[ipfire.org.git] / www / webapp / backend / stasy.py

#!/usr/bin/python

from __future__ import division

import datetime
import hwdata
import logging
import pymongo
import re

from misc import Singleton

DATABASE_HOST = ["irma.ipfire.org", "madeye.ipfire.org"]
DATABASE_NAME = "stasy"

CPU_SPEED_CONSTRAINTS = (0, 500, 1000, 1500, 2000, 2500, 3000, 3500)
MEMORY_CONSTRAINTS = (0, 128, 256, 512, 1024, 2048, 4096, 8192, 16384)

CPU_STRINGS = (
	# AMD
	(r"(AMD Athlon).*(XP).*", r"\1 \2"),
	(r"(AMD Phenom).* ([0-9]+) .*", r"\1 \2"),
	(r"(AMD Phenom).*", r"\1"),
	(r"(AMD Sempron).*", r"\1"),
	(r"AMD Athlon.* II X2 ([a-z0-9]+).*", r"AMD Athlon X2 \1"),
	(r"(Geode).*", r"\1"),

	# Intel
	(r"Intel.*(Atom|Celeron).*CPU\s*([A-Z0-9]+) .*", r"Intel \1 \2"),
	(r"(Intel).*(Celeron).*", r"\1 \2"),
	(r"Intel.* Core.*2 Duo CPU .* ([A-Z0-9]+) .*", r"Intel C2D \1"),
	(r"Intel.* Core.*2 CPU .* ([A-Z0-9]+) .*", r"Intel C2 \1"),
	(r"Intel.* Core.*2 Quad CPU .* ([A-Z0-9]+) .*", r"Intel C2Q \1"),
	(r"Intel.* Xeon.* CPU .* ([A-Z0-9]+) .*", r"Intel Xeon \1"),
	(r"(Intel).*(Xeon).*", r"\1 \2"),
	(r"Intel.* Pentium.* (D|4) .*", r"Intel Pentium \1"),
	(r"Intel.* Pentium.* Dual .* ([A-Z0-9]+) .*", r"Intel Pentium Dual \1"),
	(r"Pentium.* Dual-Core .* ([A-Z0-9]+) .*", r"Intel Pentium Dual \1"),
	(r"(Pentium I{2,3}).*", r"Intel \1"),
	(r"(Celeron \(Coppermine\))", r"Intel Celeron"),

	# VIA
	(r"(VIA \w*).*", r"\1"),
)

class ProfileDict(object):
	def __init__(self, data):
		self._data = data


class ProfileCPU(ProfileDict):
	@property
	def arch(self):
		return self._data.get("arch")

	@property
	def vendor(self):
		return self._data.get("vendor")

	@property
	def speed(self):
		return self._data.get("speed")

	@property
	def friendly_speed(self):
		if self.speed < 1000:
			return "%dMHz" % self.speed

		return "%.1fGHz" % round(self.speed / 1000, 1)

	@property
	def bogomips(self):
		return self._data.get("bogomips")

	@property
	def model(self):
		return self._data.get("model")

	@property
	def model_string(self):
		return self._data.get("model_string")

	@property
	def flags(self):
		return self._data.get("flags")

	@property
	def count(self):
		return self._data.get("count")

	@property
	def capable_64bit(self):
		return "lm" in self.flags

	@property
	def capable_pae(self):
		return "pae" in self.flags

	@property
	def capable_virt(self):
		return "vmx" in self.flags or "svm" in self.flags

	@property
	def friendly_vendor(self):
		s = self.model_string
		for pattern, repl in CPU_STRINGS:
			if re.match(pattern, s) is None:
				continue
			return re.sub(pattern, repl, s)

		return s

	@property
	def friendly_string(self):
		return "%s @ %s" % (self.friendly_vendor, self.friendly_speed)


class ProfileHypervisor(ProfileDict):
	def __repr__(self):
		return "<%s %s-%s>" % (self.__class__.__name__, self.vendor, self.type)

	@property
	def vendor(self):
		return self._data.get("vendor")

	@property
	def type(self):
		return self._data.get("type")


class ProfileNetwork(ProfileDict):
	def has_zone(self, name):
		return self._data.get(name)

	@property
	def green(self):
		return self._data.get("green")

	@property
	def red(self):
		return self._data.get("red")

	@property
	def blue(self):
		return self._data.get("blue")

	@property
	def orange(self):
		return self._data.get("orange")


class ProfileDevice(ProfileDict):
	subsystem2class = {
		"pci" : hwdata.PCI,
		"usb" : hwdata.USB,
	}

	classid2name = {
		"pci" : {
			"00" : "Unclassified",
			"01" : "Mass storage",
			"02" : "Network",
			"03" : "Display",
			"04" : "Multimedia",
			"05" : "Memory controller",
			"06" : "Bridge",
			"07" : "Communication",
			"08" : "Generic system peripheral",
			"09" : "Input device",
			"0a" : "Docking station",
			"0b" : "Processor",
			"0c" : "Serial bus",
			"0d" : "Wireless",
			"0e" : "Intelligent controller",
			"0f" : "Satellite communications controller",
			"10" : "Encryption",
			"11" : "Signal processing controller",
			"ff" : "Unassigned class",
		},
		
