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