def test_prepare_data():
"""Test prepare_data function on sample data."""
data = {
"browser_arch": "x86",
"os_name": "Windows_NT",
"os_version": "6.1",
"memory_mb": 4286,
"is_wow64": False,
"gfx0_vendor_id": "0xfeee",
"gfx0_device_id": "0xd1d1",
"screen_width": 1280,
"screen_height": 1024,
"cpu_cores": 2,
"cpu_vendor": "SomeCpuVendor",
"cpu_speed": 3261,
"has_flash": True,
}
prepared_data = summarize_json.prepare_data(data, device_map)
assert (prepared_data["browser_arch"] == "x86"
), "The browser architecture must be correct."
assert prepared_data[
"cpu_cores"] == 2, "The number of CPU cores must be correct."
assert (
prepared_data["cpu_speed"] == 3.3
), "The CPU speed must be in GHz and correctly rounded to 1 decimal."
assert (prepared_data["cpu_vendor"] == "SomeCpuVendor"
), "The CPU vendor must be correct."
assert (prepared_data["cpu_cores_speed"] == "2_3.3"
), "The CPU cores and speed must be correctly merged together."
assert (
prepared_data["gfx0_vendor_name"] == "Other"
), "The GPU vendor name must be correctly converted from the vendor id."
assert (
prepared_data["gfx0_model"] == "family-chipset"
), "The GPU family and chipset must be correctly derived from vendor and device ids."
assert (prepared_data["resolution"] == "1280x1024"
), "The screen resolution must be correctly concatenated."
assert prepared_data[
"memory_gb"] == 4, "The RAM memory must be converted to GB."
assert (prepared_data["os"] == "Windows_NT-6.1"
), "The OS string must contain the OS name and version."
assert (prepared_data["os_arch"] == "x86"
), "The OS architecture must be correctly inferred."
assert (prepared_data["has_flash"] is
True), "The flash plugin must be correctly reported."
def test_prepare_data():
"""Test prepare_data function on sample data."""
data = {
'browser_arch': 'x86',
'os_name': 'Windows_NT',
'os_version': '6.1',
'memory_mb': 4286,
'is_wow64': False,
'gfx0_vendor_id': '0xfeee',
'gfx0_device_id': '0xd1d1',
'screen_width': 1280,
'screen_height': 1024,
'cpu_cores': 2,
'cpu_vendor': 'SomeCpuVendor',
'cpu_speed': 3261,
'has_flash': True
}
prepared_data = summarize_json.prepare_data(data, device_map)
assert prepared_data["browser_arch"] == "x86", (
"The browser architecture must be correct.")
assert prepared_data["cpu_cores"] == 2, (
"The number of CPU cores must be correct.")
assert prepared_data["cpu_speed"] == 3.3, (
"The CPU speed must be in GHz and correctly rounded to 1 decimal.")
assert prepared_data["cpu_vendor"] == "SomeCpuVendor", (
"The CPU vendor must be correct.")
assert prepared_data["cpu_cores_speed"] == "2_3.3", (
"The CPU cores and speed must be correctly merged together.")
assert prepared_data["gfx0_vendor_name"] == "Other", (
"The GPU vendor name must be correctly converted from the vendor id.")
assert prepared_data["gfx0_model"] == "family-chipset", (
"The GPU family and chipset must be correctly derived from vendor and device ids."
)
assert prepared_data["resolution"] == "1280x1024", (
"The screen resolution must be correctly concatenated.")
assert prepared_data["memory_gb"] == 4, (
"The RAM memory must be converted to GB.")
assert prepared_data["os"] == "Windows_NT-6.1", (
"The OS string must contain the OS name and version.")
assert prepared_data["os_arch"] == "x86", (
"The OS architecture must be correctly inferred.")
assert prepared_data["has_flash"] is True, (
"The flash plugin must be correctly reported.")
def test_aggregate_data():
"""Test aggregate_data function on sample data."""
