fix(bigquery,impala): escape all ASCII escape sequences in string lit…

…erals

This also fixes Impala regex functions.

ibis/backends/base/sql/registry/literal.py

@@ -21,7 +21,22 @@ def _boolean_literal_format(translator, op):
 
 
 def _string_literal_format(translator, op):
-    return "'{}'".format(op.value.replace("'", "\\'"))
+    return "'{}'".format(
+        op.value
+        # Escape \ first so we don't double escape other characters.
+        .replace("\\", "\\\\")
+        # Escape ' since we're using those for the string literal.
+        .replace("'", "\\'")
+        # ASCII escape sequences that are recognized in Python:
+        # https://docs.python.org/3/reference/lexical_analysis.html#string-and-bytes-literals
+        .replace("\a", "\\a")  # Bell
+        .replace("\b", "\\b")  # Backspace
+        .replace("\f", "\\f")  # Formfeed
+        .replace("\n", "\\n")  # Newline / Linefeed
+        .replace("\r", "\\r")  # Carriage return
+        .replace("\t", "\\t")  # Tab
+        .replace("\v", "\\v")  # Vertical tab
+    )
 
 
 def _number_literal_format(translator, op):

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_string/escape_ascii_sequences/out.sql

@@ -0,0 +1 @@
+SELECT 'a\ab\bc\fd\ne\rf\tg\vh' AS `'a_x07b_x08c_x0cd_ne_rf_tg_x0bh'`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_string/escape_backslash/out.sql

@@ -0,0 +1 @@
+SELECT 'a\\b\\c' AS `'a_b_c'`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_string/escape_quote/out.sql

@@ -0,0 +1 @@
+SELECT 'a\'b"c' AS `'a_'b_c'`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_string/not_escape_special_characters/out.sql

@@ -0,0 +1 @@
+SELECT '`~!@#$%^&*()_=+-|[]{};:/?<>' AS `'_#_%_&__=+-|_:_<>'`

ibis/backends/bigquery/tests/unit/test_compiler.py

@@ -115,6 +115,20 @@ def test_integer_to_timestamp(case, unit, snapshot):
     snapshot.assert_match(to_sql(expr), "out.sql")
 
 
+@pytest.mark.parametrize(
+    ("case",),
+    [
+        param("a\\b\\c", id="escape_backslash"),
+        param("a\ab\bc\fd\ne\rf\tg\vh", id="escape_ascii_sequences"),
+        param("a'b\"c", id="escape_quote"),
+        param("`~!@#$%^&*()_=+-|[]{};:/?<>", id="not_escape_special_characters"),
+    ],
+)
+def test_literal_string(case, snapshot):
+    expr = ibis.literal(case)
+    snapshot.assert_match(to_sql(expr), "out.sql")
+
+
 @pytest.mark.parametrize(
     ("case", "dtype"),
     [

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/re_extract/out.sql

@@ -1 +1 @@
-regexp_extract(`string_col`, '[\d]+', 0)
+regexp_extract(`string_col`, '[\\d]+', 0)

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/re_replace/out.sql

@@ -1 +1 @@
-regexp_replace(`string_col`, '[\d]+', 'aaa')
+regexp_replace(`string_col`, '[\\d]+', 'aaa')

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/re_search/out.sql

@@ -1 +1 @@
-regexp_like(`string_col`, '[\d]+')
+regexp_like(`string_col`, '[\\d]+')

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/rlike/out.sql

@@ -1 +1 @@
-regexp_like(`string_col`, '[\d]+')
+regexp_like(`string_col`, '[\\d]+')

ibis/backends/tests/test_string.py

@@ -199,7 +199,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.notimpl(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -211,7 +210,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.notimpl(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -240,7 +238,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -252,7 +249,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -276,7 +272,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -290,7 +285,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -304,7 +298,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -318,7 +311,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -330,7 +322,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -342,7 +333,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(["impala"], raises=AssertionError),
             ],
         ),
         param(
@@ -369,10 +359,6 @@ def test_string_col_is_unicode(alltypes, df):
                     ["mysql", "mssql", "druid", "oracle"],
                     raises=com.OperationNotDefinedError,
                 ),
-                pytest.mark.broken(
-                    ["impala"],
-                    raises=AssertionError,
-                ),
             ],
         ),
         param(