Skip to content

Latest commit

 

History

History
914 lines (665 loc) · 31.8 KB

CHANGELOG.md

File metadata and controls

914 lines (665 loc) · 31.8 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

Changed

  • @Rsullivan00 - Do not tranform word-literal arrays when there is an escape sequence.
  • @steobrien, @kddeisz - Do not indent heredocs with calls more than they should be.
  • @jpickwell - Include .simplecov in filenames

0.19.0 - 2020-07-03

Added

  • [@ryan-hunter-pc] - Add the option to disable the Symbol#to_proc transform.
  • [@ryan-hunter-pc], [@SViccari], @kddeisz - Disable Symbol#to_proc transform when used as a key inside of a hash where the key is either :if or :unless.

0.18.2 - 2020-05-01

Changed

  • @alse - Support vscodeLanguageIds for HAML.
  • @ShayDavidson, @kddeisz - Don't allow replacing if/else with ternary if there's an assignment in the predicate.
  • @janklimo - Do not add an empty line after rescue when the block is empty.

0.18.1 - 2020-04-05

Changed

  • @petevk, @kddeisz - Use braces for block format iff it was originally a brace block, otherwise you could be changing precedence. For example:
expect do
  field 1 do
    "foo"
  end
end.to raise_error

should maintain its do...end and not switch to inline braces otherwise the brace might get associated with the 1.

  • @flyerhzm - Rewrite operators binary parser, as in:
[
  '1111111111111111111111111111111111111111111111111111111111111111111111111',
  222
] +
  [1]
  • @ftes, @kddeisz - When old-form dynamic attributes are added to a div tag in HAML, it was previously skipping printing the %div, which led to it being incorrectly displayed.
  • @ftes, @kddeisz - Previously if you had a long tag declaration with attributes that made it hit the line limit, then the content of the tag would be pushed to the next line but indented one character too many.
  • @ftes, @kddeisz - Don't explicitly require JSON if it has already been loaded, as this can lead to rubygems activation errors.
  • @mmainz, @kddeisz - Handle heredocs as the receivers of call nodes, as in:
foo = <<~TEXT.strip
  bar
TEXT
  • @github0013, @kddeisz - Leave parentheses in place if the value of a return node contains a binary with low operator precedence, as in:
return (a or b) if c?

0.18.0 - 2020-03-17

Added

  • @kddeisz - Support for the nokw_param node for specifying when methods should no accept keywords, as in:
def foo(**nil); end
  • @kddeisz - Support for the args_forward node for forwarding all types of arguments, as in:
def foo(...)
  bar(...)
end

Changed

  • @ftes, @kddeisz - Handled 3 or more classes on a node in HAML, as in:
%table.table.is-striped.is-hoverable
  • @ftes, @kddeisz - Better handling of indentation of if/elsif/else, unless/elsif/else, and case/when branches, as in:
.column.is-12
  - if true
    TRUE
  - else
    FALSE
  • @tobyndockerill - Format numbers with underscores after 4 digits, as opposed to 3.
  • @ianks - Improve performance by using --disable-gems.
  • @flyerhzm - Calls are grouped such that after an end parenthesis the following call will not be indented, as in:
Config::Download.new(
  'prettier',
  filename: 'prettier.yml', url: 'https://raw.githubusercontent.com/...'
)
  .perform

will now be printed as:

Config::Download.new(
  'prettier',
  filename: 'prettier.yml', url: 'https://raw.githubusercontent.com/...'
).perform
  • @pje, @kddeisz - Method definition bodies (on defs nodes) should dedent if a helper method is called. As in:
private def self.foo
          'bar'
        end

should instead be indented as:

private def self.foo
  'bar'
end
  • @masqita, @kddeisz - Inline variable assignment within a predicate should force the conditional to break, as in:
array.each do |element|
  if index = difference.index(element)
    difference.delete_at(index)
  end
end

0.17.0 - 2019-12-12

Added

  • @matt-wratt - Better support for explicit return nodes with empty arrays or arrays with a single element.
  • @jrdioko, @kddeisz - Alignment of not_to is explicitly allowed to not indent to better support rspec.

Changed

  • @gin0606 - The max buffer being passed into the Ruby process is now up to 10MB.

0.16.0 - 2019-11-14

Added

  • @mmainz, @kddeisz - Support for extra commas in multiple assignment, as it changes the meaning. For example,
a, = [1, 2, 3]

would previously get printed as a = [1, 2, 3], which changes the value of a from 1 to the value of the entire array.

  • @kddeisz - Experimental support for the HAMtemplate language.

