models.py

   1 import re, string
   2
   3 from django.db import models
   4 from django.contrib.contenttypes.models import ContentType
   5 from django.contrib.contenttypes import generic
   6
   7 PUNCTUATION_REGEX = re.compile('[' + re.escape(string.punctuation) + ']')
   8
   9 FULL_TEXT_STOP_WORDS = frozenset([
  10  'a', 'about', 'according', 'accordingly', 'affected', 'affecting', 'after',
  11  'again', 'against', 'all', 'almost', 'already', 'also', 'although',
  12  'always', 'am', 'among', 'an', 'and', 'any', 'anyone', 'apparently', 'are',
  13  'arise', 'as', 'aside', 'at', 'away', 'be', 'became', 'because', 'become',
  14  'becomes', 'been', 'before', 'being', 'between', 'both', 'briefly', 'but',
  15  'by', 'came', 'can', 'cannot', 'certain', 'certainly', 'could', 'did', 'do',
  16  'does', 'done', 'during', 'each', 'either', 'else', 'etc', 'ever', 'every',
  17  'following', 'for', 'found', 'from', 'further', 'gave', 'gets', 'give',
  18  'given', 'giving', 'gone', 'got', 'had', 'hardly', 'has', 'have', 'having',
  19  'here', 'how', 'however', 'i', 'if', 'in', 'into', 'is', 'it', 'itself',
  20  'just', 'keep', 'kept', 'knowledge', 'largely', 'like', 'made', 'mainly',
  21  'make', 'many', 'might', 'more', 'most', 'mostly', 'much', 'must', 'nearly',
  22  'necessarily', 'neither', 'next', 'no', 'none', 'nor', 'normally', 'not',
  23  'noted', 'now', 'obtain', 'obtained', 'of', 'often', 'on', 'only', 'or',
  24  'other', 'our', 'out', 'owing', 'particularly', 'past', 'perhaps', 'please',
  25  'poorly', 'possible', 'possibly', 'potentially', 'predominantly', 'present',
  26  'previously', 'primarily', 'probably', 'prompt', 'promptly', 'put',
  27  'quickly', 'quite', 'rather', 'readily', 'really', 'recently', 'regarding',
  28  'regardless', 'relatively', 'respectively', 'resulted', 'resulting',
  29  'results', 'said', 'same', 'seem', 'seen', 'several', 'shall', 'should',
  30  'show', 'showed', 'shown', 'shows', 'significantly', 'similar', 'similarly',
  31  'since', 'slightly', 'so', 'some', 'sometime', 'somewhat', 'soon',
  32  'specifically', 'state', 'states', 'strongly', 'substantially',
  33  'successfully', 'such', 'sufficiently', 'than', 'that', 'the', 'their',
  34  'theirs', 'them', 'then', 'there', 'therefore', 'these', 'they', 'this',
  35  'those', 'though', 'through', 'throughout', 'to', 'too', 'toward', 'under',
  36  'unless', 'until', 'up', 'upon', 'use', 'used', 'usefully', 'usefulness',
  37  'using', 'usually', 'various', 'very', 'was', 'we', 'were', 'what', 'when',
  38  'where', 'whether', 'which', 'while', 'who', 'whose', 'why', 'widely',
  39  'will', 'with', 'within', 'without', 'would', 'yet', 'you'])
  40
  41 class Index(models.Model):
  42     content_type = models.ForeignKey(ContentType)
  43     object_id = models.PositiveIntegerField()
  44     content = models.TextField()
  45
  46     content_object = generic.GenericForeignKey()
  47
  48     @staticmethod
  49     def full_text_index(text):
  50         if text:
  51             text = PUNCTUATION_REGEX.sub(' ', text)
  52             words = text.lower().split()
  53             words = set(words)
  54             words -= FULL_TEXT_STOP_WORDS
  55
  56             for word in list(words):
  57                 if len(word) < FULL_TEXT_MIN_LENGTH:
  58                     words.remove(word)
  59         else:
  60             words = set()
  61         return words
  62
  63     @staticmethod
  64     def index(obj):
  65         keywords = set()
  66
  67         properties = obj._meta.fields
  68         for property in properties:
  69             isurlfield = not isinstance(property, models.URLField)
  70             istext = isinstance(property, models.CharField) or isinstance(property, models.TextField)
  71             if istext and isurlfield:
  72                 text = property.value_from_object(obj)
  73                 mykeywords = Index.full_text_index(text)
  74                 keywords = keywords.union(mykeywords)
  75
  76         text = " ".join(keywords)
  77
  78         # Create or Update
  79         ctype = ContentType.objects.get_for_model(obj)
  80
  81         try:
  82             index = Index.objects.get(content_type__pk=ctype.id, object_id=obj.id)
  83             index.content = text
  84             index.save()
  85         except:
  86             index = Index(content_object=obj, content=text)
  87             index.save()
  88         return index
  89
  90 class Searchable(object):
  91     @staticmethod
  92     def search(cls, query):
  93         ctype = ContentType.objects.get_for_model(cls)
  94         return Index.objects.filter(content__search=query).filter(content_type=ctype)
  95
  96     @staticmethod
  97     def regex(cls, query):
  98         ctype = ContentType.objects.get_for_model(cls)
  99         return Index.objects.filter(content__regex=query).filter(content_type=ctype)
 100
 101     def index(self):
 102         return Index.index(self)
 103
 104 class Overlay(models.Model):
 105     name = models.CharField(max_length=255)
 106     description = models.TextField()
 107     link = models.URLField()
 108
 109     def __unicode__(self):
 110         return self.name
 111
 112
 113 class Ebuild(Searchable, models.Model):
 114     name = models.CharField(max_length=255)
 115     category = models.CharField(max_length=255)
 116     version = models.CharField(max_length=255)
 117     description = models.TextField()
 118     keywords = models.TextField(max_length=255)
 119     license = models.TextField(max_length=255)
 120     iuse = models.TextField(max_length=255)
 121     homepage = models.URLField()
 122     overlay = models.ForeignKey(Overlay)
 123
 124     def path(self):
 125         return "/%s/%s/%s/%s" % (self.overlay.name, self.category, self.name, self.version)
 126
 127     def get_absolute_url(self):
 128         return "/search/?q=%s" % self.name
 129
 130     def __unicode__(self):
 131         return self.name