querysource 4.5.14__tar.gz → 4.5.15__tar.gz

{querysource-4.5.14 → querysource-4.5.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: querysource
-Version: 4.5.14
+Version: 4.5.15
 Summary: Aiohttp web service for querying several databases easily
 Author-email: Jesus Lara <jesuslarag@gmail.com>
 License: BSD-3-Clause
@@ -37,7 +37,6 @@ Requires-Dist: oauth2client>=4.1.3
 Requires-Dist: google-analytics-data>=0.18.15
 Requires-Dist: google-api-python-client>=2.151.0
 Requires-Dist: google-auth-oauthlib==1.2.1
-Requires-Dist: sqloxide==0.1.39
 Requires-Dist: aiocsv>=1.3.2
 Requires-Dist: lxml>=5.3.0
 Requires-Dist: xlsxwriter==3.2.0

{querysource-4.5.14 → querysource-4.5.15}/pyproject.toml

@@ -50,7 +50,6 @@ dependencies = [
     "google-analytics-data>=0.18.15",
     "google-api-python-client>=2.151.0", # Aligned with flowtask
     "google-auth-oauthlib==1.2.1",
-    "sqloxide==0.1.39",
     "aiocsv>=1.3.2",
     "lxml>=5.3.0",
     "xlsxwriter==3.2.0",

{querysource-4.5.14 → querysource-4.5.15}/querysource/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '4.5.14'
-__version_tuple__ = version_tuple = (4, 5, 14)
+__version__ = version = '4.5.15'
+__version_tuple__ = version_tuple = (4, 5, 15)
 
-__commit_id__ = commit_id = 'g5872d5ef4'
+__commit_id__ = commit_id = 'g33895edcd'

{querysource-4.5.14 → querysource-4.5.15}/querysource/types/dt/filters.py

@@ -299,12 +299,17 @@ def drop_duplicates(df: pd.DataFrame, columns: Optional[list] = None, **kwargs):
     :return: The DataFrame with duplicates dropped.
     """
     if columns and isinstance(columns, list):
-        # NOTE: do not ``set_index(columns, drop=False)`` here — it leaves the
-        # same names as both an index level and a column, which makes the
-        # subsequent ``sort_values(by=columns)`` raise "is both an index level
-        # and a column label, which is ambiguous". Sorting on the columns
-        # directly keeps ``keep='first'`` deterministic without that clash.
-        df = df.sort_values(by=columns).drop_duplicates(subset=columns, **kwargs)
+        # NOTE: do NOT sort before dropping. ``sort_values(by=columns)`` is
+        # unnecessary for ``keep='first'`` (drop_duplicates already keeps the
+        # first occurrence in the DataFrame's existing order) and it is fragile:
+        # with a multi-column subset, if any column holds mixed/unorderable
+        # values (e.g. an object column mixing str and NaN, or ints and str),
+        # ``sort_values`` raises ``TypeError: '<' not supported between
+        # instances of ...`` — which the Filter operator swallows, so the rows
+        # are never de-duplicated and the downstream upsert then fails with a
+        # CardinalityViolation. Dropping directly on the subset is order-stable
+        # and type-agnostic.
+        df = df.drop_duplicates(subset=columns, **kwargs)
     return df
 
 

{querysource-4.5.14 → querysource-4.5.15}/querysource/version.py

@@ -6,7 +6,7 @@ __description__ = (
     'QuerySource Query parser and generator.'
     'Supports SQL, NoSQL, MongoDB, Cassandra, Elasticsearch, etc.'
 )
-__version__ = '4.5.14'
+__version__ = '4.5.15'
 __url__ = 'https://github.com/phenobarbital/querysource'
 __copyright__ = 'Copyright (c) 2018-2025 Jesus Lara'
 __author__ = 'Jesus Lara'

{querysource-4.5.14 → querysource-4.5.15}/querysource.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: querysource
-Version: 4.5.14
+Version: 4.5.15
 Summary: Aiohttp web service for querying several databases easily
 Author-email: Jesus Lara <jesuslarag@gmail.com>
 License: BSD-3-Clause
@@ -37,7 +37,6 @@ Requires-Dist: oauth2client>=4.1.3
 Requires-Dist: google-analytics-data>=0.18.15
 Requires-Dist: google-api-python-client>=2.151.0
 Requires-Dist: google-auth-oauthlib==1.2.1
-Requires-Dist: sqloxide==0.1.39
 Requires-Dist: aiocsv>=1.3.2
 Requires-Dist: lxml>=5.3.0
 Requires-Dist: xlsxwriter==3.2.0