Changed

  • @github0013, @kddeisz - Support proper string escaping when the original string in the source is wrapped in %q|...|. For example, %q|\'| should get printed as "\'", where previously it was dropping the backslash.
  • @jamescostian, @kddeisz - Force ternary breaking when using the lower-precendence operators and and or. For example,
if x.nil?
  puts 'nil' and return
else
  x
end

the previous expression was being transformed into a ternary which was invalid ruby. Instead it now stays broken out into an if/else block.

<<-HERE
  foo bar baz
  #{qux}
  foo bar baz
HERE

should remain formatted as it is. Whereas previously due to the way the lines were split, you would sometimes end up with it breaking after #{.

  • @jamescostian, @kddeisz - Fix up return node printing. When returning multiple values, you need to return an array literal as opposed to using parentheses.

0.15.1 - 2019-11-05

Changed

  • @AlanFoster - Add bin/lex for viewing the tokenized result of Ripper on Ruby code.
  • @jakeprime, @kddeisz - When predicates from within an if, unless, while, or until loop break the line, they should be aligned together. For example,
if foooooo || barrrrrr
  baz
end

If the line was set to very short, the binary node should be aligned to 3 spaces from the left of the file (which aligns with the if, it would be more for unless). So it would look like:

if foooooo ||
     barrrrrr
  baz
end
if foo?
end
  • @mmainz, @kddeisz - Hash keys are not converted to keyword syntax if they would make invalid symbols. For example,
{ :[] => nil }

cannot be translated into []: as that is an invalid symbol. Instead, it stays with the hash rocket syntax.

  • @cldevs, @kddeisz - Do not attempt to format the insides of xstring literals (string that get sent to the command line surrounded by backticks or %x).
  • @cldevs, @kddeisz - When predicates for if, unless, while, or until nodes contain an assignment, we can't know for sure that it doesn't modify the body. In this case we need to always break and form a multi-line block.
  • @MarcManiez, @kddeisz - When the return value of if, unless, while, or until nodes are assigned to anything other than a local variable, we need to wrap them in parentheses if we're changing to the modifier form. This is because the following expressions have different semantic meaning:
hash[:key] = break :value while false
hash[:key] = while false do break :value end

The first one will not result in an empty hash, whereas the second one will result in { key: nil }. In this case what we need to do for the first expression to align is wrap it in parens, as in:

hash[:key] = (break :value while false)

That will guarantee that the expressions are equivalent.

  • @AlanFoster - Fix crashes that were happening with ignored_nl nodes.

0.15.0 - 2019-08-06

Changed

  • @dudeofawesome, @kddeisz - If xstring literals (command line calls surrounded by backticks) break, then we indent and place the command on a new line. Previously, this was resulting in new lines getting added each time the code was formatted. Now this happens correctly.
  • @krachtstefan, @kddeisz - When a while or until loop modifies a begin...end statement, it must remain in the modifier form or else it changes sematic meaning. For example,
begin
  foo
end while bar

cannot be transformed into:

while bar
  foo
end

because that would never execute foo if bar is falsy, whereas in the initial example it would have.

  • @jviney, @kddeisz - When transforming a block into the Symbol#to_proc syntax from within a list of arguments inside of an aref node (i.e., foo[:bar].each), we can't put the block syntax inside the brackets.
  • @jakeprime, @kddeisz - Values for the return keyword that broke the line were previously just printed as they were, which breaks if you have a block expression like an if or while. For example,
return foo ? bar : baz

if the line was set to very short would be printed as:

return if foo
  bar
else
  baz
end

which wouldn't work. Instead, they now get printed with parentheses around the value, as in:

return(
  if foo
    bar
  else
    baz
  end
)
  • @jakeprime, @kddeisz - When switching from a double-quoted string to a single-quoted string that contained escaped double quotes, the backslashes would stay in the string. As in:
"Foo \"Bar\" Baz"

would get formatted as:

'Foo \"Bar\" Baz'

but now gets formatted as:

'Foo "Bar" Baz'

0.14.0 - 2019-07-17

Added

  • @kddeisz - Support for pattern matching for variables and array patterns. Currently waiting on Ripper support for hash patterns. For examples, check out the test/js/patterns.test.js file.

Changed

  • @jviney, @kddeisz - if/else blocks that had method calls on the end of them that were also transformed into ternaries need to have parens added to them. For example,
if foo
  1
else
  2
end.to_s

now correctly gets transformed into:

(foo ? 1 : 2).to_s
  • @acrewdson, @kddeisz - Fixed a bug where multiple newlines at the end of the file would cause a crash.
  • @jviney, @kddeisz - If a variable is assigned inside of the predicate of a conditional, then we can't change it into the single-line version as this breaks. For example,
if foo = 1
  foo
end

must stay the same.

