Merge pull request #639 from Carbaz/feature/python-c-extension-generator

Python C extension generator using OpenAI GPT-5

week4/community-contributions/c_extension_generator/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Carlos Bazaga
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

week4/community-contributions/c_extension_generator/README.md

@@ -0,0 +1,134 @@
+# Python C Extension code generator
+
+Written by Carlos Bazaga [@carbaz] based on the work of Ed Donner [@ed-donner]
+under the MIT License.
+
+This folder contains a Jupyter notebook that demonstrates how to use a Frontier model
+to generate high-performance Python C extension code from Python code.
+
+The notebook includes examples of generating C extensions for calculating Pi using the
+Leibniz formula and finding the maximum sub-array in an array.
+
+Also, it includes a Gradio app that provides an interactive interface for users to input
+Python code, generate C extension code, compile it, and test its performance against
+the original Python code.
+
+> [!CAUTION]
+>
+> **Always review the generated codes before running them, as they will be executed in
+> your local environment and may contain code that could be harmful or unwanted.**
+>
+> AI-generated code may contain errors or unsafe practices, so it's crucial to
+> thoroughly review and test any code before using it in a production environment.
+>
+> Never run code generated by AI models without understanding its implications and
+> ensuring it adheres to your security and safety standards.
+
+> [!IMPORTANT]
+>
+> **Disclaimer:** This notebook and the Gradio app are provided for educational purposes
+> only. Use them at your own risk.
+
+## Gradio app overview
+
+In this image, you can see the Gradio app dashboard whose main sections are
+described below.
+
+![Gradio app dashboard](gradio_dashboard.jpg)\
+*Image: Gradio app dashboard with default example `hello world` code loaded.*
+*(compile output redacted for privacy)*
+
+Sections:
+
+* **Dropdown selectors and input fields**:
+  * **Module name input**:
+    A text input field where users can specify the name of the C extension module to be
+    generated.
+
+    That name will be used to create the C extension file `<module_name>.c` and
+    the `setup.py` file required to compile the extension.
+
+    That name will also be used to import the compiled module as usual in Python:
+
+    ```python
+    import <module_name>
+    ```
+
+    Or
+
+    ```python
+    from <module_name> import <function_name>
+    ```
+
+  * **Model selector**:
+    A dropdown menu to select the Frontier model to use for code generation.
+
+    Currently it includes:
+    * `gpt-4o` (default)
+    * `gpt-5`
+
+* **Text input areas**:
+
+  This areas are all editable, included those filled with generated code by the model.
+  this allows users to modify and experiment with the code as needed.
+
+  * **Python code**:
+    A text area where users can input their Python code.
+  * **C extension code**:
+    A text area that displays the generated C extension code and allows to edit it.
+  * **Compilation code**:
+    A text area that shows the `setup.py` file generated,
+    this file is required to compile the C extension.
+  * **Test compare code**:
+    A text area that provides example code to run the compiled C extension.
+
+* **Output areas**:
+
+  This are non-editable areas that display the results of various operations.
+
+  * **C Extension result**:
+    A text area that displays the output of the C extension code build.
+
+    Beware that this area can contain a large amount of text including warnings during
+    the compilation process and sensible information about the local environment,
+    like: paths, Python version, etc may be included.
+
+    Redact that information if you plan to share the output.
+
+  * **Test result**:
+    A text area that displays the output of the test code run.
+
+* **Buttons**:
+  * **Generate extension code**:
+    A button that triggers the generation of the C extension code from the provided
+    Python code.
+
+    It will call the Frontier model to generate the C code, the setup.py file and
+    the test code, filling the corresponding text areas automatically.
+
+  * **Compile extension**:
+    A button that compiles the generated C extension using the provided `setup.py` file.
+    It will create the extension c file, `<module_name>.c` and the `setup.py` files in
+    the local folder, then it will run the compilation command and build the C extension.
+
+    > [!CAUTION]
+    >
+    > **Always review the `setup.py` code before running it, as it will be executed in
+    > your local environment and may contain code that could be harmful or unwanted.**
+    >
+    > **Also review the generated C code, as it will be compiled and executed in your
+    > local environment and may contain code that could be harmful or unwanted.**
+
+    It will display the compilation output in the "C Extension result" area.
+
+  * **Test code**:
+    A button that executes the test code to compare the performance of the original
+    Python code and the generated C extension.
+
+    > [!CAUTION]
+    >
+    > **Always review the test code before running it, as it will be executed in
+    > your local environment and may contain code that could be harmful or unwanted.**
+
+    Will save the test code provided in the "Test compare code" into the
+    `usage_example.py` file and execute it showing the output in the "Test result" area.

