Commit 57e30c5f by Alex Gaynor

Some minor cleanup, renaming.

parent f364ac09
...@@ -7,3 +7,4 @@ fakeempire <adam@fakeempire.com> ...@@ -7,3 +7,4 @@ fakeempire <adam@fakeempire.com>
Ben Firshman <ben@firshman.co.uk> Ben Firshman <ben@firshman.co.uk>
Alex Gaynor <alex.gaynor@gmail.com> Alex Gaynor <alex.gaynor@gmail.com>
Rob Hudson <rob@cogit8.org> Rob Hudson <rob@cogit8.org>
Frank Wiles
...@@ -3,19 +3,19 @@ django-taggit ...@@ -3,19 +3,19 @@ django-taggit
``django-taggit`` a simpler approach to tagging with Django. Add it to your ``django-taggit`` a simpler approach to tagging with Django. Add it to your
``INSTALLED_APPS` then just add a TaggableManager to your model and go: ``INSTALLED_APPS` then just add a TaggableManager to your model and go:
from django.db import models from django.db import models
from taggit.managers import TaggableManager from taggit.managers import TaggableManager
class Food(models.Model): class Food(models.Model):
# ... fields here # ... fields here
tags = TaggableManager() tags = TaggableManager()
Then you can use the API like so: Then you can use the API like so:
>>> apple = Food.objects.create(name="apple") >>> apple = Food.objects.create(name="apple")
>>> apple.tags.add("red", "green", "delicious") >>> apple.tags.add("red", "green", "delicious")
>>> apple.tags.all() >>> apple.tags.all()
...@@ -27,3 +27,5 @@ Then you can use the API like so: ...@@ -27,3 +27,5 @@ Then you can use the API like so:
Tags will show up for you automatically in forms and the admin. Tags will show up for you automatically in forms and the admin.
``django-taggit`` requires Django 1.1 or greater.
taggit.contrib.suggest
taggit.contrib.suggest
====================== ======================
This add on module allows you to easily associate keywords and regular This add on module allows you to easily associate keywords and regular
expressions with a Tag object. This is useful to help keep your database expressions with a Tag object. This is useful to help keep your database
getting filled up with several similar tags that really represent the same getting filled up with several similar tags that really represent the same
thing. thing.
For example, if your site is a humor site you might want to collapse all of For example, if your site is a humor site you might want to collapse all of
#fun, #funny, #funnies, #hilarious, #rofl, and #lol into one tag #funny. The #fun, #funny, #funnies, #hilarious, #rofl, and #lol into one tag #funny. The
suggest_tags() function in taggit.contrib.suggest.utils will give you a list ``suggest_tags()`` function in ``taggit.contrib.suggest.utils`` will give you a
of tags that seem appropriate for the text content given to it. list of tags that seem appropriate for the text content given to it.
Unlike the rest of ``django-taggit``, ``suggest`` requires Django 1.2.
Usage Usage
===== =====
Put 'taggit.contrib.suggest' into INSTALLED_APPS and run a syncdb to create Put ``'taggit.contrib.suggest'`` into ``INSTALLED_APPS`` and run a syncdb to
the necessary models. This will add Keywords and Regular Expression inlines create the necessary models. This will add ``Keywords`` and
to the default django-taggit admin. Once you've populated those based on your ``Regular Expression`` inlines to the default ``django-taggit`` admin. Once
site you can do a simple: you've populated those based on your site you can do a simple:
.. sourcecode:: python
from taggit.contrib.suggest.utils import suggest_tags
from taggit.contrib.suggest.utils import suggest_tags tags = suggest_tags(content='Some textual content...')
tags = suggest_tags(content='Some textual content...')
TODO TODO
==== ====
* In a later version I hope to a simple way to help determine keywords for you * In a later version I hope to a simple way to help determine keywords for you
automatically, by learning from your past tags and content. automatically, by learning from your past tags and content.