		"usb" : {
			"00" : "Unclassified",
			"01" : "Multimedia",
			"02" : "Communication",
			"03" : "Input device",
			"05" : "Generic system peripheral",
			"06" : "Image",
			"07" : "Printer",
			"08" : "Mass storage",
			"09" : "Hub",
			"0a" : "Communication",
			"0b" : "Smart card",
			"0d" : "Encryption",
			"0e" : "Display",
			"0f" : "Personal Healthcare",
			"dc" : "Diagnostic Device",
			"e0" : "Wireless",
			"ef" : "Unclassified",
			"fe" : "Unclassified",
			"ff" : "Unclassified",
		}
	}

	def __cmp__(self, other):
		return cmp(self.vendor, other.vendor) or \
			cmp(self.model, other.model) or \
			cmp(self.driver, other.driver)

	@property
	def model(self):
		return self._data.get("model")

	@property
	def model_string(self):
		cls = self.subsystem2class[self.subsystem]

		return cls().get_device(self.vendor, self.model)

	@property
	def subsystem(self):
		return self._data.get("subsystem")

	@property
	def vendor(self):
		return self._data.get("vendor")

	@property
	def vendor_string(self):
		cls = self.subsystem2class[self.subsystem]

		return cls().get_vendor(self.vendor)

	@property
	def driver(self):
		return self._data.get("driver")

	@property
	def cls(self):
		classid = self._data.get("deviceclass")

		if self.subsystem == "pci":
			classid = classid[:-4]
			if len(classid) == 1:
				classid = "0%s" % classid

		elif self.subsystem == "usb" and classid:
			classid = classid.split("/")[0]
			classid = "%02x" % int(classid)

		try:
			return self.classid2name[self.subsystem][classid]
		except KeyError:
			return "N/A"


class Profile(ProfileDict):
	def __init__(self, profile_blob):
		ProfileDict.__init__(self, profile_blob.get("profile"))

		self.public_id = profile_blob.get("public_id")
		self.updated = profile_blob.get("updated")
		self._geoip = profile_blob.get("geoip", None)

	def __repr__(self):
		return "<%s %s>" % (self.__class__.__name__, self.public_id)

	@property
	def cpu(self):
		return ProfileCPU(self._data.get("cpu"))

	@property
	def devices(self):
		devices = []
		for d in self._data.get("devices"):
			d = ProfileDevice(d)

			if d.driver in ("pcieport", "usb", "hub"):
				continue

			devices.append(d)

		return devices

	@property
	def hypervisor(self):
		if self.virtual:
			return ProfileHypervisor(self._data.get("hypervisor"))

	@property
	def virtual(self):
		return self.system.get("virtual")

	@property
	def system(self):
		return self._data.get("system")

	@property
	def release(self):
		return self.system.get("release")

	@property
	def kernel(self):
		return self.system.get("kernel_release")

	@property
	def memory(self):
		return self.system.get("memory")

	@property
	def friendly_memory(self):
		units = ("k", "M", "G", "T")

		mem = self.memory
		i = 0
		while mem >= 1024:
			mem /= 1024
			i += 1

		return "%d%s" % (round(mem, 0), units[i])

	@property
	def root_size(self):
		return self.system.get("root_size") or 0

	@property
	def friendly_root_size(self):
		units = ("k", "M", "G", "T")

		size = self.root_size
		if not size:
			return

		i = 0
		while size >= 1024:
			size /= 1024
			i += 1

		return "%d%s" % (round(size, 0), units[i])

	@property
	def vendor(self):
		return self.system.get("vendor")

	@property
	def model(self):
		return self.system.get("model")

	@property
	def country_code(self):
		if self._geoip:
			return self._geoip["country_code"].lower()

		return "unknown"

	@property
	def network(self):
		network = self._data.get("network", None)
		if network:
			return ProfileNetwork(network)


class Stasy(object):
	__metaclass__ = Singleton

	def __init__(self):
		# Initialize database connection
		self._conn = pymongo.Connection(DATABASE_HOST)
		self._db = self._conn[DATABASE_NAME]

	def get_profile_count(self):
		# XXX need to implement something to get profiles updated since
		# a given date

		# All distinct profiles (based on public_id)
		# XXX possibly bad performance
		return len(self._db.profiles.distinct("public_id"))

	def get_archives_count(self):
		return self._db.archives.count()

	#def _get_profile_cursor(self, public_id):
	#	c = self._db.profiles.find({ "public_id" : public_id })
	#	c.sort("updated", pymongo.ASCENDING)
	#
	#	return c

	def profile_exists(self, public_id):
		return self.query({ "public_id" : public_id }).count() >= 1

	def get_profile(self, public_id):
		p = None
		# XXX should only find one object in the end
		for p in self.query({ "public_id" : public_id }):
			p = Profile(p)

		return p

	def get_profiles(self):
		# XXX needs nicer database query
		profiles = []
		for p in self._db.profiles.find():
			if not p.get("public_id") in profiles:
				profiles.append(p.get("public_id"))

		return profiles

	def query(self, query, archives=False, no_virt=False, all=False):
		db = self._db.profiles

		if archives:
			db = self.db.archives

		if not all:
			# XXX cannot use the index?
			query.update({ "profile" : { "$exists" : True }})

		if no_virt:
			query.update({ "profile.system.virtual" : False })

		logging.debug("Executing query: %s" % query)

		return db.find(query)

