Merge pull request #265 from NHSDigital/relative-test-data

Make paths to test data relative

Author: thomasleese

libs/mavis_constants.py

@@ -83,8 +83,8 @@ class mavis_file_types(Enum):
 
 
 class test_data_file_paths:
-    PARENTAL_CONSENT_HPV: Final[str] = "test_data/ParentalConsent_HPV.xlsx"
-    PARENTAL_CONSENT_DOUBLES: Final[str] = "test_data/ParentalConsent_Doubles.xlsx"
+    PARENTAL_CONSENT_HPV: Final[str] = "ParentalConsent_HPV.xlsx"
+    PARENTAL_CONSENT_DOUBLES: Final[str] = "ParentalConsent_Doubles.xlsx"
     VACCS_POSITIVE: Final[str] = "VACCS_HPV_POSITIVE"
     VACCS_NEGATIVE: Final[str] = "VACCS_HPV_NEGATIVE"
     VACCS_HIST_POSITIVE: Final[str] = "VACCS_HIST_HPV_POSITIVE"

libs/test_data.py

@@ -1,4 +1,4 @@
-import pathlib
+from pathlib import Path
 from typing import Optional
 
 import nhs_number
@@ -18,14 +18,14 @@ class TestData:
     A class to handle operations related to test data.
     """
 
+    template_path = Path("test_data")
+    working_path = Path("working")
+
     def __init__(self):
-        """
-        Initialize the testdata_operations class.
-        """
-        self.mapping_df: pd.DataFrame = pd.read_csv("test_data/file_mapping.csv")
+        self.file_mapping = pd.read_csv(self.template_path / "file_mapping.csv")
 
-    def read_file(self, path):
-        return pathlib.Path(path).read_text(encoding="utf-8")
+    def read_file(self, filename):
+        return (self.template_path / filename).read_text(encoding="utf-8")
 
     def create_file_from_template(
         self,
@@ -85,9 +85,8 @@ def create_file_from_template(
 
         filename = f"{file_name_prefix}{get_current_datetime()}.csv"
 
-        path = pathlib.Path("working") / filename
+        path = self.working_path / filename
         path.write_text("\n".join(_file_text), encoding="utf-8")
-
         return str(path)
 
     def get_new_nhs_no(self, valid=True) -> str:
@@ -119,23 +118,27 @@ def get_expected_errors(self, file_path: str):
         return file_content.splitlines() if file_content else None
 
     def read_spreadsheet(
-        self, file_path: str, clean_df: bool = True, sheet_name: str = "Sheet1"
+        self, filename: str, sheet_name: str = "Sheet1"
     ) -> pd.DataFrame:
         """
         Read a spreadsheet into a DataFrame.
 
         Args:
-            file_path (str): Path to the spreadsheet file.
-            clean_df (bool, optional): Whether to clean the DataFrame. Defaults to True.
+            filename (str): Path to the spreadsheet file.
             sheet_name (str, optional): Name of the sheet to read. Defaults to "Sheet1".
 
         Returns:
             pd.DataFrame: DataFrame containing the spreadsheet data.
         """
-        _df = pd.read_excel(
-            file_path, sheet_name=sheet_name, header=0, dtype="str", index_col=0
+        return self.clean_df(
+            pd.read_excel(
+                self.template_path / filename,
+                sheet_name=sheet_name,
+                header=0,
+                dtype="str",
+                index_col=0,
+            )
         )
-        return self.clean_df(df=_df) if clean_df else _df
 
     def clean_df(self, df: pd.DataFrame) -> pd.DataFrame:
         """
@@ -170,7 +173,7 @@ def get_file_paths(
         Returns:
             tuple[str, str]: Input and output file paths.
         """
-        query = self.mapping_df.query("ID==@file_paths")
+        query = self.file_mapping.query("ID==@file_paths")
 
         _input_template_path: str = query["INPUT_TEMPLATE"].to_string(index=False)
         _output_template_path: str = query["OUTPUT_TEMPLATE"].to_string(index=False)
@@ -210,17 +213,16 @@ def create_child_list_from_file(self, file_path: str, file_type: mavis_file_type
         _names_list = _file_df[_cols[0]] + ", " + _file_df[_cols[1]].values.tolist()
         return _names_list
 
-    def get_session_id(self, excel_path: str) -> str:
+    def get_session_id(self, path: str) -> str:
         """
         Get the session ID from an Excel file.
 
