Handle Person.ascii in edit profile better to try to ensure that

people fill it in correctly. Blank it out if it's unchanged from name
and use unidecode to set it automatically (with a warning if it
actually converts something). Branch ready for merge.
 - Legacy-Id: 11752

ietf/ietfauth/forms.py

@@ -9,6 +9,8 @@ from django.contrib.auth.models import User
 from django.utils.html import mark_safe
 from django.core.urlresolvers import reverse as urlreverse
 
+from unidecode import unidecode
+
 import debug                            # pyflakes:ignore
 
 from ietf.person.models import Person, Email
@@ -45,7 +47,7 @@ class PasswordForm(forms.Form):
 def ascii_cleaner(supposedly_ascii):
     outside_printable_ascii_pattern = r'[^\x20-\x7F]'
     if re.search(outside_printable_ascii_pattern, supposedly_ascii):
-        raise forms.ValidationError("Please only enter ASCII characters.")
+        raise forms.ValidationError("Only unaccented Latin characters are allowed.")
     return supposedly_ascii
 
 def prevent_at_symbol(name):
@@ -69,12 +71,31 @@ def get_person_form(*args, **kwargs):
         def __init__(self, *args, **kwargs):
             super(ModelForm, self).__init__(*args, **kwargs)
 
+            # blank ascii if it's the same as name
+            self.fields["ascii"].required = self.fields["ascii"].widget.is_required = False
+            self.fields["ascii"].help_text += " " + "Leave blank to use auto-reconstructed Latin version of name."
+
+            if self.initial.get("ascii") == self.initial.get("name"):
+                self.initial["ascii"] = ""
+
+            self.unidecoded_ascii = False
+
+            if self.data and not self.data.get("ascii", "").strip():
+                self.data = self.data.copy()
+                name = self.data["name"]
+                reconstructed_name = unidecode(name)
+                self.data["ascii"] = reconstructed_name
+                self.unidecoded_ascii = name != reconstructed_name
+
         def clean_name(self):
             name = self.cleaned_data.get("name") or u""
             prevent_at_symbol(name)
             return name
 
         def clean_ascii(self):
+            if self.unidecoded_ascii:
+                raise forms.ValidationError("Name contained non-ASCII characters, and was automatically reconstructed using only Latin characters. Check the result - if you are happy, just hit Submit again.")
+
             name = self.cleaned_data.get("ascii") or u""
             prevent_at_symbol(name)
             return ascii_cleaner(name)

ietf/ietfauth/tests.py

@@ -213,6 +213,15 @@ class IetfAuthTests(TestCase):
         q = PyQuery(r.content)
         self.assertTrue(len(q("form .has-error")) > 0)
 
+        # edit details - blank ASCII
+        blank_ascii = base_data.copy()
+        blank_ascii["ascii"] = u""
+        r = self.client.post(url, blank_ascii)
+        self.assertEqual(r.status_code, 200)
+        q = PyQuery(r.content)
+        self.assertTrue(len(q("form .has-error")) > 0) # we get a warning about reconstructed name
+        self.assertEqual(q("input[name=ascii]").val(), base_data["ascii"])
+
         # edit details
         r = self.client.post(url, base_data)
         self.assertEqual(r.status_code, 200)