collect_existing_summary_text_objects(_summaries):
 
         for resource in codebase.walk(topdown=False):
             if not (hasattr(resource, 'copyrights')
-                and hasattr(resource, 'holders')):
+                and hasattr(resource, 'holders')
+                and hasattr(resource, 'authors')):
                 continue
             copyrights_summary = []
             holders_summary = []
+            authors_summary = []
             try:
-                # 1. Collect statements from this file/resource if any.
+                # Collect values from this file/resource if any.
                 copyrights_summary = [entry.get('value') for entry in resource.copyrights]
-
-                # 2. Collect holders from this file/resource if any.
                 holders_summary = [entry.get('value', []) for entry in resource.holders]
+                authors_summary = [entry.get('value') for entry in resource.authors]
 
                 if TRACE_DEEP:
                     logger_debug('process_codebase:1:from self:copyrights_summary:')
@@ -139,12 +145,19 @@ def _collect_existing_summary_text_objects(_summaries):
                     for s in holders_summary:
                         logger_debug('  ', s)
 
-                # 3. Collect direct children pre-summarized Texts
+                    logger_debug('process_codebase:1:from self:authors_summary:')
+                    for s in authors_summary:
+                        logger_debug('  ', s)
+
+
+                # Collect direct children pre-summarized Texts
                 for child in resource.children(codebase):
                     copyrights_summary.extend(
                         _collect_existing_summary_text_objects(child.copyrights_summary))
                     holders_summary.extend(
                         _collect_existing_summary_text_objects(child.holders_summary))
+                    authors_summary.extend(
+                        _collect_existing_summary_text_objects(child.authors_summary))
 
                 if TRACE_DEEP:
                     logger_debug('process_codebase:2:self+children:copyrights_summary:')
@@ -155,17 +168,25 @@ def _collect_existing_summary_text_objects(_summaries):
                     for s in holders_summary:
                         logger_debug('  ', s)
 
+                    logger_debug('process_codebase:2:self+children:authors_summary:')
+                    for s in authors_summary:
+                        logger_debug('  ', s)
+
                 # 3. summarize proper and save: expansion, cleaning and deduplication
                 summarized_copyright = summarize_copyrights(copyrights_summary, ignore_years=True)
                 summarized_holder = summarize_holders(holders_summary, expand=False)
+                summarized_authors = summarize_holders(authors_summary, expand=False)
                 resource.copyrights_summary = summarized_copyright
                 resource.holders_summary = summarized_holder
+                resource.authors_summary = summarized_authors
                 codebase.save_resource(resource)
 
             except Exception as _e:
-                msg = 'Failed to create copyrights_summary or holders_summary for resource:\n{}\n'.format(repr(resource))
+                msg = 'Failed to create copyrights, authors or holders summary '
+                'for resource:\n{}\n'.format(repr(resource))
                 msg += 'with copyrights_summary:{}\n'.format(repr(copyrights_summary))
                 msg += 'with holders_summary:{}\n'.format(repr(holders_summary))
+                msg += 'with authors_summary:{}\n'.format(repr(authors_summary))
                 import traceback
                 msg += traceback.format_exc()
                 raise Exception(msg)

tests/cluecode/cluecode_assert_utils.py

@@ -79,6 +79,7 @@ class CopyrightTest(object):
 
     holders_summary = attr.ib(default=attr.Factory(list))
     copyrights_summary = attr.ib(default=attr.Factory(list))
+    authors_summary = attr.ib(default=attr.Factory(list))
 
     expected_failures = attr.ib(default=attr.Factory(list))
     notes = attr.ib(default=None)
@@ -99,10 +100,12 @@ def __attrs_post_init__(self, *args, **kwargs):
         for holders_sum in self.holders_summary:
             holders_sum['count'] = int(holders_sum['count'])
 
-        # fix counts to be ints: sane yaml loads everything as string
         for copyrs_sum in self.copyrights_summary:
             copyrs_sum['count'] = int(copyrs_sum['count'])
 
+        for auths_sum in self.authors_summary:
+            auths_sum['count'] = int(auths_sum['count'])
+
     def to_dict(self):
         """
         Serialize self to an ordered mapping.
@@ -236,12 +239,17 @@ def closure_test_function(*args, **kwargs):
         if 'copyrights_summary' in test.what:
             copyrights_summary = summarize_copyrights(copyrights)
 
+        authors_summary = []
+        if 'authors_summary' in test.what:
+            authors_summary = summarize_holders(authors)
+
         results = dict(
             copyrights=copyrights,
             authors=authors,
             holders=holders,
             holders_summary=holders_summary,
             copyrights_summary=copyrights_summary,
+            authors_summary=authors_summary,
             )
 
         if regen:

