smarter tag-string parsing (from django-tagging)

AUTHORS.txt

@@ -9,3 +9,4 @@ Alex Gaynor <alex.gaynor@gmail.com>
 Rob Hudson <rob@cogit8.org>
 Carl Meyer <carl@oddbird.net>
 Frank Wiles
+Jonathan Buchanan

taggit/forms.py

 from django import forms
 
-from taggit.utils import parse_tags
+from taggit.utils import parse_tags, edit_string_for_tags
 
 
 class TagWidget(forms.TextInput):
     def render(self, name, value, attrs=None):
         if value is not None and not isinstance(value, basestring):
-            value = ", ".join(o.tag.name for o in value.select_related("tag"))
+            value = edit_string_for_tags([o.tag for o in value.select_related("tag")])
         return super(TagWidget, self).render(name, value, attrs)
 
 class TagField(forms.CharField):

taggit/tests/tests.py

+from unittest import TestCase as UnitTestCase
+
 from django.test import TestCase, TransactionTestCase
 
 from taggit.models import Tag, TaggedItem
+from taggit.utils import parse_tags, edit_string_for_tags
 from taggit.tests.forms import FoodForm, DirectFoodForm, CustomPKFoodForm
 from taggit.tests.models import (Food, Pet, HousePet, DirectFood, DirectPet,
     DirectHousePet, TaggedPet, CustomPKFood, CustomPKPet, CustomPKHousePet,
@@ -208,8 +211,13 @@ class TaggableFormTestCase(BaseTaggingTestCase):
         self.assert_tags_equal(raspberry.tags.all(), [])
         
         f = self.form_class(instance=apple)
-        self.assertEqual(str(f), """<tr><th><label for="id_name">Name:</label></th><td><input id="id_name" type="text" name="name" value="apple" maxlength="50" /></td></tr>\n<tr><th><label for="id_tags">Tags:</label></th><td><input type="text" name="tags" value="green, red, yummy, delicious" id="id_tags" /></td></tr>""")
+        self.assertEqual(str(f), """<tr><th><label for="id_name">Name:</label></th><td><input id="id_name" type="text" name="name" value="apple" maxlength="50" /></td></tr>\n<tr><th><label for="id_tags">Tags:</label></th><td><input type="text" name="tags" value="delicious green red yummy" id="id_tags" /></td></tr>""")
+
+        apple.tags.add('has,comma')
+        f = self.form_class(instance=apple)
+        self.assertEqual(str(f), """<tr><th><label for="id_name">Name:</label></th><td><input id="id_name" type="text" name="name" value="apple" maxlength="50" /></td></tr>\n<tr><th><label for="id_tags">Tags:</label></th><td><input type="text" name="tags" value="delicious green red yummy &quot;has,comma&quot;" id="id_tags" /></td></tr>""")
 
+        
 class TaggableFormDirectTestCase(TaggableFormTestCase):
     form_class = DirectFoodForm
     food_model = DirectFood
@@ -217,3 +225,79 @@ class TaggableFormDirectTestCase(TaggableFormTestCase):
 class TaggableFormCustomPKTestCase(TaggableFormTestCase):
     form_class = CustomPKFoodForm
     food_model = CustomPKFood
+
+
+class TagStringParseTestCase(UnitTestCase):
+    """
+    Ported from Jonathan Buchanan's `django-tagging
+    <http://django-tagging.googlecode.com/>`_
+
+    """
+    def test_with_simple_space_delimited_tags(self):
+        """ Test with simple space-delimited tags. """
+        
+        self.assertEquals(parse_tags('one'), [u'one'])
+        self.assertEquals(parse_tags('one two'), [u'one', u'two'])
+        self.assertEquals(parse_tags('one two three'), [u'one', u'three', u'two'])
+        self.assertEquals(parse_tags('one one two two'), [u'one', u'two'])
+    
+    def test_with_comma_delimited_multiple_words(self):
+        """ Test with comma-delimited multiple words.
+            An unquoted comma in the input will trigger this. """
+            
+        self.assertEquals(parse_tags(',one'), [u'one'])
+        self.assertEquals(parse_tags(',one two'), [u'one two'])
+        self.assertEquals(parse_tags(',one two three'), [u'one two three'])
+        self.assertEquals(parse_tags('a-one, a-two and a-three'),
+            [u'a-one', u'a-two and a-three'])
+    
+    def test_with_double_quoted_multiple_words(self):
+        """ Test with double-quoted multiple words.
+            A completed quote will trigger this.  Unclosed quotes are ignored. """
+            
+        self.assertEquals(parse_tags('"one'), [u'one'])
+        self.assertEquals(parse_tags('"one two'), [u'one', u'two'])
+        self.assertEquals(parse_tags('"one two three'), [u'one', u'three', u'two'])
+        self.assertEquals(parse_tags('"one two"'), [u'one two'])
+        self.assertEquals(parse_tags('a-one "a-two and a-three"'),
+            [u'a-one', u'a-two and a-three'])
+    
+    def test_with_no_loose_commas(self):
+        """ Test with no loose commas -- split on spaces. """
+        self.assertEquals(parse_tags('one two "thr,ee"'), [u'one', u'thr,ee', u'two'])
+        
+    def test_with_loose_commas(self):
+        """ Loose commas - split on commas """
+        self.assertEquals(parse_tags('"one", two three'), [u'one', u'two three'])
+        
+    def test_tags_with_double_quotes_can_contain_commas(self):
+        """ Double quotes can contain commas """
+        self.assertEquals(parse_tags('a-one "a-two, and a-three"'),
+            [u'a-one', u'a-two, and a-three'])
+        self.assertEquals(parse_tags('"two", one, one, two, "one"'),
+            [u'one', u'two'])
+    
+    def test_with_naughty_input(self):
+        """ Test with naughty input. """
+        
+        # Bad users! Naughty users!
+        self.assertEquals(parse_tags(None), [])
+        self.assertEquals(parse_tags(''), [])
+        self.assertEquals(parse_tags('"'), [])
+        self.assertEquals(parse_tags('""'), [])
+        self.assertEquals(parse_tags('"' * 7), [])
+        self.assertEquals(parse_tags(',,,,,,'), [])
+        self.assertEquals(parse_tags('",",",",",",","'), [u','])
+        self.assertEquals(parse_tags('a-one "a-two" and "a-three'),
+            [u'a-one', u'a-three', u'a-two', u'and'])
+
+    def test_recreation_of_tag_list_string_representations(self):
+        plain = Tag.objects.create(name='plain')
+        spaces = Tag.objects.create(name='spa ces')
+        comma = Tag.objects.create(name='com,ma')
+        self.assertEquals(edit_string_for_tags([plain]), u'plain')
+        self.assertEquals(edit_string_for_tags([plain, spaces]), u'plain, spa ces')
+        self.assertEquals(edit_string_for_tags([plain, spaces, comma]), u'plain, spa ces, "com,ma"')
+        self.assertEquals(edit_string_for_tags([plain, comma]), u'plain "com,ma"')
+        self.assertEquals(edit_string_for_tags([comma, spaces]), u'"com,ma", spa ces')
+    

