quotes.py

   1 #!/usr/bin/env python
   2 # -*- encoding: utf-8 -*-
   3
   4 """ Gestion des quotes """
   5
   6 import datetime
   7 import time
   8 import re
   9 import json
  10 import random
  11
  12 import config
  13
  14 quote_matcher = re.compile(config.quote_regexp, flags=re.UNICODE)
  15 quote_matcher_with_timestamp = re.compile(config.quote_regexp_with_timestamp, flags=re.UNICODE)
  16 spaces_matcher = re.compile(u"\s", flags=re.U)
  17
  18 def equivalence_partition(iterable, relation):
  19     """ Partitionne l'itérable en classes d'équivalences. """
  20     classes = []
  21     for o in iterable:
  22         # find the class it is in
  23         found = False
  24         for c in classes:
  25             if relation( iter(c).next(), o ): # is it equivalent to this class?
  26                 c.add( o )
  27                 found = True
  28                 break
  29         if not found: # it is in a new class
  30             classes.append( set( [ o ] ) )
  31     return classes
  32
  33 def get_now():
  34     """ Renvoie la date actuelle """
  35     return datetime.datetime(*time.localtime()[:6])
  36
  37 def sanitize_author(raw):
  38     """Proprifie l'auteur : enlève les espaces insécables."""
  39     return spaces_matcher.sub(u" ", raw)
  40
  41 class Quote(object):
  42     """ Une citation """
  43     def __init__(self, author, content, timestamp=None, place=None, quoter=None):
  44         if timestamp is None:
  45             timestamp = get_now()
  46         elif isinstance(timestamp, basestring):
  47             timestamp = datetime.datetime(*time.strptime(timestamp, u"%Y-%m-%d_%H:%M:%S")[:6])
  48         self.author = sanitize_author(author)
  49         self.content = content
  50         self.timestamp = timestamp
  51         self.place = place
  52         self.quoter = quoter
  53
  54     def jsonize(self):
  55         d = {"author" : self.author, "content" : self.content,
  56              "timestamp" : self.timestamp.strftime(u"%F_%T"),
  57              "place" : self.place, "quoter" : self.quoter}
  58         return d
  59
  60     def __unicode__(self):
  61         """ Retourne la quote affichable """
  62         return config.quote_template % self.__dict__
  63     def __str__(self):
  64         return unicode(self).encode("utf-8")
  65
  66     def display(self, show_context=False):
  67         """
  68         Retourne une chaîne contenant toujours la quote et l'auteur,
  69         et le contexte ssi ``show_context = True``.
  70         """
  71         s = config.quote_template % self.__dict__
  72         if show_context:
  73             s = u"%s | %s" % (s, self.place)
  74         return s.encode("utf-8")
  75
  76     def full_str(self):
  77         """ Retourne une chaîne représentant la totalité des infos de la quote,
  78             tout en étant parsable et human-readable. """
  79         place = self.place if self.place and self.place.strip() != u"" else u""
  80         quoter = self.quoter if self.quoter and self.quoter.strip() != u"" else u""
  81         return (u"%s %s | %s | %s" % (self.timestamp.strftime("%F_%T"), config.quote_template % self.__dict__
  82                   , place, quoter)).encode("utf-8")
  83
  84     def __eq__(self, otherquote):
  85         """ Vérifie si cette phrase n'a pas déjà été dite par la même personne.