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Gordon Matzigkeit <gord@gnu.ai.mit.edu>
+authors_summary:
+    -   value: Gordon Matzigkeit <gord@gnu.ai.mit.edu>
+        count: 1

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - John Doe
+authors_summary:
+    -   value: John Doe
+        count: 1

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

@@ -1,5 +1,9 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Avinash Kak (kak@purdue.edu)
     - Avinash Kak (kak@purdue.edu)
+authors_summary:
+    -   value: Avinash Kak (kak@purdue.edu)
+        count: 2

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

@@ -1,5 +1,9 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Avinash Kak (kak@purdue.edu)
     - Avinash Kak (kak@purdue.edu)
+authors_summary:
+    -   value: Avinash Kak (kak@purdue.edu)
+        count: 2

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - the University of California, Berkeley and its contributors.
+authors_summary:
+    -   value: the University of California, Berkeley and its contributors.
+        count: 1

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Gordon Matzigkeit
+authors_summary:
+    -   value: Gordon Matzigkeit
+        count: 1

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

@@ -1,5 +1,11 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Mohit Agarwal.
     - Mitsubishi and NTT.
+authors_summary:
+    -   value: Mitsubishi and NTT.
+        count: 1
+    -   value: Mohit Agarwal.
+        count: 1

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Gary O'Neall
+authors_summary:
+    -   value: Gary O'Neall
+        count: 1

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

@@ -1,5 +1,11 @@
 what:
     - authors
+    - authors_summary
 authors:
     - gthomas, sorin@netappi.com
     - gthomas, sorin@netappi.com, andrew.lunn@ascom.ch
+authors_summary:
+    -   value: gthomas, sorin@netappi.com, andrew.lunn@ascom.ch
+        count: 1
+    -   value: gthomas, sorin@netappi.com
+        count: 1

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Scott Violet
+authors_summary:
+    -   value: Scott Violet
+        count: 1

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Apple Banana Car
+authors_summary:
+    -   value: Apple Banana Car
+        count: 1

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Patrick McHardy <kaber@trash.net>
+authors_summary:
+    -   value: Patrick McHardy <kaber@trash.net>
+        count: 1

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

@@ -1,6 +1,7 @@
 what:
     - authors
+    - authors_summary
 notes: |-
     name +email is not enough to create an author
-                hence this is not detected correctly
-                Nathan Mittler <nathan.mittler@gmail.com>
+    hence this is not detected correctly
+    Nathan Mittler <nathan.mittler@gmail.com>

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,4 +1,5 @@
 what:
     - authors
+    - authors_summary
 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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

tests/cluecode/data/authors/author_snippet.yml

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

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

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - faith@dict.org
+authors_summary:
+    -   value: faith@dict.org
+        count: 1

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

@@ -1,5 +1,11 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Philip Milne
     - Scott Violet
+authors_summary:
+    -   value: Philip Milne
+        count: 1
+    -   value: Scott Violet
+        count: 1

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

@@ -1,6 +1,12 @@
 what:
     - authors
+    - authors_summary
 authors:
     - the University of California, Berkeley and its contributors.
     - UC Berkeley and its contributors.
     - the University of California, Berkeley and its contributors.
+authors_summary:
+    -   value: the University of California, Berkeley and its contributors.
+        count: 2
+    -   value: UC Berkeley and its contributors.
+        count: 1

tests/cluecode/data/authors/author_var_route_c-var_route_c.c.yml

@@ -1,6 +1,14 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Erik Schoenfelder (schoenfr@ibr.cs.tu-bs.de)
     - Jennifer Bray of Origin (jbray@origin-at.co.uk)
     - Simon Leinen (simon@switch.ch)
+authors_summary:
+    -   value: Erik Schoenfelder (schoenfr@ibr.cs.tu-bs.de)
+        count: 1
+    -   value: Jennifer Bray of Origin (jbray@origin-at.co.uk)
+        count: 1
+    -   value: Simon Leinen (simon@switch.ch)
+        count: 1

tests/cluecode/data/authors/author_vs-visual_studio.txt.yml

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary

tests/cluecode/data/authors/configure.yml

@@ -1,4 +1,8 @@
 what:
     - authors
+    - authors_summary
 authors:
     - Gordon Matzigkeit <gord@gnu.ai.mit.edu>
+authors_summary:
+    -   value: Gordon Matzigkeit <gord@gnu.ai.mit.edu>
+        count: 1

tests/cluecode/data/authors/cvs_keywords.yml

@@ -1,2 +1,3 @@
 what:
     - authors
+    - authors_summary