raw_data = [
{
"browser_arch": "x86",
"os_name": "Windows_NT",
"os_version": "6.1",
"memory_mb": 4286,
"is_wow64": False,
"gfx0_vendor_id": "0xfeee",
"gfx0_device_id": "0xd1d1",
"screen_width": 1280,
"screen_height": 1024,
"cpu_cores": 2,
"cpu_vendor": "SomeCpuVendor",
"cpu_speed": 3261,
"has_flash": True,
},
{
"browser_arch": "x86",
"os_name": "Windows_NT",
"os_version": "6.1",
"memory_mb": 4286,
"is_wow64": False,
"gfx0_vendor_id": "0xfeee",
"gfx0_device_id": "0xd1d1",
"screen_width": 1280,
"screen_height": 1024,
"cpu_cores": 2,
"cpu_vendor": "SomeCpuVendor",
"cpu_speed": 3261,
"has_flash": True,
},
{
"browser_arch": "x86-64",
"os_name": "Darwin",
"os_version": "15.3",
"memory_mb": 8322,
"is_wow64": False,
"gfx0_vendor_id": "0xfeee",
"gfx0_device_id": "0xd1d2",
"screen_width": 1320,
"screen_height": 798,
"cpu_cores": 4,
"cpu_vendor": "SomeCpuVendor",
"cpu_speed": 4211,
"has_flash": True,
},
]
spark = SparkSession.builder.appName(
"hardware_report_dashboard").getOrCreate()
# Create an rdd with the pepared data, then aggregate.
data_rdd = spark.sparkContext.parallelize(
[summarize_json.prepare_data(d, device_map) for d in raw_data])
agg_data = summarize_json.aggregate_data(data_rdd)
assert (
agg_data[("os_arch",
"x86")] == 2), "Two 'x86' OS architectures must be reported."
assert (agg_data[(
"os_arch",
"x86-64")] == 1), "One 'x86-64' OS architecture must be reported."
assert (agg_data[("has_flash", True)] == 3
), "All the entries had the flash plugin, so this must be 3."
def test_aggregate_data():
"""Test aggregate_data function on sample data."""
raw_data = [
{
'browser_arch': 'x86',
'os_name': 'Windows_NT',
'os_version': '6.1',
'memory_mb': 4286,
'is_wow64': False,
'gfx0_vendor_id': '0xfeee',
'gfx0_device_id': '0xd1d1',
'screen_width': 1280,
'screen_height': 1024,
'cpu_cores': 2,
'cpu_vendor': 'SomeCpuVendor',
'cpu_speed': 3261,
'has_flash': True
},
{
'browser_arch': 'x86',
'os_name': 'Windows_NT',
'os_version': '6.1',
'memory_mb': 4286,
'is_wow64': False,
'gfx0_vendor_id': '0xfeee',
'gfx0_device_id': '0xd1d1',
'screen_width': 1280,
'screen_height': 1024,
'cpu_cores': 2,
'cpu_vendor': 'SomeCpuVendor',
'cpu_speed': 3261,
'has_flash': True
},
{
'browser_arch': 'x86-64',
'os_name': 'Darwin',
'os_version': '15.3',
'memory_mb': 8322,
'is_wow64': False,
'gfx0_vendor_id': '0xfeee',
'gfx0_device_id': '0xd1d2',
'screen_width': 1320,
'screen_height': 798,
'cpu_cores': 4,
'cpu_vendor': 'SomeCpuVendor',
'cpu_speed': 4211,
'has_flash': True
},
]
spark = (SparkSession
.builder
.appName("hardware_report_dashboard")
.getOrCreate())
# Create an rdd with the pepared data, then aggregate.
data_rdd = spark.sparkContext.parallelize(
[summarize_json.prepare_data(d, device_map) for d in raw_data])
agg_data = summarize_json.aggregate_data(data_rdd)
assert agg_data[('os_arch', 'x86')] == 2, (
"Two 'x86' OS architectures must be reported.")
assert agg_data[('os_arch', 'x86-64')] == 1, (
"One 'x86-64' OS architecture must be reported.")
assert agg_data[('has_flash', True)] == 3, (
"All the entries had the flash plugin, so this must be 3.")