fix #50 partition by single expression raise TypeError (#52)

* fix #50 partition by single expression raise TypeError

* fix testpypi release version

Author: jayvynl

.github/workflows/base-test.yml

@@ -45,6 +45,11 @@ jobs:
         flag-name: run-${{ join(matrix.*, '-') }}
         parallel: true
 
+  release:
+
+    needs: test
+    uses: ./.github/workflows/release.yml
+
   coveralls-finish:
 
     needs: test

.github/workflows/release.yml

@@ -1,6 +1,7 @@
 name: Publish Python 🐍 distribution 📦 to PyPI and TestPyPI
 
-on: push
+on:
+  workflow_call:
 
 jobs:
   build:

.gitignore

@@ -1,5 +1,6 @@
 .coverage
 .idea
+.vscode
 .tox
 venv
 **/__pycache__

CHANGELOG.md

@@ -1,3 +1,6 @@
+### 1.1.3
+- Fix #50 partition by single expression raise TypeError.
+
 ### 1.1.2
 - Use [flake8](https://flake8.pycqa.org/) to lint code.
 - Add GitHub action which runs tests.

README.md

@@ -168,7 +168,7 @@ Notices about model definition:
 - need to specify the engine for clickhouse, specify the order_by for clickhouse order and the partition_by argument
 
 ```python
-from django.db.models import CheckConstraint, Func, IntegerChoices, Q
+from django.db.models import CheckConstraint, IntegerChoices, Q
 from django.utils import timezone
 
 from clickhouse_backend import models
@@ -180,22 +180,21 @@ class Event(models.ClickhouseModel):
         DROP = 2
         ALERT = 3
     ip = models.GenericIPAddressField(default="::")
-    ipv4 = models.GenericIPAddressField(default="127.0.0.1")
+    ipv4 = models.IPv4Field(default="127.0.0.1")
     ip_nullable = models.GenericIPAddressField(null=True)
     port = models.UInt16Field(default=0)
     protocol = models.StringField(default="", low_cardinality=True)
-    content = models.StringField(default="")
+    content = models.JSONField(default=dict)
     timestamp = models.DateTime64Field(default=timezone.now)
     created_at = models.DateTime64Field(auto_now_add=True)
     action = models.EnumField(choices=Action.choices, default=Action.PASS)
 
     class Meta:
-        verbose_name = "Network event"
-        ordering = ["-id"]
-        db_table = "event"
-        engine = models.ReplacingMergeTree(
-            order_by=["id"],
-            partition_by=Func("timestamp", function="toYYYYMMDD"),
+        ordering = ["-timestamp"]
+        engine = models.MergeTree(
+            primary_key="timestamp",
+            order_by=("timestamp", "id"),
+            partition_by=models.toYYYYMMDD("timestamp"),
             index_granularity=1024,
             index_granularity_bytes=1 << 20,
             enable_mixed_granularity_parts=1,

clickhouse_backend/__init__.py

@@ -1,3 +1,5 @@
-VERSION = (1, 1, 2)
+from clickhouse_backend.utils.version import get_version
 
-__version__ = ".".join(map(str, VERSION))
+VERSION = (1, 1, 3, "alpha", 0)
+
+__version__ = get_version(VERSION)

clickhouse_backend/backend/operations.py

@@ -5,7 +5,7 @@
 from clickhouse_backend import compat
 from clickhouse_backend.driver import JSON
 from clickhouse_backend.driver.client import insert_pattern
-from clickhouse_backend.utils import get_timezone
+from clickhouse_backend.utils.timezone import get_timezone
 
 
 class DatabaseOperations(BaseDatabaseOperations):

clickhouse_backend/models/engines.py

@@ -1,5 +1,3 @@
-from itertools import zip_longest
-
 from django.db.models import Func, Value
 from django.utils.itercompat import is_iterable
 
@@ -133,31 +131,31 @@ def __init__(
     ):
         assert (
             order_by is not None or primary_key is not None
-        ), "order_by or primary_key is missing."
+        ), "At least one of order_by or primary_key must be provided"
         self.order_by = order_by
         self.primary_key = primary_key
         self.partition_by = partition_by
 
-        if order_by is not None:
-            if isinstance(order_by, str) or not is_iterable(order_by):
-                self.order_by = [order_by]
-            if any(o is None for o in self.order_by):
-                raise ValueError("None is not allowed in order_by")
-
-        if primary_key is not None:
-            if isinstance(primary_key, str) or not is_iterable(primary_key):
-                self.primary_key = [primary_key]
-            if any(o is None for o in self.primary_key):
-                raise ValueError("None is not allowed in primary_key")
+        for key in ["order_by", "primary_key", "partition_by"]:
+            value = getattr(self, key)
+            if value is not None:
+                if isinstance(value, str) or not is_iterable(value):
+                    value = (value,)
+                    setattr(self, key, value)
+                elif not isinstance(value, tuple):
+                    value = tuple(value)
+                    setattr(self, key, value)
+                if any(i is None for i in value):
+                    raise ValueError(f"None is not allowed in {key}")
 
         # https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#choosing-a-primary-key-that-differs-from-the-sorting-key
         # primary key expression tuple must be a prefix of the sorting key expression tuple.
-        if self.order_by is not None and self.primary_key is not None:
-            for o, p in zip_longest(self.order_by, self.primary_key):
-                if p is None:
-                    break
-                if p != o:
-                    raise ValueError("primary_key must be a prefix of order_by")
+        if (
+            self.order_by is not None
+            and self.primary_key is not None
+            and self.order_by[: len(self.primary_key)] != self.primary_key
+        ):
+            raise ValueError("primary_key must be a prefix of order_by")
 
         super().__init__(*expressions, **settings)
 

clickhouse_backend/utils/__init__.py

@@ -7,10 +7,10 @@ def get_timezone():
     """Guess what timezone the user is in.
 
     Just remember, no matter USE_TZ or not,
-    clickhouse always store UNIX timestamp (which is in UTC timezone).
+    ClickHouse always store UNIX timestamp (which is in UTC timezone).
 
     Results of lookups turning datetime to date are relevant to timezone.
-    We should always pass timezone parameter to clickhouse functions like
+    We should always pass timezone parameter to ClickHouse functions like
     toDate, toStartOfMonth.
     """
     if settings.USE_TZ: