Reduce intersection between functions and keywords

Author: mattbasta

src/languages/clickhouse/clickhouse.formatter.ts

@@ -226,6 +226,8 @@ const reservedSetOperations = expandPhrases([
 const reservedJoins = expandPhrases([
   // https://clickhouse.com/docs/sql-reference/statements/select/join
   '[GLOBAL] [INNER|LEFT|RIGHT|FULL|CROSS] [OUTER|SEMI|ANTI|ANY|ALL|ASOF] JOIN',
+  // https://clickhouse.com/docs/sql-reference/statements/select/array-join
+  '[LEFT] ARRAY JOIN',
 ]);
 
 const reservedKeywordPhrases = expandPhrases([

src/languages/clickhouse/clickhouse.functions.ts

@@ -13,7 +13,6 @@ export const functions: string[] = [
   'CRC32',
   'CRC32IEEE',
   'CRC64',
-  'DATABASE',
   'DATE',
   'DATE_DIFF',
   'DATE_FORMAT',
@@ -189,7 +188,6 @@ export const functions: string[] = [
   'aggThrow',
   'alphaTokens',
   'analysisOfVariance',
-  'and',
   'anova',
   'any',
   'anyHeavy',
@@ -603,7 +601,9 @@ export const functions: string[] = [
   'flatten',
   'flattenTuple',
   'floor',
-  'format',
+  // We do not include FORMAT as a function, because it's also a keyword.
+  // FORMAT clauses are fairly common: https://clickhouse.com/docs/sql-reference/statements/select/format
+  // 'format',
   'formatDateTime',
   'formatDateTimeInJodaSyntax',
   'formatQuery',
@@ -766,7 +766,8 @@ export const functions: string[] = [
   'ifNotFinite',
   'ifNull',
   'ignore',
-  'ilike',
+  // ilike() is a function, but the ILIKE keyword is very common in SQL.
+  // 'ilike',
   'inIgnoreSet',
   'indexHint',
   'indexOf',
@@ -843,7 +844,8 @@ export const functions: string[] = [
   'levenshteinDistance',
   'levenshteinDistanceUTF8',
   'lgamma',
-  'like',
+  // like() is a function, but the ILIKE keyword is very common in SQL.
+  // 'like',
   'ln',
   'locate',
   'log',
@@ -1005,7 +1007,8 @@ export const functions: string[] = [
   'normalizeUTF8NFKD',
   'normalizedQueryHash',
   'normalizedQueryHashKeepNames',
-  'not',
+  // not() is a function, but the NOT keyword is very common in SQL.
+  // 'not',
   'notEmpty',
   'notEquals',
   'notILike',
@@ -1043,7 +1046,6 @@ export const functions: string[] = [
   'numericIndexedVectorPointwiseSubtract',
   'numericIndexedVectorShortDebugString',
   'numericIndexedVectorToMap',
-  'or',
   'overlay',
   'overlayUTF8',
   'parseDateTime',
@@ -1185,7 +1187,8 @@ export const functions: string[] = [
   'randomPrintableASCII',
   'randomString',
   'randomStringUTF8',
-  'range',
+  // range() is a function, but the RANGE keyword is important for window functions.
+  // 'range',
   'rank',
   'rankCorr',
   'readWKBLineString',
@@ -1283,7 +1286,6 @@ export const functions: string[] = [
   'serverTimeZone',
   'serverTimezone',
   'serverUUID',
-  'set',
   'shardCount',
   'shardNum',
   'showCertificate',
@@ -1629,7 +1631,8 @@ export const functions: string[] = [
   'trimLeft',
   'trimRight',
   'trunc',
-  'truncate',
+  // truncate() is a function, but the TRUNCATE keyword is a statement type.
+  // 'truncate',
   'tryBase32Decode',
   'tryBase58Decode',
   'tryBase64Decode',
@@ -1744,7 +1747,8 @@ export const functions: string[] = [
   'PostgreSQL',
   'MySQL',
   'SQLite',
-  'Backup',
+  // Disabling this because it's more likely to be used in a GRANT statement as a permission.
+  // 'Backup',
   'MaterializedPostgreSQL',
   'DataLakeCatalog',
 ];

