Adds JSON/CBOR support and an Io-type option

amazon/ionbenchmark/API.py

@@ -4,6 +4,6 @@
 # Serialization/deserialization APIs to benchmark.
 class API(Enum):
     """Enumeration of the APIs."""
-    SIMPLE_ION = 'simple_ion'
-    EVENT = 'event'
-    DEFAULT = 'simple_ion'
+    LOAD_DUMP = 'load_dump'
+    STREAMING = 'streaming'
+    DEFAULT = 'load_dump'

amazon/ionbenchmark/Command.py

@@ -0,0 +1,7 @@
+from enum import Enum
+
+
+class Command(Enum):
+    """Enumeration of the Command."""
+    READ = 'read'
+    WRITE = 'write'

amazon/ionbenchmark/Format.py

@@ -1,8 +1,39 @@
 from enum import Enum
 
 
+def format_is_ion(format_option):
+    return (format_option == Format.ION_BINARY.value) or (format_option == Format.ION_TEXT.value)
+
+
+def format_is_json(format_option):
+    return (format_option == Format.JSON.value) or (format_option == Format.SIMPLEJSON.value) \
+           or (format_option == Format.UJSON.value) or (format_option == Format.RAPIDJSON.value) or \
+           (format_option == Format.ORJSON.value)
+
+
+def format_is_cbor(format_option):
+    return (format_option == Format.CBOR.value) or (format_option == Format.CBOR2.value)
+
+
+def format_is_binary(format_option):
+    return format_is_cbor(format_option) or (format_option == Format.ION_BINARY.value) \
+           or (format_option == Format.ORJSON.value)
+
+
+def rewrite_file_to_format(file, format_option):
+    return file
+
+
 class Format(Enum):
     """Enumeration of the formats."""
     ION_TEXT = 'ion_text'
     ION_BINARY = 'ion_binary'
+    JSON = 'json'
+    SIMPLEJSON = 'simplejson'
+    UJSON = 'ujson'
+    RAPIDJSON = 'rapidjson'
+    ORJSON = 'orjson'
+    CBOR = 'cbor'
+    CBOR2 = 'cbor2'
     DEFAULT = 'ion_binary'
+

amazon/ionbenchmark/Io_type.py

@@ -0,0 +1,8 @@
+from enum import Enum
+
+
+class Io_type(Enum):
+    """Enumeration of the IO types."""
+    FILE = 'file'
+    BUFFER = 'buffer'
+    DEFAULT = 'file'

install.py

@@ -89,6 +89,9 @@ def _download_ionc():
 
         os.chdir(_CURRENT_ION_C_DIR)
 
+        # TODO Use ion-c 1.1.0 for now - https://github.com/amazon-ion/ion-python/issues/249
+        check_call(['git', 'reset', '--hard', 'v1.1.0'])
+
         # Initialize submodule.
         check_call(['git', 'submodule', 'update', '--init'])
 

requirements.txt

@@ -15,4 +15,11 @@ tox==3.23.1
 virtualenv==20.4.7
 setuptools<=60.5.0
 docopt==0.6.2
-tabulate==0.9.0
+tabulate==0.9.0
+simplejson~=3.18.3
+pip~=23.0
+six~=1.16.0
+cbor~=1.0.0
+cbor2~=5.4.6
+python-rapidjson~=1.9
+ujson~=5.7.0

tests/benchmark_sample_data/cbor/sample

@@ -0,0 +1 @@
+sthis is a test file

tests/test_benchmark_cli.py

@@ -1,13 +1,16 @@
+import json
 import time
 from itertools import chain
 from os.path import abspath, join, dirname
 
+import cbor2
 from docopt import docopt
 
 from amazon.ion import simpleion
-from amazon.ionbenchmark import ion_benchmark_cli
-from amazon.ionbenchmark.ion_benchmark_cli import generate_simpleion_load_test_code, generate_simpleion_dump_test_code,\
-    ion_python_benchmark_cli
+from amazon.ionbenchmark import ion_benchmark_cli, Format, Io_type
+from amazon.ionbenchmark.Format import format_is_ion, format_is_cbor, format_is_json
+from amazon.ionbenchmark.ion_benchmark_cli import generate_read_test_code, \
+    generate_write_test_code, ion_python_benchmark_cli
 from amazon.ionbenchmark.util import str_to_bool, TOOL_VERSION
 from tests import parametrize
 from tests.test_simpleion import generate_scalars_text
@@ -44,12 +47,46 @@ def generate_test_path(p):
 @parametrize(
     generate_test_path('integers.ion')
 )
-def test_generate_simpleion_load_test_code(path):
-    actual = generate_simpleion_load_test_code(path, memory_profiling=False, single_value=False, emit_bare_values=False)
+def test_generate_simpleion_read_test_code(path):
+    actual = generate_read_test_code(path, memory_profiling=False, single_value=False,
+                                     format_option=Format.Format.ION_TEXT.value, emit_bare_values=False,
+                                     io_type=Io_type.Io_type.FILE, binary=False)
 
