From 542781d62c1d05ed37199b9c8657357e5041772d Mon Sep 17 00:00:00 2001
From: Micah Jerome Ellison <micah.jerome.ellison@gmail.com>
Date: Sat, 29 Apr 2023 15:05:22 -0700
Subject: [PATCH] Explicitly check for valid dates in FolderJournal and add
 unit test

---
 jrnl/journals/FolderJournal.py             | 23 +++++----
 jrnl/time.py                               | 11 ++++
 tests/unit/test_journals_folder_journal.py | 60 ++++++++++++++++++++++
 3 files changed, 83 insertions(+), 11 deletions(-)
 create mode 100644 tests/unit/test_journals_folder_journal.py

diff --git a/jrnl/journals/FolderJournal.py b/jrnl/journals/FolderJournal.py
index 5ee9d0b5..020b57e6 100644
--- a/jrnl/journals/FolderJournal.py
+++ b/jrnl/journals/FolderJournal.py
@@ -13,20 +13,16 @@ from .Journal import Journal
 if TYPE_CHECKING:
     from jrnl.journals import Entry
 
-# Search patterns
+# glob search patterns
 DIGIT_PATTERN = "[0123456789]"
 YEAR_PATTERN = DIGIT_PATTERN * 4
-MONTH_PATTERN = DIGIT_PATTERN * 2
-DAY_PATTERN = (DIGIT_PATTERN * 2) + ".txt"
+MONTH_PATTERN = "[01]" + DIGIT_PATTERN
+DAY_PATTERN = "[0123]" + DIGIT_PATTERN + ".txt"
 
-# Entire range of possible folder/file names for months and days
-MONTH_FOLDERS = [str(m).zfill(2) for m in range(1, 13)]
-DAY_FILE_STEMS = [str(d).zfill(2) for d in range(1, 32)]
 
 class Folder(Journal):
     """A Journal handling multiple files in a folder"""
 
-
     def __init__(self, name: str = "default", **kwargs):
         self.entries = []
         self._diff_entry_dates = []
@@ -134,14 +130,14 @@ class Folder(Journal):
     @staticmethod
     def _get_year_folders(path: pathlib.Path) -> list[pathlib.Path]:
         for child in path.glob(YEAR_PATTERN):
-            if child.name.isdigit() and len(child.name) == 4 and child.is_dir():
+            if child.is_dir():
                 yield child
         return
 
     @staticmethod
     def _get_month_folders(path: pathlib.Path) -> list[pathlib.Path]:
         for child in path.glob(MONTH_PATTERN):
-            if child.name in MONTH_FOLDERS and path.is_dir():
+            if int(child.name) > 0 and int(child.name) <= 12 and path.is_dir():
                 yield child
         return
 
@@ -149,8 +145,13 @@ class Folder(Journal):
     def _get_day_files(path: pathlib.Path) -> list[str]:
         for child in path.glob(DAY_PATTERN):
             if (
-                child.stem in DAY_FILE_STEMS
-                and child.match("*.txt")
+                int(child.stem) > 0
+                and int(child.stem) <= 31
+                and time.is_valid_date(
+                    year=int(path.parent.name),
+                    month=int(path.name),
+                    day=int(child.stem),
+                )
                 and child.is_file()
             ):
                 yield str(child)
diff --git a/jrnl/time.py b/jrnl/time.py
index 514d94f2..15e3ec4c 100644
--- a/jrnl/time.py
+++ b/jrnl/time.py
@@ -89,3 +89,14 @@ def parse(
     if dt.days < -28 and not year_present:
         date = date.replace(date.year - 1)
     return date
+
+
+def is_valid_date(year, month, day):
+    """
+    Checks if a string can be parsed as a date
+    """
+    try:
+        datetime.datetime(year, month, day)
+        return True
+    except ValueError:
+        return False
diff --git a/tests/unit/test_journals_folder_journal.py b/tests/unit/test_journals_folder_journal.py
new file mode 100644
index 00000000..66f6e819
--- /dev/null
+++ b/tests/unit/test_journals_folder_journal.py
@@ -0,0 +1,60 @@
+# Copyright © 2012-2023 jrnl contributors
+# License: https://www.gnu.org/licenses/gpl-3.0.html
+
+import pathlib
+from unittest import mock
+
+import pytest
+
+import jrnl
+from jrnl.journals.FolderJournal import Folder
+
+
+@pytest.mark.parametrize(
+    "inputs_and_outputs",
+    [
+        [
+            "/2020/01",
+            ["02.txt", "03.txt", "31.txt"],
+            ["/2020/01/02.txt", "/2020/01/03.txt", "/2020/01/31.txt"],
+        ],
+        [
+            "/2020/02",  # leap year
+            ["02.txt", "03.txt", "28.txt", "29.txt", "31.txt", "39.txt"],
+            [
+                "/2020/02/02.txt",
+                "/2020/02/03.txt",
+                "/2020/02/28.txt",
+                "/2020/02/29.txt",
+            ],
+        ],
+        [
+            "/2100/02",  # not a leap year
+            ["01.txt", "28.txt", "29.txt", "39.txt"],
+            ["/2100/02/01.txt", "/2100/02/28.txt"],
+        ],
+        [
+            "/2023/04",
+            ["29.txt", "30.txt", "31.txt", "39.txt"],
+            ["/2023/04/29.txt", "/2023/04/30.txt"],
+        ],
+    ],
+)
+def test_get_day_files_expected_filtering(inputs_and_outputs):
+    year_month_path, glob_filenames, expected_output = inputs_and_outputs
+
+    year_month_path = pathlib.Path(year_month_path)
+
+    glob_files = map(lambda x: year_month_path / x, glob_filenames)
+    expected_output = list(map(lambda x: str(pathlib.PurePath(x)), expected_output))
+
+    with (
+        mock.patch("pathlib.Path.glob", return_value=glob_files),
+        mock.patch.object(pathlib.Path, "is_file", return_value=True),
+    ):
+        actual_output = list(Folder._get_day_files(year_month_path))
+        actual_output.sort()
+
+        expected_output.sort()
+
+        assert actual_output == expected_output