src/languages/clickhouse/clickhouse.keywords.ts

@@ -14,7 +14,6 @@ export const keywords: string[] = [
   'ANTI',
   'APPEND',
   'APPLY',
-  'ARRAY',
   'AS',
   'ASC',
   'ASCENDING',
@@ -39,11 +38,9 @@ export const keywords: string[] = [
   'CACHES',
   'CASCADE',
   'CASE',
-  'CAST',
   'CHANGE',
   'CHANGEABLE_IN_READONLY',
   'CHANGED',
-  'CHAR',
   'CHARACTER',
   'CHECK',
   'CLEANUP',
@@ -66,14 +63,10 @@ export const keywords: string[] = [
   'CROSS',
   'CUBE',
   'CURRENT',
-  'CURRENTUSER',
-  'CURRENT_USER',
   'D',
   'DATA',
   'DATABASE',
   'DATABASES',
-  'DATE',
-  'DAY',
   'DAYS',
   'DD',
   'DDL',
@@ -98,7 +91,6 @@ export const keywords: string[] = [
   'DOUBLE_SHA1_PASSWORD',
   'DROP',
   'ELSE',
-  'EMPTY',
   'ENABLED',
   'END',
   'ENFORCED',
@@ -120,7 +112,6 @@ export const keywords: string[] = [
   'FALSE',
   'FETCH',
   'FIELDS',
-  'FILE',
   'FILESYSTEM',
   'FILL',
   'FILTER',
@@ -151,15 +142,12 @@ export const keywords: string[] = [
   'HH',
   'HIERARCHICAL',
   'HOST',
-  'HOUR',
   'HOURS',
   'HTTP',
   // Disabling this because it's a keyword, but formats far more than
   // it should.
   // 'ID',
   'IDENTIFIED',
-  'IF',
-  'IGNORE',
   'ILIKE',
   'IN',
   'INDEX',
@@ -192,8 +180,6 @@ export const keywords: string[] = [
   'LAYOUT',
   'LDAP',
   'LEADING',
-  'LEFT',
-  'LESS',
   'LEVEL',
   'LIFETIME',
   'LIGHTWEIGHT',
@@ -206,26 +192,18 @@ export const keywords: string[] = [
   'LOCAL',
   'M',
   'MASK',
-  'MATCH',
-  'MATERIALIZE',
   'MATERIALIZED',
-  'MAX',
   'MCS',
   'MEMORY',
   'MERGES',
   'METRICS',
   'MI',
   'MICROSECOND',
   'MICROSECONDS',
-  'MILLISECOND',
   'MILLISECONDS',
-  'MIN',
-  'MINUTE',
   'MINUTES',
   'MM',
-  'MOD',
   'MODIFY',
-  'MONTH',
   'MONTHS',
   'MOVE',
   'MS',
@@ -284,7 +262,6 @@ export const keywords: string[] = [
   'Q',
   'QQ',
   'QUALIFY',
-  'QUARTER',
   'QUARTERS',
   'QUERY',
   'QUOTA',
@@ -300,15 +277,13 @@ export const keywords: string[] = [
   'REGEXP',
   'REMOVE',
   'RENAME',
-  'REPLACE',
   'RESET',
   'RESPECT',
   'RESTORE',
   'RESTRICT',
   'RESTRICTIVE',
   'RESUME',
   'REVOKE',
-  'RIGHT',
   'ROLE',
   'ROLES',
   'ROLLBACK',
@@ -321,7 +296,6 @@ export const keywords: string[] = [
   'SAMPLE',
   'SAN',
   'SCHEME',
-  'SECOND',
   'SECONDS',
   'SECURITY',
   'SELECT',
@@ -379,7 +353,6 @@ export const keywords: string[] = [
   'THEN',
   'TIES',
   'TIME',
-  'TIMESTAMP',
   'TO',
   'TOP',
   'TOTALS',
@@ -404,7 +377,6 @@ export const keywords: string[] = [
   'UPDATE',
   'URL',
   'USE',
-  'USER',
   'USING',
   'UUID',
   'VALID',
@@ -415,7 +387,6 @@ export const keywords: string[] = [
   'VOLUME',
   'WATCH',
   'WATERMARK',
-  'WEEK',
   'WEEKS',
   'WHEN',
   'WHERE',
@@ -425,7 +396,6 @@ export const keywords: string[] = [
   'WK',
   'WRITABLE',
   'WW',
-  'YEAR',
   'YEARS',
   'YY',
   'YYYY',

test/clickhouse.test.ts

@@ -157,6 +157,9 @@ describe('ClickhouseFormatter', () => {
       'ANY JOIN',
       'ALL JOIN',
       'ASOF JOIN',
+
+      'ARRAY JOIN',
+      'LEFT ARRAY JOIN',
     ],
   });
   supportsSetOperations(format, ['UNION', 'UNION ALL', 'UNION DISTINCT', 'PARALLEL WITH']);