Fixes small issues and code style, added unit tests.

Author: matejak

.cico.yaml

@@ -78,5 +78,5 @@
             git_username: openscap
             git_repo: scap-security-guide
             ci_project: '{name}'
-            ci_cmd: "yum -y install install cmake openscap-utils git PyYAML python-jinja2 && cd build && cmake ../ && make -j4 && CTEST_OUTPUT_ON_FAILURE=1 ctest -j4"
+            ci_cmd: "yum -y install install cmake openscap-utils git PyYAML python-jinja2 python-pytest && cd build && cmake ../ && make -j4 && CTEST_OUTPUT_ON_FAILURE=1 ctest -j4"
             timeout: '20m'

CMakeLists.txt

@@ -178,6 +178,9 @@ enable_testing()
 
 include(SSGCommon)
 
+ssg_python_unit_tests("misc")
+ssg_python_unit_tests("utils")
+
 # Targets 'stats', 'profile-stats' and 'zipfile' need to be added
 # before any product because they will receive dependencies from products added
 

cmake/SSGCommon.cmake

@@ -219,6 +219,22 @@ if (SSG_SHELLCHECK_BASH_FIXES_VALIDATION_ENABLED AND SHELLCHECK_EXECUTABLE AND "
     endif()
 endmacro()
 
+
+if (ENABLE_PYTHON_COVERAGE)
+    set(PYTEST_COVERAGE_OPTIONS --cov-append --cov-report=xml --cov "${CMAKE_SOURCE_DIR}/shared/")
+endif()
+
+macro(ssg_python_unit_tests PYTHON_COMPONENT)
+add_test(
+    NAME "python-unit-${PYTHON_COMPONENT}"
+    COMMAND "${PYTHON_EXECUTABLE}" -m pytest ${PYTEST_COVERAGE_OPTIONS} "${CMAKE_SOURCE_DIR}/tests/unit/${PYTHON_COMPONENT}"
+)
+set_tests_properties ("python-unit-${PYTHON_COMPONENT}" PROPERTIES ENVIRONMENT
+    "PYTHONPATH=${CMAKE_SOURCE_DIR}/shared/${PYTHON_COMPONENT}"
+)
+endmacro()
+
+
 macro(ssg_build_remediations PRODUCT)
     message(STATUS "Scanning for dependencies of ${PRODUCT} fixes (bash, ansible, puppet and anaconda)...")
     _ssg_build_remediations_for_language(${PRODUCT} "bash")

shared/misc/generate-contributors.py shared/misc/generate_contributors.py

@@ -9,9 +9,7 @@
 import collections
 
 
-MANUAL_EDIT_WARNING = \
-"""
-This file is generated using the %s script. DO NOT MANUALLY EDIT!!!!
+MANUAL_EDIT_WARNING = """This file is generated using the %s script. DO NOT MANUALLY EDIT!!!!
 Last Modified: %s
 """ % (os.path.basename(__file__), datetime.datetime.now().strftime("%Y-%m-%d %H:%M"))
 
@@ -68,22 +66,21 @@
 }
 
 
-def _get_contributions_by_email(output):
+def _get_contributions_by_canonical_email(output):
     contributions_by_email = collections.defaultdict(list)
     for line in output.split("\n"):
-        match = re.match(r"[\s]*([0-9]+)[\s+](.+)[\s]+\<(.+)\>", line)
+        match = re.match(r"[\s]*([0-9]+)\s+(.+)\s+\<(.+)\>", line)
         if match is None:
             continue
 
-        commits, name, email = match.groups()
+        commits_count, author_name, email = match.groups()
 
-        if email in email_mappings:
-            email = email_mappings[email]
+        canonical_email = email_mappings.get(email, email)
 
-        if email == "":
+        if canonical_email == "":
             continue  # ignored
 
-        contributions_by_email[email].append((int(commits), name))
+        contributions_by_email[canonical_email].append((int(commits_count), author_name))
     return contributions_by_email
 
 
@@ -94,23 +91,21 @@ def _get_name_used_most_in_contributions(contribution_sets):
 
 def _get_contributor_email_mapping(contributions_by_email):
     contributors = {}
-    # We will use the most used full name
     for email in contributions_by_email:
         name_used_most = _get_name_used_most_in_contributions(contributions_by_email[email])
-        if name_used_most in name_mappings:
-            name_used_most = name_mappings[name_used_most]
+        canonical_name_used_most = name_mappings.get(name_used_most, name_used_most)
 
-        contributors[name_used_most] = email
+        contributors[canonical_name_used_most] = email
     return contributors
 
 
 def _names_sorted_by_last_name(names):
-    return sorted(names, key=lambda x: x.split(" ")[-1].upper())
+    return sorted(names, key=lambda x: tuple(n.upper() for n in x.split(" "))[::-1])
 
 
 def main():
     output = subprocess.check_output(["git", "shortlog", "-se"]).decode("utf-8")
-    contributions_by_email = _get_contributions_by_email(output)
+    contributions_by_email = _get_contributions_by_canonical_email(output)
     contributors = _get_contributor_email_mapping(contributions_by_email)
 
     contributors_md = "<!---%s--->\n\n" % MANUAL_EDIT_WARNING

tests/unit/misc/test_generate_contributors.py

@@ -0,0 +1,45 @@
+import pytest
+
+import generate_contributors
+
+
+INPUT = """
+    1  David Smith <[email protected]>
+    170  David Smith <[email protected]>
+    19  Dave Smith <[email protected]>
+    2  Deric Crago abc <[email protected]>
+    1  Deric Crago aaa <[email protected]>
+    1  Aeric Erago abc <[email protected]>
+    1  Aeric Drago abc <[email protected]>
+    14  nobodyl <[email protected]>
+"""
+
+
+@pytest.fixture()
+def emails():
+    return generate_contributors._get_contributions_by_canonical_email(INPUT)
+
+
+@pytest.fixture()
+def authors(emails):
+    return generate_contributors._get_contributor_email_mapping(emails)
+
+
+def test_contributions_aggregation(emails):
+    assert "[email protected]" in emails
+    assert "[email protected]" not in emails  # <-- Not the canonical email
+    assert "[email protected]" not in emails  # <-- emails is supposed to be ignored
+
+
+def test_contributors_aggregation(authors):
+    assert "Dave Smith" not in authors
+    assert "David Smith" in authors
+
+
+def test_name_sorting(authors):
+    authors_names = list(authors.keys())
+    sorted_names = generate_contributors._names_sorted_by_last_name(authors_names)
+    assert sorted_names[0] == "Deric Crago aaa"
+    assert sorted_names[1] == "Deric Crago abc"
+    assert sorted_names[2] == "Aeric Drago abc"
+    assert sorted_names[3] == "Aeric Erago abc"

tests/unit/modules/test_ssgcommon.py

@@ -0,0 +1,5 @@
+
+
+
+def test_nothing():
+    pass

tests/unit/utils/test_relabel_ids.py

@@ -0,0 +1,4 @@
+
+
+def test_nothing():
+    pass