from django.contrib import admin from django.contrib import admin
from taggit.models import Tag from taggit.admin import TaggedItemInline
from taggit.admin import TaggedItemInline from taggit.contrib.suggest.models import TagKeyword, TagRegex
from taggit.contrib.suggest.models import TagKeyword, TagRegExp from taggit.models import Tag
class TagKeywordInline(admin.StackedInline): class TagKeywordInline(admin.StackedInline):
model = TagKeyword model = TagKeyword
class TagRegExpInline(admin.StackedInline):
model = TagRegExp
class TagSuggestAdmin(admin.ModelAdmin): class TagRegxInline(admin.StackedInline):
model = TagRegex
class TagSuggestAdmin(admin.ModelAdmin):
inlines = [ inlines = [
TaggedItemInline, TaggedItemInline,
TagKeywordInline, TagKeywordInline,
TagRegExpInline, TagRegxInline,
] ]
admin.site.unregister(Tag)
admin.site.unregister(Tag)
admin.site.register(Tag, TagSuggestAdmin) admin.site.register(Tag, TagSuggestAdmin)
import re import re
from django.db import models from django.core.exceptions import ValidationError
from django.core.exceptions import ValidationError from django.db import models
from taggit.models import Tag
HAS_PYSTEMMER = True from taggit.models import Tag
try:
try:
import Stemmer import Stemmer
except ImportError: except ImportError:
HAS_PYSTEMMER = False Stemmer = None
class TagKeyword(models.Model): class TagKeyword(models.Model):
""" Model to associate simple keywords to a Tag """ """ Model to associate simple keywords to a Tag """
tag = models.ForeignKey(Tag, related_name='keywords') tag = models.ForeignKey(Tag, related_name='keywords')
keyword = models.CharField(max_length=30) keyword = models.CharField(max_length=30)
stem = models.CharField(max_length=30) stem = models.CharField(max_length=30)
def __unicode__(self): def __unicode__(self):
return "Keyword '%s' for Tag '%s'" % (self.keyword, self.tag.name) return "Keyword '%s' for Tag '%s'" % (self.keyword, self.tag.name)
def save(self, *args, **kwargs): def save(self, *args, **kwargs):
""" Stem the keyword on save if they have PyStemmer """ """ Stem the keyword on save if they have PyStemmer """
language = kwargs.pop('stemmer-language', 'english') language = kwargs.pop('stemmer-language', 'english')
if not self.id and not self.stem and HAS_PYSTEMMER: if not self.pk and not self.stem and Stemmer:
stemmer = Stemmer.Stemmer(language) stemmer = Stemmer.Stemmer(language)
self.stem = stemmer.stemWord(self.keyword) self.stem = stemmer.stemWord(self.keyword)
super(TagKeyword,self).save(*args,**kwargs) super(TagKeyword, self).save(*args, **kwargs)
def validate_regexp(value):
""" Make sure we have a valid regular expression """ def validate_regex(value):
try: """ Make sure we have a valid regular expression """
try:
re.compile(value) re.compile(value)
except: except Exception:
raise ValidationError('Please enter a valid regular expression') raise ValidationError('Please enter a valid regular expression')
class TagRegExp(models.Model):
""" Model to associate regular expressions with a Tag """ class TagRegex(models.Model):
tag = models.ForeignKey(Tag, related_name='regexps') """ Model to associate regular expressions with a Tag """
tag = models.ForeignKey(Tag, related_name='regexes')
name = models.CharField(max_length=30) name = models.CharField(max_length=30)
regexp = models.CharField(max_length=250, regex = models.CharField(
validators=[validate_regexp], max_length=250,
help_text='Enter a valid Regular Expression. To make it case-insensitive include "(?i)" in your expression.' validators=[validate_regex],
) help_text=('Enter a valid Regular Expression. To make it '
'case-insensitive include "(?i)" in your expression.')
def __unicode__(self): )
return self.name
def __unicode__(self):
def save(self, *args, **kwargs): return self.name
""" Make sure to validate """
def save(self, *args, **kwargs):
""" Make sure to validate """
self.full_clean() self.full_clean()
super(TagRegExp,self).save(*args,**kwargs) super(TagRegex,self).save(*args, **kwargs)
from taggit.contrib.suggest.tests.tests import SuggestCase from taggit.contrib.suggest.tests.tests import SuggestCase
DATABASE_ENGINE = 'sqlite3' DATABASE_ENGINE = 'sqlite3'
INSTALLED_APPS = [ INSTALLED_APPS = [
'django.contrib.contenttypes', 'django.contrib.contenttypes',
'taggit', 'taggit',
'taggit.tests',
'taggit.contrib.suggest', 'taggit.contrib.suggest',
] ]
from django.test import TestCase from django.core.exceptions import ValidationError
from django.core.exceptions import ValidationError from django.test import TestCase
from taggit.models import Tag
from taggit.contrib.suggest.models import TagKeyword, TagRegExp from taggit.contrib.suggest.models import TagKeyword, TagRegex