	@property
	def secret_ids(self):
		return self._db.profiles.distinct("secret_id")

	@property
	def cpus(self):
		return self.query({}, no_virt=True).distinct("profile.cpu")

	@property
	def cpu_vendors(self):
		return self.query({}, no_virt=True).distinct("profile.cpu.vendor")

	@property
	def cpu_vendors_map(self):
		cpus = {}

		for vendor in self.cpu_vendors:
			cpus[vendor] = \
				self.query({
					"profile.cpu.vendor" : vendor
				}, no_virt=True).count()

		return cpus

	@property
	def cpu_speed_average(self):
		speed = 0

		all = self.query({}, no_virt=True)

		# XXX ugly. needs to be done by group()
		for m in all:
			if not m.has_key("profile"):
				continue
			speed += m.get("profile").get("cpu").get("speed")

		return (speed / all.count())

	@property
	def cpu_speed_map(self):
		cpu_speeds = {}

		for i in range(len(CPU_SPEED_CONSTRAINTS) - 1):
			min, max = CPU_SPEED_CONSTRAINTS[i:i+2]

			cpu_speeds[min, max] = \
				self.query({
					"profile.cpu.speed" : {
						"$gte" : min, "$lt" : max
					}
				}, no_virt=True).count()

		return cpu_speeds

	def get_memory_map(self):
		memory = {}

		for i in range(len(MEMORY_CONSTRAINTS) - 1):
			min, max = MEMORY_CONSTRAINTS[i:i+2]

			memory[min, max] = \
				self.query(
					{ "profile.system.memory" : {
						"$gte" : min * 1024, "$lt" : max * 1024
					}
				}, no_virt=True).count()

		return memory

	@property
	def memory_average(self):
		memory = 0

		all = self.query({}, no_virt=True)

		# XXX ugly. needs to be done by group()
		for m in all:
			if not m.has_key("profile"):
				continue
			memory += int(m.get("profile").get("system").get("memory"))

		return (memory / all.count()) / 1024

	@property
	def hypervisor_vendors(self):
		return self.query({}).distinct("profile.hypervisor.vendor")

	@property
	def hypervisor_map(self):
		hypervisors = {}

		for hypervisor in self.hypervisor_vendors:
			hypervisors[hypervisor] = \
				self.query({
					"profile.hypervisor.vendor" : hypervisor
				}).count()

		return hypervisors

	@property
	def hypervisor_models(self):
		return self.query({}).distinct("profile.hypervisor.model")

	@property
	def virtual_map(self):
		virtual = {
			True: None,
			False: None,
		}

		for k in virtual.keys():
			virtual[k] = \
				self.query({ "profile.system.virtual": k }).count()

		return virtual

	@property
	def languages(self):
		return self.query({}).distinct("profile.system.language")

	def get_language_map(self):
		languages = {}

		for language in self.languages:
			languages[language] = \
				self.query({
					"profile.system.language" : language
				}).count()

		return languages

	@property
	def vendors(self):
		return self.query({}).distinct("profile.system.vendor")

	@property
	def vendor_map(self):
		vendors = {}

		for vendor in self.vendors:
			vendors[vendor] = \
				self.query({
					"profile.system.vendor" : vendor
				}).count()

		return vendors

	@property
	def models(self):
		return self.query({}).distinct("profile.system.model")

	@property
	def model_map(self):
		models = {}

		for model in self.models:
			models[model] = \
				self.query({
					"profile.system.model" : model
				}).count()

		return models

	@property
	def arches(self):
		return self.query({}).distinct("profile.cpu.arch")

	@property
	def arch_map(self):
		arches = {}

		for arch in self.arches:
			arches[arch] = \
				self.query({
					"profile.cpu.arch" : arch
				}).count()

		return arches

	@property
	def kernels(self):
		return self.query({}).distinct("profile.system.kernel_release")

	@property
	def kernel_map(self):
		kernels = {}

		for kernel in self.kernels:
			kernels[kernel] = \
				self.query({
					"profile.system.kernel_release" : kernel
				}).count()

		return kernels

	@property
	def releases(self):
		return self.query({}).distinct("profile.system.release")

	@property
	def release_map(self):
		releases = {}

		for release in self.releases:
			releases[release] = \
				self.query({
					"profile.system.release" : release
				}).count()

		return releases

	def get_device_percentage(self, bus, vendor_id, model_id):
		profiles_with_device = self.query({
			"profile.devices.subsystem" : bus,
			"profile.devices.vendor" : vendor_id,
			"profile.devices.model" : model_id,
		})

		profiles_all = self.query({})

		if not profiles_all.count():
			return 0

		return profiles_with_device.count() / profiles_all.count()

	def get_cpu_flag_map(self, flags):
		# XXX needs a cleanup

		_flags = { True : 0 }

		if type(flags) == type("a"):
			flags = [flags]

		for flag in flags:
			_flags[True] += \
				self.query({
					"profile.cpu.flags" : flag,
				}, no_virt=True).count()

