maloja/maloja/utilities.py

import re
import os
import hashlib
from threading import Thread, Timer
import pickle
import json
import urllib
import datetime
import random
import itertools
import base64
from doreah import settings
from doreah import caching
from doreah.logging import log
from doreah.regular import yearly, daily

#from .external import api_request_track, api_request_artist
from . import thirdparty
from .__pkginfo__ import version
from . import globalconf
from .globalconf import datadir



#####
## SERIALIZE
#####

def serialize(obj):
	try:
		return serialize(obj.hashable())
	except:
		try:
			return json.dumps(obj)
		except:
			if isinstance(obj,list) or isinstance(obj,tuple):
				return "[" + ",".join(serialize(o) for o in obj) + "]"
			elif isinstance(obj,dict):
				return "{" + ",".join(serialize(o) + ":" + serialize(obj[o]) for o in obj) + "}"
			return json.dumps(obj.hashable())


	#if isinstance(obj,list) or if isinstance(obj,tuple):
	#	return "[" + ",".join(dumps(o) for o in obj) + "]"
	#if isinstance(obj,str)







#####
## RULESTATE VALIDATION
#####





def checksumTSV(folder):

	sums = ""

	for f in os.listdir(folder + "/"):
		if (f.endswith(".tsv")):
			f = open(folder + "/" + f,"rb")
			sums += hashlib.md5(f.read()).hexdigest() + "\n"
			f.close()

	return sums

# returns whether checksums match and sets the checksum to invalid if they don't (or sets the new one if no previous one exists)
def combineChecksums(filename,checksums):
	import os

	if os.path.exists(filename + ".rulestate"):
		f = open(filename + ".rulestate","r")
		oldchecksums = f.read()
		f.close()
		if oldchecksums == checksums:
		# the new checksum given by the calling db server represents the rule state that all current unsaved scrobbles were created under
		# if this is the same as the existing one, we're all good
			return True
		elif (oldchecksums != "INVALID"):
			#if not, the file is not consistent to any single rule state (some scrobbles were created with an old ruleset, some not)
			f = open(filename + ".rulestate","w")
			f.write("INVALID") # this will never match any sha256sum
			f.close()
			return False
		else:
			#if the file already says invalid, no need to open it and rewrite
			return False
	else:
		f = open(filename + ".rulestate","w")
		f.write(checksums)
		f.close()
		return True

# checks ALL files for their rule state. if they are all the same as the current loaded one, the entire database can be assumed to be consistent with the current ruleset
# in any other case, get out
def consistentRulestate(folder,checksums):

	result = []
	for scrobblefile in os.listdir(folder + "/"):

		if (scrobblefile.endswith(".tsv")):

			try:
				with open(folder + "/" + scrobblefile + ".rulestate","r") as f:
					if f.read() != checksums:
						return False
			except:
				return False

	return True













#####
## IMAGES
#####


if globalconf.USE_THUMBOR:
	def thumborize(url):
		if url.startswith("/"): url = globalconf.OWNURL + url
		encrypted_url = globalconf.THUMBOR_GENERATOR.generate(
		    width=300,
		    height=300,
		    smart=True,
		    image_url=url
		)
		return globalconf.THUMBOR_SERVER + encrypted_url

else:
	def thumborize(url):
		return url



### Caches

cacheage = settings.get_settings("CACHE_EXPIRE_POSITIVE") * 24 * 3600
cacheage_neg = settings.get_settings("CACHE_EXPIRE_NEGATIVE") * 24 * 3600

artist_cache = caching.Cache(name="imgcache_artists",maxage=cacheage,maxage_negative=cacheage_neg,persistent=True)
track_cache = caching.Cache(name="imgcache_tracks",maxage=cacheage,maxage_negative=cacheage_neg,persistent=True)


# removes emojis and weird shit from names
def clean(name):
	return "".join(c for c in name if c.isalnum() or c in []).strip()

def get_all_possible_filenames(artist=None,artists=None,title=None):
	# check if we're dealing with a track or artist, then clean up names
	# (only remove non-alphanumeric, allow korean and stuff)

	if title is not None and artists is not None:
		track = True
		title, artists = clean(title), [clean(a) for a in artists]
	elif artist is not None:
		track = False
		artist = clean(artist)
	else: return []


	superfolder = "images/tracks/" if track else "images/artists/"

	filenames = []

	if track:
		#unsafeartists = [artist.translate(None,"-_./\\") for artist in artists]
		safeartists = [re.sub("[^a-zA-Z0-9]","",artist) for artist in artists]
		#unsafetitle = title.translate(None,"-_./\\")
		safetitle = re.sub("[^a-zA-Z0-9]","",title)

		if len(artists) < 4:
			unsafeperms = itertools.permutations(artists)
			safeperms = itertools.permutations(safeartists)
		else:
			unsafeperms = [sorted(artists)]
			safeperms = [sorted(safeartists)]


