ssrjson-benchmark 0.0.1__tar.gz → 0.0.1b0__tar.gz

{ssrjson_benchmark-0.0.1 → ssrjson_benchmark-0.0.1b0}/CMakeLists.txt

@@ -66,7 +66,7 @@ add_library(ssrjson_benchmark SHARED ${SRC_FILES})
 target_link_libraries(ssrjson_benchmark PUBLIC ${Python3_LIBRARIES})
 set_target_properties(ssrjson_benchmark PROPERTIES PREFIX "")
 target_include_directories(ssrjson_benchmark PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src> ${Python3_INCLUDE_DIRS})
-
+set_target_properties(ssrjson_benchmark PROPERTIES OUTPUT_NAME "_ssrjson_benchmark")
 # ------------------------------------------------------------------------------
 if(XCODE)
     set(SSRJSON_BENCHMARK_FLAGS)

ssrjson_benchmark-0.0.1b0/PKG-INFO

@@ -0,0 +1,60 @@
+Metadata-Version: 2.4
+Name: ssrjson_benchmark
+Version: 0.0.1b0
+Summary: Benchmark for ssrJSON
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: ssrjson
+Requires-Dist: orjson
+Requires-Dist: matplotlib
+Provides-Extra: all
+Requires-Dist: svglib; extra == "all"
+Requires-Dist: reportlab; extra == "all"
+Requires-Dist: py-cpuinfo; extra == "all"
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: summary
+
+# ssrJSON-benchmark
+
+<div align="center">
+
+[![PyPI - Version](https://img.shields.io/pypi/v/ssrjson-benchmark)](https://pypi.org/project/ssrjson-benchmark/) [![PyPI - Wheel](https://img.shields.io/pypi/wheel/ssrjson-benchmark)](https://pypi.org/project/ssrjson-benchmark/)
+
+The [ssrJSON](https://github.com/Antares0982/ssrjson) benchmark repository.
+
+</div>
+
+## Benchmark Results
+
+The benchmark results can be found in [results](results). Contributing your benchmark result is welcomed.
+
+Quick jump for
+
+* [x86-64-v2, SSE4.2](results/SSE4.2)
+* [x86-64-v3, AVX2](results/AVX2)
+* [x86-64-v4, AVX512](results/AVX512)
+
+## Usage
+
+```bash
+# you may need to install `svglib`, `reportlab` and `py-cpuinfo` as well
+pip install ssrjson-benchmark
+python -m ssrjson_benchmark
+```
+
+## Benchmark options
+
+* `-m` output in Markdown instead of PDF.
+* `-f <json_path>` used exists benchmark json result.
+* `--process-bytes <bytes_num>` Total process bytes per test, default 1e8.
+
+## Notes
+
+* This repository conducts benchmarking using json, orjson, and ssrJSON. The `dumps` benchmark produces str objects, comparing three operations: `json.dumps`, `orjson.dumps` followed by decode, and `ssrjson.dumps`. The `dumps_to_bytes` benchmark produces bytes objects, comparing three functions: `json.dumps` followed by encode, `orjson.dumps`, and `ssrjson.dumps_to_bytes`.
+* When orjson handles non-ASCII strings, if the cache of the `PyUnicodeObject`’s UTF-8 representation does not exist, it invokes the `PyUnicode_AsUTF8AndSize` function to obtain the UTF-8 encoding. This function then caches the UTF-8 representation within the `PyUnicodeObject`. If the same `PyUnicodeObject` undergoes repeated encode-decode operations, subsequent calls after the initial one will execute more quickly due to this caching. However, in real-world production scenarios, it is uncommon to perform JSON encode-decode repeatedly on the exact same string object; even identical strings are unlikely to be the same object instance. To achieve benchmark results that better reflect practical use cases, we employ `ssrjson.run_unicode_accumulate_benchmark` and `benchmark_invalidate_dump_cache` functions, which ensure that new `PyUnicodeObject`s are different for each input every time. (ref: [orjson#586](https://github.com/ijl/orjson/issues/586))
+* The performance of JSON encoding is primarily constrained by the speed of writing to the buffer, whereas decoding performance is mainly limited by the frequent invocation of CPython interfaces for object creation. During decoding, both ssrJSON and orjson employ short key caching to reduce the number of object creations, and this caching mechanism is global in both cases. As a result, decoding benchmark tests may not accurately reflect the conditions encountered in real-world production environments.
+* The files simple_object.json and simple_object_zh.json do not represent real-world data; they are solely used to compare the performance of the fast path. Therefore, the benchmark results should not be interpreted as indicative of actual performance.

