Fix: Reinstate dateparser==1.2.2 with a workaround for the upstream issue

Author: erindru

pyproject.toml

@@ -11,7 +11,7 @@ dependencies = [
     "click",
     "croniter",
     "duckdb>=0.10.0,!=0.10.3",
-    "dateparser<=1.2.1",
+    "dateparser>=1.2.2",
     "hyperscript>=0.1.0",
     "importlib-metadata; python_version<'3.12'",
     "ipywidgets",

sqlmesh/__init__.py

@@ -5,6 +5,13 @@
 
 from __future__ import annotations
 
+# Work around dateparser upstream issue: https://github.com/scrapinghub/dateparser/issues/1282
+# This needs to happen before sqlmesh triggers an import of dateparser
+# This hack can be removed when the upstream issue is resolved
+from sqlmesh._hacks import fix_dateparser
+
+fix_dateparser()
+
 import glob
 import logging
 import os

sqlmesh/_hacks.py

@@ -0,0 +1,40 @@
+UPSTREAM_PICKLE_FILE_SIGNATURES = {
+    "a3a8d27b822072fa5c67c0651cb3c934"  # distributed with dateparser==1.2.2
+}
+
+
+def fix_dateparser() -> None:
+    # work around the following upstream issues in dateparser==1.2.2 (which all have the same root cause):
+    # - https://github.com/scrapinghub/dateparser/issues/1270
+    # - https://github.com/scrapinghub/dateparser/issues/1281
+    # - https://github.com/scrapinghub/dateparser/issues/1282
+    #
+    # This hack can be removed if this issue is fixed upstream.
+    # If you're removing this hack, make sure to update pyproject.toml to blacklist version 1.2.2
+
+    import importlib, hashlib
+    from pathlib import Path
+
+    spec = importlib.util.find_spec("dateparser")
+    if spec and spec.origin:
+        # spec.origin will be something like:
+        # "/path/to/venv/lib/python3.9/site-packages/dateparser/__init__.py"
+        tz_cache = Path(spec.origin).parent / "data" / "dateparser_tz_cache.pkl"
+        if tz_cache.exists():
+            # if the tz_cache file matches the signature of the buggy upstream one, delete it
+            # we dont want to delete it if it's been generated locally against our environment
+            signature = hashlib.md5(tz_cache.read_bytes()).hexdigest()
+            if signature in UPSTREAM_PICKLE_FILE_SIGNATURES:
+                try:
+                    tz_cache.unlink()
+                except Exception as e:
+                    print(f"WARNING: Unable to delete upstream dateparser cache: {str(e)}")
+
+    # Test that it actually worked
+    import dateparser
+
+    if dateparser.parse("1 minute ago") is None:
+        print(
+            "WARNING: Buggy dateparser detected; some date expressions may fail to parse.\n"
+            f"Please either delete the file '{str(tz_cache)}' manually or use dateparser<=1.2.1"
+        )