fix: handle plot file times under non-english locales

.github/workflows/ci.yml

@@ -127,6 +127,12 @@ jobs:
       with:
         python-version: ${{ matrix.python.action }}
 
+    - name: Generate extra locales
+      run: |
+        sudo apt-get update
+        sudo apt-get install --yes tzdata locales
+        sudo locale-gen en_US.UTF-8 de_DE.UTF-8
+
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip setuptools wheel

MANIFEST.in

@@ -5,3 +5,4 @@ include *.md
 include VERSION
 include tox.ini
 recursive-include src *.py
+recursive-include src/plotman/_tests/resources *

setup.cfg

@@ -28,10 +28,12 @@ project_urls =
     Changelog = https://github.com/ericaltendorf/plotman/blob/main/CHANGELOG.md
 
 [options]
+include_package_data = True
 package_dir=
     =src
 packages=find:
 install_requires =
+    pendulum
     psutil ~= 5.8
     pyyaml
     texttable

src/plotman/_tests/job_test.py

@@ -0,0 +1,56 @@
+import contextlib
+import datetime
+import locale
+import importlib.resources
+
+import pytest
+from plotman import job
+from plotman._tests import resources
+
+
+class FauxJobWithLogfile:
+    # plotman.job.Job does too much in its .__init_() so we have this to let us
+    # test its .init_from_logfile().
+
+    def __init__(self, logfile_path):
+        self.logfile = logfile_path
+
+    def update_from_logfile(self):
+        pass
+
+
+@pytest.fixture(name='logfile_path')
+def logfile_fixture(tmp_path):
+    log_name = '2021-04-04-19:00:47.log'
+    log_contents = importlib.resources.read_binary(resources, log_name)
+    log_file_path = tmp_path.joinpath(log_name)
+    log_file_path.write_bytes(log_contents)
+
+    return log_file_path
+
+
+@contextlib.contextmanager
+def set_locale(name):
+    # This is terrible and not thread safe.
+
+    original = locale.setlocale(locale.LC_ALL)
+
+    try:
+        yield locale.setlocale(locale.LC_ALL, name)
+    finally:
+        locale.setlocale(locale.LC_ALL, original)
+
+with set_locale('C'):
+    log_file_time = datetime.datetime.strptime('Sun Apr  4 19:00:50 2021', '%a %b  %d %H:%M:%S %Y')
+
+@pytest.mark.parametrize(
+    argnames=['locale_name'],
+    argvalues=[['C'], ['en_US.utf8'], ['de_DE.utf8']],
+)
+def test_job_parses_time_with_non_english_locale(logfile_path, locale_name):
+    faux_job_with_logfile = FauxJobWithLogfile(logfile_path=logfile_path)
+
+    with set_locale(locale_name):
+        job.Job.init_from_logfile(self=faux_job_with_logfile)
+
+    assert faux_job_with_logfile.start_time == log_file_time

src/plotman/_tests/resources/2021-04-04-19:00:47.notes

@@ -0,0 +1,8 @@
+This sample log file was created from a real log file.  It has had the
+following values replaced by naively randomly generated hexadecimal
+characters.
+
+pool public key
+farmer public key
+memo
+id

src/plotman/job.py

@@ -12,6 +12,7 @@
 from enum import Enum, auto
 from subprocess import call
 
+import pendulum
 import psutil
 
 
@@ -47,6 +48,10 @@ def cmdline_argfix(cmdline):
         else:
             yield i
 
+def parse_chia_plot_time(s):
+    # This will grow to try ISO8601 as well for when Chia logs that way
+    return pendulum.from_format(s, 'ddd MMM DD HH:mm:ss YYYY', locale='en', tz=None)
+
 # TODO: be more principled and explicit about what we cache vs. what we look up
 # dynamically from the logfile
 class Job:
@@ -173,7 +178,7 @@ def init_from_logfile(self):
                     m = re.match(r'^Starting phase 1/4:.*\.\.\. (.*)', line)
                     if m:
                         # Mon Nov  2 08:39:53 2020
-                        self.start_time = datetime.strptime(m.group(1), '%a %b  %d %H:%M:%S %Y')
+                        self.start_time = parse_chia_plot_time(m.group(1))
                         found_log = True
                         break  # Stop reading lines in file