taggit/utils.py

+from django.utils.encoding import force_unicode
 from django.utils.functional import wraps
 
-def parse_tags(tags):
-    if tags is None:
-        tags = ""
-    return [o.strip() for o in tags.split(',') if o.strip()]
+
+def parse_tags(tagstring):
+    """
+    Parses tag input, with multiple word input being activated and
+    delineated by commas and double quotes. Quotes take precedence, so
+    they may contain commas.
+
+    Returns a sorted list of unique tag names.
+
+    Ported from Jonathan Buchanan's `django-tagging
+    <http://django-tagging.googlecode.com/>`_
+    """
+    if not tagstring:
+        return []
+
+    tagstring = force_unicode(tagstring)
+
+    # Special case - if there are no commas or double quotes in the
+    # input, we don't *do* a recall... I mean, we know we only need to
+    # split on spaces.
+    if u',' not in tagstring and u'"' not in tagstring:
+        words = list(set(split_strip(tagstring, u' ')))
+        words.sort()
+        return words
+
+    words = []
+    buffer = []
+    # Defer splitting of non-quoted sections until we know if there are
+    # any unquoted commas.
+    to_be_split = []
+    saw_loose_comma = False
+    open_quote = False
+    i = iter(tagstring)
+    try:
+        while 1:
+            c = i.next()
+            if c == u'"':
+                if buffer:
+                    to_be_split.append(u''.join(buffer))
+                    buffer = []
+                # Find the matching quote
+                open_quote = True
+                c = i.next()
+                while c != u'"':
+                    buffer.append(c)
+                    c = i.next()
+                if buffer:
+                    word = u''.join(buffer).strip()
+                    if word:
+                        words.append(word)
+                    buffer = []
+                open_quote = False
+            else:
+                if not saw_loose_comma and c == u',':
+                    saw_loose_comma = True
+                buffer.append(c)
+    except StopIteration:
+        # If we were parsing an open quote which was never closed treat
+        # the buffer as unquoted.
+        if buffer:
+            if open_quote and u',' in buffer:
+                saw_loose_comma = True
+            to_be_split.append(u''.join(buffer))
+    if to_be_split:
+        if saw_loose_comma:
+            delimiter = u','
+        else:
+            delimiter = u' '
+        for chunk in to_be_split:
+            words.extend(split_strip(chunk, delimiter))
+    words = list(set(words))
+    words.sort()
+    return words
+
+
+def split_strip(string, delimiter=u','):
+    """
+    Splits ``string`` on ``delimiter``, stripping each resulting string
+    and returning a list of non-empty strings.
+
+    Ported from Jonathan Buchanan's `django-tagging
+    <http://django-tagging.googlecode.com/>`_
+    """
+    if not string:
+        return []
+
+    words = [w.strip() for w in string.split(delimiter)]
+    return [w for w in words if w]
+
+
+def edit_string_for_tags(tags):
+    """
+    Given list of ``Tag`` instances, creates a string representation of
+    the list suitable for editing by the user, such that submitting the
+    given string representation back without changing it will give the
+    same list of tags.
+
+    Tag names which contain commas will be double quoted.
+
+    If any tag name which isn't being quoted contains whitespace, the
+    resulting string of tag names will be comma-delimited, otherwise
+    it will be space-delimited.
+
+    Ported from Jonathan Buchanan's `django-tagging
+    <http://django-tagging.googlecode.com/>`_
+    """
+    names = []
+    use_commas = False
+    for tag in tags:
+        name = tag.name
+        if u',' in name:
+            names.append('"%s"' % name)
+            continue
+        elif u' ' in name:
+            if not use_commas:
+                use_commas = True
+        names.append(name)
+    if use_commas:
+        glue = u', '
+    else:
+        glue = u' '
+    return glue.join(names)
+
 
 def require_instance_manager(func):
     @wraps(func)