Add Werror to CI

.circleci/setup_env.sh

@@ -55,6 +55,6 @@ if pip show pandas 1>/dev/null; then
 fi
 
 echo "Install pandas"
-python -m pip install --no-build-isolation -ve .
+python -m pip install --no-build-isolation -ve . --config-settings=setup-args="--werror"
 
 echo "done"

.github/actions/build_pandas/action.yml

@@ -25,8 +25,10 @@ runs:
     - name: Build Pandas
       run: |
         if [[ ${{ inputs.editable }} == "true" ]]; then
-          pip install -e . --no-build-isolation -v --no-deps
+          pip install -e . --no-build-isolation -v --no-deps \
+            --config-settings=setup-args="--werror"
         else
-          pip install . --no-build-isolation -v --no-deps
+          pip install . --no-build-isolation -v --no-deps \
+            --config-settings=setup-args="--werror"
         fi
       shell: bash -el {0}

.github/workflows/unit-tests.yml

@@ -254,7 +254,7 @@ jobs:
           python -m pip install --no-cache-dir -U pip wheel setuptools meson[ninja]==1.2.1 meson-python==0.13.1
           python -m pip install numpy --config-settings=setup-args="-Dallow-noblas=true"
           python -m pip install --no-cache-dir versioneer[toml] cython python-dateutil pytz pytest>=7.3.2 pytest-xdist>=2.2.0 hypothesis>=6.46.1
-          python -m pip install --no-cache-dir --no-build-isolation -e .
+          python -m pip install --no-cache-dir --no-build-isolation -e . --config-settings=setup-args="--werror"
           python -m pip list --no-cache-dir
           export PANDAS_CI=1
           python -m pytest -m 'not slow and not network and not clipboard and not single_cpu' pandas --junitxml=test-data.xml
@@ -292,7 +292,7 @@ jobs:
           . ~/virtualenvs/pandas-dev/bin/activate
           python -m pip install --no-cache-dir -U pip wheel setuptools meson-python==0.13.1 meson[ninja]==1.2.1
           python -m pip install --no-cache-dir versioneer[toml] cython numpy python-dateutil pytz pytest>=7.3.2 pytest-xdist>=2.2.0 hypothesis>=6.46.1
-          python -m pip install --no-cache-dir --no-build-isolation -e .
+          python -m pip install --no-cache-dir --no-build-isolation -e . --config-settings=setup-args="--werror"
           python -m pip list --no-cache-dir
 
       - name: Run Tests
@@ -365,7 +365,7 @@ jobs:
           python -m pip install --pre --extra-index-url https://pypi.anaconda.org/scientific-python-nightly-wheels/simple numpy
           python -m pip install versioneer[toml]
           python -m pip install python-dateutil pytz tzdata cython hypothesis>=6.46.1 pytest>=7.3.2 pytest-xdist>=2.2.0 pytest-cov
-          python -m pip install -ve . --no-build-isolation --no-index --no-deps
+          python -m pip install -ve . --no-build-isolation --no-index --no-deps --config-settings=setup-args="--werror"
           python -m pip list
 
       - name: Run Tests

pandas/_libs/include/pandas/datetime/date_conversions.h

@@ -12,7 +12,7 @@ The full license is in the LICENSE file, distributed with this software.
 #include <numpy/ndarraytypes.h>
 
 // Scales value inplace from nanosecond resolution to unit resolution
-int scaleNanosecToUnit(npy_int64 *value, NPY_DATETIMEUNIT unit);
+int scaleNanosecToUnit(int64_t *value, NPY_DATETIMEUNIT unit);
 
 // Converts an int64 object representing a date to ISO format
 // up to precision `base` e.g. base="s" yields 2020-01-03T00:00:00Z

pandas/_libs/include/pandas/datetime/pd_datetime.h

@@ -33,7 +33,7 @@ extern "C" {
 typedef struct {
   npy_datetime (*npy_datetimestruct_to_datetime)(NPY_DATETIMEUNIT,
                                                  const npy_datetimestruct *);
-  int (*scaleNanosecToUnit)(npy_int64 *, NPY_DATETIMEUNIT);
+  int (*scaleNanosecToUnit)(int64_t *, NPY_DATETIMEUNIT);
   char *(*int64ToIso)(int64_t, NPY_DATETIMEUNIT, NPY_DATETIMEUNIT, size_t *);
   char *(*PyDateTimeToIso)(PyObject *, NPY_DATETIMEUNIT, size_t *);
   npy_datetime (*PyDateTimeToEpoch)(PyObject *, NPY_DATETIMEUNIT);

pandas/_libs/src/datetime/date_conversions.c

@@ -20,7 +20,7 @@ The full license is in the LICENSE file, distributed with this software.
  *
  * Mutates the provided value directly. Returns 0 on success, non-zero on error.
  */
-int scaleNanosecToUnit(npy_int64 *value, NPY_DATETIMEUNIT unit) {
+int scaleNanosecToUnit(int64_t *value, NPY_DATETIMEUNIT unit) {
   switch (unit) {
   case NPY_FR_ns:
     break;

pandas/_libs/src/datetime/pd_datetime.c

@@ -176,7 +176,7 @@ static npy_datetime PyDateTimeToEpoch(PyObject *dt, NPY_DATETIMEUNIT base) {
     }
   }
 
