@@ -56,3 +56,86 @@ describe('parser', function() {
5656 } ) ;
5757 } )
5858} )
59+
60+ describe ( 'language_classifier' , function ( ) {
61+ it ( 'should classify phrases' , function ( ) {
62+ const cases = [
63+ {
64+ text : 'street' ,
65+ expected : [ { language : 'en' , probability : 0.9975550392228959 } ]
66+ } ,
67+ {
68+ text : 'calle' ,
69+ expected : [ { language : 'es' , probability : 0.9948278315613933 } ]
70+ } ,
71+ {
72+ text : '200 santa monica pier santa monica coahuila 90401 usa' ,
73+ expected : [ { language : 'es' , probability : 0.9889375382113144 } ]
74+ } ,
75+ {
76+ text : '200 santa monica pier santa monica compania 90401 usa' ,
77+ expected : [
78+ { language : 'es' , probability : 0.6896583016342134 } ,
79+ { language : 'it' , probability : 0.12355268595569942 } ,
80+ { language : 'en' , probability : 0.09448842175870104 } ,
81+ { language : 'pt' , probability : 0.09152280207660235 }
82+ ]
83+ } ,
84+ {
85+ text : '200 santa monica pier santa monica compania anonima 90401 usa' ,
86+ expected : [
87+ { language : 'en' , probability : 0.39249744852100377 } ,
88+ { language : 'pt' , probability : 0.27690798537637573 } ,
89+ { language : 'es' , probability : 0.2631088374468901 } ,
90+ { language : 'it' , probability : 0.06642801798298495 }
91+ ]
92+ } ,
93+ {
94+ text : '200 santa monica pier santa monica calle 90401 usa' ,
95+ expected : [
96+ { language : 'es' , probability : 0.9332794851572307 } ,
97+ { language : 'it' , probability : 0.0653229950496308 }
98+ ]
99+ } ,
100+ {
101+ text : '200 santa monica pier santa monica ca 90401 usa' ,
102+ expected : [
103+ { language : 'es' , probability : 0.6860753090491215 } ,
104+ { language : 'it' , probability : 0.16420520901155986 } ,
105+ { language : 'en' , probability : 0.08797427320205269 } ,
106+ { language : 'pt' , probability : 0.061694606284459816 }
107+ ]
108+ } ,
109+ {
110+ text : '200 santa monica pier santa monica casa 90401 usa' ,
111+ expected : [
112+ { language : 'it' , probability : 0.7099331948983176 } ,
113+ { language : 'pt' , probability : 0.1484441794525817 } ,
114+ { language : 'en' , probability : 0.09171194632862785 }
115+ ]
116+ } ,
117+ {
118+ text : '200 santa monica pier santa monica cagliari 90401 usa' ,
119+ expected : [ { language : 'it' , probability : 0.9409304710772296 } ]
120+ } ,
121+ {
122+ text : '200 santa monica pier santa monica california 90401 usa' ,
123+ expected : [
124+ { language : 'en' , probability : 0.529208078531959 } ,
125+ { language : 'pt' , probability : 0.26387306487348855 } ,
126+ { language : 'es' , probability : 0.20130465470054235 }
127+ ]
128+ } ,
129+ {
130+ text : '200 santa monica pier santa monica companhia anonima 90401 usa' ,
131+ expected : [
132+ { language : 'pt' , probability : 0.8631314862441015 } ,
133+ { language : 'en' , probability : 0.1090210811124842 }
134+ ]
135+ }
136+ ] ;
137+ for ( const c of cases ) {
138+ assert . deepEqual ( postal . language_classifier ( c . text ) , c . expected , c . text ) ;
139+ }
140+ } ) ;
141+ } )
0 commit comments