week4/community-contributions/c_extension_generator/calculate_pi.c

@@ -0,0 +1,83 @@
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include <math.h>
+#include <float.h>
+#include <limits.h>
+#include <stdint.h>
+
+static PyObject* leibniz_pi(PyObject* self, PyObject* args) {
+    PyObject* iterations_obj;
+    if (!PyArg_ParseTuple(args, "O", &iterations_obj)) {
+        return NULL;
+    }
+
+    long long n_signed;
+    int overflow = 0;
+    n_signed = PyLong_AsLongLongAndOverflow(iterations_obj, &overflow);
+    if (n_signed == -1 && PyErr_Occurred() && overflow == 0) {
+        return NULL;
+    }
+
+    unsigned long long n = 0ULL;
+    if (overflow < 0) {
+        n = 0ULL;
+    } else if (overflow > 0) {
+        unsigned long long tmp = PyLong_AsUnsignedLongLong(iterations_obj);
+        if (tmp == (unsigned long long)-1 && PyErr_Occurred()) {
+            return NULL;
+        }
+        n = tmp;
+    } else {
+        if (n_signed <= 0) {
+            n = 0ULL;
+        } else {
+            n = (unsigned long long)n_signed;
+        }
+    }
+
+    double result = 1.0;
+    if (n == 0ULL) {
+        return PyFloat_FromDouble(result * 4.0);
+    }
+
+    Py_BEGIN_ALLOW_THREADS
+    for (unsigned long long i = 1ULL; i <= n; ++i) {
+        double jd1;
+        if (i <= ULLONG_MAX / 4ULL) {
+            unsigned long long j1 = i * 4ULL - 1ULL;
+            jd1 = (double)j1;
+        } else {
+            jd1 = (double)i * 4.0 - 1.0;
+        }
+        result -= 1.0 / jd1;
+
+        double jd2;
+        if (i <= (ULLONG_MAX - 1ULL) / 4ULL) {
+            unsigned long long j2 = i * 4ULL + 1ULL;
+            jd2 = (double)j2;
+        } else {
+            jd2 = (double)i * 4.0 + 1.0;
+        }
+        result += 1.0 / jd2;
+    }
+    Py_END_ALLOW_THREADS
+
+    return PyFloat_FromDouble(result * 4.0);
+}
+
+static PyMethodDef CalculatePiMethods[] = {
+    {"leibniz_pi", leibniz_pi, METH_VARARGS, "Compute pi using the Leibniz series with the given number of iterations."},
+    {NULL, NULL, 0, NULL}
+};
+
+static struct PyModuleDef calculate_pimodule = {
+    PyModuleDef_HEAD_INIT,
+    "calculate_pi",
+    "High-performance Leibniz pi calculation.",
+    -1,
+    CalculatePiMethods
+};
+
+PyMODINIT_FUNC PyInit_calculate_pi(void) {
+    return PyModule_Create(&calculate_pimodule);
+}