{ssrjson_benchmark-0.0.1 → ssrjson_benchmark-0.0.1b0}/README.md

@@ -1,7 +1,13 @@
 # ssrJSON-benchmark
 
+<div align="center">
+
+[![PyPI - Version](https://img.shields.io/pypi/v/ssrjson-benchmark)](https://pypi.org/project/ssrjson-benchmark/) [![PyPI - Wheel](https://img.shields.io/pypi/wheel/ssrjson-benchmark)](https://pypi.org/project/ssrjson-benchmark/)
+
 The [ssrJSON](https://github.com/Antares0982/ssrjson) benchmark repository.
 
+</div>
+
 ## Benchmark Results
 
 The benchmark results can be found in [results](results). Contributing your benchmark result is welcomed.
@@ -14,17 +20,10 @@ Quick jump for
 
 ## Usage
 
-To generate a benchmark report, you need to install `ssrJSON` either by fetched [PyPi](https://pypi.org/project/ssrjson/) or built from [source](https://github.com/Antares0982/ssrjson), and toolkit(`ssrjson_benchmark`) from this repo by:
-
-```bash
-python -m build
-pip install dist/*.whl
-```
-
-Then run the benchmark script:
-
 ```bash
-python benchmark.py
+# you may need to install `svglib`, `reportlab` and `py-cpuinfo` as well
+pip install ssrjson-benchmark
+python -m ssrjson_benchmark
 ```
 
 ## Benchmark options
@@ -36,9 +35,6 @@ python benchmark.py
 ## Notes
 
 * This repository conducts benchmarking using json, orjson, and ssrJSON. The `dumps` benchmark produces str objects, comparing three operations: `json.dumps`, `orjson.dumps` followed by decode, and `ssrjson.dumps`. The `dumps_to_bytes` benchmark produces bytes objects, comparing three functions: `json.dumps` followed by encode, `orjson.dumps`, and `ssrjson.dumps_to_bytes`.
-* The ssrJSON built with the `BUILD_BENCHMARK` option includes several additional C functions specifically designed for executing benchmarks. These functions utilize high-precision timing APIs, and within the loop, only the time spent on the actual `PyObject_Call` invocations is measured.
-* When orjson handles non-ASCII strings, if the cache of the `PyUnicodeObject`’s UTF-8 representation does not exist, it invokes the `PyUnicode_AsUTF8AndSize` function to obtain the UTF-8 encoding. This function then caches the UTF-8 representation within the `PyUnicodeObject`. If the same `PyUnicodeObject` undergoes repeated encode-decode operations, subsequent calls after the initial one will execute more quickly due to this caching. However, in real-world production scenarios, it is uncommon to perform JSON encode-decode repeatedly on the exact same string object; even identical strings are unlikely to be the same object instance. To achieve benchmark results that better reflect practical use cases, we employ `ssrjson.run_unicode_accumulate_benchmark` and `benchmark_invalidate_dump_cache` functions, which ensure that new `PyUnicodeObject`s are different for each input every time.
-
+* When orjson handles non-ASCII strings, if the cache of the `PyUnicodeObject`’s UTF-8 representation does not exist, it invokes the `PyUnicode_AsUTF8AndSize` function to obtain the UTF-8 encoding. This function then caches the UTF-8 representation within the `PyUnicodeObject`. If the same `PyUnicodeObject` undergoes repeated encode-decode operations, subsequent calls after the initial one will execute more quickly due to this caching. However, in real-world production scenarios, it is uncommon to perform JSON encode-decode repeatedly on the exact same string object; even identical strings are unlikely to be the same object instance. To achieve benchmark results that better reflect practical use cases, we employ `ssrjson.run_unicode_accumulate_benchmark` and `benchmark_invalidate_dump_cache` functions, which ensure that new `PyUnicodeObject`s are different for each input every time. (ref: [orjson#586](https://github.com/ijl/orjson/issues/586))
 * The performance of JSON encoding is primarily constrained by the speed of writing to the buffer, whereas decoding performance is mainly limited by the frequent invocation of CPython interfaces for object creation. During decoding, both ssrJSON and orjson employ short key caching to reduce the number of object creations, and this caching mechanism is global in both cases. As a result, decoding benchmark tests may not accurately reflect the conditions encountered in real-world production environments.
-
 * The files simple_object.json and simple_object_zh.json do not represent real-world data; they are solely used to compare the performance of the fast path. Therefore, the benchmark results should not be interpreted as indicative of actual performance.