         Args:
-            excel_path (str): Path to the Excel file.
+            path (str): Path to the Excel file.
 
         Returns:
             str: Session ID.
         """
-        _df = self.read_spreadsheet(
-            file_path=excel_path, clean_df=False, sheet_name="Vaccinations"
-        )
-        return _df["SESSION_ID"].iloc[0]
+
+        data_frame = pd.read_excel(path, sheet_name="Vaccinations")
+        return data_frame["SESSION_ID"].iloc[0]

pages/sessions.py

@@ -313,7 +313,7 @@ def download_offline_recording_excel(self) -> str:
 
     def get_session_id_from_offline_excel(self):
         file_path = self.download_offline_recording_excel()
-        return self.tdo.get_session_id(excel_path=file_path)
+        return self.tdo.get_session_id(file_path)
 
     def add_gillick_competence(
         self, is_competent: bool, competence_details: str

test_data/file_mapping.csv

@@ -1,54 +1,54 @@
 ID,INPUT_TEMPLATE,OUTPUT_TEMPLATE,FILE_PREFIX
-VACCS_HPV_POSITIVE,test_data/vaccs/i_positive.csv,test_data/vaccs/o_positive.csv,hpvpos
-VACCS_HPV_NEGATIVE,test_data/vaccs/i_negative.csv,test_data/vaccs/o_negative.csv,hpvneg
-VACCS_HIST_HPV_POSITIVE,test_data/vaccs/i_hist_positive.csv,test_data/vaccs/o_hist_positive.csv,hpvhistpos
-VACCS_HIST_HPV_NEGATIVE,test_data/vaccs/i_hist_negative.csv,test_data/vaccs/o_hist_negative.csv,hpvhistneg
-VACCS_HPV_DUP_1,test_data/vaccs/i_dup_1.csv,test_data/vaccs/o_dup_1.csv,hpvdup1
-VACCS_HPV_DUP_2,test_data/vaccs/i_dup_2.csv,test_data/vaccs/o_dup_2.csv,hpvdup2
-VACCS_HPV_INVALID_STRUCTURE,test_data/vaccs/i_invalid_structure.csv,test_data/vaccs/o_invalid_structure.csv,hpvstruc
-VACCS_HPV_EMPTY_FILE,test_data/vaccs/i_empty.csv,test_data/vaccs/o_empty.csv,hpvempty
-VACCS_HPV_HEADER_ONLY,test_data/vaccs/i_header_only.csv,test_data/vaccs/o_header_only.csv,hpvheadonly
-VACCS_HPV_MAV_853,test_data/vaccs/i_mav_853.csv,test_data/vaccs/o_mav_853.csv,hpvmav853
-VACCS_HPV_MAV_855,test_data/vaccs/i_mav_855.csv,test_data/vaccs/o_mav_855.csv,hpvmav855
-VACCS_HPV_DOSE_TWO,test_data/vaccs/i_dose_two.csv,test_data/vaccs/o_dose_two.csv,hpvdose2
-VACCS_MAV_1080,test_data/vaccs/i_mav_1080.csv,test_data/vaccs/o_mav_1080.csv,hpvmav1080
-VACCS_SYSTMONE_POSITIVE,test_data/vaccs/i_systmone_positive.csv,test_data/vaccs/o_systmone_positive.csv,systmonepos
-VACCS_SYSTMONE_NEGATIVE,test_data/vaccs/i_systmone_negative.csv,test_data/vaccs/o_systmone_negative.csv,systmoneneg
-VACCS_SYSTMONE_HIST_NEGATIVE,test_data/vaccs/i_systmone_hist_negative.csv,test_data/vaccs/o_systmone_hist_negative.csv,systmonehistneg
-VACCS_SYSTMONE_MAV_1080,test_data/vaccs/i_systmone_mav_1080.csv,test_data/vaccs/o_systmone_mav_1080.csv,systmonemav1080
-COHORTS_POSITIVE,test_data/cohorts/i_positive.csv,test_data/cohorts/o_positive.csv,cohortpos
-COHORTS_NEGATIVE,test_data/cohorts/i_negative.csv,test_data/cohorts/o_negative.csv,cohortneg
