mirror of
https://github.com/Hopiu/wagtail.git
synced 2026-05-21 21:41:55 +00:00
5c3908968c
A `LookupError` is raised by `apps.get_model()` for an unknown model, but this `LookupError` was also being raised when values with unhandled types were passed in. This should be a `ValueError` instead. The docstring has been amended to document the errors raised.
90 lines
3.5 KiB
Python
90 lines
3.5 KiB
Python
import re
|
|
import unicodedata
|
|
|
|
from django.db.models import Model
|
|
from django.apps import apps
|
|
from django.utils.encoding import force_text
|
|
from django.utils.text import slugify
|
|
from django.utils.six import string_types
|
|
|
|
|
|
def camelcase_to_underscore(str):
|
|
# http://djangosnippets.org/snippets/585/
|
|
return re.sub('(((?<=[a-z])[A-Z])|([A-Z](?![A-Z]|$)))', '_\\1', str).lower().strip('_')
|
|
|
|
|
|
def resolve_model_string(model_string, default_app=None):
|
|
"""
|
|
Resolve an 'app_label.model_name' string into an actual model class.
|
|
If a model class is passed in, just return that.
|
|
|
|
Raises a LookupError if a model can not be found, or ValueError if passed
|
|
something that is neither a model or a string.
|
|
"""
|
|
if isinstance(model_string, string_types):
|
|
try:
|
|
app_label, model_name = model_string.split(".")
|
|
except ValueError:
|
|
if default_app is not None:
|
|
# If we can't split, assume a model in current app
|
|
app_label = default_app
|
|
model_name = model_string
|
|
else:
|
|
raise ValueError("Can not resolve {0!r} into a model. Model names "
|
|
"should be in the form app_label.model_name".format(
|
|
model_string), model_string)
|
|
|
|
return apps.get_model(app_label, model_name)
|
|
|
|
elif isinstance(model_string, type) and issubclass(model_string, Model):
|
|
return model_string
|
|
|
|
else:
|
|
raise ValueError("Can not resolve {0!r} into a model".format(model_string), model_string)
|
|
|
|
|
|
SCRIPT_RE = re.compile(r'<(-*)/script>')
|
|
|
|
|
|
def escape_script(text):
|
|
"""
|
|
Escape `</script>` tags in 'text' so that it can be placed within a `<script>` block without
|
|
accidentally closing it. A '-' character will be inserted for each time it is escaped:
|
|
`<-/script>`, `<--/script>` etc.
|
|
"""
|
|
return SCRIPT_RE.sub(r'<-\1/script>', text)
|
|
|
|
|
|
SLUGIFY_RE = re.compile(r'[^\w\s-]', re.UNICODE)
|
|
|
|
|
|
def cautious_slugify(value):
|
|
"""
|
|
Convert a string to ASCII exactly as Django's slugify does, with the exception
|
|
that any non-ASCII alphanumeric characters (that cannot be ASCIIfied under Unicode
|
|
normalisation) are escaped into codes like 'u0421' instead of being deleted entirely.
|
|
|
|
This ensures that the result of slugifying e.g. Cyrillic text will not be an empty
|
|
string, and can thus be safely used as an identifier (albeit not a human-readable one).
|
|
"""
|
|
value = force_text(value)
|
|
|
|
# Normalize the string to decomposed unicode form. This causes accented Latin
|
|
# characters to be split into 'base character' + 'accent modifier'; the latter will
|
|
# be stripped out by the regexp, resulting in an ASCII-clean character that doesn't
|
|
# need to be escaped
|
|
value = unicodedata.normalize('NFKD', value)
|
|
|
|
# Strip out characters that aren't letterlike, underscores or hyphens,
|
|
# using the same regexp that slugify uses. This ensures that non-ASCII non-letters
|
|
# (e.g. accent modifiers, fancy punctuation) get stripped rather than escaped
|
|
value = SLUGIFY_RE.sub('', value)
|
|
|
|
# Encode as ASCII, escaping non-ASCII characters with backslashreplace, then convert
|
|
# back to a unicode string (which is what slugify expects)
|
|
value = value.encode('ascii', 'backslashreplace').decode('ascii')
|
|
|
|
# Pass to slugify to perform final conversion (whitespace stripping, applying
|
|
# mark_safe); this will also strip out the backslashes from the 'backslashreplace'
|
|
# conversion
|
|
return slugify(value)
|