1
0
mirror of https://github.com/krateng/maloja.git synced 2023-08-10 21:12:55 +03:00
maloja/cleanup.py
2019-04-08 13:38:47 +02:00

183 lines
5.9 KiB
Python

import re
import utilities
from doreah import tsv
# need to do this as a class so it can retain loaded settings from file
# apparently this is not true
# I'm dumb
class CleanerAgent:
def __init__(self):
self.updateRules()
def updateRules(self):
raw = tsv.parse_all("rules","string","string","string")
self.rules_belongtogether = [b for [a,b,c] in raw if a=="belongtogether"]
self.rules_notanartist = [b for [a,b,c] in raw if a=="notanartist"]
self.rules_replacetitle = {b.lower():c for [a,b,c] in raw if a=="replacetitle"}
self.rules_replaceartist = {b.lower():c for [a,b,c] in raw if a=="replaceartist"}
# we always need to be able to tell if our current database is made with the current rules
self.checksums = utilities.checksumTSV("rules")
def fullclean(self,artist,title):
artists = self.parseArtists(self.removespecial(artist))
title = self.parseTitle(self.removespecial(title))
(title,moreartists) = self.parseTitleForArtists(title)
artists += moreartists
artists = list(set(artists))
artists.sort()
return (artists,title)
def removespecial(self,s):
s = s.replace("\t","").replace("","").replace("\n","")
s = re.sub(" +"," ",s)
return s
# if an artist appears in any created rule, we can assume that artist is meant to exist and be spelled like that
def confirmedReal(self,a):
confirmed = self.rules_belongtogether + [self.rules_replaceartist[r] for r in self.rules_replaceartist]
return (a in confirmed)
delimiters_feat = ["ft.","ft","feat.","feat","featuring","Ft.","Ft","Feat.","Feat","Featuring"] #Delimiters used for extra artists, even when in the title field
delimiters = ["vs.","vs","&"] #Delimiters in informal artist strings, spaces expected around them
delimiters_formal = ["; ",";","/"] #Delimiters used specifically to tag multiple artists when only one tag field is available, no spaces used
def parseArtists(self,a):
if a.strip() == "":
return []
if a.strip() in self.rules_notanartist:
return []
if " performing " in a.lower():
return self.parseArtists(re.split(" [Pp]erforming",a)[0])
if a.strip() in self.rules_belongtogether:
return [a.strip()]
if a.strip().lower() in self.rules_replaceartist:
return self.rules_replaceartist[a.strip().lower()].split("")
for d in self.delimiters_feat:
if re.match(r"(.*) \(" + d + " (.*)\)",a) is not None:
return self.parseArtists(re.sub(r"(.*) \(" + d + " (.*)\)",r"\1",a)) + self.parseArtists(re.sub(r"(.*) \(" + d + " (.*)\)",r"\2",a))
for d in self.delimiters_formal:
if (d in a):
ls = []
for i in a.split(d):
ls += self.parseArtists(i)
return ls
for d in (self.delimiters_feat + self.delimiters):
if ((" " + d + " ") in a):
ls = []
for i in a.split(" " + d + " "):
ls += self.parseArtists(i)
return ls
return [a.strip()]
def parseTitle(self,t):
if t.strip().lower() in self.rules_replacetitle:
return self.rules_replacetitle[t.strip().lower()]
t = t.replace("[","(").replace("]",")")
t = re.sub(r" \(as made famous by .*?\)","",t)
t = re.sub(r" \(originally by .*?\)","",t)
t = re.sub(r" \(.*?Remaster.*?\)","",t)
return t.strip()
def parseTitleForArtists(self,t):
for d in self.delimiters_feat:
if re.match(r"(.*) \(" + d + " (.*?)\)",t) is not None:
(title,artists) = self.parseTitleForArtists(re.sub(r"(.*) \(" + d + " (.*?)\)",r"\1",t))
artists += self.parseArtists(re.sub(r"(.*) \(" + d + " (.*?)\).*",r"\2",t))
return (title,artists)
if re.match(r"(.*) - " + d + " (.*)",t) is not None:
(title,artists) = self.parseTitleForArtists(re.sub(r"(.*) - " + d + " (.*)",r"\1",t))
artists += self.parseArtists(re.sub(r"(.*) - " + d + " (.*).*",r"\2",t))
return (title,artists)
if re.match(r"(.*) " + d + " (.*)",t) is not None:
(title,artists) = self.parseTitleForArtists(re.sub(r"(.*) " + d + " (.*)",r"\1",t))
artists += self.parseArtists(re.sub(r"(.*) " + d + " (.*).*",r"\2",t))
return (title,artists)
return (t,[])
#this is for all the runtime changes (counting Trouble Maker as HyunA for charts etc)
class CollectorAgent:
def __init__(self):
self.updateRules()
def updateRules(self):
raw = tsv.parse_all("rules","string","string","string")
self.rules_countas = {b:c for [a,b,c] in raw if a=="countas"}
self.rules_include = {} #Twice the memory, double the performance! (Yes, we're saving redundant information here, but it's not unelegant if it's within a closed object!)
for a in self.rules_countas:
self.rules_include[self.rules_countas[a]] = self.rules_include.setdefault(self.rules_countas[a],[]) + [a]
# this agent needs to be aware of the current id assignment in the main program. unelegant, but the best way i can think of
def updateIDs(self,artistlist):
self.rules_countas_id = {artistlist.index(a):artistlist.index(self.rules_countas[a]) for a in self.rules_countas}
#self.rules_include_id = {artistlist.index(a):artistlist.index(self.rules_include[a]) for a in self.rules_include}
#this needs to take lists into account
def getCredited(self,artist):
if artist in self.rules_countas_id:
return self.rules_countas_id[artist]
if artist in self.rules_countas:
return self.rules_countas[artist]
else:
return artist
def getCreditedList(self,artists):
updatedArtists = []
for artist in artists:
updatedArtists.append(self.getCredited(artist))
return list(set(updatedArtists))
def getAllAssociated(self,artist):
return self.rules_include.get(artist,[])
# this function is there to check for artists that we should include in the database even though they never have any scrobble. important to avoid bugs when
# countas rules are declared preemptively
def getAllArtists(self):
return list(set([a for a in self.rules_countas] + [self.rules_countas[a] for a in self.rules_countas]))
def flatten(lis):
newlist = []
for l in lis:
if isinstance(l, str):
newlist.append(l)
else:
newlist = newlist + l
return list(set(newlist))