ssrjson_benchmark-0.0.1b0/setup.py

@@ -0,0 +1,100 @@
+import os
+import shutil
+import subprocess
+
+from setuptools import Extension, find_packages, setup
+from setuptools.command.build_ext import build_ext
+from pathlib import Path
+
+
+def find_version(src_file_content: str):
+    # find macro SSRJSON_BENCHMARK_VERSION
+    prefix = "#define SSRJSON_BENCHMARK_VERSION"
+    for line in src_file_content.splitlines():
+        if line.startswith(prefix):
+            version = line[len(prefix) :].strip()[1:-1]
+            return version
+    raise RuntimeError("Cannot find SSRJSON_BENCHMARK_VERSION in source file")
+
+
+with open("./src/benchmark.c", "r", encoding="utf-8") as f:
+    version_string = find_version(f.read())
+
+
+class CMakeBuild(build_ext):
+    def run(self):
+        build_dir = os.path.abspath("build")
+        if not os.path.exists(build_dir):
+            os.makedirs(build_dir)
+        cmake_cmd = [
+            "cmake",
+            "-DCMAKE_BUILD_TYPE=Release",
+            ".",
+            "-B",
+            "build",
+        ]
+        subprocess.check_call(cmake_cmd)
+
+        if os.name == "nt":
+            build_cmd = ["cmake", "--build", "build", "--config", "Release"]
+        else:
+            build_cmd = ["cmake", "--build", "build"]
+        subprocess.check_call(build_cmd)
+
+        if os.name == "nt":
+            built_filename = "Release/_ssrjson_benchmark.dll"
+            target_filename = "_ssrjson_benchmark.pyd"
+        else:
+            built_filename = "_ssrjson_benchmark.so"
+            target_filename = built_filename
+
+        built_path = os.path.join(build_dir, built_filename)
+        if not os.path.exists(built_path):
+            raise RuntimeError(f"Built library not found: {built_path}")
+
+        target_dir = self.build_lib + "/ssrjson_benchmark"
+        if not os.path.exists(target_dir):
+            os.makedirs(target_dir)
+
+        target_path = os.path.join(target_dir, target_filename)
+        self.announce(f"Copying {built_path} to {target_path}")
+        print(f"Copying {built_path} to {target_path}")
+        shutil.copyfile(built_path, target_path)
+
+
+setup(
+    name="ssrjson_benchmark",
+    version=version_string,
+    description="Benchmark for ssrJSON",
+    long_description=Path("README.md").read_text(encoding="utf-8"),
+    long_description_content_type="text/markdown",
+    ext_modules=[
+        Extension(
+            "_ssrjson_benchmark",
+            sources=["src/benchmark.c"],
+            language="c",
+        )
+    ],
+    packages=["ssrjson_benchmark", "ssrjson_benchmark._files"],
+    package_dir={"": "src"},
+    package_data={
+        "ssrjson_benchmark": ["template.md"],
+        "ssrjson_benchmark._files": ["*.json"],
+    },
+    include_package_data=True,
+    install_requires=[
+        "ssrjson",
+        "orjson",
+        "matplotlib",
+    ],
+    extras_require={
+        "all": [
+            "svglib",
+            "reportlab",
+            "py-cpuinfo",
+        ],
+    },
+    cmdclass={
+        "build_ext": CMakeBuild,
+    },
+)