		_flags[False] = self.query({}, no_virt=True).count() - _flags[True]

		return _flags

	@property
	def geo_locations(self):
		return [code.lower() for code in self.query({}, all=True).distinct("geoip.country_code")]

	def get_geo_location_map(self):
		geo_locations = {}

		count = 0
		for geo_location in self.geo_locations:
			geo_locations[geo_location] = \
				self.query({
					"geoip.country_code" : geo_location.upper()
				}, all=True).count()

			count += geo_locations[geo_location]

		for geo_location in geo_locations.keys():
			geo_locations[geo_location] /= count

		return geo_locations

	def get_models_by_vendor(self, subsystem, vendor_id):
		devices = []

		profiles_all = self.query({})

		for profile in profiles_all:
			if not profile.has_key("profile"):
				continue

			profile = Profile(profile)

			for device in profile.devices:
				if not device.vendor == vendor_id:
					continue

				if not device in devices:
					devices.append(device)

		return devices

	def get_network_zones_map(self):
		zones = { "green" : 0, "blue" : 0, "orange" : 0, "red" : 0 }

		all = self.query({ "profile.network" : { "$exists" : True }})

		for zone in zones.keys():
			zones[zone] = self.query({
				"profile.network.%s" % zone : True,
			}).count() / all.count()

		return zones

	def get_profile_ratio(self):
		return (self.query({}).count(), self.get_profile_count())

	def get_updated_since(self, since, _query={}):
		since = datetime.datetime.utcnow() - datetime.timedelta(**since)

		query = { "updated" : { "$gte" : since }}
		query.update(_query)

		return self.query(query)

	def get_updates_by_release_since(self, since):
		updates = {}

		for release in self.releases:
			updates[release] = self.get_updated_since(since,
				{ "profile.system.release" : release }).count()