-  npy_datetime npy_dt = npy_datetimestruct_to_datetime(NPY_FR_ns, &dts);
+  int64_t npy_dt = npy_datetimestruct_to_datetime(NPY_FR_ns, &dts);
   if (scaleNanosecToUnit(&npy_dt, base) == -1) {
     PyErr_Format(PyExc_ValueError,
                  "Call to scaleNanosecToUnit with value %" NPY_DATETIME_FMT

pandas/_libs/src/vendored/ujson/python/objToJSON.c

@@ -1236,11 +1236,11 @@ static char **NpyArr_encodeLabels(PyArrayObject *labels, PyObjectEncoder *enc,
     }
 
     int is_datetimelike = 0;
-    npy_int64 i8date;
+    int64_t i8date;
     NPY_DATETIMEUNIT dateUnit = NPY_FR_ns;
     if (PyTypeNum_ISDATETIME(type_num)) {
       is_datetimelike = 1;
-      i8date = *(npy_int64 *)dataptr;
+      i8date = *(int64_t *)dataptr;
       dateUnit = get_datetime_metadata_from_dtype(dtype).base;
     } else if (PyDate_Check(item) || PyDelta_Check(item)) {
       is_datetimelike = 1;
@@ -1250,7 +1250,11 @@ static char **NpyArr_encodeLabels(PyArrayObject *labels, PyObjectEncoder *enc,
         i8date = get_long_attr(item, "_value");
       } else {
         if (PyDelta_Check(item)) {
-          i8date = total_seconds(item) * 1000000000LL; // nanoseconds per second
+          // TODO(anyone): cast below loses precision if total_seconds return
+          // value exceeds number of bits that significand can hold
+          // also liable to overflow
+          i8date = (int64_t)(total_seconds(item) *
+                             1000000000LL); // nanoseconds per second
         } else {
           // datetime.* objects don't follow above rules
           i8date = PyDateTimeToEpoch(item, NPY_FR_ns);
@@ -1291,7 +1295,7 @@ static char **NpyArr_encodeLabels(PyArrayObject *labels, PyObjectEncoder *enc,
             ret = 0;
             break;
           }
-          snprintf(cLabel, size_of_cLabel, "%" NPY_DATETIME_FMT, i8date);
+          snprintf(cLabel, size_of_cLabel, "%" PRId64, i8date);
           len = strlen(cLabel);
         }
       }
@@ -1495,10 +1499,14 @@ static void Object_beginTypeContext(JSOBJ _obj, JSONTypeContext *tc) {
     }
     return;
   } else if (PyDelta_Check(obj)) {
-    npy_int64 value =
-        PyObject_HasAttrString(obj, "_value") ? get_long_attr(obj, "_value")
-                                              : // pd.Timedelta object or pd.NaT
-            total_seconds(obj) * 1000000000LL;  // nanoseconds per sec
+    // pd.Timedelta object or pd.NaT should evaluate true here
+    // fallback to nanoseconds per sec for other objects
+    // TODO(anyone): cast below loses precision if total_seconds return
+    // value exceeds number of bits that significand can hold
+    // also liable to overflow
+    int64_t value = PyObject_HasAttrString(obj, "_value")
+                        ? get_long_attr(obj, "_value")
+                        : (int64_t)(total_seconds(obj) * 1000000000LL);
 
     if (value == get_nat()) {
       tc->type = JT_NULL;

pandas/tests/io/json/test_pandas.py

@@ -1144,6 +1144,18 @@ def test_timedelta_to_json(self, as_object, date_format, timedelta_typ):
         result = ser.to_json(date_format=date_format)
         assert result == expected
 
+    @pytest.mark.parametrize("as_object", [True, False])
+    @pytest.mark.parametrize("timedelta_typ", [pd.Timedelta, timedelta])
+    def test_timedelta_to_json_fractional_precision(self, as_object, timedelta_typ):
+        data = [timedelta_typ(milliseconds=42)]
+        ser = Series(data, index=data)
+        if as_object:
+            ser = ser.astype(object)
+
+        result = ser.to_json()
+        expected = '{"42":42}'
+        assert result == expected
+
     def test_default_handler(self):
         value = object()
         frame = DataFrame({"a": [7, value]})