Add dataTypeCase & functionCase

README.md

@@ -151,6 +151,8 @@ All fields are optional and all fields that are not specified will be filled wit
 - [**`useTabs`**](docs/useTabs.md) to use tabs for indentation.
 - [**`keywordCase`**](docs/keywordCase.md) uppercases or lowercases keywords.
 - [**`identifierCase`**](docs/identifierCase.md) uppercases or lowercases identifiers. (**experimental!**)
+- [**`dataTypeCase`**](docs/dataTypeCase.md) uppercases or lowercases data types. (**experimental!**)
+- [**`functionCase`**](docs/functionCase.md) uppercases or lowercases function names. (**experimental!**)
 - [**`indentStyle`**](docs/indentStyle.md) defines overall indentation style.
 - [**`logicalOperatorNewline`**](docs/logicalOperatorNewline.md) newline before or after boolean operator (AND, OR, XOR).
 - [**`expressionWidth`**](docs/expressionWidth.md) maximum number of characters in parenthesized expressions to be kept on single line.

docs/dataTypeCase.md

@@ -0,0 +1,45 @@
+# dataTypeCase (experimental)
+
+Converts data types to upper- or lowercase.
+
+## Options
+
+- `"preserve"` (default) preserves the original case.
+- `"upper"` converts to uppercase.
+- `"lower"` converts to lowercase.
+
+### preserve
+
+```sql
+CREATE TABLE user (
+  id InTeGeR PRIMARY KEY,
+  first_name VarChaR(30) NOT NULL,
+  bio teXT,
+  is_email_verified BooL,
+  created_timestamp timestamP
+);
+```
+
+### upper
+
+```sql
+CREATE TABLE user (
+  id INTEGER PRIMARY KEY,
+  first_name VARCHAR(30) NOT NULL,
+  bio TEXT,
+  is_email_verified BOOL,
+  created_timestamp TIMESTAMP
+);
+```
+
+### lower
+
+```sql
+CREATE TABLE user (
+  id integer PRIMARY KEY,
+  first_name varchar(30) NOT NULL,
+  bio text,
+  is_email_verified bool,
+  created_timestamp timestamp
+);
+```

docs/functionCase.md

@@ -0,0 +1,48 @@
+# functionCase (experimental)
+
+Converts function names to upper- or lowercase.
+
+## Options
+
+- `"preserve"` (default) preserves the original case.
+- `"upper"` converts to uppercase.
+- `"lower"` converts to lowercase.
+
+### preserve
+
+```sql
+SELECT
+  Concat(Trim(first_name), ' ', Trim(last_name)) AS name,
+  Max(salary) AS max_pay,
+  Cast(ssid AS INT)
+FROM
+  employee
+WHERE
+  expires_at > Now()
+```
+
+### upper
+
+```sql
+SELECT
+  CONCAT(TRIM(first_name), ' ', TRIM(last_name)) AS name,
+  MAX(salary) AS max_pay,
+  CAST(ssid AS INT)
+FROM
+  employee
+WHERE
+  expires_at > NOW()
+```
+
+### lower
+
+```sql
+SELECT
+  concat(trim(first_name), ' ', trim(last_name)) AS name,
+  max(salary) AS max_pay,
+  cast(ssid AS INT)
+FROM
+  employee
+WHERE
+  expires_at > now()
+```

docs/identifierCase.md

@@ -6,7 +6,6 @@ This option doesn't yet support all types of identifiers:
 
 - prefixed variables like `@my_var` are not converted.
 - parameter placeholders like `:param` are not converted.
-- function names like `count(*)` are not converted (this are currently governed by `keywordCase` option instead.)
 
 ## Options
 
@@ -26,7 +25,8 @@ from
 where
   Column6
   and Column7
-group by Column4
+group by
+  Column4
 ```
 
 ### upper
@@ -41,7 +41,8 @@ from
 where
   COLUMN6
   and COLUMN7
-group by COLUMN4
+group by
+  COLUMN4
 ```
 
