Improve dateparse and make string_utils.extract_date.

1 files changed, 23 insertions, 0 deletions

diff --git a/string_utils.py b/string_utils.py
index 4127079..7c40dc9 100644
--- a/string_utils.py
+++ b/string_utils.py
@@ -1388,6 +1388,29 @@ def to_date(in_str: str) -> Optional[datetime.date]:
     return None
 
 
+def extract_date(in_str: Any) -> Optional[str]:
+    import itertools
+
+    import dateparse.dateparse_utils as du
+
+    d = du.DateParser()  # type: ignore
+    chunks = in_str.split()
+    for ngram in itertools.chain(
+        list_utils.ngrams(chunks, 5),
+        list_utils.ngrams(chunks, 4),
+        list_utils.ngrams(chunks, 3),
+        list_utils.ngrams(chunks, 2),
+    ):
+        try:
+            expr = " ".join(ngram)
+            logger.debug(f"Trying {expr}")
+            if d.parse(expr):
+                return d.get_date()
+        except du.ParseException:  # type: ignore
+            pass
+    return None
+
+
 def is_valid_date(in_str: str) -> bool:
     """
     Args: