Fix/stringify codes

Documentation/debug_n_transform_scripts/stringify_rule_codes.py

@@ -0,0 +1,31 @@
+import json
+import re
+
+
+def extract_rule_code(filepath):
+    pattern = r"rule_(\d+)\.py"
+    match = re.search(pattern, filepath)
+    if match:
+        return match.group(1)
+    else:
+        return None
+
+
+with open("files_failed.json", "r") as f:
+    filepaths = json.load(f)
+# filepath = "C:\\Users\\tambe.tabitha\\CIN-validator\\cin_validator\\rules\\cin2022_23\\rule_8940.py"
+
+for filepath in filepaths:
+    print(f"+++++++++++++++++++++++++++++++{filepath}+++++++++++++++++++++++++++++++")
+    try:
+        rule_code = extract_rule_code(filepath)
+
+        with open(filepath, "r") as f:
+            file_text = f.read()
+            file_text = file_text.replace(f"{rule_code}", f"'{rule_code}'")
+        with open(filepath, "w") as f:
+            f.write(file_text)
+            print(f"updated {rule_code} to '{rule_code}'")
+    except Exception as e:
+        print(f"Error updating {filepath}: {e}")
+        continue

cin_validator/__main__.py

@@ -1,5 +1,6 @@
 import datetime
 import importlib
+import json
 import os
 import xml.etree.ElementTree as ET
 from pathlib import Path
@@ -144,7 +145,23 @@ def test_cmd(rule, ruleset):
             for p in module_folder.glob("*.py")
             if p.stem != "__init__"
         ]
-    pytest.main(test_files)
+
+    failed_files = []
+    for file_path in test_files:
+        result = pytest.main([file_path])
+        if result != pytest.ExitCode.OK:
+            failed_files.append(file_path)
+
+    with open("files_failed.json", "w") as f:
+        json.dump(failed_files, f)
+    # pytest.main(test_files)
+
+
+@cli.command(name="retest")
+def retest():
+    with open("files_failed.json", "r") as f:
+        filepaths = json.load(f)
+    pytest.main(filepaths)
 
 
 @cli.command(name="xmltocsv")

cin_validator/cin_validator.py

@@ -282,6 +282,7 @@ def __init__(
         self.full_issue_df.drop_duplicates(
             ["child_id", "rule_code", "columns_affected", "row_id"], inplace=True
         )
+        self.full_issue_df.reset_index(drop=True, inplace=True)
 
     def get_rules_to_run(
         self, registry, selected_rules: Optional[list[str]] = None

cin_validator/rule_engine/__registry.py

@@ -5,7 +5,7 @@
 
 
 def rule_definition(
-    code: int,
+    code: str,
     module: CINTable,
     rule_type: RuleType = RuleType.ERROR,
     message: Optional[str] = None,
@@ -29,7 +29,7 @@ def wrapper(*args, **kwargs):
             return func(*args, **kwargs)
 
         definition = RuleDefinition(
-            code=code,
+            code=str(code),
             func=func,
             rule_type=rule_type,
             module=module,

cin_validator/rules/cin2022_23/rule_100.py

@@ -16,7 +16,7 @@
 
 
 @rule_definition(
-    code=100,
+    code="100",
     module=CINTable.Header,
     message="Reference Date is incorrect",
     affected_fields=[ReferenceDate],
@@ -72,5 +72,5 @@ def test_validate():
     ]
 
     # Checks rule code and message are correct.
-    assert result.definition.code == 100
+    assert result.definition.code == "100"
     assert result.definition.message == "Reference Date is incorrect"

cin_validator/rules/cin2022_23/rule_1103.py

@@ -22,7 +22,7 @@
 
 
 @rule_definition(
-    code=1103,
+    code="1103",
     module=CINTable.Assessments,
     message="The assessment start date cannot be before the referral date",
     affected_fields=[
@@ -209,7 +209,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 1103
+    assert result.definition.code == "1103"
     assert (
         result.definition.message
         == "The assessment start date cannot be before the referral date"

cin_validator/rules/cin2022_23/rule_1104.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=1104,
+    code="1104",
     module=CINTable.Section47,
     message="The date of the initial child protection conference cannot be before the referral date",
     affected_fields=[
@@ -200,7 +200,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 1104
+    assert result.definition.code == "1104"
     assert (
         result.definition.message
         == "The date of the initial child protection conference cannot be before the referral date"

cin_validator/rules/cin2022_23/rule_1105.py

@@ -1,5 +1,5 @@
 """
-Rule number: 1105
+Rule number: '1105'
 Module: Child protection plans
 Rule details: Where present, the <CPPStartDate> (N00105) must be on or after the <CINReferralDate> (N00100)
 Rule message: The child protection plan start date cannot be before the referral date
@@ -29,7 +29,7 @@
 
