Test file for text2int

.gitignore

@@ -103,3 +103,4 @@ docs/**/*.html
 /data/call_history_text2int_1_bash.csv
 /data/processed_call_history_sentiment_1_bash.csv
 /data/processed_call_history_text2int_1_bash.csv
+/data/text2int_results.csv

.gitlab-ci.yml

@@ -0,0 +1,15 @@
+# .gitlab-ci.yml keyword reference
+# https://docs.gitlab.com/ee/ci/yaml/
+
+stages:
+  - test
+
+test:
+  stage: test
+  image: python:3.10-slim-buster
+  before_script:
+    - echo "Test stage has started!"
+  script:
+    - pip install -r requirements.txt && pytest --verbose
+  after_script:
+    - echo "Test stage has finished!"

app.py

@@ -293,4 +293,5 @@ with gr.Blocks() as html_block:
 
 html_block.predict_durations = []
 
-html_block.launch()
+if __name__ == "__main__":
+    html_block.launch()

data/test_text2int.all.csv

@@ -0,0 +1,42 @@
+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
+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

data/text2int_results.csv

@@ -0,0 +1,42 @@
+input,output,text2int,score
+fourteen,14,14.0,True
+one thousand four hundred ninety two,1492,1492.0,True
+Fourteen Hundred Ninety-Two,1492,1492.0,True
+forteen,14,14.0,True
+seventeen-thousand and seventy two,17072,17072.0,True
+two hundred and nine,209,209.0,True
+ninety nine hundred and seventy seven,9977,276.0,False
+seven thousands,7000,217.0,False
+2 hundreds,200,60.0,False
+99 thousands and one,9901,162.0,False
+"forty-five thousand, seven hundred and nine",45709,192.0,False
+eighty eight hundred eighty,8880,268.0,False
+a hundred hundred,10000,3200.0,False
+a hundred thousand,100000,4100.0,False
+a hundred million,100000000,3100000000.0,False
+nineteen ninety nine,1999,1809.0,False
+forteen twenty seven,1427,307.0,False
+one nine eight five,1985,1985.0,True
+nineteen eighty-five,1985,1605.0,False
+six,6,6.0,True
+eighty,80,80.0,True
+eight three seven five three zero nine,8375308,8375309.0,False
+oh one,1,1.0,True
+six oh 1,601,601.0,True
+sex,6,6.0,True
+eight oh,80,8.0,False
+ate,8,8.0,True
+double eight,88,,False
+eight three seven five three O nine,8375308,8375309.0,False
+eight three seven five three oh nine,8375308,8375309.0,False
+eight three seven five three 0 nine,8375308,8375309.0,False
+eight three seven five three oh ni-ee-ine,8375308,115.0,False
+two thousand ten,2010,2010.0,True
+two thousand and ten,2010,2010.0,True
+twelve million,12000000,12000000.0,True
+8 billion,8000000000,8000000000.0,True
+twenty ten,2010,2010.0,True
+thirty-two hundred,3200,3200.0,True
+nine,9,9.0,True
+forty two,42,42.0,True
+1 2 three,123,123.0,True

requirements.txt

@@ -1,9 +1,10 @@
-spacy
-pandas
-pandas-gbq
-gradio==3.14.0
-python-dotenv
-transformers
-torch
-httpx
-matplotlib
+spacy==3.4.*
+pandas==1.5.*
+pandas-gbq==0.19.*
+gradio==3.14.*
+python-dotenv==0.21.*
+transformers==4.24.*
+torch==1.12.*
+httpx==0.23.*
+matplotlib==3.6.*
+pytest==7.2.*

tests/test_text2int.py

@@ -0,0 +1,43 @@
+import unittest
+
+import pandas as pd
+
+from app import text2int
+
+
+TEST_DATA_FILE = "data/test_text2int.all.csv"
+
+
+class TestStringMethods(unittest.TestCase):
+
+    def setUp(self):
+        pass
+
+    @staticmethod
+    def get_text2int(text):
+        """Makes a post request to the endpoint"""
+        r = None
+        try:
+            r = text2int(text)
+        except:
+            pass
+        return r
+
+
+    def test_acc_score_text2int(self):
+        """Calculates accuracy score for endpoint"""
+        df = pd.read_csv(TEST_DATA_FILE)
+
+        df["text2int"] = df["input"].apply(func=self.get_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)
+
+        self.assertGreaterEqual(acc_score, 0.8, f"Accuracy score: '{acc_score}'. Value is too low!")
+
+
+if __name__ == '__main__':
+    unittest.main()