{ssrjson_benchmark-0.0.1 → ssrjson_benchmark-0.0.1b0}/src/benchmark.c

@@ -23,7 +23,7 @@
 #include <Python.h>
 #include <stdbool.h>
 
-#define SSRJSON_BENCHMARK_VERSION "0.0.1"
+#define SSRJSON_BENCHMARK_VERSION "0.0.1b0"
 
 /** compiler builtin check (since gcc 10.0, clang 2.6, icc 2021) */
 #ifndef has_builtin
@@ -319,7 +319,7 @@ static PyMethodDef ssrjson_benchmark_methods[] = {
 
 static struct PyModuleDef moduledef = {
         PyModuleDef_HEAD_INIT,
-        "ssrjson_benchmark",       /* m_name */
+        "_ssrjson_benchmark",       /* m_name */
         0,                         /* m_doc */
         0,                         /* m_size */
         ssrjson_benchmark_methods, /* m_methods */
@@ -329,7 +329,7 @@ static struct PyModuleDef moduledef = {
         NULL                       /* m_free */
 };
 
-PyMODINIT_FUNC PyInit_ssrjson_benchmark(void) {
+PyMODINIT_FUNC PyInit__ssrjson_benchmark(void) {
     PyObject *module;
     // check if module already exists
     if ((module = PyState_FindModule(&moduledef)) != NULL) {

ssrjson_benchmark-0.0.1b0/src/ssrjson_benchmark/__init__.py

@@ -0,0 +1,15 @@
+from .benchmark_main import (
+    run_benchmark,
+    generate_report_markdown,
+    generate_report,
+    run_benchmark_default,
+)
+from ._ssrjson_benchmark import __version__
+
+__all__ = [
+    "run_benchmark",
+    "generate_report_markdown",
+    "generate_report",
+    "run_benchmark_default",
+    "__version__",
+]

ssrjson_benchmark-0.0.1b0/src/ssrjson_benchmark/__main__.py

@@ -0,0 +1,54 @@
+import argparse
+import json
+import os
+from ssrjson_benchmark import (
+    run_benchmark,
+    generate_report_markdown,
+    generate_report,
+)
+
+
+def main():
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "-f", "--file", help="record JSON file", required=False, default=None
+    )
+    parser.add_argument(
+        "-m",
+        "--markdown",
+        help="Generate markdown report",
+        required=False,
+        action="store_true",
+    )
+    parser.add_argument(
+        "--process-bytes",
+        help="Total process bytes per test, default 1e8",
+        required=False,
+        default=1e8,
+        type=int,
+    )
+    parser.add_argument(
+        "--out-dir",
+        help="Output directory for reports",
+        required=False,
+        default=os.getcwd(),
+    )
+    args = parser.parse_args()
+
+    if args.file:
+        with open(args.file, "r") as f:
+            j = json.load(f)
+        file = args.file.split("/")[-1]
+    else:
+        j, file = run_benchmark(args.process_bytes)
+        file = file.split("/")[-1]
+
+    if args.markdown:
+        generate_report_markdown(j, file, args.out_dir)
+    else:
+        generate_report(j, file, args.out_dir)
+
+
+if __name__ == "__main__":
+    main()