 
 @rule_definition(
-    code=1105,
+    code="1105",
     module=CINTable.ChildProtectionPlans,
     message="The child protection plan start date cannot be before the referral date",
     affected_fields=[
@@ -208,7 +208,7 @@ def test_validate():
 
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 1105
+    assert result.definition.code == "1105"
     assert (
         result.definition.message
         == "The child protection plan start date cannot be before the referral date"

cin_validator/rules/cin2022_23/rule_1510.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=1510,
+    code="1510",
     module=CINTable.ChildIdentifiers,
     message="UPN invalid (wrong check letter at character 1)",
     affected_fields=[UPN],
@@ -107,7 +107,7 @@ def test_validate():
         IssueLocator(CINTable.ChildIdentifiers, UPN, 6),
     ]
 
-    assert result.definition.code == 1510
+    assert result.definition.code == "1510"
     assert (
         result.definition.message == "UPN invalid (wrong check letter at character 1)"
     )

cin_validator/rules/cin2022_23/rule_1520.py

@@ -1,5 +1,5 @@
 """
-Rule number: 1520
+Rule number: '1520'
 Module: Child idenitifiers
 Rule details: Each pupil <UPN> (N00001) must be unique across all pupils in the extract. 
 Note: This rule should be evaluated at LA-level for imported data                                                                     
@@ -17,7 +17,7 @@
 
 
 @rule_definition(
-    code=1520,
+    code="1520",
     module=CINTable.ChildIdentifiers,
     message="More than one record with the same UPN.",
     affected_fields=[UPN],
@@ -125,5 +125,5 @@ def test_validate():
     # Check that the rule definition is what you wrote in the context above.
 
     # replace 8840 with the rule code and put the appropriate message in its place too.
-    assert result.definition.code == 1520
+    assert result.definition.code == "1520"
     assert result.definition.message == "More than one record with the same UPN."

cin_validator/rules/cin2022_23/rule_1530.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=1530,
+    code="1530",
     module=CINTable.ChildIdentifiers,
     message="UPN invalid (characters 2-4 not a recognised LA code)",
     affected_fields=[UPN],
@@ -98,7 +98,6 @@ def validate(
 
 
 def test_validate():
-
     child_identifiers = pd.DataFrame(
         {
             "UPN": [
@@ -127,7 +126,7 @@ def test_validate():
         IssueLocator(CINTable.ChildIdentifiers, UPN, 6),
     ]
 
-    assert result.definition.code == 1530
+    assert result.definition.code == "1530"
     assert (
         result.definition.message
         == "UPN invalid (characters 2-4 not a recognised LA code)"

cin_validator/rules/cin2022_23/rule_1540.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=1540,
+    code="1540",
     module=CINTable.ChildIdentifiers,
     message="UPN invalid (characters 5-12 not all numeric)",
     affected_fields=[UPN],
@@ -57,5 +57,5 @@ def test_validate():
         IssueLocator(CINTable.ChildIdentifiers, UPN, 3),
     ]
 
-    assert result.definition.code == 1540
+    assert result.definition.code == "1540"
     assert result.definition.message == "UPN invalid (characters 5-12 not all numeric)"

cin_validator/rules/cin2022_23/rule_1550.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=1550,
+    code="1550",
     module=CINTable.ChildIdentifiers,
     message="UPN invalid (character 13 not a recognised value)",
     affected_fields=[UPN],
@@ -85,7 +85,7 @@ def test_validate():
         IssueLocator(CINTable.ChildIdentifiers, UPN, 2),
     ]
 
