[MOV][FIX] rating: move rating values, thresholds and converters into a single file
Purpose is to try to have a single entry point when dealing with rating values. Also fix incoherency introduced with odoo/odoo@32ba417549 where grades are not coherent in all computation. This new computation seems better as bad is now limited to <= 1 everywhere. Task-2728564 Part-of: odoo/odoo#82792
This commit is contained in:
@@ -0,0 +1,73 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
# Part of Odoo. See LICENSE file for full copyright and licensing details.
|
||||
|
||||
import operator
|
||||
|
||||
from odoo.tools.float_utils import float_compare
|
||||
|
||||
RATING_AVG_TOP = 3.66
|
||||
RATING_AVG_OK = 2.33
|
||||
RATING_AVG_MIN = 1
|
||||
|
||||
RATING_LIMIT_SATISFIED = 4
|
||||
RATING_LIMIT_OK = 3
|
||||
RATING_LIMIT_MIN = 1
|
||||
RATING_TEXT = [
|
||||
('top', 'Satisfied'),
|
||||
('ok', 'Okay'),
|
||||
('ko', 'Dissatisfied'),
|
||||
('none', 'No Rating yet'),
|
||||
]
|
||||
|
||||
OPERATOR_MAPPING = {
|
||||
'=': operator.eq,
|
||||
'!=': operator.ne,
|
||||
'<': operator.lt,
|
||||
'<=': operator.le,
|
||||
'>': operator.gt,
|
||||
'>=': operator.ge,
|
||||
}
|
||||
|
||||
def _rating_avg_to_text(rating_avg):
|
||||
if float_compare(rating_avg, RATING_AVG_TOP, 2) >= 0:
|
||||
return 'top'
|
||||
if float_compare(rating_avg, RATING_AVG_OK, 2) >= 0:
|
||||
return 'ok'
|
||||
if float_compare(rating_avg, RATING_AVG_MIN, 2) >= 0:
|
||||
return 'ko'
|
||||
return 'none'
|
||||
|
||||
def _rating_assert_value(rating_value):
|
||||
assert 0 <= rating_value <= 5
|
||||
|
||||
def _rating_to_grade(rating_value):
|
||||
""" From a rating value give a text-based mean value. """
|
||||
_rating_assert_value(rating_value)
|
||||
if rating_value >= RATING_LIMIT_SATISFIED:
|
||||
return 'great'
|
||||
if rating_value >= RATING_LIMIT_OK:
|
||||
return 'okay'
|
||||
return 'bad'
|
||||
|
||||
def _rating_to_text(rating_value):
|
||||
""" From a rating value give a text-based mean value. """
|
||||
_rating_assert_value(rating_value)
|
||||
if rating_value >= RATING_LIMIT_SATISFIED:
|
||||
return 'top'
|
||||
if rating_value >= RATING_LIMIT_OK:
|
||||
return 'ok'
|
||||
if rating_value >= RATING_LIMIT_MIN:
|
||||
return 'ko'
|
||||
return 'none'
|
||||
|
||||
def _rating_to_threshold(rating_value):
|
||||
""" From a rating value, return the thresholds in form of 0-1-3-5 used
|
||||
notably for images. """
|
||||
_rating_assert_value(rating_value)
|
||||
if rating_value >= RATING_LIMIT_SATISFIED:
|
||||
return 5
|
||||
if rating_value >= RATING_LIMIT_OK:
|
||||
return 3
|
||||
if rating_value >= RATING_LIMIT_MIN:
|
||||
return 1
|
||||
return 0
|
||||
Reference in New Issue
Block a user