chore: bump sqlglot 20.4.0

and adjust index to accomodate change from tobymao/sqlglot#2619

.pre-commit-config.yaml

@@ -5,8 +5,6 @@ repos:
     rev: v1.16.23
     hooks:
       - id: typos
-        # needed because .typos.toml is ignored see https://github.com/crate-ci/typos/issues/347
-        exclude: CHANGELOG.md
   # formatters and linters are available in the virtualenv so they can be run from the makefile & vscode
   - repo: local
     hooks:

fakesnow/transforms.py

@@ -355,9 +355,13 @@ def regex_substr(expression: exp.Expression) -> exp.Expression:
 
         # which occurrence of the pattern to match
         try:
-            occurrence = expression.args["occurrence"]
+            occurrence = int(expression.args["occurrence"].this)
         except KeyError:
-            occurrence = exp.Literal(this="1", is_string=False)
+            occurrence = 1
+
+        # the duckdb dialect increments bracket (ie: index) expressions by 1 because duckdb is 1-indexed,
+        # so we need to compensate by subtracting 1
+        occurrence = exp.Literal(this=str(occurrence - 1), is_string=False)
 
         try:
             regex_parameters_value = str(expression.args["parameters"].this)

pyproject.toml

@@ -11,7 +11,7 @@ dependencies = [
     "duckdb~=0.9.2",
     "pyarrow",
     "snowflake-connector-python",
-    "sqlglot~=19.5.1",
+    "sqlglot~=20.4.0",
 ]
 
 [project.urls]

tests/test_transforms.py

@@ -150,7 +150,7 @@ def test_regex_substr() -> None:
     assert (
         sqlglot.parse_one("SELECT regexp_substr(string1, 'the\\\\W+\\\\w+')", read="snowflake")
         .transform(regex_substr)
-        .sql()
+        .sql(dialect="duckdb")
         == "SELECT REGEXP_EXTRACT_ALL(string1[1 : ], 'the\\W+\\w+', 0, '')[1]"
     )