diff options
| -rw-r--r-- | numpy/core/setup.py | 2 | ||||
| -rw-r--r-- | numpy/core/src/common/npy_argparse.c | 421 | ||||
| -rw-r--r-- | numpy/core/src/common/npy_argparse.h | 96 |
3 files changed, 519 insertions, 0 deletions
diff --git a/numpy/core/setup.py b/numpy/core/setup.py index 822f9f580..514ec1f23 100644 --- a/numpy/core/setup.py +++ b/numpy/core/setup.py @@ -731,6 +731,7 @@ def configuration(parent_package='',top_path=None): join('src', 'common', 'cblasfuncs.h'), join('src', 'common', 'lowlevel_strided_loops.h'), join('src', 'common', 'mem_overlap.h'), + join('src', 'common', 'npy_argparse.h'), join('src', 'common', 'npy_cblas.h'), join('src', 'common', 'npy_config.h'), join('src', 'common', 'npy_ctypes.h'), @@ -749,6 +750,7 @@ def configuration(parent_package='',top_path=None): common_src = [ join('src', 'common', 'array_assign.c'), join('src', 'common', 'mem_overlap.c'), + join('src', 'common', 'npy_argparse.c'), join('src', 'common', 'npy_longdouble.c'), join('src', 'common', 'templ_common.h.src'), join('src', 'common', 'ucsnarrow.c'), diff --git a/numpy/core/src/common/npy_argparse.c b/numpy/core/src/common/npy_argparse.c new file mode 100644 index 000000000..3df780990 --- /dev/null +++ b/numpy/core/src/common/npy_argparse.c @@ -0,0 +1,421 @@ +#include "Python.h" + +#define NPY_NO_DEPRECATED_API NPY_API_VERSION +#define _MULTIARRAYMODULE + +#include "numpy/ndarraytypes.h" +#include "npy_argparse.h" +#include "npy_pycompat.h" +#include "npy_import.h" + +#include "arrayfunction_override.h" + + +/** + * Small wrapper converting to array just like CPython does. + * + * We could use our own PyArray_PyIntAsInt function, but it handles floats + * differently. + * A disadvantage of this function compared to ``PyArg_*("i")`` code is that + * it will not say which parameter is wrong. + * + * @param obj The python object to convert + * @param value The output value + * + * @returns 0 on failure and 1 on success (`NPY_FAIL`, `NPY_SUCCEED`) + */ +NPY_NO_EXPORT int +PyArray_PythonPyIntFromInt(PyObject *obj, int *value) +{ + /* Pythons behaviour is to check only for float explicitly... */ + if (NPY_UNLIKELY(PyFloat_Check(obj))) { + PyErr_SetString(PyExc_TypeError, + "integer argument expected, got float"); + return NPY_FAIL; + } + + long result = PyLong_AsLong(obj); + if (NPY_UNLIKELY((result == -1) && PyErr_Occurred())) { + return NPY_FAIL; + } + if (NPY_UNLIKELY((result > INT_MAX) || (result < INT_MIN))) { + PyErr_SetString(PyExc_OverflowError, + "Python int too large to convert to C int"); + return NPY_FAIL; + } + else { + *value = (int)result; + return NPY_SUCCEED; + } +} + + +typedef int convert(PyObject *, void *); + +/** + * Internal function to initialize keyword argument parsing. + * + * This does a few simple jobs: + * + * * Check the input for consistency to find coding errors, for example + * a parameter not marked with | after one marked with | (optional). + * 2. Find the number of positional-only arguments, the number of + * total, required, and keyword arguments. + * 3. Intern all keyword arguments strings to allow fast, identity based + * parsing and avoid string creation overhead on each call. + * + * @param funcname Name of the function, mainly used for errors. + * @param cache A cache object stored statically in the parsing function + * @param va_orig Argument list to npy_parse_arguments + * @return 0 on success, -1 on failure + */ +static int +initialize_keywords(const char *funcname, + _NpyArgParserCache *cache, va_list va_orig) { + va_list va; + int nargs = 0; + int nkwargs = 0; + int npositional_only = 0; + int nrequired = 0; + int npositional = 0; + char state = '\0'; + + va_copy(va, va_orig); + while (1) { + /* Count length first: */ + char *name = va_arg(va, char *); + convert *converter = va_arg(va, convert *); + void *data = va_arg(va, void *); + + /* Check if this is the sentinel, only converter may be