-COHORTS_INVALID_STRUCTURE,test_data/cohorts/i_invalid_structure.csv,test_data/cohorts/o_invalid_structure.csv,cohortstruc
-COHORTS_EMPTY_FILE,test_data/cohorts/i_empty.csv,test_data/cohorts/o_empty.csv,cohortempty
-COHORTS_HEADER_ONLY,test_data/cohorts/i_header_only.csv,test_data/cohorts/o_header_only.csv,cohortheadonly
-CHILD_POSITIVE,test_data/child/i_positive.csv,test_data/child/o_positive.csv,childpos
-CHILD_NEGATIVE,test_data/child/i_negative.csv,test_data/child/o_negative.csv,childneg
-CHILD_INVALID_STRUCTURE,test_data/child/i_invalid_structure.csv,test_data/child/o_invalid_structure.csv,childstruc
-CHILD_EMPTY_FILE,test_data/child/i_empty.csv,test_data/child/o_empty.csv,childempty
-CHILD_HEADER_ONLY,test_data/child/i_header_only.csv,test_data/child/o_header_only.csv,childheadonly
-CHILD_MAV_1080,test_data/child/i_mav_1080.csv,test_data/child/o_mav_1080.csv,childmav1080
-CLASS_POSITIVE,test_data/class_list/i_positive.csv,test_data/class_list/o_positive.csv,classpos
-CLASS_NEGATIVE,test_data/class_list/i_negative.csv,test_data/class_list/o_negative.csv,classneg
-CLASS_INVALID_STRUCTURE,test_data/class_list/i_invalid_structure.csv,test_data/class_list/o_invalid_structure.csv,classstruc
-CLASS_EMPTY_FILE,test_data/class_list/i_empty.csv,test_data/class_list/o_empty.csv,classempty
-CLASS_HEADER_ONLY,test_data/class_list/i_header_only.csv,test_data/class_list/o_header_only.csv,classheadonly
-CLASS_CHILDREN_FILTER,test_data/class_list/i_children_filter.csv,test_data/class_list/o_children_filter.csv,classfilter
-CLASS_YEAR_GROUP,test_data/class_list/i_year_group.csv,test_data/class_list/o_year_group.csv,classyeargrp
-CLASS_MAV_854,test_data/class_list/i_mav_854.csv,test_data/class_list/o_mav_854.csv,classmav854
-CLASS_SESSION_ID,test_data/class_list/i_session_id.csv,test_data/class_list/o_session_id.csv,classsessid
-CLASS_SINGLE_VACC,test_data/class_list/i_single_vacc.csv,test_data/class_list/o_single_vacc.csv,classsinglevacc
-CLASS_MAV_965,test_data/class_list/i_mav_965.csv,test_data/class_list/o_mav_965.csv,classmav965
-CLASS_CHANGE_NHSNO,test_data/class_list/i_change_nhsno.csv,test_data/class_list/o_change_nhsno.csv,classchangenhsno
-CLASS_MAV_1080,test_data/class_list/i_mav_1080.csv,test_data/class_list/o_mav_1080.csv,classmav1080
-COHORTS_NO_CONSENT,test_data/cohorts/i_no_consent.csv,test_data/cohorts/o_no_consent.csv,cohortnoconsent
-COHORTS_CONFLICTING_CONSENT,test_data/cohorts/i_conflicting_consent.csv,test_data/cohorts/o_conflicting_consent.csv,cohortconflict
-COHORTS_E2E_1,test_data/cohorts/i_e2e_1.csv,test_data/cohorts/o_e2e_1.csv,cohort_e2e
-COHORTS_UCR_MATCH,test_data/cohorts/i_ucr_match.csv,test_data/cohorts/o_ucr_match.csv,cohortmatch
-COHORTS_CONSENT_TWICE,test_data/cohorts/i_consent_twice.csv,test_data/cohorts/o_consent_twice.csv,cohorttwice
-COHORTS_CONFLICTING_GILLICK,test_data/cohorts/i_conflicting_gillick.csv,test_data/cohorts/o_conflicting_gillick.csv,cohortgillick