week4/community-contributions/c_extension_generator/python_hard.c

@@ -0,0 +1,244 @@
+#include <Python.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <math.h>
+
+// LCG step with 32-bit wrap-around
+static inline uint32_t lcg_next(uint32_t *state) {
+    *state = (uint32_t)(1664525u * (*state) + 1013904223u);
+    return *state;
+}
+
+static inline int add_overflow_int64(int64_t a, int64_t b, int64_t *res) {
+    if ((b > 0 && a > INT64_MAX - b) || (b < 0 && a < INT64_MIN - b)) return 1;
+    *res = a + b;
+    return 0;
+}
+
+// Kadane for int64 array with overflow detection; returns PyLong or NULL (on overflow -> signal via *overflowed)
+static PyObject* kadane_int64(const int64_t *arr, Py_ssize_t n, int *overflowed) {
+    if (n <= 0) {
+        return PyFloat_FromDouble(-INFINITY);
+    }
+    int64_t meh = arr[0];
+    int64_t msf = arr[0];
+    for (Py_ssize_t i = 1; i < n; ++i) {
+        int64_t x = arr[i];
+        if (meh > 0) {
+            int64_t tmp;
+            if (add_overflow_int64(meh, x, &tmp)) { *overflowed = 1; return NULL; }
+            meh = tmp;
+        } else {
+            meh = x;
+        }
+        if (meh > msf) msf = meh;
+    }
+    return PyLong_FromLongLong(msf);
+}
+
+// Kadane for PyObject* integer array
+static PyObject* kadane_big(PyObject **arr, Py_ssize_t n) {
+    if (n <= 0) {
+        return PyFloat_FromDouble(-INFINITY);
+    }
+    PyObject *meh = arr[0]; Py_INCREF(meh);
+    PyObject *msf = arr[0]; Py_INCREF(msf);
+    PyObject *zero = PyLong_FromLong(0);
+    if (!zero) { Py_DECREF(meh); Py_DECREF(msf); return NULL; }
+
+    for (Py_ssize_t i = 1; i < n; ++i) {
+        int cmp = PyObject_RichCompareBool(meh, zero, Py_GT);
+        if (cmp < 0) { Py_DECREF(meh); Py_DECREF(msf); Py_DECREF(zero); return NULL; }
+        if (cmp == 1) {
+            PyObject *t = PyNumber_Add(meh, arr[i]);
+            if (!t) { Py_DECREF(meh); Py_DECREF(msf); Py_DECREF(zero); return NULL; }
+            Py_DECREF(meh);
+            meh = t;
+        } else {
+            Py_DECREF(meh);
+            meh = arr[i]; Py_INCREF(meh);
+        }
+        int cmp2 = PyObject_RichCompareBool(meh, msf, Py_GT);
+        if (cmp2 < 0) { Py_DECREF(meh); Py_DECREF(msf); Py_DECREF(zero); return NULL; }
+        if (cmp2 == 1) {
+            Py_DECREF(msf);
+            msf = meh; Py_INCREF(msf);
+        }
+    }
+    Py_DECREF(meh);
+    Py_DECREF(zero);
+    return msf; // new reference
+}
+
+// Generate int64 array fast path; returns 0 on success
+static int gen_array_int64(Py_ssize_t n, uint32_t seed, int64_t min_v, int64_t max_v, int64_t *out) {
+    uint32_t state = seed;
+    uint64_t umax = (uint64_t)max_v;
+    uint64_t umin = (uint64_t)min_v;
+    uint64_t range = (umax - umin) + 1ULL; // max>=min guaranteed by caller
+    for (Py_ssize_t i = 0; i < n; ++i) {
+        state = lcg_next(&state);
+        uint32_t r32 = state;
+        uint64_t r = (range > 0x100000000ULL) ? (uint64_t)r32 : ((uint64_t)r32 % range);
+        int64_t val = (int64_t)(min_v + (int64_t)r);
+        out[i] = val;
+    }
+    return 0;
+}
+
+// Generate PyObject* int array general path using Python arithmetic
+static PyObject** gen_array_big(Py_ssize_t n, uint32_t seed, PyObject *min_val, PyObject *max_val) {
+    PyObject **arr = (PyObject**)PyMem_Malloc((n > 0 ? n : 1) * sizeof(PyObject*));
+    if (!arr) {
+        PyErr_NoMemory();
+        return NULL;
+    }
+    PyObject *one = PyLong_FromLong(1);
+    if (!one) { PyMem_Free(arr); return NULL; }
+    PyObject *diff = PyNumber_Subtract(max_val, min_val);
+    if (!diff) { Py_DECREF(one); PyMem_Free(arr); return NULL; }
+    PyObject *range_obj = PyNumber_Add(diff, one);
+    Py_DECREF(diff);
+    Py_DECREF(one);
+    if (!range_obj) { PyMem_Free(arr); return NULL; }
+
+    uint32_t state = seed;
+    for (Py_ssize_t i = 0; i < n; ++i) {
+        state = lcg_next(&state);
+        PyObject *v = PyLong_FromUnsignedLong((unsigned long)state);
+        if (!v) {
+            Py_DECREF(range_obj);
+            for (Py_ssize_t k = 0; k < i; ++k) Py_DECREF(arr[k]);
+            PyMem_Free(arr);
+            return NULL;
+        }
+        PyObject *r = PyNumber_Remainder(v, range_obj);
+        Py_DECREF(v);
+        if (!r) {
+            Py_DECREF(range_obj);
+            for (Py_ssize_t k = 0; k < i; ++k) Py_DECREF(arr[k]);
+            PyMem_Free(arr);
+            return NULL;
+        }
+        PyObject *val = PyNumber_Add(r, min_val);
+        Py_DECREF(r);
