Separate data types from generic keywords

Author: karlhorky

src/languages/bigquery/bigquery.formatter.ts

@@ -1,8 +1,8 @@
 import { DialectOptions } from '../../dialect.js';
-import { EOF_TOKEN, isToken, TokenType, Token } from '../../lexer/token.js';
 import { expandPhrases } from '../../expandPhrases.js';
-import { keywords } from './bigquery.keywords.js';
+import { EOF_TOKEN, isToken, Token, TokenType } from '../../lexer/token.js';
 import { functions } from './bigquery.functions.js';
+import { dataTypes, keywords } from './bigquery.keywords.js';
 
 const reservedSelect = expandPhrases(['SELECT [ALL | DISTINCT] [AS STRUCT | AS VALUE]']);
 
@@ -163,7 +163,9 @@ export const bigquery: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords: keywords,
+    reservedKeywords:
+      // Temporary, will be replaced by reservedDataTypes
+      [...new Set(keywords.concat(dataTypes))],
     reservedFunctionNames: functions,
     extraParens: ['[]'],
     stringTypes: [

src/languages/bigquery/bigquery.keywords.ts

@@ -96,8 +96,9 @@ export const keywords: string[] = [
   'WINDOW',
   'WITH',
   'WITHIN',
+];
 
-  // datatypes
+export const dataTypes: string[] = [
   'ARRAY', // parametric, ARRAY<T>
   'BOOL',
   'BYTES', // parameterised, BYTES(Length)

src/languages/postgresql/postgresql.formatter.ts

@@ -1,7 +1,7 @@
 import { DialectOptions } from '../../dialect.js';
 import { expandPhrases } from '../../expandPhrases.js';
 import { functions } from './postgresql.functions.js';
-import { keywords } from './postgresql.keywords.js';
+import { dataTypes, keywords } from './postgresql.keywords.js';
 
 const reservedSelect = expandPhrases(['SELECT [ALL | DISTINCT]']);
 
@@ -260,7 +260,9 @@ export const postgresql: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords: keywords,
+    reservedKeywords:
+      // Temporary, will be replaced by reservedDataTypes
+      [...new Set(keywords.concat(dataTypes))],
     reservedFunctionNames: functions,
     nestedBlockComments: true,
     extraParens: ['[]'],