-COHORTS_FULL_NAME,test_data/cohorts/i_full_name.csv,test_data/cohorts/o_full_name.csv,cohortfullname
-COHORTS_MAV_927_PERF,test_data/cohorts/i_mav_927_perf.csv,test_data/cohorts/o_mav_927_perf.csv,cohortmav927perf
-COHORTS_MAV_909,test_data/cohorts/i_mav_909.csv,test_data/cohorts/o_mav_909.csv,cohortmav909
-COHORTS_MAV_853,test_data/cohorts/i_mav_853.csv,test_data/cohorts/o_mav_853.csv,cohortmav853
-CLASS_MOVES_CONFIRM_IGNORE,test_data/class_list/i_moves_1.csv,test_data/class_list/o_moves_1.csv,classmoves1
-CLASS_MOVES_UNKNOWN_HOMESCHOOLED,test_data/class_list/i_moves_2.csv,test_data/class_list/o_moves_2.csv,classmoves2
+VACCS_HPV_POSITIVE,vaccs/i_positive.csv,vaccs/o_positive.csv,hpvpos
+VACCS_HPV_NEGATIVE,vaccs/i_negative.csv,vaccs/o_negative.csv,hpvneg
+VACCS_HIST_HPV_POSITIVE,vaccs/i_hist_positive.csv,vaccs/o_hist_positive.csv,hpvhistpos
+VACCS_HIST_HPV_NEGATIVE,vaccs/i_hist_negative.csv,vaccs/o_hist_negative.csv,hpvhistneg
+VACCS_HPV_DUP_1,vaccs/i_dup_1.csv,vaccs/o_dup_1.csv,hpvdup1
+VACCS_HPV_DUP_2,vaccs/i_dup_2.csv,vaccs/o_dup_2.csv,hpvdup2
+VACCS_HPV_INVALID_STRUCTURE,vaccs/i_invalid_structure.csv,vaccs/o_invalid_structure.csv,hpvstruc
+VACCS_HPV_EMPTY_FILE,vaccs/i_empty.csv,vaccs/o_empty.csv,hpvempty
+VACCS_HPV_HEADER_ONLY,vaccs/i_header_only.csv,vaccs/o_header_only.csv,hpvheadonly
+VACCS_HPV_MAV_853,vaccs/i_mav_853.csv,vaccs/o_mav_853.csv,hpvmav853
+VACCS_HPV_MAV_855,vaccs/i_mav_855.csv,vaccs/o_mav_855.csv,hpvmav855
+VACCS_HPV_DOSE_TWO,vaccs/i_dose_two.csv,vaccs/o_dose_two.csv,hpvdose2
+VACCS_MAV_1080,vaccs/i_mav_1080.csv,vaccs/o_mav_1080.csv,hpvmav1080
+VACCS_SYSTMONE_POSITIVE,vaccs/i_systmone_positive.csv,vaccs/o_systmone_positive.csv,systmonepos
+VACCS_SYSTMONE_NEGATIVE,vaccs/i_systmone_negative.csv,vaccs/o_systmone_negative.csv,systmoneneg
+VACCS_SYSTMONE_HIST_NEGATIVE,vaccs/i_systmone_hist_negative.csv,vaccs/o_systmone_hist_negative.csv,systmonehistneg
+VACCS_SYSTMONE_MAV_1080,vaccs/i_systmone_mav_1080.csv,vaccs/o_systmone_mav_1080.csv,systmonemav1080
+COHORTS_POSITIVE,cohorts/i_positive.csv,cohorts/o_positive.csv,cohortpos
+COHORTS_NEGATIVE,cohorts/i_negative.csv,cohorts/o_negative.csv,cohortneg
+COHORTS_INVALID_STRUCTURE,cohorts/i_invalid_structure.csv,cohorts/o_invalid_structure.csv,cohortstruc
+COHORTS_EMPTY_FILE,cohorts/i_empty.csv,cohorts/o_empty.csv,cohortempty
+COHORTS_HEADER_ONLY,cohorts/i_header_only.csv,cohorts/o_header_only.csv,cohortheadonly
+CHILD_POSITIVE,child/i_positive.csv,child/o_positive.csv,childpos
+CHILD_NEGATIVE,child/i_negative.csv,child/o_negative.csv,childneg
+CHILD_INVALID_STRUCTURE,child/i_invalid_structure.csv,child/o_invalid_structure.csv,childstruc
+CHILD_EMPTY_FILE,child/i_empty.csv,child/o_empty.csv,childempty
+CHILD_HEADER_ONLY,child/i_header_only.csv,child/o_header_only.csv,childheadonly
+CHILD_MAV_1080,child/i_mav_1080.csv,child/o_mav_1080.csv,childmav1080
+CLASS_POSITIVE,class_list/i_positive.csv,class_list/o_positive.csv,classpos