+        if (!val) {
+            Py_DECREF(range_obj);
+            for (Py_ssize_t k = 0; k < i; ++k) Py_DECREF(arr[k]);
+            PyMem_Free(arr);
+            return NULL;
+        }
+        arr[i] = val;
+    }
+    Py_DECREF(range_obj);
+    return arr;
+}
+
+static PyObject* max_subarray_sum_internal(Py_ssize_t n, uint32_t seed, PyObject *min_val, PyObject *max_val) {
+    if (n <= 0) {
+        return PyFloat_FromDouble(-INFINITY);
+    }
+
+    if (PyLong_Check(min_val) && PyLong_Check(max_val)) {
+        int overflow1 = 0, overflow2 = 0;
+        long long min64 = PyLong_AsLongLongAndOverflow(min_val, &overflow1);
+        if (overflow1) goto BIGINT_PATH;
+        long long max64 = PyLong_AsLongLongAndOverflow(max_val, &overflow2);
+        if (overflow2) goto BIGINT_PATH;
+        if (max64 >= min64) {
+            int64_t *arr = (int64_t*)PyMem_Malloc((size_t)n * sizeof(int64_t));
+            if (!arr) { PyErr_NoMemory(); return NULL; }
+            if (gen_array_int64(n, seed, (int64_t)min64, (int64_t)max64, arr) != 0) {
+                PyMem_Free(arr);
+                return NULL;
+            }
+            int overflowed = 0;
+            PyObject *res = kadane_int64(arr, n, &overflowed);
+            if (!res && overflowed) {
+                // fallback to big-int Kadane
+                PyObject **arr_obj = (PyObject**)PyMem_Malloc((size_t)n * sizeof(PyObject*));
+                if (!arr_obj) { PyMem_Free(arr); PyErr_NoMemory(); return NULL; }
+                for (Py_ssize_t i = 0; i < n; ++i) {
+                    arr_obj[i] = PyLong_FromLongLong(arr[i]);
+                    if (!arr_obj[i]) {
+                        for (Py_ssize_t k = 0; k < i; ++k) Py_DECREF(arr_obj[k]);
+                        PyMem_Free(arr_obj);
+                        PyMem_Free(arr);
+                        return NULL;
+                    }
+                }
+                PyObject *bires = kadane_big(arr_obj, n);
+                for (Py_ssize_t i = 0; i < n; ++i) Py_DECREF(arr_obj[i]);
+                PyMem_Free(arr_obj);
+                PyMem_Free(arr);
+                return bires;
+            }
+            PyMem_Free(arr);
+            return res;
+        }
+    }
+BIGINT_PATH: ;
+    PyObject **arr_obj = gen_array_big(n, seed, min_val, max_val);
+    if (!arr_obj) return NULL;
+    PyObject *res = kadane_big(arr_obj, n);
+    for (Py_ssize_t i = 0; i < n; ++i) Py_DECREF(arr_obj[i]);
+    PyMem_Free(arr_obj);
+    return res;
+}
+
+static PyObject* py_max_subarray_sum(PyObject *self, PyObject *args) {
+    Py_ssize_t n;
+    PyObject *seed_obj, *min_val, *max_val;
+    if (!PyArg_ParseTuple(args, "nOOO", &n, &seed_obj, &min_val, &max_val)) return NULL;
+    if (n < 0) n = 0;
+    uint32_t seed = (uint32_t)(PyLong_AsUnsignedLongLongMask(seed_obj) & 0xFFFFFFFFULL);
+    if (PyErr_Occurred()) return NULL;
+    return max_subarray_sum_internal(n, seed, min_val, max_val);
+}
+
+static PyObject* py_total_max_subarray_sum(PyObject *self, PyObject *args) {
+    Py_ssize_t n;
+    PyObject *init_seed_obj, *min_val, *max_val;
+    if (!PyArg_ParseTuple(args, "nOOO", &n, &init_seed_obj, &min_val, &max_val)) return NULL;
+    if (n < 0) n = 0;
+    uint32_t state = (uint32_t)(PyLong_AsUnsignedLongLongMask(init_seed_obj) & 0xFFFFFFFFULL);
+    if (PyErr_Occurred()) return NULL;
+
+    PyObject *total = PyLong_FromLong(0);
+    if (!total) return NULL;
+
+    for (int i = 0; i < 20; ++i) {
+        uint32_t seed = lcg_next(&state);
+        PyObject *part = max_subarray_sum_internal(n, seed, min_val, max_val);
+        if (!part) { Py_DECREF(total); return NULL; }
+        PyObject *new_total = PyNumber_Add(total, part);
+        Py_DECREF(part);
+        if (!new_total) { Py_DECREF(total); return NULL; }
+        Py_DECREF(total);
+        total = new_total;
+    }
+    return total;
+}
+
+static PyMethodDef module_methods[] = {
+    {"max_subarray_sum", (PyCFunction)py_max_subarray_sum, METH_VARARGS, "Compute maximum subarray sum using LCG-generated array."},
+    {"total_max_subarray_sum", (PyCFunction)py_total_max_subarray_sum, METH_VARARGS, "Compute total of maximum subarray sums over 20 LCG seeds."},
+    {NULL, NULL, 0, NULL}
+};
+
+static struct PyModuleDef moduledef = {
+    PyModuleDef_HEAD_INIT,
+    "python_hard",
+    NULL,
+    -1,
+    module_methods,
+    NULL,
+    NULL,
+    NULL,
+    NULL
+};
+
+PyMODINIT_FUNC PyInit_python_hard(void) {
+    return PyModule_Create(&moduledef);
+}

