Added project construct to get_cloc function

Also moved the utility FakeProject from test_core to utils.

codemetrics/cloc.py

@@ -9,7 +9,7 @@
 
 import pandas as pd
 
-from . import internals
+from . import internals, scm
 
 __all__ = ["get_cloc"]
 
@@ -24,7 +24,7 @@ class ClocEntry:
 
 
 def get_cloc(
- path: str = ".", cloc_program: str = "cloc", cwd: pl.Path = None
+ project: scm.Project, path: str = ".", cloc_program: str = "cloc"
 ) -> pd.DataFrame:
  """Retrieve lines of code (LOC) using cloc.pl
 
@@ -40,11 +40,11 @@ def get_cloc(
  comment and code counts.
 
  """
- internals.check_run_in_root(path, cwd)
+ internals.check_run_in_root(path, project.cwd)
  cmdline = [cloc_program, "--csv", "--by-file", path]
  cloc_entries = []
  try:
- output = internals.run(cmdline, cwd=cwd).split("\n")
+ output = internals.run(cmdline, cwd=project.cwd).split("\n")
  except FileNotFoundError as err:
  msg = (
  f"{err}. Is {cloc_program} available? Please pass "

tests/test_cloc.py

@@ -4,22 +4,24 @@
 """Tests for loc (lines of code) module."""
 
 import io
+import pathlib as pl
 import textwrap
 import unittest
 from unittest import mock
 
 import pandas as pd
 
 import codemetrics.cloc as loc
-from tests.utils import add_data_frame_equality_func
+
+from . import utils
 
 
 class SimpleDirectory(unittest.TestCase):
  """Given a simple directory."""
 
  def setUp(self):
  """Mocks the internal run command."""
- add_data_frame_equality_func(self)
+ utils.add_data_frame_equality_func(self)
  self.run_output = textwrap.dedent(
  """\
  language,filename,blank,comment,code,"http://cloc.sourceforge.net"
@@ -43,8 +45,8 @@ def tearDown(self):
 
  def test_cloc_reads_files(self):
  """cloc is called and reads the output csv file."""
- actual = loc.get_cloc()
- self.run_.assert_called_with("cloc --csv --by-file .".split(), cwd=None)
+ actual = loc.get_cloc(utils.FakeProject())
+ self.run_.assert_called_with("cloc --csv --by-file .".split(), cwd=pl.Path("."))
  expected = pd.read_csv(
  io.StringIO(
  textwrap.dedent(
@@ -65,7 +67,7 @@ def test_cloc_not_found(self):
  """Clean error message when cloc is not found in the path."""
  self.run_.side_effect = [FileNotFoundError]
  with self.assertRaises(FileNotFoundError) as context:
- _ = loc.get_cloc()
+ _ = loc.get_cloc(utils.FakeProject())
  self.assertIn("cloc", str(context.exception))
 
 
@@ -76,13 +78,15 @@ class TestClocCall(unittest.TestCase):
  def test_cloc_fails_if_not_in_root(self, path_glob):
  """Make sure that command line call checks it is run from the root."""
  with self.assertRaises(ValueError) as context:
- loc.get_cloc()
+ loc.get_cloc(utils.FakeProject())
  path_glob.assert_called_with(pattern=".svn")
  self.assertIn("git or svn root", str(context.exception))
 
  @mock.patch("codemetrics.internals.check_run_in_root", autospec=True)
  @mock.patch("codemetrics.internals.run", autospec=True)
  def test_cloc_called_with_path(self, run, _):
  """Make sure the path is passed as argument to cloc when passed to the function."""
- loc.get_cloc(path="some-path")
- run.assert_called_with(["cloc", "--csv", "--by-file", "some-path"], cwd=None)
+ loc.get_cloc(utils.FakeProject(), path="some-path")
+ run.assert_called_with(
+ ["cloc", "--csv", "--by-file", "some-path"], cwd=pl.Path(".")
+ )

tests/test_core.py

@@ -5,15 +5,13 @@
 
 import datetime as dt
 import io
-import pathlib as pl
 import textwrap
 import unittest
 import unittest.mock as mock
 
 import lizard as lz
 import numpy as np
 import pandas as pd
-import tqdm
 
 import codemetrics as cm
 import codemetrics.scm as scm
@@ -403,23 +401,6 @@ def other():
  """
  )
 
- class FakeProject(scm.Project):
- """Fake project with pre-determined values for the download return values."""
-
- def download(self, data: pd.DataFrame) -> scm.DownloadResult:
- pass
-
- def get_log(
- self,
- path: pl.Path = pl.Path("."),
- after: dt.datetime = None,
- before: dt.datetime = None,
- progress_bar: tqdm.tqdm = None,
- # FIXME: Why do we need path _and_ relative_url
- relative_url: str = None,
- ) -> pd.DataFrame:
- pass
-
  def setUp(self):
  super().setUp()
  self.log = pd.read_csv(
@@ -435,9 +416,9 @@ def setUp(self):
 
  def get_complexity(self):
  """Factor retrieval of complexity"""
- project = self.FakeProject()
+ project = utils.FakeProject()
  with mock.patch.object(
- self.FakeProject,
+ utils.FakeProject,
  "download",
  autospec=True,
  side_effect=[
@@ -470,9 +451,9 @@ def test_handles_no_function(self):
  + cm.core._lizard_fields
  + "file_tokens file_nloc".split()
  )
- project = self.FakeProject()
+ project = utils.FakeProject()
  with mock.patch.object(
- self.FakeProject,
+ utils.FakeProject,
  "download",
  autospec=True,
  return_value=scm.DownloadResult(1, "f.py", ""),
@@ -490,14 +471,14 @@ def test_handles_no_function(self):
  def test_analysis_empty_input_return_empty_output(self, _):
  """Empty input returns and empty dataframe."""
  self.log = self.log.iloc[:0]
- actual = cm.get_complexity(self.log, self.FakeProject())
+ actual = cm.get_complexity(self.log, utils.FakeProject())
  self.assertTrue(actual.empty)
 
  def test_use_default_download(self):
  """When the context.downlad_funcc is defined, use it."""
- project = self.FakeProject()
+ project = utils.FakeProject()
  with mock.patch.object(
- self.FakeProject, "download", return_value=scm.DownloadResult(1, "/", "")
+ utils.FakeProject, "download", return_value=scm.DownloadResult(1, "/", "")
  ) as download:
  _ = cm.get_complexity(self.log, project)
  download.assert_called_with(self.log)

tests/utils.py

@@ -5,11 +5,14 @@
 """Test utility functions and wrappers."""
 
 
+import datetime as dt
 import io
+import pathlib as pl
 import unittest
 
 import pandas as pd
 import pandas.testing as pdt
+import tqdm
 
 import codemetrics.scm as scm
 
@@ -86,3 +89,21 @@ def csvlog_to_dataframe(csv_log: str) -> pd.DataFrame:
  # Reorder columns.
  df = df[scm.LogEntry.__slots__].pipe(scm.normalize_log)
  return df
+
+
+class FakeProject(scm.Project):
+ """Fake project with pre-determined values for the download return values."""
+
+ def download(self, data: pd.DataFrame) -> scm.DownloadResult:
+ pass
+
+ def get_log(
+ self,
+ path: pl.Path = pl.Path("."),
+ after: dt.datetime = None,
+ before: dt.datetime = None,
+ progress_bar: tqdm.tqdm = None,
+ # FIXME: Why do we need path _and_ relative_url
+ relative_url: str = None,
+ ) -> pd.DataFrame:
+ pass