-    assert result.definition.code == 1550
+    assert result.definition.code == "1550"
     assert (
         result.definition.message == "UPN invalid (character 13 not a recognised value)"
     )

cin_validator/rules/cin2022_23/rule_2883.py

@@ -18,9 +18,10 @@
 
 Section47 = CINTable.Section47
 
+
 # define characteristics of rule
 @rule_definition(
-    code=2883,
+    code="2883",
     # module is table that seems central to the condition.
     module=CINTable.ChildProtectionPlans,
     message="There are more child protection plans starting than initial conferences taking place",
@@ -150,6 +151,6 @@ def test_validate():
 
     issues = result.la_issues
     assert issues == (
-        2883,
+        "2883",
         "There are more child protection plans starting than initial conferences taking place",
     )

cin_validator/rules/cin2022_23/rule_2884.py

@@ -15,7 +15,7 @@
 
 
 @rule_definition(
-    code=2884,
+    code="2884",
     module=CINTable.Section47,
     message="An initial child protection conference is recorded at both the S47 and CIN Details level and it should only be recorded in one",
     affected_fields=[
@@ -25,7 +25,6 @@
 def validate(
     data_container: Mapping[CINTable, pd.DataFrame], rule_context: RuleContext
 ):
-
     df_47 = data_container[Section47].copy()
     df_cin = data_container[CINdetails].copy()
 
@@ -221,7 +220,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 2884
+    assert result.definition.code == "2884"
     assert (
         result.definition.message
         == "An initial child protection conference is recorded at both the S47 and CIN Details level and it should only be recorded in one"

cin_validator/rules/cin2022_23/rule_2885.py

@@ -22,8 +22,8 @@
 
 # define characteristics of rule
 @rule_definition(
-    # write the rule code here, in place of 2885
-    code=2885,
+    # write the rule code here, in place of '2885'
+    code="2885",
     # replace ChildProtectionPlans with the value in the module column of the excel sheet corresponding to this rule .
     # Note that even if multiple tables are involved, one table will be named in the module column.
     module=CINTable.ChildProtectionPlans,
@@ -606,7 +606,7 @@ def test_validate():
     assert issue_rows0.equals(expected_df0)
 
     # Confirm that the rule details were properly pushed through.
-    assert result.definition.code == 2885
+    assert result.definition.code == "2885"
     assert (
         result.definition.message
         == "Child protection plan shown as starting a different day to the initial child protection conference."

cin_validator/rules/cin2022_23/rule_2889.py

@@ -14,15 +14,14 @@
 
 
 @rule_definition(
-    code=2889,
+    code="2889",
     module=CINTable.Section47,
     message="The S47 start date cannot be before the referral date.",
     affected_fields=[S47ActualStartDate, CINreferralDate],
 )
 def validate(
     data_container: Mapping[CINTable, pd.DataFrame], rule_context: RuleContext
 ):
-
     df_s47 = data_container[Section47]
     df_cin = data_container[CINdetails]
 
@@ -162,7 +161,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 2889
+    assert result.definition.code == "2889"
     assert (
         result.definition.message
         == "The S47 start date cannot be before the referral date."

cin_validator/rules/cin2022_23/rule_2990.py

@@ -19,7 +19,7 @@
 
 # define characteristics of rule
 @rule_definition(
-    code=2990,
+    code="2990",
     module=CINTable.CINdetails,
     message="Activity is recorded against a case marked as ‘Case closed after assessment, no further action’ or 'case closed after assessment, referred to early help'.",
     affected_fields=[
@@ -382,7 +382,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 2990
+    assert result.definition.code == "2990"
     assert (
         result.definition.message
         == "Activity is recorded against a case marked as ‘Case closed after assessment, no further action’ or 'case closed after assessment, referred to early help'."

cin_validator/rules/cin2022_23/rule_4000.py

@@ -14,7 +14,7 @@
 
 
 @rule_definition(
-    code=4000,
+    code="4000",
     module=CINTable.CINplanDates,
     message="CIN Plan details provided for a referral with no further action",
     affected_fields=[
@@ -156,7 +156,7 @@ def test_validate():
     )
     assert issue_rows.equals(expected_df)
 
-    assert result.definition.code == 4000
+    assert result.definition.code == "4000"
     assert (
         result.definition.message
         == "CIN Plan details provided for a referral with no further action"