+CLASS_NEGATIVE,class_list/i_negative.csv,class_list/o_negative.csv,classneg
+CLASS_INVALID_STRUCTURE,class_list/i_invalid_structure.csv,class_list/o_invalid_structure.csv,classstruc
+CLASS_EMPTY_FILE,class_list/i_empty.csv,class_list/o_empty.csv,classempty
+CLASS_HEADER_ONLY,class_list/i_header_only.csv,class_list/o_header_only.csv,classheadonly
+CLASS_CHILDREN_FILTER,class_list/i_children_filter.csv,class_list/o_children_filter.csv,classfilter
+CLASS_YEAR_GROUP,class_list/i_year_group.csv,class_list/o_year_group.csv,classyeargrp
+CLASS_MAV_854,class_list/i_mav_854.csv,class_list/o_mav_854.csv,classmav854
+CLASS_SESSION_ID,class_list/i_session_id.csv,class_list/o_session_id.csv,classsessid
+CLASS_SINGLE_VACC,class_list/i_single_vacc.csv,class_list/o_single_vacc.csv,classsinglevacc
+CLASS_MAV_965,class_list/i_mav_965.csv,class_list/o_mav_965.csv,classmav965
+CLASS_CHANGE_NHSNO,class_list/i_change_nhsno.csv,class_list/o_change_nhsno.csv,classchangenhsno
+CLASS_MAV_1080,class_list/i_mav_1080.csv,class_list/o_mav_1080.csv,classmav1080
+COHORTS_NO_CONSENT,cohorts/i_no_consent.csv,cohorts/o_no_consent.csv,cohortnoconsent
+COHORTS_CONFLICTING_CONSENT,cohorts/i_conflicting_consent.csv,cohorts/o_conflicting_consent.csv,cohortconflict
+COHORTS_E2E_1,cohorts/i_e2e_1.csv,cohorts/o_e2e_1.csv,cohort_e2e
+COHORTS_UCR_MATCH,cohorts/i_ucr_match.csv,cohorts/o_ucr_match.csv,cohortmatch
+COHORTS_CONSENT_TWICE,cohorts/i_consent_twice.csv,cohorts/o_consent_twice.csv,cohorttwice
+COHORTS_CONFLICTING_GILLICK,cohorts/i_conflicting_gillick.csv,cohorts/o_conflicting_gillick.csv,cohortgillick
+COHORTS_FULL_NAME,cohorts/i_full_name.csv,cohorts/o_full_name.csv,cohortfullname
+COHORTS_MAV_927_PERF,cohorts/i_mav_927_perf.csv,cohorts/o_mav_927_perf.csv,cohortmav927perf
+COHORTS_MAV_909,cohorts/i_mav_909.csv,cohorts/o_mav_909.csv,cohortmav909
+COHORTS_MAV_853,cohorts/i_mav_853.csv,cohorts/o_mav_853.csv,cohortmav853
+CLASS_MOVES_CONFIRM_IGNORE,class_list/i_moves_1.csv,class_list/o_moves_1.csv,classmoves1
+CLASS_MOVES_UNKNOWN_HOMESCHOOLED,class_list/i_moves_2.csv,class_list/o_moves_2.csv,classmoves2

tests/helpers/parental_consent_helper_doubles.py

@@ -6,7 +6,7 @@
 class ParentalConsentHelper:
     def __init__(self):
         self.df = TestData().read_spreadsheet(
-            file_path=test_data_file_paths.PARENTAL_CONSENT_DOUBLES
+            test_data_file_paths.PARENTAL_CONSENT_DOUBLES
         )
 
     def read_data_for_scenario(self, scenario_data) -> None:

tests/helpers/parental_consent_helper_hpv.py

@@ -5,9 +5,7 @@
 
 class ParentalConsentHelper:
     def __init__(self):
-        self.df = TestData().read_spreadsheet(
-            file_path=test_data_file_paths.PARENTAL_CONSENT_HPV
-        )
+        self.df = TestData().read_spreadsheet(test_data_file_paths.PARENTAL_CONSENT_HPV)
 
     def read_data_for_scenario(self, scenario_data) -> None:
         _, _row = scenario_data