		return updates


if __name__ == "__main__":
	s = Stasy()
Commit	Line	Data
372efc19 MT	1	#!/usr/bin/python
372efc19 MT	2
91a446f0 MT	3	from __future__ import division
91a446f0 MT	4
cbb9726a	5	import datetime
372efc19 MT	6	import hwdata
	7	import logging
	8	import pymongo
abc3bb31	9	import re
372efc19 MT	10
	11	from misc import Singleton
	12
	13	DATABASE_HOST = ["irma.ipfire.org", "madeye.ipfire.org"]
	14	DATABASE_NAME = "stasy"
	15
91a446f0	16	CPU_SPEED_CONSTRAINTS = (0, 500, 1000, 1500, 2000, 2500, 3000, 3500)
8e413e4f	17	MEMORY_CONSTRAINTS = (0, 128, 256, 512, 1024, 2048, 4096, 8192, 16384)
372efc19	18
abc3bb31 MT	19	CPU_STRINGS = (
	20	# AMD
	21	(r"(AMD Athlon).(XP).", r"\1 \2"),
	22	(r"(AMD Phenom).* ([0-9]+) .*", r"\1 \2"),
	23	(r"(AMD Phenom).*", r"\1"),
	24	(r"(AMD Sempron).*", r"\1"),
	25	(r"AMD Athlon.* II X2 ([a-z0-9]+).*", r"AMD Athlon X2 \1"),
	26	(r"(Geode).*", r"\1"),
	27
	28	# Intel
	29	(r"Intel.(Atom\|Celeron).CPU\s([A-Z0-9]+) .", r"Intel \1 \2"),
	30	(r"(Intel).(Celeron).", r"\1 \2"),
	31	(r"Intel.* Core.2 Duo CPU . ([A-Z0-9]+) .*", r"Intel C2D \1"),
	32	(r"Intel.* Core.2 CPU . ([A-Z0-9]+) .*", r"Intel C2 \1"),
	33	(r"Intel.* Core.2 Quad CPU . ([A-Z0-9]+) .*", r"Intel C2Q \1"),
	34	(r"Intel.* Xeon.* CPU .* ([A-Z0-9]+) .*", r"Intel Xeon \1"),
	35	(r"(Intel).(Xeon).", r"\1 \2"),
	36	(r"Intel.* Pentium.* (D\|4) .*", r"Intel Pentium \1"),
	37	(r"Intel.* Pentium.* Dual .* ([A-Z0-9]+) .*", r"Intel Pentium Dual \1"),
	38	(r"Pentium.* Dual-Core .* ([A-Z0-9]+) .*", r"Intel Pentium Dual \1"),
	39	(r"(Pentium I{2,3}).*", r"Intel \1"),
	40	(r"(Celeron \(Coppermine\))", r"Intel Celeron"),
	41
	42	# VIA
	43	(r"(VIA \w).", r"\1"),
	44	)
	45
372efc19 MT	46	class ProfileDict(object):
	47	def __init__(self, data):
	48	self._data = data
	49
	50
	51	class ProfileCPU(ProfileDict):
	52	@property
	53	def arch(self):
	54	return self._data.get("arch")
	55
	56	@property
	57	def vendor(self):
	58	return self._data.get("vendor")
	59
91a446f0 MT	60	@property
	61	def speed(self):
	62	return self._data.get("speed")
	63
abc3bb31 MT	64	@property
	65	def friendly_speed(self):
	66	if self.speed < 1000:
	67	return "%dMHz" % self.speed
	68
	69	return "%.1fGHz" % round(self.speed / 1000, 1)
	70
91a446f0 MT	71	@property
	72	def bogomips(self):
	73	return self._data.get("bogomips")
	74
372efc19 MT	75	@property
	76	def model(self):
	77	return self._data.get("model")
	78
	79	@property
	80	def model_string(self):
	81	return self._data.get("model_string")
	82
	83	@property
	84	def flags(self):
	85	return self._data.get("flags")
	86
91a446f0 MT	87	@property
	88	def count(self):
	89	return self._data.get("count")
	90
372efc19 MT	91	@property
	92	def capable_64bit(self):
	93	return "lm" in self.flags
	94
	95	@property
	96	def capable_pae(self):
	97	return "pae" in self.flags
	98
	99	@property
	100	def capable_virt(self):
	101	return "vmx" in self.flags or "svm" in self.flags
	102
abc3bb31 MT	103	@property
	104	def friendly_vendor(self):
	105	s = self.model_string
	106	for pattern, repl in CPU_STRINGS:
	107	if re.match(pattern, s) is None:
	108	continue
	109	return re.sub(pattern, repl, s)
	110
	111	return s
	112
	113	@property
	114	def friendly_string(self):
	115	return "%s @ %s" % (self.friendly_vendor, self.friendly_speed)
	116
372efc19 MT	117
	118	class ProfileHypervisor(ProfileDict):
	119	def __repr__(self):
	120	return "<%s %s-%s>" % (self.__class__.__name__, self.vendor, self.type)
	121
	122	@property
	123	def vendor(self):
	124	return self._data.get("vendor")
	125
372efc19 MT	126	@property
	127	def type(self):
	128	return self._data.get("type")
	129
	130
07785a9c MT	131	class ProfileNetwork(ProfileDict):
	132	def has_zone(self, name):
	133	return self._data.get(name)
	134
	135	@property
	136	def green(self):
	137	return self._data.get("green")
	138
	139	@property
	140	def red(self):
	141	return self._data.get("red")
	142
	143	@property
	144	def blue(self):
	145	return self._data.get("blue")
	146
	147	@property
	148	def orange(self):
	149	return self._data.get("orange")
	150
	151
372efc19 MT	152	class ProfileDevice(ProfileDict):
	153	subsystem2class = {
	154	"pci" : hwdata.PCI,
	155	"usb" : hwdata.USB,
	156	}
	157
91a446f0 MT	158	classid2name = {
91a446f0 MT	159	"pci" : {
bc19cabd	160	"00" : "Unclassified",
91a446f0 MT	161	"01" : "Mass storage",
	162	"02" : "Network",
	163	"03" : "Display",
	164	"04" : "Multimedia",
	165	"05" : "Memory controller",
	166	"06" : "Bridge",
	167	"07" : "Communication",
	168	"08" : "Generic system peripheral",
	169	"09" : "Input device",
	170	"0a" : "Docking station",
	171	"0b" : "Processor",
	172	"0c" : "Serial bus",
	173	"0d" : "Wireless",
	174	"0e" : "Intelligent controller",
	175	"0f" : "Satellite communications controller",
bc19cabd	176	"10" : "Encryption",
91a446f0 MT	177	"11" : "Signal processing controller",
	178	"ff" : "Unassigned class",
	179	},
	180
	181	"usb" : {
bc19cabd	182	"00" : "Unclassified",
