SERVER-39413: Add script to analyze evergreen task tag usage

4 files changed, 210 insertions, 23 deletions

diff --git a/buildscripts/ciconfig/evergreen.py b/buildscripts/ciconfig/evergreen.py
index 1cbe11fd320..55818b73322 100644
--- a/buildscripts/ciconfig/evergreen.py
+++ b/buildscripts/ciconfig/evergreen.py
@@ -140,6 +140,11 @@ class Task(object):
             return ResmokeArgs.get_arg(args, "suites")
         return None
 
+    @property
+    def tags(self):
+        """Get a set of tags this task has been marked with."""
+        return set(self.raw.get("tags", []))
+
     def __str__(self):
         return self.name
 
diff --git a/buildscripts/evergreen_task_tags.py b/buildscripts/evergreen_task_tags.py
new file mode 100755
index 00000000000..8144f048e51
--- /dev/null
+++ b/buildscripts/evergreen_task_tags.py
@@ -0,0 +1,91 @@
+#!/usr/bin/env python
+"""Script to gather information about how tags are used in evergreen tasks."""
+
+from __future__ import absolute_import
+from __future__ import print_function
+
+import argparse
+import os
+import sys
+
+# Get relative imports to work when the package is not installed on the PYTHONPATH.
+if __name__ == "__main__" and __package__ is None:
+    sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from buildscripts.ciconfig import evergreen  # pylint: disable=wrong-import-position
+
+DEFAULT_EVERGREEN_FILE = "etc/evergreen.yml"
+
+
+def parse_command_line():
+    """Parse command line options."""
+    parser = argparse.ArgumentParser(description=main.__doc__)
+
+    parser.add_argument("--list-tags", action="store_true", default=False,
+                        help="List all tags used by tasks in evergreen yml.")
+    parser.add_argument("--tasks-for-tag", type=str, default=None,
+                        help="List all tasks that use the given tag.")
+    parser.add_argument("--evergreen-file", type=str, default=DEFAULT_EVERGREEN_FILE,
+                        help="Location of evergreen file.")
+
+    options = parser.parse_args()
+
+    return options
+
+
+def get_all_task_tags(evg_config):
+    """Get all task tags used in the evergreen configuration."""
+    if evg_config.tasks:
+        return sorted(set.union(*[task.tags for task in evg_config.tasks]))
+    return set()
+
+
+def list_all_tags(evg_config):
+    """
+    Print all task tags found in the evergreen configuration.
+
+    :param evg_config: evergreen configuration.
+    """
+    all_tags = get_all_task_tags(evg_config)
+    for tag in all_tags:
+        print(tag)
+
+
+def get_tasks_with_tag(evg_config, tag):
+    """
+    Get all tasks marked with the given tag in the evergreen configuration.
+
+    :param evg_config: evergreen configuration.
+    :param tag: tag to search for.
+    :return: list of tasks marked with the given tag.
+    """
+    return sorted([task.name for task in evg_config.tasks if tag in task.tags])
+
+
+def list_tasks_with_tag(evg_config, tag):
+    """
+    Print all tasks that are marked with the given tag.
+
+    :param evg_config: evergreen configuration.
+    :param tag: tag to search for.
+    """
+    task_list = get_tasks_with_tag(evg_config, tag)
+    for task in task_list:
+        print(task)
+
+
+def main():
+    """Gather information about how tags are used in evergreen tasks."""
+    options = parse_command_line()
+
+    evg_config = evergreen.parse_evergreen_file(options.evergreen_file)
+
+    if options.list_tags:
+        list_all_tags(evg_config)
+
+    if options.tasks_for_tag:
+        list_tasks_with_tag(evg_config, options.tasks_for_tag)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/buildscripts/tests/ciconfig/test_evergreen.py b/buildscripts/tests/ciconfig/test_evergreen.py
index 8fd3d25ed9a..3fb06c7a67c 100644
--- a/buildscripts/tests/ciconfig/test_evergreen.py
+++ b/buildscripts/tests/ciconfig/test_evergreen.py
@@ -76,12 +76,12 @@ class TestTask(unittest.TestCase):
 
     def test_from_dict(self):
         task_dict = {
-            "name":
-                "compile", "depends_on": [],
+            "name": "compile",
+            "depends_on": [],
             "commands": [{"func": "fetch source"}, {"func": "run a task that passes"},
                          {"func": "run a function with an arg", "vars": {"foobar": "TESTING: ONE"}},
                          {"func": "run a function with an arg", "vars": {"foobar": "TESTING: TWO"}}]
-        }
+        }  # yapf: disable
         task = _evergreen.Task(task_dict)
 
         self.assertEqual("compile", task.name)
@@ -90,12 +90,12 @@ class TestTask(unittest.TestCase):
 
     def test_resmoke_args(self):
         task_dict = {
-            "name":
-                "jsCore", "commands": [{
-                    "func": "run tests",
-                    "vars": {"resmoke_args": "--suites=core --shellWriteMode=commands"}
-                }]
-        }
+            "name": "jsCore",
+            "commands": [{
+                "func": "run tests",
+                "vars": {"resmoke_args": "--suites=core --shellWriteMode=commands"}
+            }]
+        }  # yapf: disable
         task = _evergreen.Task(task_dict)
 
         self.assertEqual("--suites=core --shellWriteMode=commands", task.resmoke_args)
@@ -103,12 +103,12 @@ class TestTask(unittest.TestCase):
 
