Capitalisation: Add camelCase

src/sqlfluff/core/rules/config_info.py

@@ -60,10 +60,18 @@
  "definition": "The capitalisation policy to enforce.",
  },
  "extended_capitalisation_policy": {
- "validation": ["consistent", "upper", "lower", "pascal", "capitalise", "snake"],
+ "validation": [
+ "consistent",
+ "upper",
+ "lower",
+ "pascal",
+ "capitalise",
+ "snake",
+ "camel",
+ ],
  "definition": (
- "The capitalisation policy to enforce, extended with PascalCase "
- "and snake_case. "
+ "The capitalisation policy to enforce, extended with PascalCase, "
+ "snake_case, and camelCase. "
  "This is separate from ``capitalisation_policy`` as it should not be "
  "applied to keywords."
  ),

src/sqlfluff/rules/capitalisation/CP01.py

@@ -136,17 +136,20 @@ def _handle_segment(self, segment: BaseSegment, context: RuleContext) -> LintRes
  refuted_cases = memory.get("refuted_cases", set())
 
  # Which cases are definitely inconsistent with the segment?
+ first_letter_is_lowercase = False
  for character in segment.raw:
  if is_capitalizable(character):
  first_letter_is_lowercase = character != character.upper()
  break
- # If none of the characters are letters there will be a parsing
- # error, so not sure we need this statement
- first_letter_is_lowercase = False
 
+ # We refute inference of camel, pascal, and snake case.
+ # snake, if not explicitly set, can be destructive to
+ # variable names, adding underscores.
+ # camel and Pascal could allow poorly linted code in,
+ # so must be explicitly chosen.
+ refuted_cases.update(["camel", "pascal", "snake"])
  if first_letter_is_lowercase:
- # snake added here as it cannot be inferred (presents as lower)
- refuted_cases.update(["upper", "capitalise", "pascal", "snake"])
+ refuted_cases.update(["upper", "capitalise"])
  if segment.raw != segment.raw.lower():
  refuted_cases.update(["lower"])
  else:
@@ -155,8 +158,6 @@ def _handle_segment(self, segment: BaseSegment, context: RuleContext) -> LintRes
  refuted_cases.update(["upper"])
  if segment.raw != segment.raw.capitalize():
  refuted_cases.update(["capitalise"])
- if not segment.raw.isalnum():
- refuted_cases.update(["pascal", "snake"])
 
  # Update the memory
  memory["refuted_cases"] = refuted_cases
@@ -219,6 +220,14 @@ def _handle_segment(self, segment: BaseSegment, context: RuleContext) -> LintRes
  lambda match: match.group(1) + match.group(2).upper() + match.group(3),
  segment.raw,
  )
+ elif concrete_policy == "camel":
+ # Similar to Pascal, for Camel, we can only do a best efforts approach.
+ # This presents as us never changing case mid-string.
+ fixed_raw = regex.sub(
+ "([^a-zA-Z0-9]+|^)([a-zA-Z0-9])([a-zA-Z0-9]*)",
+ lambda match: match.group(1) + match.group(2).lower() + match.group(3),
+ segment.raw,
+ )
  elif concrete_policy == "snake":
  if segment.raw.isupper():
  fixed_raw = segment.raw.lower()
@@ -240,14 +249,10 @@ def _handle_segment(self, segment: BaseSegment, context: RuleContext) -> LintRes
  # build description based on the policy in use
  consistency = "consistently " if cap_policy == "consistent" else ""
 
- if concrete_policy in ["upper", "lower"]:
+ if concrete_policy in ["upper", "lower", "pascal", "camel", "snake"]:
  policy = f"{concrete_policy} case."
  elif concrete_policy == "capitalise":
  policy = "capitalised."
- elif concrete_policy == "pascal":
- policy = "pascal case."
- elif concrete_policy == "snake":
- policy = "snake case."
 
  # Return the fixed segment
  self.logger.debug(

test/fixtures/rules/std_rule_cases/CP02.yml

@@ -15,16 +15,18 @@ test_pass_consistent_capitalisation_with_single_letter_upper:
  pass_str: SELECT A, Boo
 
 test_pass_consistent_capitalisation_with_single_word_snake:
- # Single-word ambiguity: Pascal vs Capitalise
- pass_str: SELECT Apple, Banana_split
+ # Snake is refuted as ambiguous
+ pass_str: SELECT apple, banana_split
 
-test_pass_consistent_capitalisation_with_single_word_pascal:
+test_fail_consistent_capitalisation_with_single_word_pascal:
  # Single-word ambiguity: Pascal vs Capitalise
- pass_str: SELECT AppleFritter, Banana
+ fail_str: SELECT AppleFritter, Banana
+ fix_str: SELECT APPLEFRITTER, BANANA
 
