add method POST terms validation along with tests and utility functions

gargantools/utils/terms_utils.py

+import petl
+
+from gargantools.utils.utils import check_unacceptedCharactersQuote, check_unacceptedCharactersTab
+
+def check_statusCell(cell):
+    return (cell in ["MapTerm", "CandidateTerm", "StopTerm"]) == True
+
+def check_fileContent(fileContent):
+    header = ("status", "label")
+    constraints = [
+        dict(
+            name = "acceptedVariable", field = "status", assertion = lambda x : check_statusCell(x)
+        ),
+        dict(
+            name = "str_formatQuote", field = "label", assertion = lambda x : check_unacceptedCharactersQuote(x)
+        ),
+        dict(
+            name = "str_formatTab", field = "label", assertion = lambda x : check_unacceptedCharactersTab(x)
+        )
+    ]
+    
+    dataTable = fileContent.values.tolist()
+    dataTable.insert(0, fileContent.columns.to_list())
+
+    problemCells = petl.validate(dataTable, constraints, header)
+
+    if problemCells.len() > 1:
+        return False, problemCells
+    else:
+        return True, None

tests/test_files/terms/correct.csv

+status	label
+MapTerm	a
+CandidateTerm	b
+StopTerm	c

tests/test_files/terms/incorrect.csv

+status	label
+MapTerm	a"""
+CandidateTerm	b
+StopTerm	"c	"

tests/test_files/terms/incorrectDelimiter.csv

+status,label
+MapTerm,a
+CandidateTerm,b
+StopTerm,c

tests/test_files/terms/incorrectHeader.csv

+status	labels
+MapTerm	a
+CandidateTerm	b
+StopTerm	c

tests/test_files/terms/malformed.csv

+status	label
+MapTerm
+CandidateTerm	b
+StopTerm	s	c

tests/test_terms/test_termsValidation.py

+from io import BytesIO
+from itertools import product
+from werkzeug.datastructures import FileStorage
+
+import pytest
+
+from tests.conftest import id_files
+
+
+def test_missingFile(client):
+    response = client.post('/terms')
+    
+    assert response.data == b"Bad request, missing file\n"
+    assert response.status_code == 400
+
+@pytest.mark.parametrize("test_file_copy", product(["incorrect.csv", "malformed.csv", "incorrectDelimiter.csv", "incorrectHeader.csv", "incorrectExtension"], ["terms"]), indirect=True, ids=lambda param : id_files(param))
+def test_incorrectFile(client, test_file_copy):
+    with open(test_file_copy, "rb") as f:
+        f_data = BytesIO(f.read())
+        file_storage = FileStorage(f_data, filename=test_file_copy)
+        response = client.post('/terms', data={'file': file_storage}, content_type='multipart/form-data')
+
+        print(test_file_copy)
+
+        if "incorrect.csv" in test_file_copy:
+            assert response.data == b'Incorrect file - File is not compatible with GarganText\n'
+        elif "malformed.csv" in test_file_copy:
+            assert response.data == b"Error while getting file delimiter.\n Can be the following : \n - File delimiter not found \n - Incorrect file delimiter, should be a tabulation \n - File is malformed\n"
+        elif "incorrectDelimiter.csv" in test_file_copy:
+            assert response.data == b"Error while getting file delimiter.\n Can be the following : \n - File delimiter not found \n - Incorrect file delimiter, should be a tabulation \n - File is malformed\n"
+        elif "incorrectHeader.csv" in test_file_copy:
+            assert b"Some column names were not found" in response.data
+        elif "incorrectExtension" in test_file_copy:
+            assert response.data == b"Incorrect file format or file format not found\n"
+            assert response.status_code == 400
+        else:
+            assert False
+
+        if "incorrectExtension" not in test_file_copy:
+            assert response.status_code == 422
+
+@pytest.mark.parametrize("test_file_copy", product(["correct.csv"], ["terms"]), indirect=True, ids=lambda param : id_files(param))
+def test_correctFile(client, test_file_copy):
+    with open(test_file_copy, "rb") as f:
+        f_data = BytesIO(f.read())
+        file_storage = FileStorage(f_data, filename=test_file_copy)
+        response = client.post('/terms', data={'file': file_storage}, content_type='multipart/form-data')
+
+        assert response.data == b"Correct file\n"
+        assert response.status_code == 200

tests/test_terms/test_terms_utils.py

+from io import BytesIO
+from itertools import product
+import pytest
+
+from gargantools.utils.terms_utils import check_fileContent
+from gargantools.utils.utils import get_fileContent
+from tests.conftest import id_files
+
+
+@pytest.mark.parametrize("test_file_copy", product(["correct.csv"], ["terms"]), indirect=True, ids=lambda param : id_files(param))
+def test_check_fileContent(test_file_copy):
+    with open(test_file_copy, "r") as f:
+        f_data = BytesIO(f.read().encode("utf-8"))
+        assert check_fileContent(get_fileContent(f_data, '\t')) == (True, None)