     def test_resmoke_args_gen(self):
         task_dict = {
-            "name":
-                "jsCore", "commands": [{
-                    "func": "generate resmoke tasks",
-                    "vars": {"task": "core", "resmoke_args": "--shellWriteMode=commands"}
-                }]
-        }
+            "name": "jsCore",
+            "commands": [{
+                "func": "generate resmoke tasks",
+                "vars": {"task": "core", "resmoke_args": "--shellWriteMode=commands"}
+            }]
+        }  # yapf: disable
         task = _evergreen.Task(task_dict)
 
         self.assertEqual("--suites=core --shellWriteMode=commands", task.resmoke_args)
@@ -116,19 +116,47 @@ class TestTask(unittest.TestCase):
 
     def test_resmoke_args_gen_with_suite(self):
         task_dict = {
-            "name":
-                "jsCore", "commands": [{
-                    "func": "generate resmoke tasks", "vars": {
-                        "task": "jsCore", "suite": "core",
-                        "resmoke_args": "--shellWriteMode=commands"
-                    }
-                }]
-        }
+            "name": "jsCore",
+            "commands": [{
+                "func": "generate resmoke tasks", "vars": {
+                    "task": "jsCore", "suite": "core",
+                    "resmoke_args": "--shellWriteMode=commands"
+                }
+            }]
+        }  # yapf: disable
         task = _evergreen.Task(task_dict)
 
         self.assertEqual("--suites=core --shellWriteMode=commands", task.resmoke_args)
         self.assertEqual("core", task.resmoke_suite)
 
+    def test_tags_with_no_tags(self):
+        task_dict = {
+            "name": "jsCore",
+            "commands": [{
+                "func": "run tests",
+                "vars": {"resmoke_args": "--suites=core --shellWriteMode=commands"}
+            }]
+        }  # yapf: disable
+        task = _evergreen.Task(task_dict)
+
+        self.assertEqual(0, len(task.tags))
+
+    def test_tags_with_tags(self):
+        task_dict = {
+            "name": "jsCore",
+            "tags": ["tag 0", "tag 1", "tag 2"],
+            "commands": [{
+                "func": "run tests",
+                "vars": {"resmoke_args": "--suites=core --shellWriteMode=commands"}
+            }]
+        }  # yapf: disable
+        task = _evergreen.Task(task_dict)
+
+        tag_set = task.tags
+        for tag in task_dict["tags"]:
+            self.assertIn(tag, tag_set)
+        self.assertEqual(len(task_dict["tags"]), len(tag_set))
+
 
 class TestTaskGroup(unittest.TestCase):
     """Unit tests for the TaskGroup class."""
diff --git a/buildscripts/tests/test_evergreen_task_tags.py b/buildscripts/tests/test_evergreen_task_tags.py
new file mode 100644
index 00000000000..582bd93c87b
--- /dev/null
+++ b/buildscripts/tests/test_evergreen_task_tags.py
@@ -0,0 +1,63 @@
+"""Unit tests for the evergreen_task_tags script."""
+
+from __future__ import absolute_import
+
+import unittest
+
+from mock import MagicMock
+
+from buildscripts import evergreen_task_tags as ett
+
+# pylint: disable=missing-docstring,no-self-use
+
+
+def gen_tag_set(prefix, size):
+    return set([prefix + " " + str(index) for index in range(size)])
+
+
+class TestGetAllTaskTags(unittest.TestCase):
+    def test_with_no_tasks(self):
+        evg_config_mock = MagicMock(tasks=[])
+        self.assertEqual(0, len(ett.get_all_task_tags(evg_config_mock)))
+
+    def test_with_no_tags(self):
+        n_tasks = 5
+        task_list_mock = [MagicMock(tags=set()) for _ in range(n_tasks)]
+        evg_config_mock = MagicMock(tasks=task_list_mock)
+        self.assertEqual(0, len(ett.get_all_task_tags(evg_config_mock)))
+
+    def test_with_some_tags(self):
+        task_prefixes = ["b", "a", "q", "v"]
+        n_tags = 3
+        task_list_mock = [MagicMock(tags=gen_tag_set(prefix, n_tags)) for prefix in task_prefixes]
+        evg_config_mock = MagicMock(tasks=task_list_mock)
+
+        tag_list = ett.get_all_task_tags(evg_config_mock)
+        self.assertEqual(n_tags * len(task_prefixes), len(tag_list))
+        self.assertEqual(sorted(tag_list), tag_list)
+
+
+class TestGetTasksWithTag(unittest.TestCase):
+    def test_with_no_tasks(self):
+        evg_config_mock = MagicMock(tasks=[])
+        self.assertEqual(0, len(ett.get_tasks_with_tag(evg_config_mock, 'tag')))
+
+    def test_with_no_tags(self):
+        n_tasks = 5
+        task_list_mock = [MagicMock(tags=set()) for _ in range(n_tasks)]
+        evg_config_mock = MagicMock(tasks=task_list_mock)
+        self.assertEqual(0, len(ett.get_tasks_with_tag(evg_config_mock, 'tag')))
+
+    def test_with_one_tag_each(self):
+        task_prefixes = ["b", "a", "b", "v"]
+        n_tags = 3
+        task_list_mock = [MagicMock(tags=gen_tag_set(prefix, n_tags)) for prefix in task_prefixes]
+        for index, task in enumerate(task_list_mock):
+            task.name = "task " + str(index)
+        evg_config_mock = MagicMock(tasks=task_list_mock)
+
+        task_list = ett.get_tasks_with_tag(evg_config_mock, "b 0")
+        self.assertIn("task 0", task_list)
+        self.assertIn("task 2", task_list)
+        self.assertEqual(2, len(task_list))
+        self.assertEqual(sorted(task_list), task_list)