The new keywordCompletion option can be used to define what kind of completions are generated for keywords.
Remove single-letter words from the list of Postgres keywords, since they interfere with alias-based autocompletion.
Dialects can now disable quoting of identifiers containing upper-case characters with the caseInsensitiveIdentifiers option.
Fix a bug that broke tokenizing of e'\n'-style strings.
Make statement folding leave the entire first line visible.
Fix a null dereference in schema-based autocompletion.
Fix a bug where Postgres-style dollar-quoted strings were enabled for all dialects, and the doubleDollarQuotedStrings options was ignored.
Properly support tags in PostgreSQL 4073 quoted strings.
Fix an issue that caused completions to be missing when using the defaultSchema option.
Don't tokenize identifiers after periods as anything but plain identifiers.
The schema option now allows nested objects to define multiple levels of completions, as well as self completion options for specific levels. The old format (using tables/schemas) continues to work but is deprecated.
Make sure table and column completions with upper-case characters are quoted.
Tag comments and strings as isolating for the purpose of bidirectional text.
Remove use of negative lookbehind in a regular expression, which recent versions of Safari still don't support.
The PL/SQL dialect now correctly handles q'[]'-quoting syntax.
Allow table names to contain multiple dots in the schema passed to schemaCompletionSource.
schemaCompletionSource now adds quotes around non-word identifiers even if the user didn't type a starting quote.
Dialect objects now have a public spec property holding their configuration.
Fix a bug where tokenizing of block comments got confused when nested comment start/end markers appeared directly next to each other.
Fix syntax tree node names for curly and square brackets, which had their names swapped.
The new schemas config option can be used to provide custom completion objects for schema completions.
Fix tokenizing of double-$ strings in SQL dialects that support them.
Make sure the language object has a name.
Fix tokenizing of -- line comments.
Schema-based completion now understands basic table alias syntax, and will take it into account when looking up completions.
The new unquotedBitLiterals dialect option controls whether 0b01 syntax is recognized.
Dialects now allow a treatBitsAsBytes option to allow any characters inside quoted strings prefixed with b.
The new doubleDollarQuotedStrings options to SQL dialects allows parsing of text delimited by $$ as strings. Regenerate readme
Update dependencies to 6.0.0
Schema completion descriptions may now include dots in table names to indicate nested schemas.
Fix a bug where the slash at the end of block comments wasn't considered part of the comment token.
Fix an infinite recursion bug in schemaCompletionSource.
The schemaCompletion and keywordCompletion exports, which returned extensions, have been replaced with schemaCompletionSource and keywordCompletionSource, which return completion sources. The old exports will remain available until the next major version.
Fix autocompletion on columns when the table name is written with upper-case letters. Move to @lezer/highlight
Remove duplicate keywords/types in dialect configurations.
Fix a bug that caused characters directly before a space to be tokenized incorrectly.
Fix a bug that broke tokenization of keywords.
Fix incorrect versions for @lezer dependencies.
Update dependencies to 0.19.0
Update dependencies to 0.18.
Fix bad syntax tree creation when the input ends with an unfinished quoted identifier.
The package now also exports a CommonJS module.
First numbered release.