0.13.0 - 2019-07-05

Added

  • @kddeisz - Added locStart and locEnd functions to support --cursor-offset.

Changed

  • @xipgroc, @kddeisz - Comments inside of do...end blocks that preceeded call nodes were associating the comment with the var_ref instead of the call itself. For example,
foo.each do |bar|
  # comment
  bar.baz
  bar.baz
end

would get printed as

foo.each do |bar|
  # comment
  bar
    .baz
  bar.baz
end

but now gets printed correctly.

  • @petevk, @kddeisz - Double splats inside a hash were previously failing to print. For example,
{ foo: "bar", **baz }

would fail to print, but now works.

0.12.3 - 2019-05-16

Changed

  • @kddeisz - Move arg, assign, constant, flow, massign, operator, scope, and statement nodes into their own files.
  • @kddeisz - Move @int, access_ctrl, assocsplat, block_var, else, number_arg, super, undef, var_ref, and var_ref as well as various call and symbol nodes into appropriate files.
  • @kddeisz - Better support for excessed commas in block args. Previously proc { |x,| } would add an extra space, but now it does not.
  • @kddeisz - Add a lot more documentation to the parser.
  • @glejeune, @kddeisz - Previously, the unary not operator inside a ternary (e.g., a ? not(b) : c) would break because it wouldn't add parentheses, but now it adds them.
  • @kddeisz - if and unless nodes used to not be able to handle if a comment was the only statement in the body. For example,
if foo
  # comment
end

would get printed as

# comment if foo

Now the if and unless printers check for the presence of single comments.

  • @JoshuaKGoldberg, @kddeisz - Fixes an error where command nodes within def nodes would fail to format if it was only a single block argument. For example,
def curry(&block)
  new &block
end

would fail, but now works.

  • @xipgroc, @kddeisz - Comments on lines with array references were previously deleting the array references entirely. For example,
array[index] # comment

would previously result in array[], but now prints properly.

0.12.2 - 2019-04-30

Changed

  • @kddeisz - When symbol literal hash keys end with =, they cannot be transformed into hash labels.
  • @xipgroc, @kddeisz - Fixed when blocks on methods with no arguments are transformed into to_proc syntax.

0.12.1 - 2019-04-22

Changed

  • @kddeisz - If a lambda literal is nested under a command or command_call node anywhere in the heirarchy, then it needs to use the higher-precedence { ... } braces as opposed to the do ... end delimiters.
  • @jpickwell, @kddeisz - Calling super with a block and no args was causing the parser to fail when attempting to inspect lambda nodes.
  • @kddeisz - Support better breaking within interpolation by grouping the interpolated content.

0.12.0 - 2019-04-18

Added

  • @kddeisz - Automatically convert lambda { ... } method calls into -> { ... } literals.

0.11.0 - 2019-04-18

Added

  • @kddeisz - Support for parsing things with a ruby shebang (e.g., #!/usr/bin/env ruby or #!/usr/bin/ruby).
  • @kddeisz - Big tests refactor.
  • @kddeisz - Make multiple when predicates break at 80 chars and then wrap to be inline with the other predicates.
  • @kddeisz - Automatically add underscores in large numbers that aren't already formatted.
  • @AlanFoster - Better support for inline access control modifiers.
  • @jpickwell, @kddeisz - Better support for heredocs in hash literals.
  • @kddeisz - Better support for heredocs in array literals.
  • @kddeisz - Support automatically transforming def/begin/rescue/end/end into def/rescue/end.

Changed

  • @deecewan - Fixed support for dynamic string hash keys.
  • @kddeisz - Moved case/when into its own file and added better documentation.
  • @kddeisz - Moved begin/rescue into its own file.
  • @AlanFoster - Automatically add newlines around access modifiers.
  • @kddeisz - Alignment of command calls with arguments is fixed.
  • @aaronjensen, @kddeisz - Alignment of to is explicitly allowed to not indent to better support rspec.
  • @kddeisz - Fix up the to_proc transform so that it works with other argument handling appropriately.
  • @kddeisz - Fixed regression on regexp comments.
  • @CodingItWrong, @kddeisz - Fix up block delimiters when nested inside a command or command_call node.
  • @kddeisz - Moved hashes into its own file.

0.10.0 - 2019-03-25

Added

  • @kddeisz - Support for block-local variables.
  • @kddeisz - Support for dyna-symbols that are using single quotes.