week4/community-contributions/c_extension_generator/setup_calculate_pi.py

@@ -0,0 +1,25 @@
+from setuptools import setup, Extension
+import sys
+import os
+
+extra_compile_args = []
+extra_link_args = []
+
+if os.name == 'nt':
+    extra_compile_args.extend(['/O2', '/fp:precise'])
+else:
+    extra_compile_args.extend(['-O3', '-fno-strict-aliasing'])
+
+module = Extension(
+    'calculate_pi',
+    sources=['calculate_pi.c'],
+    extra_compile_args=extra_compile_args,
+    extra_link_args=extra_link_args,
+)
+
+setup(
+    name='calculate_pi',
+    version='1.0.0',
+    description='High-performance C extension for computing pi via the Leibniz series',
+    ext_modules=[module],
+)

week4/community-contributions/c_extension_generator/setup_python_hard.py

@@ -0,0 +1,25 @@
+from setuptools import setup, Extension
+import sys
+
+extra_compile_args = []
+extra_link_args = []
+if sys.platform == 'win32':
+    extra_compile_args = ['/O2', '/Ot', '/GL', '/fp:fast']
+    extra_link_args = ['/LTCG']
+else:
+    extra_compile_args = ['-O3', '-march=native']
+
+module = Extension(
+    name='python_hard',
+    sources=['python_hard.c'],
+    extra_compile_args=extra_compile_args,
+    extra_link_args=extra_link_args,
+    language='c'
+)
+
+setup(
+    name='python_hard',
+    version='1.0.0',
+    description='High-performance C extension reimplementation',
+    ext_modules=[module]
+)

week4/community-contributions/c_extension_generator/setup_zz_my_module.py

@@ -0,0 +1,14 @@
+
+from setuptools import setup, Extension
+
+module = Extension(
+    'zz_my_module',
+    sources=['zz_my_module.c'],
+)
+
+setup(
+    name='zz_my_module',
+    version='1.0',
+    description='This is a custom C extension module.',
+    ext_modules=[module]
+)

week4/community-contributions/c_extension_generator/usage_example_calculate_pi.py

@@ -0,0 +1,38 @@
+# Build first: python setup.py build_ext --inplace
+import time
+import math
+import calculate_pi
+
+# Original Python implementation
+def py_leibniz_pi(iterations):
+    result = 1.0
+    for i in range(1, iterations + 1):
+        j = i * 4 - 1
+        result -= (1 / j)
+        j = i * 4 + 1
+        result += (1 / j)
+    return result * 4
+
+iters = 5_000_000
+
+# Warm-up
+calculate_pi.leibniz_pi(10)
+py_leibniz_pi(10)
+
+start = time.perf_counter()
+res_c = calculate_pi.leibniz_pi(iters)
+end = time.perf_counter()
+ctime = end - start
+
+start = time.perf_counter()
+res_py = py_leibniz_pi(iters)
+end = time.perf_counter()
+pytime = end - start
+
+print(f"Iterations: {iters}")
+print(f"C extension result: {res_c}")
+print(f"Python result:      {res_py}")
+print(f"Absolute difference: {abs(res_c - res_py)}")
+print(f"C extension time: {ctime:.6f} s")
+print(f"Python time:      {pytime:.6f} s")
+print(f"Speedup: {pytime/ctime if ctime > 0 else float('inf'):.2f}x")