91a446f0 MT	183	"01" : "Multimedia",
	184	"02" : "Communication",
	185	"03" : "Input device",
bc19cabd MT	186	"05" : "Generic system peripheral",
bc19cabd MT	187	"06" : "Image",
91a446f0 MT	188	"07" : "Printer",
	189	"08" : "Mass storage",
	190	"09" : "Hub",
bc19cabd MT	191	"0a" : "Communication",
	192	"0b" : "Smart card",
	193	"0d" : "Encryption",
91a446f0	194	"0e" : "Display",
bc19cabd	195	"0f" : "Personal Healthcare",
91a446f0 MT	196	"dc" : "Diagnostic Device",
91a446f0 MT	197	"e0" : "Wireless",
bc19cabd MT	198	"ef" : "Unclassified",
	199	"fe" : "Unclassified",
	200	"ff" : "Unclassified",
91a446f0 MT	201	}
	202	}
	203
	204	def __cmp__(self, other):
	205	return cmp(self.vendor, other.vendor) or \
	206	cmp(self.model, other.model) or \
	207	cmp(self.driver, other.driver)
	208
372efc19 MT	209	@property
	210	def model(self):
	211	return self._data.get("model")
	212
	213	@property
	214	def model_string(self):
	215	cls = self.subsystem2class[self.subsystem]
	216
	217	return cls().get_device(self.vendor, self.model)
	218
	219	@property
	220	def subsystem(self):
	221	return self._data.get("subsystem")
	222
	223	@property
	224	def vendor(self):
	225	return self._data.get("vendor")
	226
	227	@property
	228	def vendor_string(self):
	229	cls = self.subsystem2class[self.subsystem]
	230
	231	return cls().get_vendor(self.vendor)
	232
	233	@property
	234	def driver(self):
	235	return self._data.get("driver")
	236
91a446f0 MT	237	@property
	238	def cls(self):
	239	classid = self._data.get("deviceclass")
	240
	241	if self.subsystem == "pci":
	242	classid = classid[:-4]
	243	if len(classid) == 1:
	244	classid = "0%s" % classid
	245
	246	elif self.subsystem == "usb" and classid:
	247	classid = classid.split("/")[0]
	248	classid = "%02x" % int(classid)
	249
	250	try:
	251	return self.classid2name[self.subsystem][classid]
	252	except KeyError:
	253	return "N/A"
	254
372efc19 MT	255
372efc19 MT	256	class Profile(ProfileDict):
91a446f0 MT	257	def __init__(self, profile_blob):
91a446f0 MT	258	ProfileDict.__init__(self, profile_blob.get("profile"))
372efc19	259
91a446f0 MT	260	self.public_id = profile_blob.get("public_id")
	261	self.updated = profile_blob.get("updated")
	262	self._geoip = profile_blob.get("geoip", None)
372efc19 MT	263
	264	def __repr__(self):
	265	return "<%s %s>" % (self.__class__.__name__, self.public_id)
	266
	267	@property
	268	def cpu(self):
	269	return ProfileCPU(self._data.get("cpu"))
	270
	271	@property
	272	def devices(self):
	273	devices = []
	274	for d in self._data.get("devices"):
	275	d = ProfileDevice(d)
	276
91a446f0	277	if d.driver in ("pcieport", "usb", "hub"):
372efc19 MT	278	continue
	279
	280	devices.append(d)
	281
	282	return devices
	283
	284	@property
	285	def hypervisor(self):
	286	if self.virtual:
	287	return ProfileHypervisor(self._data.get("hypervisor"))
	288
	289	@property
	290	def virtual(self):
	291	return self.system.get("virtual")
	292
	293	@property
	294	def system(self):
	295	return self._data.get("system")
	296
	297	@property
	298	def release(self):
	299	return self.system.get("release")
	300
	301	@property
	302	def kernel(self):
	303	return self.system.get("kernel_release")
	304
	305	@property
	306	def memory(self):
	307	return self.system.get("memory")
	308
abc3bb31 MT	309	@property
	310	def friendly_memory(self):
	311	units = ("k", "M", "G", "T")
	312
	313	mem = self.memory
	314	i = 0
	315	while mem >= 1024:
	316	mem /= 1024
	317	i += 1
	318
	319	return "%d%s" % (round(mem, 0), units[i])
	320
372efc19 MT	321	@property
372efc19 MT	322	def root_size(self):
91a446f0 MT	323	return self.system.get("root_size") or 0
91a446f0 MT	324
abc3bb31 MT	325	@property
	326	def friendly_root_size(self):
	327	units = ("k", "M", "G", "T")
	328
	329	size = self.root_size
	330	if not size:
	331	return
	332
	333	i = 0
	334	while size >= 1024:
	335	size /= 1024
	336	i += 1
	337
	338	return "%d%s" % (round(size, 0), units[i])
	339
91a446f0 MT	340	@property
	341	def vendor(self):
	342	return self.system.get("vendor")
	343
	344	@property
	345	def model(self):
	346	return self.system.get("model")
	347
	348	@property
	349	def country_code(self):
	350	if self._geoip:
	351	return self._geoip["country_code"].lower()
372efc19	352
91a446f0	353	return "unknown"
372efc19	354
07785a9c MT	355	@property
	356	def network(self):
	357	network = self._data.get("network", None)
	358	if network:
	359	return ProfileNetwork(network)
	360
372efc19 MT	361
	362	class Stasy(object):
	363	__metaclass__ = Singleton
	364
	365	def __init__(self):
	366	# Initialize database connection
	367	self._conn = pymongo.Connection(DATABASE_HOST)
	368	self._db = self._conn[DATABASE_NAME]
	369
	370	def get_profile_count(self):
	371	# XXX need to implement something to get profiles updated since
	372	# a given date
	373
	374	# All distinct profiles (based on public_id)
cbb9726a MT	375	# XXX possibly bad performance