     # make sure we generated the desired load function
     with open(path) as fp:
-        expect = simpleion.load(fp, single_value=False, parse_eagerly=True)
+        expect = simpleion.load(fp, single_value=False, emit_bare_values=False, parse_eagerly=True)
+
+    # make sure the return values are same
+    assert actual() == expect
+
+
+@parametrize(
+    generate_test_path('integers.ion')
+)
+def test_generate_json_read_test_code(path):
+    actual = generate_read_test_code(path, memory_profiling=False, single_value=False,
+                                     format_option=Format.Format.JSON.value, emit_bare_values=False,
+                                     io_type=Io_type.Io_type.FILE, binary=False)
+
+    # make sure we generated the desired load function
+    with open(path) as fp:
+        expect = json.load(fp)
+
+    # make sure the return values are same
+    assert actual() == expect
+
+
+@parametrize(
+    generate_test_path('integers.ion')
+)
+def test_generate_cbor_read_test_code(path):
+    actual = generate_read_test_code(path, memory_profiling=False, single_value=False,
+                                     format_option=Format.Format.CBOR2.value, emit_bare_values=False,
+                                     io_type=Io_type.Io_type.FILE, binary=False)
+
+    # make sure we generated the desired load function
+    with open(path) as fp:
+        expect = cbor2.load(fp)
 
     # make sure the return values are same
     assert actual() == expect
@@ -60,8 +97,9 @@ def test_generate_simpleion_load_test_code(path):
         generate_scalars_text(SIMPLE_SCALARS_MAP_TEXT),
     ))
 )
-def test_generate_simpleion_dump_test_code(obj):
-    actual = generate_simpleion_dump_test_code(obj, memory_profiling=False, binary=False)
+def test_generate_simpleion_write_test_code(obj):
+    actual = generate_write_test_code(obj, format_option=Format.Format.ION_TEXT.value, memory_profiling=False,
+                                      binary=False, io_type=Io_type.Io_type.BUFFER.value)
 
     # make sure we generated the desired dumps function
     expect = simpleion.dumps(obj, binary=False)
@@ -70,6 +108,38 @@ def test_generate_simpleion_dump_test_code(obj):
     assert actual() == expect
 
 
+@parametrize(
+    generate_test_path('./json/object.json'),
+)
+def test_generate_json_write_test_code(file):
+    with open(file) as fp:
+        obj = json.load(fp)
+    actual = generate_write_test_code(obj, format_option=Format.Format.JSON.value, memory_profiling=False, binary=False,
+                                      io_type=Io_type.Io_type.BUFFER.value)
+
+    # make sure we generated the desired dumps function
+    expect = json.dumps(obj)
+
+    # make sure the return values are same
+    assert actual() == expect
+
+
+@parametrize(
+    generate_test_path('./cbor/sample')
+)
+def test_generate_cbor_write_test_code(file):
+    with open(file, 'br') as fp:
+        obj = cbor2.load(fp)
+    actual = generate_write_test_code(obj, format_option=Format.Format.CBOR2.value, memory_profiling=False,
+                                      binary=False, io_type=Io_type.Io_type.BUFFER.value)
+
+    # make sure we generated the desired dumps function
+    expect = cbor2.dumps(obj)
+
+    # make sure the return values are same
+    assert actual() == expect
+
+
 def execution_with_command(c):
     return ion_python_benchmark_cli(docopt(doc, argv=c))
 
@@ -134,40 +204,127 @@ def gather_all_options_in_list(table):
 
 
 def test_read_multi_api(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['read', file, '--api', 'simple_ion', '--api', 'event'])
-    assert gather_all_options_in_list(table) == sorted([('event', 'ion_binary'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(['read', file, '--api', 'load_dump', '--api', 'streaming'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('streaming', 'ion_binary', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_write_multi_api(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['write', file, '--api', 'simple_ion', '--api', 'event'])
-    assert gather_all_options_in_list(table) == sorted([('event', 'ion_binary'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(['write', file, '--api', 'load_dump', '--api', 'streaming'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('streaming', 'ion_binary', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_read_multi_duplicated_api(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['read', file, '--api', 'simple_ion', '--api', 'event', '--api', 'event'])
-    assert gather_all_options_in_list(table) == sorted([('event', 'ion_binary'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(['read', file, '--api', 'load_dump', '--api', 'streaming', '--api', 'streaming'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('streaming', 'ion_binary', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_write_multi_duplicated_api(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['write', file, '--api', 'simple_ion', '--api', 'event', '--api', 'event'])
-    assert gather_all_options_in_list(table) == sorted([('event', 'ion_binary'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(['write', file, '--api', 'load_dump', '--api', 'streaming', '--api', 'streaming'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('streaming', 'ion_binary', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_read_multi_format(file=generate_test_path('integers.ion')):
     table = execution_with_command(['read', file, '--format', 'ion_text', '--format', 'ion_binary'])
-    assert gather_all_options_in_list(table) == sorted([('simple_ion', 'ion_binary'), ('simple_ion', 'ion_text')])
+    assert gather_all_options_in_list(table) == sorted(
+        [('load_dump', 'ion_binary', 'file'), ('load_dump', 'ion_text', 'file')])
 