 ### lower
@@ -56,5 +57,6 @@ from
 where
   column6
   and column7
-group by column4
+group by
+  column4
 ```

src/FormatOptions.ts

@@ -8,13 +8,19 @@ export type KeywordCase = 'preserve' | 'upper' | 'lower';
 
 export type IdentifierCase = 'preserve' | 'upper' | 'lower';
 
+export type DataTypeCase = 'preserve' | 'upper' | 'lower';
+
+export type FunctionCase = 'preserve' | 'upper' | 'lower';
+
 export type LogicalOperatorNewline = 'before' | 'after';
 
 export interface FormatOptions {
   tabWidth: number;
   useTabs: boolean;
   keywordCase: KeywordCase;
   identifierCase: IdentifierCase;
+  dataTypeCase: DataTypeCase;
+  functionCase: FunctionCase;
   indentStyle: IndentStyle;
   logicalOperatorNewline: LogicalOperatorNewline;
   expressionWidth: number;

src/formatter/ExpressionFormatter.ts

@@ -28,6 +28,8 @@ import {
   CaseExpressionNode,
   CaseWhenNode,
   CaseElseNode,
+  DataTypeNode,
+  ParameterizedDataTypeNode,
 } from '../parser/ast.js';
 
 import Layout, { WS } from './Layout.js';
@@ -94,6 +96,8 @@ export default class ExpressionFormatter {
     switch (node.type) {
       case NodeType.function_call:
         return this.formatFunctionCall(node);
+      case NodeType.parameterized_data_type:
+        return this.formatParameterizedDataType(node);
       case NodeType.array_subscript:
         return this.formatArraySubscript(node);
       case NodeType.property_access:
@@ -130,26 +134,46 @@ export default class ExpressionFormatter {
         return this.formatLineComment(node);
       case NodeType.block_comment:
         return this.formatBlockComment(node);
+      case NodeType.data_type:
+        return this.formatDataType(node);
       case NodeType.keyword:
         return this.formatKeywordNode(node);
     }
   }
 
   private formatFunctionCall(node: FunctionCallNode) {
     this.withComments(node.nameKw, () => {
-      this.layout.add(this.showKw(node.nameKw));
+      this.layout.add(this.showFunctionKw(node.nameKw));
+    });
+    this.formatNode(node.parenthesis);
+  }
+
+  private formatParameterizedDataType(node: ParameterizedDataTypeNode) {
+    this.withComments(node.dataType, () => {
+      this.layout.add(this.showDataType(node.dataType));
     });
     this.formatNode(node.parenthesis);
   }
 
   private formatArraySubscript(node: ArraySubscriptNode) {
+    let formattedArray: string;
+
+    switch (node.array.type) {
+      case NodeType.data_type:
+        formattedArray = this.showDataType(node.array);
+        break;
+      case NodeType.keyword:
+        formattedArray = this.showKw(node.array);
+        break;
+      default:
+        formattedArray = this.showIdentifier(node.array);
+        break;
+    }
+
     this.withComments(node.array, () => {
-      this.layout.add(
-        node.array.type === NodeType.keyword
-          ? this.showKw(node.array)
-          : this.showIdentifier(node.array)
-      );
+      this.layout.add(formattedArray);
     });
+
     this.formatNode(node.parenthesis);
   }
 
@@ -489,6 +513,10 @@ export default class ExpressionFormatter {
     }
   }
 
+  private formatDataType(node: DataTypeNode) {
+    this.layout.add(this.showDataType(node), WS.SPACE);
+  }
+
   private showKw(node: KeywordNode): string {
     if (isTabularToken(node.tokenType)) {
       return toTabularFormat(this.showNonTabularKw(node), this.cfg.indentStyle);
@@ -509,6 +537,26 @@ export default class ExpressionFormatter {
     }
   }
 
+  private showFunctionKw(node: KeywordNode): string {
+    if (isTabularToken(node.tokenType)) {
+      return toTabularFormat(this.showNonTabularFunctionKw(node), this.cfg.indentStyle);
+    } else {
+      return this.showNonTabularFunctionKw(node);
+    }
+  }
+
+  // Like showFunctionKw(), but skips tabular formatting
+  private showNonTabularFunctionKw(node: KeywordNode): string {
+    switch (this.cfg.functionCase) {
+      case 'preserve':
+        return equalizeWhitespace(node.raw);
+      case 'upper':
+        return node.text;
+      case 'lower':
+        return node.text.toLowerCase();
+    }
+  }
+
   private showIdentifier(node: IdentifierNode): string {
     if (node.quoted) {
       return node.text;
@@ -523,4 +571,15 @@ export default class ExpressionFormatter {
       }
     }
   }
+
+  private showDataType(node: DataTypeNode): string {
+    switch (this.cfg.dataTypeCase) {
+      case 'preserve':
+        return equalizeWhitespace(node.raw);
+      case 'upper':
+        return node.text;
+      case 'lower':
+        return node.text.toLowerCase();
+    }
+  }
 }

src/languages/bigquery/bigquery.formatter.ts

@@ -164,9 +164,8 @@ export const bigquery: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords:
-      // Temporary, will be replaced by reservedDataTypes
-      [...new Set(keywords.concat(dataTypes))],
+    reservedKeywords: keywords,
+    reservedDataTypes: dataTypes,
     reservedFunctionNames: functions,
     extraParens: ['[]'],
     stringTypes: [

src/languages/bigquery/bigquery.functions.ts

@@ -585,13 +585,4 @@ export const functions: string[] = [
   // pivot
   'PIVOT',
   'UNPIVOT',
-
-  // Data types with parameters like VARCHAR(100)
-  // https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#parameterized_data_types
-  'BYTES',
-  'NUMERIC',
-  'DECIMAL',
-  'BIGNUMERIC',
-  'BIGDECIMAL',
-  'STRING',
 ];

src/languages/bigquery/bigquery.keywords.ts

@@ -111,6 +111,7 @@ export const keywords: string[] = [
 ];
 
 export const dataTypes: string[] = [
+  // https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types
   'ARRAY', // parametric, ARRAY<T>
   'BOOL',
   'BYTES', // parameterised, BYTES(Length)
@@ -134,12 +135,4 @@ export const dataTypes: string[] = [
   'STRUCT', // parametric, STRUCT<T>
   'TIME',
   'TIMEZONE',
-
-  // https://cloud.google.com/bigquery/docs/reference/standard-sql/conversion_functions#formatting_syntax
-  'HEX',
-  'BASEX',
-  'BASE64M',
-  'ASCII',
-  'UTF-8',
-  'UTF8',
 ];

src/languages/db2/db2.formatter.ts

@@ -266,9 +266,8 @@ export const db2: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords:
-      // Temporary, will be replaced by reservedDataTypes
-      [...new Set(keywords.concat(dataTypes))],
+    reservedKeywords: keywords,
+    reservedDataTypes: dataTypes,
     reservedFunctionNames: functions,
     extraParens: ['[]'],
     stringTypes: [