week4/community-contributions/c_extension_generator/usage_example_python_hard.py

@@ -0,0 +1,69 @@
+import time
+
+# Original Python code
+
+def lcg(seed, a=1664525, c=1013904223, m=2**32):
+    value = seed
+    while True:
+        value = (a * value + c) % m
+        yield value
+
+def max_subarray_sum_py(n, seed, min_val, max_val):
+    lcg_gen = lcg(seed)
+    random_numbers = [next(lcg_gen) % (max_val - min_val + 1) + min_val for _ in range(n)]
+    max_sum = float('-inf')
+    for i in range(n):
+        current_sum = 0
+        for j in range(i, n):
+            current_sum += random_numbers[j]
+            if current_sum > max_sum:
+                max_sum = current_sum
+    return max_sum
+
+def total_max_subarray_sum_py(n, initial_seed, min_val, max_val):
+    total_sum = 0
+    lcg_gen = lcg(initial_seed)
+    for _ in range(20):
+        seed = next(lcg_gen)
+        total_sum += max_subarray_sum_py(n, seed, min_val, max_val)
+    return total_sum
+
+# Build and import extension (after running: python setup.py build && install or develop)
+import python_hard as ext
+
+# Example parameters
+n = 600
+initial_seed = 12345678901234567890
+min_val = -1000
+max_val = 1000
+
+# Time Python
+t0 = time.perf_counter()
+py_res1 = max_subarray_sum_py(n, (initial_seed * 1664525 + 1013904223) % (2**32), min_val, max_val)
+t1 = time.perf_counter()
+py_time1 = t1 - t0
+
+# Time C extension
+t0 = time.perf_counter()
+ext_res1 = ext.max_subarray_sum(n, (initial_seed * 1664525 + 1013904223) % (2**32), min_val, max_val)
+t1 = time.perf_counter()
+ext_time1 = t1 - t0
+
+print('max_subarray_sum equality:', py_res1 == ext_res1)
+print('Python time:', py_time1)
+print('C ext time:', ext_time1)
+
+# Total over 20 seeds
+t0 = time.perf_counter()
+py_res2 = total_max_subarray_sum_py(n, initial_seed, min_val, max_val)
+t1 = time.perf_counter()
+py_time2 = t1 - t0
+
+t0 = time.perf_counter()
+ext_res2 = ext.total_max_subarray_sum(n, initial_seed, min_val, max_val)
+t1 = time.perf_counter()
+ext_time2 = t1 - t0
+
+print('total_max_subarray_sum equality:', py_res2 == ext_res2)
+print('Python total time:', py_time2)
+print('C ext total time:', ext_time2)

week4/community-contributions/c_extension_generator/usage_example_zz_my_module.py

@@ -0,0 +1,16 @@
+
+import time
+import zz_my_module
+
+def python_hello_world():
+    print("Hello, World!")
+
+start = time.time()
+python_hello_world()
+end = time.time()
+print(f"Python function execution time: {end - start:.6f} seconds")
+
+start = time.time()
+zz_my_module.hello_world()
+end = time.time()
+print(f"C extension execution time: {end - start:.6f} seconds")

week4/community-contributions/c_extension_generator/zz_my_module.c

@@ -0,0 +1,28 @@
+
+#include <Python.h>
+
+// Function to be called from Python
+static PyObject* zz_hello_world(PyObject* self, PyObject* args) {
+    printf("Hello, World!\n");
+    Py_RETURN_NONE;
+}
+
+// Method definition structure
+static PyMethodDef zz_my_methods[] = {
+    {"hello_world", zz_hello_world, METH_VARARGS, "Print 'Hello, World!'"},
+    {NULL, NULL, 0, NULL}  // Sentinel
+};
+
+// Module definition
+static struct PyModuleDef zz_my_module = {
+    PyModuleDef_HEAD_INIT,
+    "zz_my_module",
+    "Extension module that prints Hello, World!",
+    -1,
+    zz_my_methods
+};
+
+// Module initialization function
+PyMODINIT_FUNC PyInit_zz_my_module(void) {
+    return PyModule_Create(&zz_my_module);
+}