 
 def test_write_multi_format(file=generate_test_path('integers.ion')):
     table = execution_with_command(['write', file, '--format', 'ion_text', '--format', 'ion_binary'])
-    assert gather_all_options_in_list(table) == sorted([('simple_ion', 'ion_text'), ('simple_ion', 'ion_binary')])
+    assert gather_all_options_in_list(table) == sorted(
+        [('load_dump', 'ion_text', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_read_multi_duplicated_format(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['read', file, '--format', 'ion_text', '--format', 'ion_binary', '--format', 'ion_text'])
-    assert gather_all_options_in_list(table) == sorted([('simple_ion', 'ion_text'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(
+        ['read', file, '--format', 'ion_text', '--format', 'ion_binary', '--format', 'ion_text'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('load_dump', 'ion_text', 'file'), ('load_dump', 'ion_binary', 'file')])
 
 
 def test_write_multi_duplicated_format(file=generate_test_path('integers.ion')):
-    table = execution_with_command(['write', file, '--format', 'ion_text', '--format', 'ion_binary', '--format', 'ion_text',])
-    assert gather_all_options_in_list(table) == sorted([('simple_ion', 'ion_text'), ('simple_ion', 'ion_binary')])
+    table = execution_with_command(
+        ['write', file, '--format', 'ion_text', '--format', 'ion_binary', '--format', 'ion_text', ])
+    assert gather_all_options_in_list(table) == sorted(
+        [('load_dump', 'ion_text', 'file'), ('load_dump', 'ion_binary', 'file')])
+
+
+@parametrize(
+    *tuple((f.value for f in Format.Format if Format.format_is_json(f.value)))
+)
+def test_write_json_format(f):
+    table = execution_with_command(['write', generate_test_path('integers.ion'), '--format', f'{f}'])
+    assert gather_all_options_in_list(table) == sorted([('load_dump', f'{f}', 'file')])
+
+
+@parametrize(
+    *tuple((f.value for f in Format.Format if Format.format_is_json(f.value)))
+)
+def test_read_json_format(f):
+    table = execution_with_command(['read', generate_test_path('integers.ion'), '--format', f'{f}'])
+    assert gather_all_options_in_list(table) == sorted([('load_dump', f'{f}', 'file')])
+
+
+@parametrize(
+    *tuple((f.value for f in Format.Format if Format.format_is_cbor(f.value)))
+)
+def test_write_cbor_format(f):
+    table = execution_with_command(['write', generate_test_path('integers.ion'), '--format', f'{f}'])
+    assert gather_all_options_in_list(table) == sorted([('load_dump', f'{f}', 'file')])
+
+
+@parametrize(
+    *tuple((f.value for f in Format.Format if Format.format_is_cbor(f.value)))
+)
+def test_read_cbor_format(f):
+    table = execution_with_command(['read', generate_test_path('integers.ion'), '--format', f'{f}'])
+    assert gather_all_options_in_list(table) == sorted([('load_dump', f'{f}', 'file')])
+
+
+@parametrize(
+    *tuple((io.value for io in Io_type.Io_type))
+)
+def test_write_io_type(f):
+    table = execution_with_command(
+        ['write', generate_test_path('integers.ion'), '--io-type', f'{f}', '--format', 'json'])
+    assert gather_all_options_in_list(table) == sorted([('load_dump', 'json', f'{f}')])
+
+
+@parametrize(
+    *tuple((io.value for io in Io_type.Io_type))
+)
+def test_read_io_type(f):
+    table = execution_with_command(
+        ['read', generate_test_path('integers.ion'), '--io-type', f'{f}', '--format', 'json', '--format', 'ion_binary'])
+    assert gather_all_options_in_list(table) == sorted(
+        [('load_dump', 'json', f'{f}'), ('load_dump', 'ion_binary', f'{f}')])
+
+
+@parametrize(
+    *tuple((Format.Format.ION_TEXT, Format.Format.ION_BINARY))
+)
+def test_format_is_ion(f):
+    assert format_is_ion(f.value) is True
+
+
+@parametrize(
+    *tuple((Format.Format.JSON,
+            Format.Format.UJSON,
+            Format.Format.RAPIDJSON,
+            Format.Format.SIMPLEJSON
+            ))
+)
+def test_format_is_json(f):
+    assert format_is_json(f.value) is True
+
+
+@parametrize(
+    Format.Format.CBOR,
+    Format.Format.CBOR2
+)
+def test_format_is_cbor(f):
+    assert format_is_cbor(f.value) is True