-test_pass_consistent_capitalisation_with_multiple_words_with_numbers:
+test_fail_consistent_capitalisation_with_multiple_words_with_numbers:
  # Numbers count as part of words so following letter can be upper or lower
- pass_str: SELECT AppleFritter, Apple123fritter, Apple123Fritter
+ fail_str: SELECT AppleFritter, Apple123fritter, Apple123Fritter
+ fix_str: SELECT APPLEFRITTER, APPLE123FRITTER, APPLE123FRITTER
 
 test_pass_consistent_capitalisation_with_leading_underscore:
  pass_str: SELECT _a, b
@@ -133,6 +135,36 @@ test_fail_inconsistent_capitalisation_pascal_v_capitalise:
  # Pascal vs Capitalise
  fail_str: SELECT AppleFritter, Banana_split
  fix_str: SELECT AppleFritter, Banana_Split
+ configs:
+ rules:
+ capitalisation.identifiers:
+ extended_capitalisation_policy: pascal
+
+test_fail_inconsistent_capitalisation_policy_camel_1:
+ fail_str: SELECT Camelcase
+ fix_str: SELECT camelcase
+ configs:
+ rules:
+ capitalisation.identifiers:
+ extended_capitalisation_policy: camel
+
+test_fail_inconsistent_capitalisation_policy_camel_2:
+ fail_str: SELECT Camel_Case
+ fix_str: SELECT camel_case
+ configs:
+ rules:
+ capitalisation.identifiers:
+ extended_capitalisation_policy: camel
+
+test_fail_inconsistent_capitalisation_policy_camel_3:
+ # Similar to above, you could argue the fixed string is
+ # Not really Pascal Case, but it's closer than it was!
+ fail_str: SELECT cAMEL_CASE
+ fix_str: SELECT cAMEL_cASE
+ configs:
+ rules:
+ capitalisation.identifiers:
+ extended_capitalisation_policy: camel
 
 test_pass_policy_unquoted_identifiers_aliases_1:
  pass_str: SELECT a, B
@@ -167,7 +199,7 @@ test_pass_policy_unquoted_identifiers_aliases_4:
 
 test_policy_unquoted_identifiers_aliases_5:
  fail_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS lower_case
- fix_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS Lower_Case
+ fix_str: SELECT UPPER_CASE AS PASCALCASE, PascalCase AS LOWER_CASE
  configs:
  rules:
  capitalisation.identifiers:
@@ -192,7 +224,7 @@ test_policy_unquoted_identifiers_aliases_7:
 
 test_policy_unquoted_identifiers_aliases_8:
  fail_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS lower_case FROM lower_case AS lower_case
- fix_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS Lower_Case FROM lower_case AS Lower_Case
+ fix_str: SELECT UPPER_CASE AS PASCALCASE, PascalCase AS LOWER_CASE FROM lower_case AS LOWER_CASE
  configs:
  rules:
  capitalisation.identifiers:
@@ -207,7 +239,7 @@ test_policy_unquoted_identifiers_column_aliases_1:
 
 test_policy_unquoted_identifiers_aliases_2:
  fail_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS lower_case FROM lower_case AS lower_case
- fix_str: SELECT UPPER_CASE AS PascalCase, PascalCase AS Lower_Case FROM lower_case AS lower_case
+ fix_str: SELECT UPPER_CASE AS PASCALCASE, PascalCase AS LOWER_CASE FROM lower_case AS lower_case
  configs:
  rules:
  capitalisation.identifiers:
@@ -281,3 +313,31 @@ test_fail_snake_aliases:
  rules:
  capitalisation.identifiers:
  extended_capitalisation_policy: snake
+
+test_fail_camel_aliases:
+ # Test for issue #5470
+ # similar to PascalCase, case logic defined in CP01, but tested in CP02
+ fail_str: |
+ SELECT
+ test1,
+ test_2,
+ testColumn3,
+ TestColumn4,
+ TESTCOLUMN5,
+ TEST_COLUMN6,
+ test_colUmn_7
+ fix_str: |
+ SELECT
+ test1,
+ test_2,
+ testColumn3,
+ testColumn4,
+ tESTCOLUMN5,
+ tEST_cOLUMN6,
+ test_colUmn_7
+ configs:
+ core:
+ dialect: tsql
+ rules:
+ capitalisation.identifiers:
+ extended_capitalisation_policy: camel

test/fixtures/rules/std_rule_cases/CP05.yml

@@ -27,14 +27,20 @@ test_pass_default_consistent_capitalised:
  ts Time With Time Zone
  );
 
-test_pass_default_consistent_pascal:
- # Test that we don't have the "inconsistent" bug
- pass_str: |
+test_fail_default_consistent_pascal:
+ # Attempting Pascal without config defaults to capitalize
+ fail_str: |
  CREATE TABLE distributors (
  did Integer,
  name VarChar(40),
  ts Time With Time Zone
  );
+ fix_str: |
+ CREATE TABLE distributors (
+ did Integer,
+ name Varchar(40),
+ ts Time With Time Zone
+ );
 
 test_fail_data_type_inconsistent_capitalisation_1:
  # Test that we don't have the "inconsistent" bug