		for unsafeartistlist in unsafeperms:
			filename = "-".join(unsafeartistlist) + "_" + title
			if filename != "":
				filenames.append(filename)
				filenames.append(filename.lower())
		for safeartistlist in safeperms:
			filename = "-".join(safeartistlist) + "_" + safetitle
			if filename != "":
				filenames.append(filename)
				filenames.append(filename.lower())
		filenames = list(set(filenames))
		if len(filenames) == 0: filenames.append(str(hash((frozenset(artists),title))))
	else:
		#unsafeartist = artist.translate(None,"-_./\\")
		safeartist = re.sub("[^a-zA-Z0-9]","",artist)

		filename = artist
		if filename != "":
			filenames.append(filename)
			filenames.append(filename.lower())
		filename = safeartist
		if filename != "":
			filenames.append(filename)
			filenames.append(filename.lower())

		filenames = list(set(filenames))
		if len(filenames) == 0: filenames.append(str(hash(artist)))

	return [superfolder + name for name in filenames]

def local_files(artist=None,artists=None,title=None):


	filenames = get_all_possible_filenames(artist,artists,title)

	images = []

	for purename in filenames:
		# direct files
		for ext in ["png","jpg","jpeg","gif"]:
			#for num in [""] + [str(n) for n in range(0,10)]:
			if os.path.exists(datadir(purename + "." + ext)):
				images.append("/" + purename + "." + ext)

		# folder
		try:
			for f in os.listdir(datadir(purename)):
				if f.split(".")[-1] in ["png","jpg","jpeg","gif"]:
					images.append("/" + purename + "/" + f)
		except:
			pass

	return images



# these caches are there so we don't check all files every time, but return the same one
local_cache_age = settings.get_settings("LOCAL_IMAGE_ROTATE")
local_artist_cache = caching.Cache(maxage=local_cache_age)
local_track_cache = caching.Cache(maxage=local_cache_age)

def getTrackImage(artists,title,fast=False):

	if settings.get_settings("USE_LOCAL_IMAGES"):

		try:
			return local_track_cache.get((frozenset(artists),title))
		except:
			images = local_files(artists=artists,title=title)
			if len(images) != 0:
				#return random.choice(images)
				res = random.choice(images)
				local_track_cache.add((frozenset(artists),title),res)
				return urllib.parse.quote(res)


	try:
		# check our cache
		# if we have cached the nonexistence of that image, we immediately return the redirect to the artist and let the resolver handle it
		# (even if we're not in a fast lookup right now)
		#result = cachedTracks[(frozenset(artists),title)]
		result = track_cache.get((frozenset(artists),title)) #track_from_cache(artists,title)
		if result is not None: return result
		else:
			for a in artists:
				res = getArtistImage(artist=a,fast=True)
				if res != "": return res
			return ""
	except:
		pass

	# do we have an api key?
#	apikey = settings.get_settings("LASTFM_API_KEY")
#	if apikey is None: return "" # DO NOT CACHE THAT


	# fast request only retuns cached and local results, generates redirect link for rest
	if fast: return "/image?title=" + urllib.parse.quote(title) + "&" + "&".join(["artist=" + urllib.parse.quote(a) for a in artists])

	# non-fast lookup (essentially only the resolver lookup)
	result = thirdparty.get_image_track_all((artists,title))

	# cache results (even negative ones)
	#cachedTracks[(frozenset(artists),title)] = result
	track_cache.add((frozenset(artists),title),result) #cache_track(artists,title,result)

	# return either result or redirect to artist
	if result is not None: return result
	else:
		for a in artists:
			res = getArtistImage(artist=a,fast=False)
			if res != "": return res
		return ""


def getArtistImage(artist,fast=False):

	if settings.get_settings("USE_LOCAL_IMAGES"):

		try:
			return thumborize(local_artist_cache.get(artist))
			# Local cached image
		except:
			# Get all local images, select one if present
			images = local_files(artist=artist)
			if len(images) != 0:
				#return random.choice(images)
				res = random.choice(images)
				local_artist_cache.add(artist,res)
				return thumborize(urllib.parse.quote(res))


	# if no local images (or setting to not use them)
	try:
		# check cache for foreign image
		result = artist_cache.get(artist)
		if result is not None: return thumborize(result)
		else: return ""
		# none means non-existence is cached, return empty
	except:
		pass
		# no cache entry, go on



	# do we have an api key?
#	apikey = settings.get_settings("LASTFM_API_KEY")
#	if apikey is None: return "" # DO NOT CACHE THAT


	# fast request only retuns cached and local results, generates redirect link for rest
	if fast: return "/image?artist=" + urllib.parse.quote(artist)

	# non-fast lookup (essentially only the resolver lookup)
	result = thirdparty.get_image_artist_all(artist)

	# cache results (even negative ones)
	#cachedArtists[artist] = result
	artist_cache.add(artist,result) #cache_artist(artist,result)

	if result is not None: return thumborize(result)
	else: return ""

def getTrackImages(trackobjectlist,fast=False):

	threads = []

	for track in trackobjectlist:
		t = Thread(target=getTrackImage,args=(track["artists"],track["title"],),kwargs={"fast":fast})
		t.start()
		threads.append(t)

	for t in threads:
		t.join()


	return [getTrackImage(t["artists"],t["title"]) for t in trackobjectlist]

def getArtistImages(artistlist,fast=False):

	threads = []

	for artist in artistlist:
		t = Thread(target=getArtistImage,args=(artist,),kwargs={"fast":fast})
		t.start()
		threads.append(t)

	for t in threads:
		t.join()