NULL */ + if ((name == NULL) && (converter == NULL) && (data == NULL)) { + break; + } + + if (name == NULL) { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: name is NULL in %s() at " + "argument %d.", funcname, nargs); + va_end(va); + return -1; + } + if (data == NULL) { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: data is NULL in %s() at " + "argument %d.", funcname, nargs); + va_end(va); + return -1; + } + + nargs += 1; + if (*name == '|') { + if (state == '$') { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: positional argument `|` " + "after keyword only `$` one to %s() at argument %d.", + funcname, nargs); + va_end(va); + return -1; + } + state = '|'; + name++; /* advance to actual name. */ + npositional += 1; + } + else if (*name == '$') { + state = '$'; + name++; /* advance to actual name. */ + } + else { + if (state != '\0') { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: non-required argument after " + "required | or $ one to %s() at argument %d.", + funcname, nargs); + va_end(va); + return -1; + } + + nrequired += 1; + npositional += 1; + } + + if (*name == '\0') { + /* Empty string signals positional only */ + if (state != '\0') { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: non-kwarg marked with | or $ " + "to %s() at argument %d.", funcname, nargs); + va_end(va); + return -1; + } + npositional_only += 1; + } + else { + nkwargs += 1; + } + } + va_end(va); + + if (npositional == -1) { + npositional = nargs; + } + + if (nargs > _NPY_MAX_KWARGS) { + PyErr_Format(PyExc_SystemError, + "NumPy internal error: function %s() has %d arguments, but " + "the maximum is currently limited to %d for easier parsing; " + "it can be increased by modifying `_NPY_MAX_KWARGS`.", + funcname, nargs, _NPY_MAX_KWARGS); + return -1; + } + + /* + * Do any necessary string allocation and interning, + * creating a caching object. + */ + cache->nargs = nargs; + cache->npositional_only = npositional_only; + cache->npositional = npositional; + cache->nrequired = nrequired; + + /* NULL kw_strings for easier cleanup (and NULL termination) */ + memset(cache->kw_strings, 0, sizeof(PyObject *) * (nkwargs + 1)); + + va_copy(va, va_orig); + for (int i = 0; i < nargs; i++) { + /* Advance through non-kwargs, which do not require setup. */ + char *name = va_arg(va, char *); + va_arg(va, convert *); + va_arg(va, void *); + + if (*name == '|' || *name == '$') { + name++; /* ignore | and $ */ + } + if (i >= npositional_only) { + int i_kwarg = i - npositional_only; + cache->kw_strings[i_kwarg] = PyUString_InternFromString(name); + if (cache->kw_strings[i_kwarg] == NULL) { + va_end(va); + goto error; + } + } + } + + va_end(va); + return 0; + +error: + for (int i = 0; i < nkwargs; i++) { + Py_XDECREF(cache->kw_strings[i]); + } + cache->npositional = -1; /* not initialized */ + return -1; +} + + +static int +raise_incorrect_number_of_positional_args(const char *funcname, + const _NpyArgParserCache *cache, Py_ssize_t len_args) +{ + if (cache->npositional == cache->nrequired) { + PyErr_Format(PyExc_TypeError, + "%s() takes %d positional arguments but %zd were given", + funcname, cache->npositional, len_args); + } + else { + PyErr_Format(PyExc_TypeError, + "%s() takes from %d to %d positional arguments but " + "%zd were given", + funcname, cache->nrequired, cache->npositional, len_args); + } + return -1; +} + +static void +raise_missing_argument(const char *funcname, + const _NpyArgParserCache *cache, int i) +{ + if (i < cache->npositional_only) { + PyErr_Format(PyExc_TypeError, + "%s() missing required positional argument %d", + funcname, i); + } + else { + PyObject *kw = cache->kw_strings[i - cache->npositional_only]; + PyErr_Format(PyExc_TypeError, + "%s() missing required argument '%S' (pos %d)", + funcname, kw, i); + } +} + + +/** + * Generic helper for argument parsing + * + * See macro version for an example pattern of how