	376	return len(self._db.profiles.distinct("public_id"))
	377
	378	def get_archives_count(self):
	379	return self._db.archives.count()
372efc19 MT	380
	381	#def _get_profile_cursor(self, public_id):
	382	# c = self._db.profiles.find({ "public_id" : public_id })
	383	# c.sort("updated", pymongo.ASCENDING)
	384	#
	385	# return c
	386
91a446f0	387	def profile_exists(self, public_id):
910be037	388	return self.query({ "public_id" : public_id }).count() >= 1
91a446f0	389
372efc19	390	def get_profile(self, public_id):
54af860e	391	p = None
372efc19	392	# XXX should only find one object in the end
910be037	393	for p in self.query({ "public_id" : public_id }):
91a446f0	394	p = Profile(p)
372efc19 MT	395
	396	return p
	397
	398	def get_profiles(self):
	399	# XXX needs nicer database query
	400	profiles = []
	401	for p in self._db.profiles.find():
	402	if not p.get("public_id") in profiles:
	403	profiles.append(p.get("public_id"))
	404
	405	return profiles
	406
910be037 MT	407	def query(self, query, archives=False, no_virt=False, all=False):
	408	db = self._db.profiles
	409
	410	if archives:
	411	db = self.db.archives
	412
	413	if not all:
	414	# XXX cannot use the index?
	415	query.update({ "profile" : { "$exists" : True }})
	416
	417	if no_virt:
	418	query.update({ "profile.system.virtual" : False })
	419
	420	logging.debug("Executing query: %s" % query)
	421
	422	return db.find(query)
	423
372efc19 MT	424	@property
	425	def secret_ids(self):
	426	return self._db.profiles.distinct("secret_id")
	427
	428	@property
	429	def cpus(self):
910be037	430	return self.query({}, no_virt=True).distinct("profile.cpu")
372efc19 MT	431
	432	@property
	433	def cpu_vendors(self):
910be037	434	return self.query({}, no_virt=True).distinct("profile.cpu.vendor")
372efc19 MT	435
372efc19 MT	436	@property
91a446f0	437	def cpu_vendors_map(self):
372efc19 MT	438	cpus = {}
	439
	440	for vendor in self.cpu_vendors:
	441	cpus[vendor] = \
910be037	442	self.query({
372efc19	443	"profile.cpu.vendor" : vendor
910be037	444	}, no_virt=True).count()
372efc19 MT	445
	446	return cpus
	447
	448	@property
91a446f0 MT	449	def cpu_speed_average(self):
	450	speed = 0
	451
910be037	452	all = self.query({}, no_virt=True)
91a446f0 MT	453
	454	# XXX ugly. needs to be done by group()
	455	for m in all:
	456	if not m.has_key("profile"):
	457	continue
	458	speed += m.get("profile").get("cpu").get("speed")
	459
	460	return (speed / all.count())
	461
	462	@property
	463	def cpu_speed_map(self):
	464	cpu_speeds = {}
	465
	466	for i in range(len(CPU_SPEED_CONSTRAINTS) - 1):
	467	min, max = CPU_SPEED_CONSTRAINTS[i:i+2]
	468
	469	cpu_speeds[min, max] = \
910be037 MT	470	self.query({
910be037 MT	471	"profile.cpu.speed" : {
91a446f0 MT	472	"$gte" : min, "$lt" : max
91a446f0 MT	473	}
910be037	474	}, no_virt=True).count()
91a446f0 MT	475
	476	return cpu_speeds
	477
	478	def get_memory_map(self):
372efc19 MT	479	memory = {}
	480
	481	for i in range(len(MEMORY_CONSTRAINTS) - 1):
	482	min, max = MEMORY_CONSTRAINTS[i:i+2]
	483
	484	memory[min, max] = \
910be037	485	self.query(
372efc19 MT	486	{ "profile.system.memory" : {
	487	"$gte" : min * 1024, "$lt" : max * 1024
	488	}
910be037	489	}, no_virt=True).count()
372efc19 MT	490
	491	return memory
	492
	493	@property
	494	def memory_average(self):
	495	memory = 0
	496
910be037	497	all = self.query({}, no_virt=True)
372efc19 MT	498
	499	# XXX ugly. needs to be done by group()
	500	for m in all:
	501	if not m.has_key("profile"):
	502	continue
	503	memory += int(m.get("profile").get("system").get("memory"))
	504
	505	return (memory / all.count()) / 1024
	506
	507	@property
	508	def hypervisor_vendors(self):
910be037	509	return self.query({}).distinct("profile.hypervisor.vendor")
372efc19 MT	510
	511	@property
	512	def hypervisor_map(self):
	513	hypervisors = {}
	514
	515	for hypervisor in self.hypervisor_vendors:
	516	hypervisors[hypervisor] = \
910be037	517	self.query({
372efc19 MT	518	"profile.hypervisor.vendor" : hypervisor
	519	}).count()
	520
	521	return hypervisors
	522
	523	@property
	524	def hypervisor_models(self):
910be037	525	return self.query({}).distinct("profile.hypervisor.model")
372efc19 MT	526
	527	@property
	528	def virtual_map(self):
	529	virtual = {
	530	True: None,
	531	False: None,
	532	}
	533
	534	for k in virtual.keys():
	535	virtual[k] = \
910be037	536	self.query({ "profile.system.virtual": k }).count()
372efc19 MT	537
	538	return virtual
	539
	540	@property
	541	def languages(self):
910be037	542	return self.query({}).distinct("profile.system.language")
372efc19	543
91a446f0 MT	544	def get_language_map(self):
	545	languages = {}
	546
	547	for language in self.languages:
	548	languages[language] = \
910be037	549	self.query({
91a446f0 MT	550	"profile.system.language" : language
	551	}).count()
	552
	553	return languages
	554
372efc19 MT	555	@property
372efc19 MT	556	def vendors(self):
910be037	557	return self.query({}).distinct("profile.system.vendor")
