tortoise / tortoise-orm / 34444158173

85%

master: 94%

push

github

fix(queryset): make distinct().count() ignore join-duplicated rows (#2255)

* fix(queryset): make distinct().count() ignore join-duplicated rows

QuerySet.count() always emitted COUNT(*), which counts every row the
query produces. When the queryset is distinct() and the filter spans a
m2m (or any other row-multiplying join), the returned objects are
de-duplicated by the SELECT DISTINCT but the count was not, so count()
reported more rows than the queryset actually yields.

CountQuery now receives the _distinct flag and counts distinct primary
keys instead. Grouped queries keep the existing windowed COUNT(*).

Closes #2178

* Address review feedback from @waketzheng: move changelog entry to 1.1.9 and reference #2255

* Fix lint issue and improve type hints

---------

Co-authored-by: Sanjay Santhanam <Sanjays2402@users.noreply.github.com>
Co-authored-by: Waket Zheng <waketzheng@gmail.com>

3193 of 4240 branches covered (75.31%)

Branch coverage included in aggregate %.

20 of 22 new or added lines in 1 file covered. (90.91%)

11607 of 13280 relevant lines covered (87.4%)

0.87 hits per line