Enable rules removal: do not validate deleted rules

ci/validate_metadata.sh

@@ -7,10 +7,16 @@ base="$(git merge-base FETCH_HEAD HEAD)"
 echo "Comparing against the merge-base: ${base}"
 if ! git diff --name-only --exit-code "${base}" -- rspec-tools/
 then
-  # Revalidate all rules
   basename --multiple rules/* | mapfile -t affected_rules
+  echo "Change in the tools, revalidating all rules"
 else
-  git diff --name-only "${base}" -- rules/ | sed -Ee 's#rules/(S[0-9]+)/.*#\1#' | sort -u | mapfile -t affected_rules
+  git diff --name-only "${base}" -- rules/ | # Get all the changes in rules
+    sed -Ee 's#(rules/S[0-9]+)/.*#\1#' | # extract the rule directories
+    sort -u | # deduplicate
+    while IFS= read -r rule; do [[ -d "$rule" ]] && echo "$rule" || true; done |  # filter non-deleted rules
+    sed 's#rules/##' | # get rule ids
+    mapfile -t affected_rules # store them in the `affected_rules` array
+  echo "Validating ${affected_rules[@]}"
 fi
 
 # Validate metadata
@@ -19,4 +25,6 @@ then
   cd rspec-tools
   pipenv install
   pipenv run rspec-tools validate-rules-metadata "${affected_rules[@]}"
+else
+  echo "No rule changed or added"
 fi

rspec-tools/rspec_tools/errors.py

@@ -13,3 +13,9 @@ class RuleValidationError(ClickException):
 
     def __init__(self, message):
         super().__init__(message)
+
+class RuleNotFoundError(ClickException):
+    '''Exception raised when a rule does not exist in the repository.'''
+
+    def __init__(self, message):
+        super().__init__(message)

rspec-tools/rspec_tools/rules.py

@@ -3,6 +3,7 @@ import json
 from pathlib import Path
 from typing import Final, Generator, Iterable, Optional
 from bs4 import BeautifulSoup
+from rspec_tools.errors import RuleNotFoundError
 
 
 METADATA_FILE_NAME: Final[str] = 'metadata.json'
@@ -89,4 +90,7 @@ class RulesRepository:
     return (GenericRule(child) for child in self.rules_path.glob('S*') if child.is_dir())
     
   def get_rule(self, ruleid: str):
-    return GenericRule(self.rules_path.joinpath(ruleid))
+    rulepath = self.rules_path.joinpath(ruleid)
+    if not rulepath.is_dir():
+      raise RuleNotFoundError('Cannot find rule ' + ruleid + ' in ' + str(self.rules_path))
+    return GenericRule(rulepath)

rspec-tools/tests/test_rules_repository.py

@@ -1,6 +1,8 @@
 from pathlib import Path
 
+import pytest
 from rspec_tools.rules import RulesRepository
+from rspec_tools.errors import RuleNotFoundError
 
 def test_list_rules(mockrules: Path):
   '''Check that rules are all listed.'''
@@ -22,3 +24,9 @@ def test_get_metadata(mockrules: Path):
   assert plsql.metadata['sqKey'] == 'PlSql.PackageNaming'
   java = rule.get_language('java')
   assert java.metadata['sqKey'] == 'S120'
+
+
+def test_nonexisting_rule(mockrules: Path):
+  '''Check that a nonexisting rule is reported.'''
+  with pytest.raises(RuleNotFoundError, match=fr'^Cannot find rule S200'):
+    RulesRepository(rules_path=mockrules).get_rule('S200')

rspec-tools/tests/validation/test_metadata_validation.py

@@ -127,4 +127,4 @@ def test_rule_with_complete_list_of_security_standard_passes_validation(rule_lan
   metadata['securityStandards'] = {'ASVS 4': [], 'OWASP': [], "OWASP Top 10 2021": []}
   with patch.object(LanguageSpecificRule, 'metadata', new_callable=PropertyMock) as mock:
     mock.return_value = metadata
-    validate_rule_specialization_metadata(rule_language)
+    validate_rule_specialization_metadata(rule_language)