2.0.10

CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## 2.0.19 - 2019-09-25
+
+### Changed
+
+- Performance improvements and lower memory usage in deserialization
+by creating only one `str` object for repeated map keys.
+
 ## 2.0.9 - 2019-09-22
 
 ### Changed

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "orjson"
-version = "2.0.9"
+version = "2.0.10"
 authors = ["ijl <ijl@mailbox.org>"]
 description = "Fast, correct Python JSON library"
 edition = "2018"

bench/mem_usage

@@ -0,0 +1,37 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: (Apache-2.0 OR MIT)
+
+import io
+import subprocess
+
+from tabulate import tabulate
+
+buf = io.StringIO()
+
+headers = ("Library", "RSS change after loads() (MiB)")
+
+LIBRARIES = ("orjson", "ujson", "rapidjson", "json", "simplejson")
+
+FIXTURES = ("canada.json", "citm_catalog.json", "github.json", "twitter.json")
+
+for fixture in sorted(FIXTURES, reverse=True):
+    table = []
+    buf.write("\n" + "#### " + fixture + "\n\n")
+    for lib_name in LIBRARIES:
+        proc = subprocess.Popen(
+            ("bench/run_mem", f"data/{fixture}.xz", lib_name), stdout=subprocess.PIPE
+        )
+        output = proc.stdout.readline().decode("utf-8").split(",")
+        mem_base = int(output[0]) / 1024 / 1024
+        mem_diff = int(output[1]) / 1024 / 1024
+        table.append((lib_name, f"+{mem_diff:,.1f}"))
+    buf.write(tabulate(table, headers, tablefmt="grid") + "\n")
+
+print(
+    buf.getvalue()
+    .replace("-", "")
+    .replace("=", "-")
+    .replace("+", "|")
+    .replace("||||", "")
+    .replace("\n\n", "\n")
+)

bench/run_mem

@@ -10,7 +10,7 @@ import psutil
 filename = sys.argv[1]
 
 with lzma.open(filename, "r") as fileh:
-    file_utf8 = fileh.read()
+    fixture = fileh.read()
 
 proc = psutil.Process()
 
@@ -32,7 +32,8 @@ gc.collect()
 
 mem_before = proc.memory_info().rss
 
-val = loads(file_utf8)
+for _ in range(100):
+    val = loads(fixture)
 
 mem_after = proc.memory_info().rss