Add check-tmp step to local repo tests

- check-tmp checks disk usage in a collection of directories,
  and fails if they exceed a certain size, reporting on contents
- re-use images and skip build stage in verify/check-tmp tests.
  Should save some time,
  even though the build cache would have been used.
  This way, we don't even check.

Author: minrk

tests/check-tmp

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""
+Script to check for leftover files
+
+Checks a collection of temporary or cache directories,
+to ensure we aren't wasting image size by forgetting cleanup steps.
+
+This script is run in every local repo image we test
+"""
+
+import os
+import sys
+from subprocess import check_output
+from textwrap import indent
+
+# directories larger than this are considered a failure
+# a few little files here aren't a problem
+THRESHOLD = 1  # in MB
+
+MB = 1024 * 1024
+
+# the paths to check
+# all of these locations
+# should be cleaned up
+# missing is okay
+PATHS = [
+    "/tmp/",
+    "~/",
+    "~/.cache/",
+    # not running with read permissions on root
+    # "/root/",
+]
+
+
+def du(path):
+    """Return disk usage in megabytes of a path"""
+    # -ks: get total size, reported in kilobytes
+    out = check_output(["du", "-Hks", path])
+    return int(out.split(None, 1)[0]) / 1024
+
+
+def check_dir_size(path):
+    """Check the size of a directory
+
+    Returns:
+
+    True: directory size is below THRESHOLD or is missing
+    False: directory is larger than THRESHOLD
+    """
+    path = os.path.expanduser(path)
+
+    if not os.path.exists(path):
+        print(f"{path}: missing OK")
+        return True
+
+    size_mb = du(path)
+    print(f"{path}: {size_mb:.1f} MB", end=" ")
+    if size_mb <= THRESHOLD:
+        print("OK")
+        return True
+    else:
+        print("FAIL")
+        # check size of files one-level deep (du only reports dirs)
+        for name in os.listdir(path):
+            subpath = os.path.join(path, name)
+            if os.path.isfile(subpath):
+                file_sz = os.stat(subpath).st_size / MB
+                if file_sz > 0.1:
+                    print(f"  {file_sz:.1f}M {subpath}")
+        # get report on all subdirs that are at least 100k
+        print(
+            indent(
+                check_output(["du", "-Hh", "-t", "100000", path]).decode("utf8"), "  "
+            )
+        )
+        return False
+
+
+def main():
+    results = [check_dir_size(path) for path in PATHS]
+    if not all(results):
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

tests/conftest.py

@@ -22,14 +22,17 @@
 import requests
 import subprocess
 import time
-
 from tempfile import TemporaryDirectory
 
+
+import escapism
 import pytest
 import yaml
 
 from repo2docker.__main__ import make_r2d
 
+TESTS_DIR = os.path.abspath(os.path.dirname(__file__))
+
 
 def pytest_collect_file(parent, path):
     if path.basename == "verify":
@@ -38,12 +41,18 @@ def pytest_collect_file(parent, path):
         return RemoteRepoList.from_parent(parent, fspath=path)
 
 
-def make_test_func(args):
+def make_test_func(args, skip_build=False):
     """Generate a test function that runs repo2docker"""
 
     def test():
         app = make_r2d(args)
         app.initialize()
+        if skip_build:
+
+            def build_noop():
+                print("Skipping build")
+
+            app.skip_build = build_noop
         if app.run_cmd:
             # verify test, run it
             app.start()
@@ -184,14 +193,14 @@ def repo_with_submodule():
 class Repo2DockerTest(pytest.Function):
     """A pytest.Item for running repo2docker"""
 
-    def __init__(self, name, parent, args=None):
+    def __init__(self, name, parent, args=None, skip_build=False):
         self.args = args
         self.save_cwd = os.getcwd()
-        f = parent.obj = make_test_func(args)
+        f = parent.obj = make_test_func(args, skip_build=skip_build)
         super().__init__(name, parent, callobj=f)
 
     def reportinfo(self):
-        return self.parent.fspath, None, ""
+        return (self.parent.fspath, None, "")
 
     def repr_failure(self, excinfo):
         err = excinfo.value
@@ -217,11 +226,34 @@ def collect(self):
                 extra_args = yaml.safe_load(f)
             args += extra_args
 
+        print(self.fspath.basename, self.fspath.dirname, str(self.fspath))
+        # re-use image name for multiple tests of the same image
+        # so we don't run through the build twice
+        rel_repo_dir = os.path.relpath(self.fspath.dirname, TESTS_DIR)
+        image_name = f"r2d-tests-{escapism.escape(rel_repo_dir, escape_char='-').lower()}-{int(time.time())}"
+        args.append(f"--image-name={image_name}")
         args.append(self.fspath.dirname)
-
         yield Repo2DockerTest.from_parent(self, name="build", args=args)
+
+        yield Repo2DockerTest.from_parent(
+            self,
+            name=self.fspath.basename,
+            args=args + ["./verify"],
+            skip_build=True,
+        )
+
+        # mount the tests dir as a volume
+        check_tmp_args = (
+            args[:-1]
+            + ["--volume", f"{TESTS_DIR}:/io/tests"]
+            + [args[-1], "/io/tests/check-tmp"]
+        )
+
         yield Repo2DockerTest.from_parent(
-            self, name=self.fspath.basename, args=args + ["./verify"]
+            self,
+            name="check-tmp",
+            args=check_tmp_args,
+            skip_build=True,
         )