src/languages/postgresql/postgresql.keywords.ts

@@ -1,4 +1,4 @@
-export const keywords: string[] = [
+export const keywords = [
   // https://www.postgresql.org/docs/14/sql-keywords-appendix.html
   'ABORT',
   'ABSOLUTE',
@@ -32,10 +32,7 @@ export const keywords: string[] = [
   'BEFORE',
   'BEGIN',
   'BETWEEN', // (cannot be function or type)
-  'BIGINT', // (cannot be function or type)
   'BINARY', // reserved (can be function or type)
-  'BIT', // (cannot be function or type)
-  'BOOLEAN', // (cannot be function or type)
   'BOTH', // reserved
   'BREADTH',
   'BY',
@@ -48,8 +45,6 @@ export const keywords: string[] = [
   'CAST', // reserved
   'CATALOG',
   'CHAIN',
-  'CHAR', // (cannot be function or type), requires AS
-  'CHARACTER', // (cannot be function or type), requires AS
   'CHARACTERISTICS',
   'CHECK', // reserved
   'CHECKPOINT',
@@ -96,7 +91,6 @@ export const keywords: string[] = [
   'DAY', // requires AS
   'DEALLOCATE',
   'DEC', // (cannot be function or type)
-  'DECIMAL', // (cannot be function or type)
   'DECLARE',
   'DEFAULT', // reserved
   'DEFAULTS',
@@ -117,7 +111,6 @@ export const keywords: string[] = [
   'DO', // reserved
   'DOCUMENT',
   'DOMAIN',
-  'DOUBLE',
   'DROP',
   'EACH',
   'ELSE', // reserved
@@ -145,7 +138,6 @@ export const keywords: string[] = [
   'FILTER', // requires AS
   'FINALIZE',
   'FIRST',
-  'FLOAT', // (cannot be function or type)
   'FOLLOWING',
   'FOR', // reserved, requires AS
   'FORCE',
@@ -192,10 +184,7 @@ export const keywords: string[] = [
   'INSENSITIVE',
   'INSERT',
   'INSTEAD',
-  'INT', // (cannot be function or type)
-  'INTEGER', // (cannot be function or type)
   'INTERSECT', // reserved, requires AS
-  'INTERVAL', // (cannot be function or type)
   'INTO', // reserved, requires AS
   'INVOKER',
   'IS', // reserved (can be function or type)
@@ -257,7 +246,6 @@ export const keywords: string[] = [
   'NULL', // reserved
   'NULLIF', // (cannot be function or type)
   'NULLS',
-  'NUMERIC', // (cannot be function or type)
   'OBJECT',
   'OF',
   'OFF',
@@ -292,7 +280,6 @@ export const keywords: string[] = [
   'POLICY',
   'POSITION', // (cannot be function or type)
   'PRECEDING',
-  'PRECISION', // (cannot be function or type), requires AS
   'PREPARE',
   'PREPARED',
   'PRESERVE',
@@ -307,7 +294,6 @@ export const keywords: string[] = [
   'QUOTE',
   'RANGE',
   'READ',
-  'REAL', // (cannot be function or type)
   'REASSIGN',
   'RECHECK',
   'RECURSIVE',
@@ -360,7 +346,6 @@ export const keywords: string[] = [
   'SIMILAR', // reserved (can be function or type)
   'SIMPLE',
   'SKIP',
-  'SMALLINT', // (cannot be function or type)
   'SNAPSHOT',
   'SOME', // reserved
   'SQL',
@@ -388,11 +373,8 @@ export const keywords: string[] = [
   'TEMP',
   'TEMPLATE',
   'TEMPORARY',
-  'TEXT',
   'THEN', // reserved
   'TIES',
-  'TIME', // (cannot be function or type)
-  'TIMESTAMP', // (cannot be function or type)
   'TO', // reserved, requires AS
   'TRAILING', // reserved
   'TRANSACTION',
@@ -424,9 +406,7 @@ export const keywords: string[] = [
   'VALIDATOR',
   'VALUE',
   'VALUES', // (cannot be function or type)
-  'VARCHAR', // (cannot be function or type)
   'VARIADIC', // reserved
-  'VARYING', // requires AS
   'VERBOSE', // reserved (can be function or type)
   'VERSION',
   'VIEW',
@@ -442,7 +422,6 @@ export const keywords: string[] = [
   'WORK',
   'WRAPPER',
   'WRITE',
-  'XML',
   'XMLATTRIBUTES', // (cannot be function or type)
   'XMLCONCAT', // (cannot be function or type)
   'XMLELEMENT', // (cannot be function or type)
@@ -456,5 +435,32 @@ export const keywords: string[] = [
   'XMLTABLE', // (cannot be function or type)
   'YEAR', // requires AS
   'YES',
+];
+
+export const dataTypes: string[] = [
+  // https://www.postgresql.org/docs/current/datatype.html
+  'BIGINT', // (cannot be function or type)
+  'BIT', // (cannot be function or type)
+  'BOOL', // (cannot be function or type)
+  'BOOLEAN', // (cannot be function or type)
+  'CHAR', // (cannot be function or type), requires AS
+  'CHARACTER', // (cannot be function or type), requires AS
+  'DECIMAL', // (cannot be function or type)
+  'DOUBLE',
+  'FLOAT', // (cannot be function or type)
+  'INT', // (cannot be function or type)
+  'INTEGER', // (cannot be function or type)
+  'INTERVAL', // (cannot be function or type)
+  'NUMERIC', // (cannot be function or type)
+  'PRECISION', // (cannot be function or type), requires AS
+  'REAL', // (cannot be function or type)
+  'SMALLINT', // (cannot be function or type)
+  'TEXT',
+  'TIME', // (cannot be function or type)
+  'TIMESTAMP', // (cannot be function or type)
+  'TIMESTAMPTZ', // (cannot be function or type)
+  'VARCHAR', // (cannot be function or type)
+  'VARYING', // requires AS
+  'XML',
   'ZONE',
 ];

src/languages/redshift/redshift.formatter.ts

@@ -1,7 +1,7 @@
 import { DialectOptions } from '../../dialect.js';
 import { expandPhrases } from '../../expandPhrases.js';
 import { functions } from './redshift.functions.js';
-import { keywords } from './redshift.keywords.js';
+import { dataTypes, keywords } from './redshift.keywords.js';
 
 const reservedSelect = expandPhrases(['SELECT [ALL | DISTINCT]']);
 