{querysource-4.5.14 → querysource-4.5.15}/querysource.egg-info/SOURCES.txt

@@ -542,6 +542,7 @@ tests/test_elastic_parser.py
 tests/test_error_formatter.py
 tests/test_eval.py
 tests/test_feat102_error_redaction.py
+tests/test_filters_drop_duplicates.py
 tests/test_grouping_sync.py
 tests/test_info_eda.py
 tests/test_info_eda_e2e.py

{querysource-4.5.14 → querysource-4.5.15}/querysource.egg-info/requires.txt

@@ -6,7 +6,6 @@ oauth2client>=4.1.3
 google-analytics-data>=0.18.15
 google-api-python-client>=2.151.0
 google-auth-oauthlib==1.2.1
-sqloxide==0.1.39
 aiocsv>=1.3.2
 lxml>=5.3.0
 xlsxwriter==3.2.0

{querysource-4.5.14 → querysource-4.5.15}/querysource.egg-info/scm_file_list.json

@@ -911,6 +911,7 @@
     "tests/test_error_formatter.py",
     "tests/test_eval.py",
     "tests/test_feat102_error_redaction.py",
+    "tests/test_filters_drop_duplicates.py",
     "tests/test_grouping_sync.py",
     "tests/test_info_eda.py",
     "tests/test_info_eda_e2e.py",

querysource-4.5.15/querysource.egg-info/scm_version.json

@@ -0,0 +1,8 @@
+{
+  "tag": "4.5.15",
+  "distance": 0,
+  "node": "g33895edcdbbdcf345220a28cb04c608a24524c37",
+  "dirty": false,
+  "branch": "HEAD",
+  "node_date": "2026-09-14"
+}

querysource-4.5.15/tests/test_filters_drop_duplicates.py

@@ -0,0 +1,55 @@
+"""Regression tests for filters.drop_duplicates (multi-column, order-stable).
+
+Background: drop_duplicates previously did
+``df.sort_values(by=columns).drop_duplicates(subset=columns, ...)``. The
+``sort_values`` is unnecessary for ``keep='first'`` (drop_duplicates already
+keeps the first occurrence in the frame's existing order) and is fragile on a
+multi-column subset when a column holds mixed/unorderable values. The Filter
+operator swallows the resulting error, so the rows are never de-duplicated and
+the downstream upsert fails with a CardinalityViolation.
+
+These tests pin the intended behaviour: de-duplicate on the *combination* of
+the given columns, keep the first occurrence, and preserve the original order.
+"""
+import numpy as np
+import pandas as pd
+
+from querysource.types.dt.filters import drop_duplicates
+
+
+def test_multicolumn_dedups_on_combination():
+    df = pd.DataFrame({
+        "user": ["u1", "u1", "u2", "u2", "u3"],
+        "content_id": [5, 5, 7, 8, 9],  # (u1,5) is the only real duplicate
+    })
+    out = drop_duplicates(df, columns=["user", "content_id"], keep="first")
+    # 5 rows, one duplicate pair (u1,5) collapses -> 4 unique combinations
+    assert len(out) == 4
+    assert out.duplicated(subset=["user", "content_id"]).sum() == 0
+
+
+def test_keep_first_preserves_original_order():
+    df = pd.DataFrame({
+        "user": ["b", "a", "b", "a"],
+        "content_id": [2, 1, 2, 1],  # rows 2 and 3 duplicate rows 0 and 1
+    })
+    out = drop_duplicates(df, columns=["user", "content_id"], keep="first")
+    # first occurrences kept, in their original order (b,2) then (a,1)
+    assert list(zip(out["user"], out["content_id"])) == [("b", 2), ("a", 1)]
+
+
+def test_multicolumn_with_mixed_and_null_values_does_not_raise():
+    # An object column mixing ints, strings and NaN must not blow up the dedup.
+    df = pd.DataFrame({
+        "user": ["u1", "u1", "u2", "u2"],
+        "content_id": [10, 10, np.nan, "abc"],
+    })
+    out = drop_duplicates(df, columns=["user", "content_id"], keep="first")
+    # only (u1,10) repeats -> 3 rows remain
+    assert len(out) == 3
+
+
+def test_single_column_still_works():
+    df = pd.DataFrame({"user": ["u1", "u1", "u2"], "content_id": [1, 2, 3]})
+    out = drop_duplicates(df, columns=["user"], keep="first")
+    assert len(out) == 2  # one row per user

querysource-4.5.14/querysource.egg-info/scm_version.json

@@ -1,8 +0,0 @@
-{
-  "tag": "4.5.14",
-  "distance": 0,
-  "node": "g5872d5ef4fdafc79f9636c954f28e1e2144a7a63",
-  "dirty": false,
-  "branch": "HEAD",
-  "node_date": "2026-09-11"
-}