from taggit.contrib.suggest.utils import suggest_tags from taggit.contrib.suggest.utils import suggest_tags
from taggit.models import Tag
class SuggestCase(TestCase):
def test_simple_suggest(self):
class SuggestCase(TestCase):
def test_simple_suggest(self):
ku_tag = Tag.objects.create(name='ku') ku_tag = Tag.objects.create(name='ku')
ku_keyword1 = TagKeyword.objects.create( ku_keyword1 = TagKeyword.objects.create(
tag=ku_tag, tag=ku_tag,
keyword='kansas university') keyword='kansas university'
)
suggested_tags = suggest_tags(content='I used to be a student at kansas university') suggested_tags = suggest_tags('I used to be a student at kansas university')
self.assertTrue(ku_tag in suggested_tags) self.assertTrue(ku_tag in suggested_tags)
def test_regexp_suggest(self): def test_regex_suggest(self):
ku_tag = Tag.objects.create(name='ku') ku_tag = Tag.objects.create(name='ku')
new_regexp = TagRegExp.objects.create( TagRegex.objects.create(
tag=ku_tag, tag=ku_tag,
name='Find University of Kansas', name='Find University of Kansas',
regexp='University\s+of\s+Kansas') regex='University\s+of\s+Kansas'
)
suggested_tags = suggest_tags(content='I was once a student at the University of Kansas') suggested_tags = suggest_tags('I was once a student at the University of Kansas')
self.assertTrue(ku_tag in suggested_tags) self.assertTrue(ku_tag in suggested_tags)
def test_bad_regexp(self): def test_bad_regex(self):
ku_tag = Tag.objects.create(name='ku') ku_tag = Tag.objects.create(name='ku')
ku_keyword1 = TagKeyword.objects.create( ku_keyword1 = TagKeyword.objects.create(
tag=ku_tag, tag=ku_tag,
keyword='kansas university') keyword='kansas university'
new_regexp = TagRegExp( )
tag=ku_tag, new_regex = TagRegex(
name='Find University of Kansas', tag=ku_tag,
regexp='University\s+of(\s+Kansas') name='Find University of Kansas',
self.assertRaises(ValidationError, new_regexp.save) regex='University\s+of(\s+Kansas'
)
suggested_tags = suggest_tags(content='I was once a student at the University of Kansas. Also known as kansas university by the way.') self.assertRaises(ValidationError, new_regex.save)
suggested_tags = suggest_tags('I was once a student at the University '
'of Kansas. Also known as kansas university by the way.')
self.assertTrue(ku_tag in suggested_tags) self.assertTrue(ku_tag in suggested_tags)
import re import re
from taggit.models import Tag
from taggit.contrib.suggest.models import TagKeyword, TagRegExp
from django.conf import settings
from django.conf import settings
def _suggest_keywords(content=None): from taggit.contrib.suggest.models import TagKeyword, TagRegex
""" Suggest by keywords """ from taggit.models import Tag
def _suggest_keywords(content):
""" Suggest by keywords """
suggested_keywords = set() suggested_keywords = set()
keywords = TagKeyword.objects.values_list('keyword', 'stem', 'tag') keywords = TagKeyword.objects.all()
for k in keywords: for k in keywords:
# Use the stem if available, otherwise use the whole keyword # Use the stem if available, otherwise use the whole keyword
if k[1]: if k.stem:
if k[1] in content: if k.stem in content:
suggested_keywords.add(k[2]) suggested_keywords.add(k.tag_id)
elif k[0] in content: elif k.keyword in content:
suggested_keywords.add(k[2]) suggested_keywords.add(k.tag_id)
return suggested_keywords return suggested_keywords
def _suggest_regexps(content=None): def _suggest_regexes(content):
""" Suggest by regular expressions """ """ Suggest by regular expressions """
# Grab all regular expressions and compile them # Grab all regular expressions and compile them
suggested_regexps = set() suggested_regexes = set()
regexps = set() regex_keywords = TagRegex.objects.all()
regexp_keywords = TagRegExp.objects.values_list(
'regexp', # Look for our regular expressions in the content
'tag', for r in regex_keywords:
) if re.search(r.regex, content):
suggested_regexes.add(r.tag_id)
for r in regexp_keywords:
regexps.add((re.compile(r[0]), r[1])) return suggested_regexes
# Look for our regular expressions in the content def suggest_tags(content):
for r in regexps: """ Suggest tags based on text content """
if r[0].search(content):
suggested_regexps.add(r[1])
return suggested_regexps
def suggest_tags(content=None):
""" Suggest tags based on text content """
suggested_keywords = _suggest_keywords(content) suggested_keywords = _suggest_keywords(content)
suggested_regexps = _suggest_regexps(content) suggested_regexes = _suggest_regexes(content)
suggested_tag_ids = suggested_keywords | suggested_regexps suggested_tag_ids = suggested_keywords | suggested_regexes
return Tag.objects.filter(id__in=suggested_tag_ids) return Tag.objects.filter(id__in=suggested_tag_ids)
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or sign in to comment