@@ -147,7 +147,9 @@ export const redshift: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords: keywords,
+    reservedKeywords:
+      // Temporary, will be replaced by reservedDataTypes
+      [...new Set(keywords.concat(dataTypes))],
     reservedFunctionNames: functions,
     stringTypes: ["''-qq"],
     identTypes: [`""-qq`],

src/languages/redshift/redshift.keywords.ts

@@ -197,7 +197,9 @@ export const keywords: string[] = [
    * SVL: https://docs.aws.amazon.com/redshift/latest/dg/svl_views.html
    * SVV: https://docs.aws.amazon.com/redshift/latest/dg/svv_views.html
    */
+];
 
+export const dataTypes: string[] = [
   // https://docs.aws.amazon.com/redshift/latest/dg/r_Character_types.html#r_Character_types-text-and-bpchar-types
   'BPCHAR',
   'TEXT',

src/languages/snowflake/snowflake.formatter.ts

@@ -1,7 +1,7 @@
 import { DialectOptions } from '../../dialect.js';
 import { expandPhrases } from '../../expandPhrases.js';
 import { functions } from './snowflake.functions.js';
-import { keywords } from './snowflake.keywords.js';
+import { dataTypes, keywords } from './snowflake.keywords.js';
 
 const reservedSelect = expandPhrases(['SELECT [ALL | DISTINCT]']);
 
@@ -303,7 +303,9 @@ export const snowflake: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords: keywords,
+    reservedKeywords:
+      // Temporary, will be replaced by reservedDataTypes
+      [...new Set(keywords.concat(dataTypes))],
     reservedFunctionNames: functions,
     stringTypes: ['$$', `''-qq-bs`],
     identTypes: ['""-qq'],

src/languages/snowflake/snowflake.keywords.ts

@@ -96,8 +96,9 @@ export const keywords: string[] = [
 
   // These are definitely keywords, but haven't found a definite list in the docs
   'COMMENT',
+];
 
-  // datatypes
+export const dataTypes: string[] = [
   'NUMBER',
   'DECIMAL',
   'NUMERIC',

src/languages/trino/trino.formatter.ts

@@ -1,7 +1,7 @@
 import { DialectOptions } from '../../dialect.js';
 import { expandPhrases } from '../../expandPhrases.js';
 import { functions } from './trino.functions.js';
-import { keywords } from './trino.keywords.js';
+import { dataTypes, keywords } from './trino.keywords.js';
 
 const reservedSelect = expandPhrases(['SELECT [ALL | DISTINCT]']);
 
@@ -132,7 +132,9 @@ export const trino: DialectOptions = {
     reservedSetOperations,
     reservedJoins,
     reservedPhrases,
-    reservedKeywords: keywords,
+    reservedKeywords:
+      // Temporary, will be replaced by reservedDataTypes
+      [...new Set(keywords.concat(dataTypes))],
     reservedFunctionNames: functions,
     // Trino also supports {- ... -} parenthesis.
     // The formatting of these currently works out as a result of { and -

src/languages/trino/trino.keywords.ts

@@ -236,7 +236,9 @@ export const keywords: string[] = [
   'WRAPPER',
   'WRITE',
   'ZONE',
+];
 
+export const dataTypes: string[] = [
   // https://github.com/trinodb/trino/blob/432d2897bdef99388c1a47188743a061c4ac1f34/core/trino-main/src/main/java/io/trino/metadata/TypeRegistry.java#L131-L168
   // or https://trino.io/docs/current/language/types.html
   'BIGINT',