to use this function. + * + * @param funcname + * @param cache + * @param args Python passed args (METH_FASTCALL) + * @param len_args + * @param kwnames + * @param ... List of arguments (see macro version). + * + * @return Returns 0 on success and -1 on failure. + */ +NPY_NO_EXPORT int +_npy_parse_arguments(const char *funcname, + /* cache_ptr is a NULL initialized persistent storage for data */ + _NpyArgParserCache *cache, + PyObject *const *args, Py_ssize_t len_args, PyObject *kwnames, + /* ... is NULL, NULL, NULL terminated: name, converter, value */ + ...) +{ + if (NPY_UNLIKELY(cache->npositional == -1)) { + va_list va; + va_start(va, kwnames); + + int res = initialize_keywords(funcname, cache, va); + va_end(va); + if (res < 0) { + return -1; + } + } + + if (NPY_UNLIKELY(len_args > cache->npositional)) { + return raise_incorrect_number_of_positional_args( + funcname, cache, len_args); + } + + /* NOTE: Could remove the limit but too many kwargs are slow anyway. */ + PyObject *all_arguments[NPY_MAXARGS]; + + for (Py_ssize_t i = 0; i < len_args; i++) { + all_arguments[i] = args[i]; + } + + /* Without kwargs, do not iterate all converters. */ + int max_nargs = (int)len_args; + Py_ssize_t len_kwargs = 0; + + /* If there are any kwargs, first handle them */ + if (NPY_LIKELY(kwnames != NULL)) { + len_kwargs = PyTuple_GET_SIZE(kwnames); + max_nargs = cache->nargs; + + for (int i = len_args; i < cache->nargs; i++) { + all_arguments[i] = NULL; + } + + for (Py_ssize_t i = 0; i < len_kwargs; i++) { + PyObject *key = PyTuple_GET_ITEM(kwnames, i); + PyObject *value = args[i + len_args]; + PyObject *const *name; + + /* Super-fast path, check identity: */ + for (name = cache->kw_strings; *name != NULL; name++) { + if (*name == key) { + break; + } + } + if (NPY_UNLIKELY(*name == NULL)) { + /* Slow fallback, if identity checks failed for some reason */ + for (name = cache->kw_strings; *name != NULL; name++) { + int eq = PyObject_RichCompareBool(*name, key, Py_EQ); + if (eq == -1) { + return -1; + } + else if (eq) { + break; + } + } + if (NPY_UNLIKELY(*name == NULL)) { + /* Invalid keyword argument. */ + PyErr_Format(PyExc_TypeError, + "%s() got an unexpected keyword argument '%S'", + funcname, key); + return -1; + } + } + + ssize_t param_pos = ( + (name - cache->kw_strings) + cache->npositional_only); + + /* There could be an identical positional argument */ + if (NPY_UNLIKELY(all_arguments[param_pos] != NULL)) { + PyErr_Format(PyExc_TypeError, + "argument for %s() given by name ('%S') and position " + "(position %zd)", funcname, key, param_pos); + return -1; + } + + all_arguments[param_pos] = value; + } + } + + /* + * There cannot be too many args, too many kwargs would find an + * incorrect one above. + */ + assert(len_args + len_kwargs <= cache->nargs); + + /* At this time `all_arguments` holds either NULLs or the objects */ + va_list va; + va_start(va, kwnames); + + for (int i = 0; i < max_nargs; i++) { + va_arg(va, char *); + convert *converter = va_arg(va, convert *); + void *data = va_arg(va, void *); + + if (all_arguments[i] == NULL) { + continue; + } + + int res; + if (converter == NULL) { + *((PyObject **) data) = all_arguments[i]; + continue; + } + res = converter(all_arguments[i], data); + + if (NPY_UNLIKELY(res == NPY_SUCCEED)) { + continue; + } + else if (NPY_UNLIKELY(res == NPY_FAIL)) { + /* It is usually the users responsibility to clean up. */ + goto converting_failed; + } + else if (NPY_UNLIKELY(res == Py_CLEANUP_SUPPORTED)) { + /* TODO: Implementing cleanup if/when needed should not be hard */ + PyErr_Format(PyExc_SystemError, + "converter cleanup of parameter %d to %s() not supported.", + i, funcname); + goto converting_failed; + } + assert(0); + } + + /* Required arguments are typically not passed as keyword arguments */ + if (NPY_UNLIKELY(len_args < cache->nrequired)) { + /* (PyArg_* also does this after the actual parsing