From e88c2885954e929bcad939f3f72b04594320bd56 Mon Sep 17 00:00:00 2001
From: RichieHakim <RichHakim@gmail.com>
Date: Fri, 23 Feb 2024 16:33:13 -0500
Subject: [PATCH] Add find_date_in_path function to path_helpers.py

---
 bnpm/path_helpers.py   | 85 +++++++++++++++++++++++++++++++++++++++++-
 bnpm/tests/test_all.py | 29 ++++++++++++++
 2 files changed, 113 insertions(+), 1 deletion(-)

diff --git a/bnpm/path_helpers.py b/bnpm/path_helpers.py
index 905c0d9..654d165 100644
--- a/bnpm/path_helpers.py
+++ b/bnpm/path_helpers.py
@@ -251,4 +251,87 @@ def fix_spaces_in_unix_path(path):
             Unix path with spaces replaced by r'\ '
     """
     from pathlib import Path
-    return Path(path).as_posix().replace(' ', r'\ ')
\ No newline at end of file
+    return Path(path).as_posix().replace(' ', r'\ ')
+
+
+def find_date_in_path(
+    path, 
+    regex_date_formats=[
+        r'\d{4}\d{2}\d{2}',     # 20220203
+        r'\d{4}\D\d{2}\D\d{2}', # 2022_02_03
+        r'\d{2}\D\d{2}\D\d{4}', # 02_03_2022
+        r'\d{1}\D\d{1}\D\d{4}', # 2_3_2022
+        r'\d{1}\D\d{2}\D\d{4}', # 2_03_2022
+        r'\d{2}\D\d{1}\D\d{4}', # 02_3_2022
+        r'\d{2}\D\d{2}\D\d{2}', # 02_03_22
+        r'\d{1}\D\d{1}\D\d{2}', # 2_3_22
+        r'\d{1}\D\d{2}\D\d{2}', # 2_03_22
+        r'\d{2}\D\d{1}\D\d{2}', # 02_3_22
+    ],
+    reverse_path_order=True,
+):
+    """
+    Searches a file or directory path for a date string matching one of several
+    regex patterns and returns the first match.
+    
+    RH 2024
+
+    Args:
+        path (str):
+            The file or directory path in which to search for a date.
+        regex_date_formats (List[str]):
+            A list of regex patterns to match against parts of the path.\n
+            Search goes in order of the list and stops at the first match.\n
+            (Default is a list of common date formats)
+        reverse_path_order (bool):
+            If True, search from the end of the path backwards.
+    
+    Returns:
+        str or None:
+            The first matching date string found, or None if no match is found.
+    """
+    ## make a list of strings
+    regex_date_formats = [regex_date_formats] if isinstance(regex_date_formats, str) else regex_date_formats
+
+    ## Dictionary to modify regex based on the presence of separators at start/end of the date.
+    modifiers = {
+        (0, 0): [r''  , r''  ],
+        (1, 0): [r'\D', r''  ],
+        (0, 1): [r''  , r'\D'],
+        (1, 1): [r'\D', r'\D'],
+    }
+
+    ## Split the path into components and optionally reverse the order of search.
+    parts = Path(path).parts
+    parts = parts[::-1] if reverse_path_order else parts
+
+    def _finder(regexs, parts):
+        """Inner function to find the first date in the path parts based on
+        provided regex patterns."""
+        date = []
+        for part in parts:
+            for regex in regexs:
+                date = re.findall(regex, part)
+                if len(date) > 0:
+                    ## Return the last match found in the current part.
+                    date = date[-1]
+                    break
+            if isinstance(date, str):
+                break
+        date = None if isinstance(date, list) else date
+        return date
+    
+    ## Run the finder with each modifier and stop at the first match.
+    date = None
+    for num, mod in modifiers.items():
+        ## Apply modifiers to each regex pattern and search the path parts.
+        date = _finder(
+            regexs=[mod[0] + regex + mod[1] for regex in regex_date_formats], 
+            parts=parts,
+        )
+        if date is not None:
+            ## Remove the modifiers from the date string.
+            date = date[num[0]:-num[1] if num[1]>0 else None]
+            break
+
+    return date
diff --git a/bnpm/tests/test_all.py b/bnpm/tests/test_all.py
index ec0c209..6a8aa50 100644
--- a/bnpm/tests/test_all.py
+++ b/bnpm/tests/test_all.py
@@ -4,6 +4,9 @@
 import numpy as np
 import scipy.signal
 
+import pytest
+import hypothesis
+
 from .. import math_functions
 
 from ..optimization import Convergence_checker
@@ -12,6 +15,8 @@
 
 from ..clustering import cluster_similarity_matrices
 
+from ..path_helpers import find_date_in_path
+
 
 
 def test_Convergence_checker():
@@ -199,3 +204,27 @@ def test_cluster_similarity_matrices():
     assert np.allclose(cs_mean, cs_mean_expected, atol=1e-6)
     assert np.allclose(cs_max, cs_max_expected, atol=1e-6)
     assert np.allclose(cs_min, cs_min_expected, atol=1e-6)
+
+
+@pytest.mark.parametrize("path,expected_date,reverse_path_order", [
+    (r"/home/user/documents/20220203/report.txt", "20220203", True),
+    (r"/home/user/docs/2022_02_03/report.pdf", "2022_02_03", True),
+    (r"/home/02_03_22/data.txt", "02_03_22", True),
+    (r"/2022-02-03/home/data.txt", "2022-02-03", True),
+    (r"/home/data_2_3_2022.txt", "2_3_2022", True),
+    (r"/home/docs/data_2-03-22.txt", '2-03-22', True),
+    (r"C:\home\docs\data_2-03-22.txt", '2-03-22', True),
+    (r"/home/docs/19900101/data_2-03-22.txt", '2-03-22', True),
+    (r"/home/docs/19900101/data_2-03-22.txt", '19900101', False),
+    (r"/home/docs/_19900101/data_2-03-22.txt", '19900101', False),
+    (r"/home/docs/19900101_/data_2-03-22.txt", '19900101', False),
+    (r"/home/docs/_19900101_/data_2-03-22.txt", '19900101', False),
+    (r"/home/docs/_19900101_/data_020322.txt", '19900101', False),
+    (r"/home/docs/_19900101_/data_20230322.txt", '19900101', False),
+    (r"/home/docs/_19900101_/data_20230322.txt", '20230322', True),
+    (r"/home/docs/_1990010_/data_2023032.txt", None, True),
+    (r"/home/docs/_1990010_/data_0 32.txt", None, True),
+    (r"/home/docs/_/data_.txt", None, True),
+])
+def test_known_dates_in_path(path, expected_date, reverse_path_order):
+    assert find_date_in_path(path, reverse_path_order=reverse_path_order) == expected_date