Merge branch 'main' into 50040-add-math-mode-formatter-escape=latex-part2

Author: natmokval

.github/workflows/ubuntu.yml

@@ -142,7 +142,7 @@ jobs:
           - 5432:5432
 
       moto:
-        image: motoserver/moto
+        image: motoserver/moto:4.1.4
         env:
           AWS_ACCESS_KEY_ID: foobar_key
           AWS_SECRET_ACCESS_KEY: foobar_secret

MANIFEST.in

@@ -58,3 +58,5 @@ prune pandas/tests/io/parser/data
 # Selectively re-add *.cxx files that were excluded above
 graft pandas/_libs/src
 graft pandas/_libs/tslibs/src
+include pandas/_libs/pd_parser.h
+include pandas/_libs/pd_parser.c

doc/source/_static/css/getting_started.css

@@ -236,11 +236,11 @@ ul.task-bullet > li > p:first-child {
 
 .tutorial-card .card-header {
   cursor: pointer;
-  background-color: white;
+  background-color: transparent;
 }
 
 .tutorial-card .card-body {
-  background-color: #F0F0F0;
+  background-color: transparent;
 }
 
 .tutorial-card .badge {

doc/source/ecosystem.rst

@@ -377,8 +377,8 @@ aggregations for step functions defined over real numbers, datetime and timedelt
 
 xarray brings the labeled data power of pandas to the physical sciences by
 providing N-dimensional variants of the core pandas data structures. It aims to
-provide a pandas-like and pandas-compatible toolkit for analytics on multi-
-dimensional arrays, rather than the tabular data for which pandas excels.
+provide a pandas-like and pandas-compatible toolkit for analytics on
+multi-dimensional arrays, rather than the tabular data for which pandas excels.
 
 
 .. _ecosystem.io:

doc/source/user_guide/integer_na.rst

@@ -58,8 +58,8 @@ with the dtype.
 .. warning::
 
    Currently :meth:`pandas.array` and :meth:`pandas.Series` use different
-   rules for dtype inference. :meth:`pandas.array` will infer a nullable-
-   integer dtype
+   rules for dtype inference. :meth:`pandas.array` will infer a
+   nullable-integer dtype
 
    .. ipython:: python
 

doc/source/whatsnew/v2.1.0.rst

@@ -86,7 +86,7 @@ See :ref:`install.dependencies` and :ref:`install.optional_dependencies` for mor
 
 Other API changes
 ^^^^^^^^^^^^^^^^^
--
+- :class:`~arrays.ArrowExtensionArray` comparison methods now return data with :class:`ArrowDtype` with ``pyarrow.bool_`` type instead of ``"boolean"`` dtype (:issue:`51643`)
 -
 
 .. ---------------------------------------------------------------------------
@@ -115,6 +115,7 @@ Performance improvements
 - Performance improvement in :meth:`~arrays.ArrowExtensionArray.isna` when array has zero nulls or is all nulls (:issue:`51630`)
 - Performance improvement when parsing strings to ``boolean[pyarrow]`` dtype (:issue:`51730`)
 - Performance improvement when searching an :class:`Index` sliced from other indexes (:issue:`51738`)
+- Performance improvement in :meth:`Series.combine_first` (:issue:`51777`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_210.bug_fixes:
@@ -202,7 +203,9 @@ Groupby/resample/rolling
 
 Reshaping
 ^^^^^^^^^
+- Bug in :meth:`DataFrame.stack` losing extension dtypes when columns is a :class:`MultiIndex` and frame contains mixed dtypes (:issue:`45740`)
 - Bug in :meth:`DataFrame.transpose` inferring dtype for object column (:issue:`51546`)
+- Bug in :meth:`Series.combine_first` converting ``int64`` dtype to ``float64`` and losing precision on very large integers (:issue:`51764`)
 -
 
 Sparse

pandas/_libs/__init__.py

@@ -10,6 +10,11 @@
 ]
 
 
+# Below imports needs to happen first to ensure pandas top level
+# module gets monkeypatched with the pandas_datetime_CAPI
+# see pandas_datetime_exec in pd_datetime.c
+import pandas._libs.pandas_parser  # noqa # isort: skip # type: ignore[reportUnusedImport]
+import pandas._libs.pandas_datetime  # noqa # isort: skip # type: ignore[reportUnusedImport]
 from pandas._libs.interval import Interval
 from pandas._libs.tslibs import (
     NaT,

pandas/_libs/index.pyx

@@ -20,7 +20,12 @@ from pandas._libs.tslibs.nattype cimport c_NaT as NaT
 from pandas._libs.tslibs.np_datetime cimport (
     NPY_DATETIMEUNIT,
     get_unit_from_dtype,
+    import_pandas_datetime,
 )
+
+import_pandas_datetime()
+
+
 from pandas._libs.tslibs.period cimport is_period_object
 from pandas._libs.tslibs.timedeltas cimport _Timedelta
 from pandas._libs.tslibs.timestamps cimport _Timestamp

pandas/_libs/lib.pyx

@@ -88,9 +88,11 @@ cdef extern from "numpy/arrayobject.h":
 cdef extern from "numpy/ndarrayobject.h":
     bint PyArray_CheckScalar(obj) nogil
 
-
-cdef extern from "src/parse_helper.h":
+cdef extern from "pd_parser.h":
     int floatify(object, float64_t *result, int *maybe_int) except -1
+    void PandasParser_IMPORT()
+
+PandasParser_IMPORT
 
 from pandas._libs cimport util
 from pandas._libs.util cimport (

pandas/_libs/missing.pyx

@@ -34,8 +34,11 @@ from pandas._libs.tslibs.np_datetime cimport (
     get_datetime64_unit,
     get_datetime64_value,
     get_timedelta64_value,
+    import_pandas_datetime,
 )
 