is finished) */ + if (NPY_UNLIKELY(max_nargs < cache->nrequired)) { + raise_missing_argument(funcname, cache, max_nargs); + goto converting_failed; + } + for (int i = 0; i < cache->nrequired; i++) { + if (NPY_UNLIKELY(all_arguments[i] == NULL)) { + raise_missing_argument(funcname, cache, i); + goto converting_failed; + } + } + } + + va_end(va); + return 0; + +converting_failed: + va_end(va); + return -1; + +} diff --git a/numpy/core/src/common/npy_argparse.h b/numpy/core/src/common/npy_argparse.h new file mode 100644 index 000000000..5da535c91 --- /dev/null +++ b/numpy/core/src/common/npy_argparse.h @@ -0,0 +1,96 @@ +#ifndef _NPY_ARGPARSE_H +#define _NPY_ARGPARSE_H + +#include "Python.h" +#include "numpy/ndarraytypes.h" + +/* + * This file defines macros to help with keyword argument parsing. + * This solves two issues as of now: + * 1. Pythons C-API PyArg_* keyword argument parsers are slow, due to + * not caching the strings they use. + * 2. It allows the use of METH_ARGPARSE (and `tp_vectorcall`) + * when available in Python, which removes a large chunk of overhead. + * + * Internally CPython achieves similar things by using a code generator + * argument clinic. NumPy may well decide to use argument clinic or a different + * solution in the future. + */ + +NPY_NO_EXPORT int +PyArray_PythonPyIntFromInt(PyObject *obj, int *value); + + +#define _NPY_MAX_KWARGS 15 + +typedef struct { + int npositional; + int nargs; + int npositional_only; + int nrequired; + /* Null terminated list of keyword argument name strings */ + PyObject *kw_strings[_NPY_MAX_KWARGS+1]; +} _NpyArgParserCache; + + +/* + * The sole purpose of this macro is to hide the argument parsing cache. + * Since this cache must be static, this also removes a source of error. + */ +#define NPY_PREPARE_ARGPARSER static _NpyArgParserCache __argparse_cache = {-1} + +/** + * Macro to help with argument parsing. + * + * The pattern for using this macro is by defining the method as: + * + * @code + * static PyObject * + * my_method(PyObject *self, + * PyObject *const *args, Py_ssize_t len_args, PyObject *kwnames) + * { + * NPY_PREPARE_ARGPARSER; + * + * PyObject *argument1, *argument3; + * int argument2 = -1; + * if (npy_parse_arguments("method", args, len_args, kwnames), + * "argument1", NULL, &argument1, + * "|argument2", &PyArray_PythonPyIntFromInt, &argument2, + * "$argument3", NULL, &argument3, + * NULL, NULL, NULL) < 0) { + * return NULL; + * } + * } + * @endcode + * + * The `NPY_PREPARE_ARGPARSER` macro sets up a static cache variable necessary + * to hold data for speeding up the parsing. `npy_parse_arguments` must be + * used in cunjunction with the macro defined in the same scope. + * (No two `npy_parse_arguments` may share a single `NPY_PREPARE_ARGPARSER`.) + * + * @param funcname + * @param args Python passed args (METH_FASTCALL) + * @param len_args Number of arguments (not flagged) + * @param kwnames Tuple as passed by METH_FASTCALL or NULL. + * @param ... List of arguments must be param1_name, param1_converter, + * *param1_outvalue, param2_name, ..., NULL, NULL, NULL. + * Where name is ``char *``, ``converter`` a python converter + * function or NULL and ``outvalue`` is the ``void *`` passed to + * the converter (holding the converted data or a borrowed + * reference if converter is NULL). + * + * @return Returns 0 on success and -1 on failure. + */ +NPY_NO_EXPORT int +_npy_parse_arguments(const char *funcname, + /* cache_ptr is a NULL initialized persistent storage for data */ + _NpyArgParserCache *cache_ptr, + PyObject *const *args, Py_ssize_t len_args, PyObject *kwnames, + /* va_list is NULL, NULL, NULL terminated: name, converter, value */ + ...) NPY_GCC_NONNULL(1); + +#define npy_parse_arguments(funcname, args, len_args, kwnames, ...) \ + _npy_parse_arguments(funcname, &__argparse_cache, \ + args, len_args, kwnames, __VA_ARGS__) + +#endif /* _NPY_ARGPARSE_H */ |