372efc19 MT	558
	559	@property
	560	def vendor_map(self):
	561	vendors = {}
	562
	563	for vendor in self.vendors:
	564	vendors[vendor] = \
910be037	565	self.query({
372efc19 MT	566	"profile.system.vendor" : vendor
	567	}).count()
	568
	569	return vendors
	570
	571	@property
	572	def models(self):
910be037	573	return self.query({}).distinct("profile.system.model")
372efc19	574
91a446f0 MT	575	@property
	576	def model_map(self):
	577	models = {}
	578
	579	for model in self.models:
	580	models[model] = \
910be037	581	self.query({
91a446f0 MT	582	"profile.system.model" : model
	583	}).count()
	584
	585	return models
	586
	587	@property
	588	def arches(self):
910be037	589	return self.query({}).distinct("profile.cpu.arch")
91a446f0 MT	590
	591	@property
	592	def arch_map(self):
	593	arches = {}
	594
	595	for arch in self.arches:
	596	arches[arch] = \
910be037	597	self.query({
91a446f0 MT	598	"profile.cpu.arch" : arch
	599	}).count()
	600
	601	return arches
	602
	603	@property
	604	def kernels(self):
910be037	605	return self.query({}).distinct("profile.system.kernel_release")
91a446f0 MT	606
	607	@property
	608	def kernel_map(self):
	609	kernels = {}
	610
	611	for kernel in self.kernels:
	612	kernels[kernel] = \
910be037	613	self.query({
91a446f0 MT	614	"profile.system.kernel_release" : kernel
	615	}).count()
	616
	617	return kernels
	618
	619	@property
	620	def releases(self):
910be037	621	return self.query({}).distinct("profile.system.release")
91a446f0 MT	622
	623	@property
	624	def release_map(self):
	625	releases = {}
	626
	627	for release in self.releases:
	628	releases[release] = \
910be037	629	self.query({
91a446f0 MT	630	"profile.system.release" : release
	631	}).count()
	632
	633	return releases
	634
	635	def get_device_percentage(self, bus, vendor_id, model_id):
910be037	636	profiles_with_device = self.query({
91a446f0 MT	637	"profile.devices.subsystem" : bus,
	638	"profile.devices.vendor" : vendor_id,
	639	"profile.devices.model" : model_id,
	640	})
	641
910be037	642	profiles_all = self.query({})
91a446f0 MT	643
	644	if not profiles_all.count():
	645	return 0
	646
	647	return profiles_with_device.count() / profiles_all.count()
	648
910be037 MT	649	def get_cpu_flag_map(self, flags):
910be037 MT	650	# XXX needs a cleanup
91a446f0	651
910be037	652	_flags = { True : 0 }
91a446f0	653
910be037 MT	654	if type(flags) == type("a"):
	655	flags = [flags]
	656
	657	for flag in flags:
	658	_flags[True] += \
	659	self.query({
	660	"profile.cpu.flags" : flag,
	661	}, no_virt=True).count()
91a446f0	662
910be037	663	_flags[False] = self.query({}, no_virt=True).count() - _flags[True]
91a446f0	664
910be037	665	return _flags
91a446f0 MT	666
	667	@property
	668	def geo_locations(self):
5914104c	669	return [code.lower() for code in self.query({}, all=True).distinct("geoip.country_code")]
91a446f0 MT	670
	671	def get_geo_location_map(self):
	672	geo_locations = {}
	673
	674	count = 0
	675	for geo_location in self.geo_locations:
	676	geo_locations[geo_location] = \
910be037	677	self.query({
bc1a5c5e	678	"geoip.country_code" : geo_location.upper()
5914104c	679	}, all=True).count()
91a446f0 MT	680
	681	count += geo_locations[geo_location]
	682
638e9782 MT	683	for geo_location in geo_locations.keys():
	684	geo_locations[geo_location] /= count
	685
91a446f0 MT	686	return geo_locations
	687
	688	def get_models_by_vendor(self, subsystem, vendor_id):
	689	devices = []
	690
910be037	691	profiles_all = self.query({})
91a446f0 MT	692
	693	for profile in profiles_all:
	694	if not profile.has_key("profile"):
	695	continue
	696
	697	profile = Profile(profile)
	698
	699	for device in profile.devices:
	700	if not device.vendor == vendor_id:
	701	continue
	702
	703	if not device in devices:
	704	devices.append(device)
	705
	706	return devices
372efc19	707
07785a9c MT	708	def get_network_zones_map(self):
	709	zones = { "green" : 0, "blue" : 0, "orange" : 0, "red" : 0 }
	710
e6cff1b1 MT	711	all = self.query({ "profile.network" : { "$exists" : True }})
e6cff1b1 MT	712
07785a9c MT	713	for zone in zones.keys():
	714	zones[zone] = self.query({
	715	"profile.network.%s" % zone : True,
e6cff1b1	716	}).count() / all.count()
07785a9c MT	717
	718	return zones
	719
cbb9726a MT	720	def get_profile_ratio(self):
	721	return (self.query({}).count(), self.get_profile_count())
	722
	723	def get_updated_since(self, since, _query={}):
	724	since = datetime.datetime.utcnow() - datetime.timedelta(**since)
	725
	726	query = { "updated" : { "$gte" : since }}
	727	query.update(_query)
	728
	729	return self.query(query)
	730
	731	def get_updates_by_release_since(self, since):
	732	updates = {}
	733
	734	for release in self.releases:
	735	updates[release] = self.get_updated_since(since,
	736	{ "profile.system.release" : release }).count()
	737
	738	return updates
	739
372efc19 MT	740
	741	if __name__ == "__main__":
	742	s = Stasy()
	743