Merge pull request #253 from mabel-dev/FIX/#252

FIX/#252

docs/Release Notes/Change Log.md

@@ -15,6 +15,9 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - [[#35](https://github.com/mabel-dev/opteryx/issues/35)] Table scan planning done during query planning. ([@joocer](https://github.com/joocer))   
 - [[#173](https://github.com/mabel-dev/opteryx/issues/173)] Data not found raises different errors under different scenarios. ([@joocer](https://github.com/joocer))
 
+**Fixed**
+
+- [[#252](https://github.com/mabel-dev/opteryx/issues/252)] Planner should gracefully convert byte strings to ascii strings. ([@joocer](https://github.com/joocer))
 
 ### [0.1.0] - 2022-07-02
 

docs/SQL Reference/Working with SQL/10 Working with Dates.md

@@ -0,0 +1,5 @@
+# Working with Dates
+
+
+- EXTRACT is usually faster than DATE_TRUNC but DATE_TRUNC is more functionally rich
+- EXTRACT is preferred over the individual extractors (YEAR, MONTH, DAY etc)

opteryx/engine/functions/__init__.py

@@ -21,7 +21,7 @@
 from pyarrow import compute
 
 import opteryx
-from opteryx.engine.functions import other_functions, date_functions
+from opteryx.engine.functions import other_functions, date_functions, string_functions
 from opteryx.exceptions import SqlError
 from opteryx.third_party.date_trunc import date_trunc
 from opteryx.utils.dates import parse_iso
@@ -151,7 +151,7 @@ def get_len(obj):
     "UPPER": compute.utf8_upper,  # UPPER(str) -> str
     "LOWER": compute.utf8_lower,  # LOWER(str) -> str
     "TRIM": compute.utf8_trim_whitespace,  # TRIM(str) -> str
-    "LEFT": _iterate_double_parameter(lambda x, y: str(x)[: int(y)]),
+    "LEFT": string_functions._string_slicer_left,
     "RIGHT": _iterate_double_parameter(lambda x, y: str(x)[-int(y) :]),
     # HASHING & ENCODING
     "HASH": _iterate_single_parameter(lambda x: format(CityHash64(str(x)), "X")),

opteryx/engine/functions/string_functions.py

@@ -9,3 +9,12 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
+import numpy
+
+
+def _string_slicer_left(arr, length):
+    length = int(length)  # it's probably a float64
+    arr = arr.astype(str)  # it's probably an array of objects
+    interim = arr.view((str, 1)).reshape(len(arr), -1)[:, 0:length]
+    return [numpy.array(interim).view((str, length)).flatten()]

opteryx/engine/planner/planner.py

@@ -105,6 +105,11 @@ def copy(self):
     def create_plan(self, sql: str = None, ast: dict = None):
 
         if sql:
+
+            # if it's a byte string, convert to an ascii string
+            if isinstance(sql, bytes):
+                sql = sql.decode()
+
             import sqloxide
 
             # extract temporal filters, this isn't supported by sqloxide
@@ -116,7 +121,7 @@ def create_plan(self, sql: str = None, ast: dict = None):
                 # identifiers to start with _ (underscore) and $ (dollar sign)
                 # https://github.com/sqlparser-rs/sqlparser-rs/blob/main/src/dialect/mysql.rs
             except ValueError as exception:  # pragma: no cover
-                raise SqlError(exception)
+                raise SqlError from exception
         else:
             self._ast = ast
 

opteryx/version.py

@@ -16,4 +16,4 @@
 2) we can import it in setup.py for the same reason
 """
 
-__version__ = "0.1.1-beta.2"
+__version__ = "0.1.1-beta.3"

tests/sql_battery/test_battery_shape.py

@@ -402,6 +402,8 @@
         ("SELECT * FROM tests.data.nulls WHERE username NOT ILIKE 'BBC%' FOR '2000-01-01'", 21, 5),
         ("SELECT * FROM tests.data.nulls WHERE username ~ 'BBC.+' FOR '2000-01-01'", 3, 5),
         ("SELECT * FROM tests.data.nulls WHERE tweet ILIKE '%Trump%' FOR '2000-01-01'", 0, 5),
+        # BYTE-ARRAY FAILS #252
+        (b"SELECT * FROM $satellites", 177, 8),
     ]
 # fmt:on