Merge pull request #26 from thucar/develop

Initial support for Estonian language.

resources/default-config.clj

@@ -45,6 +45,12 @@
                     "en.finance"
                     "en.communication"
                     ]}
+:et$core {:corpus [
+                    "et.numbers"
+                    ]
+           :rules  [
+                    "et.numbers"
+                    ]}
 :it$core {:corpus [
                     "it.time"
                     "it.numbers"

resources/duckling/corpus/et.communication.clj

@@ -0,0 +1,28 @@
+(
+ {}
+
+ "650-283-4757"
+ "+1 6502834757"
+ "+33 4 76095663"
+ "06 2070 2220"
+ "(650)-283-4757 ext 897"
+ (fn [token _] (and (= :phone-number (:dim token))
+                    (= (:text token) (:val token))))
+
+ "http://www.bla.com"
+ "www.bla.com:8080/path"
+ "https://myserver?foo=bar"
+ "cnn.com/info"
+ "bla.com/path/path?ext=%23&foo=bla"
+ "localhost"
+ "localhost:8000"
+ "http://kimchi" ; local url
+ (fn [token _] (and (= :url (:dim token))
+                    (= (:text token) (:val token))))
+
+ "alex@wit.ai"
+ "alex.lebrun@mail.wit.com"
+ (fn [token _] (and (= :email (:dim token))
+                    (= (:text token) (:val token))))
+
+ )

resources/duckling/corpus/et.finance.clj

@@ -0,0 +1,60 @@
+(
+
+{}
+
+"$10"
+"10$"
+"kümme dollarit"
+(money 10 "$")
+
+;"alla $15"
+;"kuni 15$"
+;"vähem kui 15$"
+;"maksimaalselt 15$"
+;"vähem kui viisteist dollarit"
+;(money 15 "$" "<")
+
+"kümme senti"
+(money 10 "cent")
+
+"$10,000"
+"$10 000"
+"10K$"
+"$10k"
+(money 10000 "$")
+
+"USD1.23"
+(money 1.23 "USD")
+
+"2 dollarit ja 23 senti"
+"kaks dollarit 23 senti"
+"2 dollarit 23"
+"kaks dollarit ja 23"
+(money 2.23 "$")
+
+"20€"
+"20 eurot"
+"20 Euri"
+"20 Eurot"
+"EUR 20"
+(money 20 "EUR")
+
+"EUR29.99"
+(money 29.99 "EUR")
+
+"3 taala"
+(money 3) ; unknown unit
+
+"£9"
+"üheksa naela"
+"üheksa naelsterlingit"
+(money 9 "£")
+
+"GBP3.01"
+"GBP 3.01"
+(money 3.01 "GBP")
+
+)
+
+;umbes $200
+;$200 ja $300 vahel

resources/duckling/corpus/et.measure.clj

@@ -0,0 +1,60 @@
+(
+	; Context map
+	{}
+
+	; Distance
+
+	"3 kilomeetrit"
+	"3 km"
+	"3km"
+	"3k"
+	(distance 3 "kilometre" {:value 3000 :unit "metre"})
+
+	"3.0 km"
+	(distance 3.0 "kilometre" {:value 3000.0 :unit "metre"})
+
+	"8 miili"
+	(distance 8 "mile" {:value 12872 :unit "metre"})
+
+	"9m"
+	(distance 9 "m")
+
+	"2cm"
+	"2 sentimeetrit"
+	(distance 2 "centimetre" {:value 0.02 :unit "metre"})
+
+	; Volume
+
+	"250 milliliitrit"
+	"250ml"
+	"250 ml"
+	(volume  250 "millilitre" {:value 0.25 :unit "litre"})
+
+	"2 liitrit"
+	(volume 2 "litre")
+
+	"3 gallonit"
+	"3 gal"
+	(volume 3 "gallon" {:value 11.355 :unit "litre"})
+
+	"3 hektoliitrit"
+	"3 kuupliitrit"
+	(volume 300 "litre" {:value 30 :unit "litre"})
+
+	"pool liitrit"
+	(volume 0.5 "litre")
+
+	; Quantity
+	"kaks naela liha"
+	(quantity 2 "pound" "meat")
+
+	"nael"
+	(quantity 1 "pound")
+
+	"liha"
+	(quantity 1 nil "meat")
+
+	"3 klaasi suhkrut"
+	(quantity 3 "cup" "sugar")
+)
+

resources/duckling/corpus/et.metrics.clj

@@ -0,0 +1,23 @@
+(
+  ; Context map
+  {}
+
+  "miili"
+  "miil"
+  (metric :length 1609.34)
+
+  "km"
+  "kilomeeter"
+  "kms"
+  "kilomeetrit"
+  (metric :length 1000)
+
+  "meetrit"
+  "meeter"
+  (metric :length 1)
+
+  "4 miili"
+  "4 miili pikkune"
+  (metric :length 6437.36)
+
+  )

resources/duckling/corpus/et.numbers.clj

@@ -0,0 +1,92 @@
+(
+  ; Context map
+  {}
+
+  "0"
+  "null"
+  (number 0)
+
+  "1"
+  "üks"
+  (number 1)
+
+  "33"
+  "kolmkümmend kolm"
+  "kolmkend kolm"
+  (number 33)
+
+  "14"
+  "neliteist"
+  (number 14)
+
+  "16"
+  "kuusteist"
+  (number 16)
+
+  "17"
+  "seitseteist"
+  (number 17)
+
+  "18"
+  "kaheksateist"
+  (number 18)
+
+  "1.1"
+  "1.10"
+  "01.10"
+  (number 1.1)
+
+  "0.77"
+  ".77"
+  (number 0.77)
+
+  "100,000"
+  "100000"
+  "100K"
+  "100k"
+  "100 000"
+  (number 100000)
+
+  "3M"
+  "3000K"
+  "3000000"
+  "3,000,000"
+  "3 000 000"
+  (number 3000000)
+
+  "1,200,000"
+  "1200000"
+  "1.2M"
+  "1200K"
+  ".0012G"
+  "1 200 000"
+  (number 1200000)
+
+  "- 1,200,000"
+  "-1200000"
+  "miinus 1,200,000"
+  "-1.2M"
+  "-1200K"
+  "-.0012G"
+  (number -1200000)
+
+  "viis tuhat"
+  (number 5000)
+
+  "kakssada tuhat"
+  (number 200000)
+
+  "kakskümmend üks tuhat üksteist"
+  (number 21011)
+
+  "seitsesada kakskümmend üks tuhat kaksteist"
+  (number 721012)
+
+  "kolmkümmend üks miljonit kakssada viiskümmend kuus tuhat seitsesada kakskümmend üks"
+  (number 31256721)
+
+  "4s"
+  "neljas"
+  (ordinal 4)
+)
+

resources/duckling/corpus/et.temperature.clj

@@ -0,0 +1,27 @@
+(
+	; Context map
+	{}
+
+	; Temperature
+
+	"37°C"
+	"37° celsiuse kraadi"
+	"37 kraadi Celsiuse järgi"
+	"kolmkümmend seitse celsiust"
+	"37 kraadi celsiust"
+	"37 kraadi C"
+	"kolmkümmend seitse kraadi celsiuse skaalal"
+	(temperature 37 "celsius")
+
+	"70°F"
+	"70° Fahrenheiti kraadi"
+	"70 kraadi F"
+	"seitsekümmend Fahrenheiti"
+	(temperature 70 "fahrenheit")
+
+	"45°"
+	"45 kraadi"
+	(temperature 45 "degree")
+
+)
+