7.0.1

Optimized tokenizer. With large inputs the formatter is now considerably faster.

7.0.0

This release contains a major rewrite of how strings, identifiers, variables and parameters are handled internally. What is and isn't supported regarding these things in various dialects was also re-evaluated, resulting in removal of support for various bits of syntax that shouldn't have been there in the first place, along with addition of missing bits of syntax.

Breaking changes

• The params option is no more supported for Hive and Spark. These dialects don't actually have prepared statements - the values of which params option was meant to provide. Instead the params there was used to replace substitution variables ${var_name}, which however did not work correctly either (e.g. inside strings).
• Dropped @".." parameter placeholder support from SQLite
• Dropped :name parameter placeholders from PostgreSQL
• No more allowing # and $ chars in PL/SQL parameter placeholder names.
• No more allowing . inside parameter placeholder names.
• No more allowing # and @ in DB2 identifiers.
• No more allowing $ and @ in BigQuery identifiers.
• Dropped `backtick quoted identifiers` from PL/SQL, DB2, PostgreSQL, Redshift, Transact-SQL.
• Dropped "double quoted identifiers" from Spark.
• Dropped [bracket quoted identifiers] from DB2.
• Removed lots of incorrect reserved keywords from SQLite (this mainly affects keywordCase option).
• Dropped the sql-formatter.js bundle from dist/ dir. Now only sql-formatter.min.js and sql-formatter.min.js.map are left there.

New features and improvements

• Added names of Oracle SQL functions (which are now also effected by keywordCase option).
• Added names of SQLite functions (which are now also effected by keywordCase option).
• Added support for Array literals in BigQuery
• Added support for @, #, $ characters in DB2 parameters
• Added support for $ char in PostgreSQL, MariaDB, MySQL and Transact-SQL identifiers
• Added support for Oracle &name substitution variables
• Added ${name} variable support for Hive and Spark
• Added support for @name named and quoted parameter placeholders to BigQuery
• Added support for :1, :2 parameter placeholders to PL/SQL
• Added support for $1, $2 parameter placeholders to N1QL
• Implemented proper dashes support for BigQuery identifiers (e.g. my-little-ident is valid identifier, but my--little--ident is treated as my followed by comment).
• Added triple-quoted strings support for BigQuery (e.g. """some string""" and '''some string''')
• Added hex-strings support to standard SQL, SQLite, Spark, PostgreSQL, MariaDB and MySQL
• Added support for PostgreSQL bit-strings.
• Added several prefixed string types for DB2 (N'...', G'...', GX'...', UX'...', U&'...')
• Added several prefixed strings to BigQuery (R"...", B"...", RB"...")

Bugfixes

• Fixed incorrectly inserted spaces inside map keys #230
• Fixed incorrect string token regexes #211
• All MySQL function names are now properly uppercased with keywordCase:upper option.

Performance improvements

• Formatting long SQL strings is now much faster, fixing a long-time performance problem. #70
• Calling the format() function repeatedly is now also much faster as we're caching the construction of tokenizer.

6.1.5

Fixes error with absolute imports in 6.1.3 and 6.1.4.

6.1.4

Broken release. Tried to fix the previous release... didn't succeed :(

6.1.3

Mostly a changes related to repository transfer from github.com/zeroturnaround to github.com/sql-formatter-org.

• Dropped . and @ characters support from PL/SQL identifiers (which weren't valid anyway).

• Dropped support for TypeScript < 4.5.

Broken release because of internal import paths refactoring. Fixed in 6.1.5.

6.1.2

Bugfixes:

• Add missing PostgreSQL JSON operators #> and #>> #217
• Don't add extra newline before a lonely semicolon when using newlineBeforeSemicolon: true

Internals:

• Rewrite of SQL string building logic
• Remove ZWS-replacement hack from indentStyle: tabular* token formatting
• Store both raw and clean values inside tokens

6.1.1

Fixes:

• Add CASE expression support for DB2 and Redshift. This means, all SQL dialects now support CASE..END syntax.

Internals:

• Reorganization of CASE expression formatting code.
• Reorganization of token preprocessing.

6.1.0

New language: SQLite

• Merged in support for SQLite formatting #189

Fixes

• Restore regression in DELETE FROM formatting

6.0.2

Minor enhancements:

• Don't crash with default "sql" language when encountering "@" character #214
• Better error message when tokenizer fails

6.0.1

Single bugfix:

• Avoid breaking up ON UPDATE and ON DELETE #215