  86             Indépendamment de la date et de la casse. """
  87         return [self.author.lower(), self.content.lower()] == [otherquote.author.lower(), otherquote.content.lower()]
  88
  89
  90 def parse(text, date=None):
  91     """ Parse le ``text`` et renvoie une quote ou None. """
  92     if date == None:
  93         date = get_now()
  94     get = quote_matcher.match(text)
  95     if not get is None:
  96         d = get.groupdict()
  97         return Quote(d["author"], d["content"], date)
  98
  99 def load_file(filename):
 100     """ Récupère les quotes depuis le fichier """
 101     with open(filename) as f:
 102         jsonquotes = json.load(f)
 103         quotes = [Quote(**q) for q in jsonquotes]
 104     return quotes
 105
 106 def save_file(quotes, filename):
 107     """ Enregistre les quotes dans le fichier """
 108     with open(filename, "w") as f:
 109         raws = [q.jsonize() for q in quotes]
 110         json.dump(raws, f)
 111
 112 class QuoteDB(object):
 113     """ Stocke et distribue des quotes. """
 114     def __init__(self):
 115         self.quotelist = []
 116
 117     def load(self):
 118         """ Charge le fichier de quotes dans la DB """
 119         self.quotelist = load_file(config.quote_file)
 120
 121     def save(self):
 122         """ Sauvegarde la DB dans le fichier de quotes """
 123         save_file(self.quotelist, config.quote_file)
 124
 125     def _collapse_author(self, author):
 126         """ Renvoie ``author`` avec la casse déjà utilisée si il a déjà été quoté
 127             sinon, le renvoie sans le modifier. """
 128         authors = list(set([q.author for q in self.quotelist if q.author.lower() == author.lower()]))
 129         if len(authors) > 1:
 130             print "Warning : authors %s" % authors
 131         if authors:
 132             return authors[0]
 133         else:
 134             return author
 135
 136     def get_clash_authors(self):
 137         """ Renvoie une liste de liste d'auteurs qui sont enregistrés avec des casses différentes. """
 138         authors = list(set([q.author for q in self.quotelist]))
 139         authors = equivalence_partition(authors, lambda x,y: x.lower() == y.lower())
 140         authors = [list(c) for c in authors if len(c) > 1]
 141         return authors
 142
 143     def store(self, timestamp=None, **kwargs):
 144         """ Enregistre une nouvelle quote, sauf si elle existe déjà.
 145             Force l'auteur à utiliser la même casse si un auteur de casse différente existait déjà.
 146             Renvoie ``True`` si elle a été ajoutée, ``False`` si elle existait. """
 147         kwargs["author"] = self._collapse_author(kwargs["author"])
 148         kwargs["timestamp"] = timestamp
 149         newquote = Quote(**kwargs)
 150         if not newquote in self.quotelist:
 151             self.quotelist.append(newquote)
 152             return True
 153         return False
 154
 155     def __repr__(self):
 156         return repr(self.quotelist)
 157
 158     def random(self):
 159         """ Sort une quote aléatoire """
 160         return random.choice(self.quotelist)
 161
 162     def quotesfrom(self, author):
 163         """ Sort toutes les quotes de ``author`` """
 164         return [q for q in self.quotelist if q.author == author]
 165     def randomfrom(self, author):
 166         """ Sort une quote aléatoire de ``author`` """
 167         return random.choice(self.quotesfrom(author))
 168
 169     def search(self, inquote=None, author=None, regexp=False):
 170         """Fait une recherche dans les quotes."""
 171         if regexp:
 172             if inquote is None:
 173                 inquote = ".*"
 174             if author is None:
 175                 author = ".*"
 176             qreg = re.compile(inquote, flags=re.UNICODE)
 177             areg = re.compile(author, flags=re.UNICODE)
 178             l = [q for q in self.quotelist if qreg.match(q.content) and areg.match(q.author)]
 179         else:
 180             if inquote is None:
 181                 inquote = ""
 182             if author is None:
 183                 author = ""
 184             l = [q for q in self.quotelist if inquote in q.content and author in q.author]
 185         return l
 186
 187     def search_authors(self, author=None, regexp=False):
 188         """Renvoie la liste des auteurs contenant ``author`` ou qui matchent la regexp."""
 189         if regexp:
 190             if author is None:
 191                 author = ".*"
 192             areg = re.compile(author, flags=re.UNICODE)
 193             l = list(set([q.author for q in self.quotelist if areg.match(q.author)]))
 194         else:
 195             if author is None:
 196                 author = ""
 197             l = list(set([q.author for q in self.quotelist if author in q.author]))
 198         return l
 199
 200 def dump(quotedb, dump_file=None):
 201     """Pour exporter les quotes dans un format readable vers un fichier."""
 202     if dump_file is None:
 203         dump_file = config.quote_dump_file
 204     t = "\n".join([q.full_str() for q in quotedb.quotelist]) + "\n"
 205     with open(dump_file, "w") as f:
 206         f.write(t)
 207
 208 def restore(dump_file=None):
 209     """Crée un DB de quotes en parsant le contenu d'un fichier de dump."""
 210     if dump_file is None:
 211         dump_file = config.quote_dump_file
 212     with open(dump_file) as f:
 213         t = f.read()
 214     t = t.decode("utf-8") # Oui, ça peut fail, mais on ne doit alors pas continuer
 215     l = [m.groupdict() for m in quote_matcher_with_timestamp.finditer(t)]
 216     # On instancie les quotes grâce aux dicos qui ont déjà la bonne tronche
 217     l = [Quote(**q) for q in l]
 218     newquotedb = QuoteDB()
 219     newquotedb.quotelist = l
 220     return newquotedb