+import_pandas_datetime()
+
 from pandas._libs.ops_dispatch import maybe_dispatch_ufunc_to_dunder_op
 
 cdef:

pandas/_libs/parsers.pyx

@@ -229,9 +229,9 @@ cdef extern from "parser/tokenizer.h":
         int64_t skip_first_N_rows
         int64_t skipfooter
         # pick one, depending on whether the converter requires GIL
-        float64_t (*double_converter)(const char *, char **,
-                                      char, char, char,
-                                      int, int *, int *) nogil
+        double (*double_converter)(const char *, char **,
+                                   char, char, char,
+                                   int, int *, int *) nogil
 
         #  error handling
         char *warn_msg
@@ -249,6 +249,16 @@ cdef extern from "parser/tokenizer.h":
         int seen_uint
         int seen_null
 
+    void COLITER_NEXT(coliter_t, const char *) nogil
+
+cdef extern from "pd_parser.h":
+    void *new_rd_source(object obj) except NULL
+
+    int del_rd_source(void *src)
+
+    void* buffer_rd_bytes(void *source, size_t nbytes,
+                          size_t *bytes_read, int *status, const char *encoding_errors)
+
     void uint_state_init(uint_state *self)
     int uint64_conflict(uint_state *self)
 
@@ -279,26 +289,49 @@ cdef extern from "parser/tokenizer.h":
     uint64_t str_to_uint64(uint_state *state, char *p_item, int64_t int_max,
                            uint64_t uint_max, int *error, char tsep) nogil
 
-    float64_t xstrtod(const char *p, char **q, char decimal,
+    double xstrtod(const char *p, char **q, char decimal,
+                   char sci, char tsep, int skip_trailing,
+                   int *error, int *maybe_int) nogil
+    double precise_xstrtod(const char *p, char **q, char decimal,
+                           char sci, char tsep, int skip_trailing,
+                           int *error, int *maybe_int) nogil
+    double round_trip(const char *p, char **q, char decimal,
                       char sci, char tsep, int skip_trailing,
                       int *error, int *maybe_int) nogil
-    float64_t precise_xstrtod(const char *p, char **q, char decimal,
-                              char sci, char tsep, int skip_trailing,
-                              int *error, int *maybe_int) nogil
-    float64_t round_trip(const char *p, char **q, char decimal,
-                         char sci, char tsep, int skip_trailing,
-                         int *error, int *maybe_int) nogil
 
     int to_boolean(const char *item, uint8_t *val) nogil
 
+    void PandasParser_IMPORT()
 
-cdef extern from "parser/io.h":
-    void *new_rd_source(object obj) except NULL
+PandasParser_IMPORT
 
-    int del_rd_source(void *src)
+# When not invoked directly but rather assigned as a function,
+# cdef extern'ed declarations seem to leave behind an undefined symbol
+cdef double xstrtod_wrapper(const char *p, char **q, char decimal,
+                            char sci, char tsep, int skip_trailing,
+                            int *error, int *maybe_int) nogil:
+    return xstrtod(p, q, decimal, sci, tsep, skip_trailing, error, maybe_int)
 
-    void* buffer_rd_bytes(void *source, size_t nbytes,
-                          size_t *bytes_read, int *status, const char *encoding_errors)
+
+cdef double precise_xstrtod_wrapper(const char *p, char **q, char decimal,
+                                    char sci, char tsep, int skip_trailing,
+                                    int *error, int *maybe_int) nogil:
+    return precise_xstrtod(p, q, decimal, sci, tsep, skip_trailing, error, maybe_int)
+
+
+cdef double round_trip_wrapper(const char *p, char **q, char decimal,
+                               char sci, char tsep, int skip_trailing,
+                               int *error, int *maybe_int) nogil:
+    return round_trip(p, q, decimal, sci, tsep, skip_trailing, error, maybe_int)
+
+
+cdef void* buffer_rd_bytes_wrapper(void *source, size_t nbytes,
+                                   size_t *bytes_read, int *status,
+                                   const char *encoding_errors) noexcept:
+    return buffer_rd_bytes(source, nbytes, bytes_read, status, encoding_errors)
+
+cdef int del_rd_source_wrapper(void *src) noexcept:
+    return del_rd_source(src)
 
 
 cdef class TextReader:
@@ -487,11 +520,11 @@ cdef class TextReader:
 
         if float_precision == "round_trip":
             # see gh-15140
-            self.parser.double_converter = round_trip
+            self.parser.double_converter = round_trip_wrapper
         elif float_precision == "legacy":
-            self.parser.double_converter = xstrtod
+            self.parser.double_converter = xstrtod_wrapper
         elif float_precision == "high" or float_precision is None:
-            self.parser.double_converter = precise_xstrtod
+            self.parser.double_converter = precise_xstrtod_wrapper
         else:
             raise ValueError(f"Unrecognized float_precision option: "
                              f"{float_precision}")
@@ -610,8 +643,8 @@ cdef class TextReader:
 
         ptr = new_rd_source(source)
         self.parser.source = ptr
-        self.parser.cb_io = &buffer_rd_bytes
-        self.parser.cb_cleanup = &del_rd_source
+        self.parser.cb_io = buffer_rd_bytes_wrapper
+        self.parser.cb_cleanup = del_rd_source_wrapper
 
     cdef _get_header(self, list prelim_header):
         # header is now a list of lists, so field_count should use header[0]