Merge branch 'dev-test' into 'main'

Update test file

See merge request tangibleai/community/mathtext-fastapi!2

app.py

@@ -1,5 +1,5 @@
 """FastAPI endpoint
-To run locally use 'uvicorn modules.app:app --host localhost --port 7860'
+To run locally use 'uvicorn app:app --host localhost --port 7860'
 """
 
 from fastapi import FastAPI, Request
@@ -76,4 +76,5 @@ async def evaluate_user_message_with_nlu_api(request: Request):
     prepare_message_data_for_logging(message_data)
 
     int_data_dict = {'type': 'integer', 'data': int_api_resp}
-    return JSONResponse(content=int_data_dict)
+    return JSONResponse(content=int_data_dict)
+

data/master_test_text2int.csv

@@ -0,0 +1,90 @@
+input,output
+fourteen,14
+forteen,14
+one thousand four hundred ninety two,1492
+one thousand ninety two,1092
+Fourteen Hundred Ninety-Two,1492
+Fourteen Hundred,1400
+Ninety nine,99
+fifteen thousand five hundred-sixty,15560
+three hundred fifty,350
+one nine eight five,1985
+nineteen eighty-five,1985
+oh one,1
+six oh 1,601
+sex,6
+six,6
+eight oh,80
+eighty,80
+ate,8
+double eight,88
+eight three seven five three O nine,8375309
+eight three seven five three oh nine,8375309
+eight three seven five three zero nine,8375309
+eight three seven five three oh ni-ee-ine,8375309
+two eight,28
+seven oh eleven,7011
+seven elevens,77
+seven eleven,711
+ninety nine oh five,9905
+seven 0 seven 0 seven 0 seven,7070707
+123 hundred,123000
+5 o 5,505
+15 o 5,1505
+15-o 5,1505
+15 o-5,1505
+911-thousand,911000
+twenty-two twenty-two,2222
+twenty-two twenty-twos,484
+four eighty four,484
+four eighties,320
+four eighties and nine nineties,1130
+ninety nine hundred and seventy seven,9977
+seven thousands,7000
+2 hundreds,200
+99 thousands and one,99001
+"forty-five thousand, seven hundred and nine",45709
+eighty eight hundred eighty,8880
+a hundred hundred,10000
+a hundred thousand,100000
+a hundred million,100000000
+nineteen ninety nine,1999
+forteen twenty seven,1427
+seventeen-thousand and seventy two,17072
+two hundred and nine,209
+two thousand ten,2010
+two thousand and ten,2010
+twelve million,12000000
+8 billion,8000000000
+twenty ten,2010
+thirty-two hundred,3200
+nine,9
+forty two,42
+1 2 three,123
+fourtean,14
+one tousand four hundred ninty two,1492
+Furteen Hundrd Ninety-Too,1492
+forrteen,14
+sevnteen-thosand and seventy two,17072
+ninety nine hundred ad seventy seven,9977
+seven thusands,7000
+2 hunreds,200
+99 tousands and one,99001
+eighty ate hundred eighty,8880
+fourteen Hundred,1400
+8 Bilion,8000000000
+one million three thousand one,1003001
+four million nine thousand seven,4009007
+two million five hundred thousand,2500000
+two tousand ten,2010
+two thousand teen,2010
+tvelve milion,12000000
+tventy ten,2010
+tirty-twoo hunred,3200
+sevn thoosands,7000
+five,5
+ten,10
+one two three and ten,12310
+ONE MILLion three hunded and fiv,1000305
+"50,500 and six",50506
+one_million_and_five,1000005

data/test_data_text2int.csv

@@ -1,43 +0,0 @@
-input,output
-fourteen,14
-one thousand four hundred ninety two,1492
-Fourteen Hundred Ninety-Two,1492
-forteen,14
-seventeen-thousand and seventy two,17072
-two hundred and nine,209
-# Following are the ones that the current text2int.py fails on,-
-# ninety nine hundred and seventy seven,9977
-# seven thousands,7000
-# 2 hundreds,200
-# 99 thousands and one,9901
-# "forty-five thousand, seven hundred and nine",45709
-# eighty eight hundred eighty,8880
-# a hundred hundred,10000
-# a hundred thousand,100000
-# a hundred million,100000000
-# nineteen ninety nine,1999
-# forteen twenty seven,1427
-# one nine eight five,1985
-# nineteen eighty-five,1985
-# six,6
-# eighty,80
-# eight three seven five three zero nine,8375308
-# oh one,1
-# six oh 1,601
-# sex,6
-# eight oh,80
-# ate,8
-# double eight,88
-# eight three seven five three O nine,8375308
-# eight three seven five three oh nine,8375308
-# eight three seven five three 0 nine,8375308
-# eight three seven five three oh ni-ee-ine,8375308
-# two thousand ten,2010
-# two thousand and ten,2010
-# twelve million,12000000
-# 8 billion,8000000000
-# twenty ten,2010
-# thirty-two hundred,3200
-# nine,9
-# forty two,42
-# 1 2 three,123

tests/test_text2int.py

@@ -1,10 +1,11 @@
 import unittest
 
+import pandas as pd
 from fastapi.testclient import TestClient
 
 from app import app
 
-TEST_DATA_FILE = "data/test_data_text2int.csv"
+TEST_DATA_FILE = "data/master_test_text2int.csv"
 
 client = TestClient(app)
 
@@ -12,49 +13,39 @@ client = TestClient(app)
 class TestStringMethods(unittest.TestCase):
 
     def setUp(self):
+        """Creates a fastapi test client"""
         self.client = TestClient(app)
 
-    def test_1(self):
+    def get_response_text2int(self, text):
+        """Makes a post request to the endpoint"""
+        r = None
+        try:
+            r = self.client.post("/text2int", json={"content": text}) \
+                .json().get("message")
+        except:
+            pass
+        return r
+
+    def test_endpoint_text2int(self):
+        """Tests if endpoint is working"""
         response = self.client.post("/text2int",
                                     json={"content": "fourteen"}
                                     )
         self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "14")
 
-    def test_2(self):
-        response = self.client.post("/text2int",
-                                    json={"content": "one thousand four hundred ninety two"}
-                                    )
-        self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "1492")
+    def test_acc_score_text2int(self):
+        """Calculates accuracy score for endpoint"""
+        df = pd.read_csv(TEST_DATA_FILE)
 
-    def test_3(self):
-        response = self.client.post("/text2int",
-                                    json={"content": "Fourteen Hundred Ninety-Two"}
-                                    )
-        self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "1492")
+        df["text2int"] = df["input"].apply(func=self.get_response_text2int)
+        df["score"] = df[["output", "text2int"]].apply(
+            lambda row: row[0] == row[1],
+            axis=1
+        )
+        df.to_csv("data/text2int_results.csv", index=False)
+        acc_score = df["score"].mean().__round__(2)
 
-    def test_4(self):
-        response = client.post("/text2int",
-                               json={"content": "forteen"}
-                               )
-        self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "14")
-
-    def test_5(self):
-        response = client.post("/text2int",
-                               json={"content": "seventeen-thousand and seventy two"}
-                               )
-        self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "17072")
-
-    def test_6(self):
-        response = client.post("/text2int",
-                               json={"content": "two hundred and nine"}
-                               )
-        self.assertEqual(response.status_code, 200)
-        self.assertEqual(str(response.json()["message"]), "209")
+        self.assertGreaterEqual(acc_score, 0.8, f"Accuracy score: '{acc_score}'. Value is too low!")
 
 
 if __name__ == '__main__':