	# async calls only cached results, now we need to get them
	return [getArtistImage(a) for a in artistlist]



# new way of serving images
# instead always generate a link locally, but redirect that on the fly
# this way the page can load faster and images will trickle in without having to resort to XHTTP requests

def resolveImage(artist=None,track=None):
	if track is not None:
		return getTrackImage(track["artists"],track["title"])
	elif artist is not None:
		return getArtistImage(artist)







def set_image(b64,**keys):
	track = "title" in keys

	regex = r"data:image/(\w+);base64,(.+)"
	type,b64 = re.fullmatch(regex,b64).groups()
	print(b64[:40])
	b64 = base64.b64decode(b64)
	filename = "webupload" + str(int(datetime.datetime.now().timestamp())) + "." + type
	for folder in get_all_possible_filenames(**keys):
		if os.path.exists(datadir(folder)):
			with open(datadir(folder,filename),"wb") as f:
				f.write(b64)

			# set as current picture in rotation
			if track: local_track_cache.add((frozenset(keys["artists"]),keys["title"]),os.path.join(folder,filename))
			else: local_artist_cache.add(keys["artist"],os.path.join(folder,filename))
			return

	folder = get_all_possible_filenames(**keys)[0]
	os.makedirs(datadir(folder))
	with open(datadir(folder,filename),"wb") as f:
		f.write(b64)

	# set as current picture in rotation
	if track: local_track_cache.add((frozenset(keys["artists"]),keys["title"]),os.path.join(folder,filename))
	else: local_artist_cache.add(keys["artist"],os.path.join(folder,filename))









#####
## PULSE MAINTENANCE
#####



@yearly
def update_medals():


	from .database import MEDALS, MEDALS_TRACKS, STAMPS, get_charts_artists, get_charts_tracks

	currentyear = datetime.datetime.utcnow().year
	try:
		firstyear = datetime.datetime.utcfromtimestamp(STAMPS[0]).year
	except:
		firstyear = currentyear


	MEDALS.clear()
	for year in range(firstyear,currentyear):

		charts = get_charts_artists(within=[year])
		for a in charts:

			artist = a["artist"]
			if a["rank"] == 1: MEDALS.setdefault(artist,{}).setdefault("gold",[]).append(year)
			elif a["rank"] == 2: MEDALS.setdefault(artist,{}).setdefault("silver",[]).append(year)
			elif a["rank"] == 3: MEDALS.setdefault(artist,{}).setdefault("bronze",[]).append(year)
			else: break

	MEDALS_TRACKS.clear()
	for year in range(firstyear,currentyear):

		charts = get_charts_tracks(within=[year])
		for t in charts:

			track = (frozenset(t["track"]["artists"]),t["track"]["title"])
			if t["rank"] == 1: MEDALS_TRACKS.setdefault(track,{}).setdefault("gold",[]).append(year)
			elif t["rank"] == 2: MEDALS_TRACKS.setdefault(track,{}).setdefault("silver",[]).append(year)
			elif t["rank"] == 3: MEDALS_TRACKS.setdefault(track,{}).setdefault("bronze",[]).append(year)
			else: break

@daily
def update_weekly():

	from .database import WEEKLY_TOPTRACKS, WEEKLY_TOPARTISTS, get_charts_artists, get_charts_tracks
	from .malojatime import ranges, thisweek


	WEEKLY_TOPARTISTS.clear()
	WEEKLY_TOPTRACKS.clear()

	for week in ranges(step="week"):
		if week == thisweek(): break
		for a in get_charts_artists(timerange=week):
			artist = a["artist"]
			if a["rank"] == 1: WEEKLY_TOPARTISTS[artist] = WEEKLY_TOPARTISTS.setdefault(artist,0) + 1

		for t in get_charts_tracks(timerange=week):
			track = (frozenset(t["track"]["artists"]),t["track"]["title"])
			if t["rank"] == 1: WEEKLY_TOPTRACKS[track] = WEEKLY_TOPTRACKS.setdefault(track,0) + 1


@daily
def send_stats():
	if settings.get_settings("SEND_STATS"):

		log("Sending daily stats report...")

		from .database import ARTISTS, TRACKS, SCROBBLES

		keys = {
			"url":"https://myrcella.krateng.ch/malojastats",
			"method":"POST",
			"headers":{"Content-Type": "application/json"},
			"data":json.dumps({
				"name":settings.get_settings("NAME"),
				"url":settings.get_settings("PUBLIC_URL"),
				"version":".".join(str(d) for d in version),
				"artists":len(ARTISTS),
				"tracks":len(TRACKS),
				"scrobbles":len(SCROBBLES)
			}).encode("utf-8")
		}
		try:
			req = urllib.request.Request(**keys)
			response = urllib.request.urlopen(req)
			log("Sent daily report!")
		except:
			log("Could not send daily report!")