Changed

  • @kddeisz - Force method calls after arrays, blocks, hashes, and xstrings to hang onto the end of the previous nodes.
  • @kddeisz - Check before anything else for an invalid ruby version.

0.9.1 - 2019-03-24

Changed

  • @kddeisz - Better support string quotes by favoring what the user chose if the string contains escape patterns.
  • @kddeisz - Better support heredocs within method calls.

0.9.0 - 2019-03-18

Added

  • @kddeisz - Support the hasPragma function.
  • @kddeisz - Support the new number_arg node type in Ruby 2.7.

Changed

  • @kddeisz - Limit the number of nodes that are allowed to turn into ternary expressions.

0.8.0 - 2019-03-08

Added

  • @kddeisz - Add eslint and fix up existing violations.
  • @AlanFoster - Add the infra for the prettier ruby gem.
  • @kddeisz - Add a rake task for easier process integration for the ruby gem.
  • @kddeisz - Handle direct interpolation of strings with %w array literals (i.e., ["#{foo}"] should not be transformed into a %w array).

Changed

  • @kddeisz - Fix string escaping for hex digit bit patterns when there's only one character after the "x".
  • @AlanFoster - Don't allow line breaks between brace block params.
  • @johnschoeman - Switch over the array.rb test case to minitest.
  • @AlanFoster - Test improvements to allow running in parallel.
  • @johnschoeman - Switch over assign.rb test case to minitest.
  • @AlanFoster - Add a contributing guide.
  • @AlanFoster - Handle longer command nodes.
  • @kddeisz - Changed the ruby executable within the prettier gem to rbprettier for easier autocomplete.

Removed

  • @kddeisz - All instances of the spread (...) operator so that we can support older versions of node.

0.7.0 - 2019-02-24

Changed

  • @kddeisz - Support checking for escaping within strings to force double quotes (e.g., "\n").
  • @RossKinsella, @kddeisz - Handle cases with empty class and module declarations that need to break.
  • @AlanFoster - Align the bin/print and bin/sexp APto support bin/print taking a filepath.
  • @AndrewRayCode, @kddeisz - Support lambdas that don't break and are inline.
  • @AlanFoster - Switch over the numbers.rb test to minitest.
  • @AlanFoster - Switch over the kwargs.rb test to minitest.
  • @AlanFoster - Bail out early if the Ruby input is invalid.
  • @kddeisz - Support __END__ content.
  • @AlanFoster - Fix up issue with whitespace being added within regexp that are multiline.
  • @AlanFoster - Better support for destructuring within multi assignment.
  • @kddeisz - Switch next test over to minitest.
  • @kddeisz - Handle multiple arguments to next with a space between.
  • @AndrewRayCode, @kddeisz - Handle multi-line conditional predicate (should align with keyword).
  • @aaronjensen, @kddeisz - Properly support adding trailing commas with and without blocks.
  • @AlanFoster, @kddeisz - Fix regression of handling comments within arrays on array literals.
  • @AlanFoster - Support multiple arguments to undef.

0.6.3 - 2019-02-18

Changed

  • @kddeisz - Switch over binary fixture to minitest.
  • @kddeisz - Reconfigure parser into multiple layer modules so that it's easier to understand and manage.
  • @kddeisz - Handle comments from within begin, rescue, ensure, while, and until nodes.
  • @kddeisz - Properly indent heredocs without taking into account current indentation level.

0.6.2 - 2019-02-17

Changed

0.6.1 - 2019-02-15

Changed

0.6.0 - 2019-02-14

Added

  • @kddeisz - Handle non UTF-8 comments.
  • @kddeisz - Handle non UTF-8 identifiers.
  • @kddeisz - Handle non UTF-8 strings.
  • @kddeisz - Handle empty parens.
  • @kddeisz - Handle rescue with splats preceeding the exception names.

Changed

  • @kddeisz - Use JSON::fast_generate to get the s-expressions back from the parser.
  • @NoahTheDuke, @kddeisz - Handle broken lambdas from within command and command_call nodes.

0.5.2 - 2019-02-13

Changed

  • @kddeisz - Support embedded expressions within strings that contain only keywords, as in "#{super}".

0.5.1 - 2019-02-13

Changed

  • @yuki24, @kddeisz - Force do blocks that we know have to be do blocks to break.
  • @kmcq, @kddeisz - Handle command and command_call nodes do blocks by forcing them to break.
  • @ashfurrow, @kddeisz - Attach comments to full hash association nodes, not just the value.

0.5.0 - 2019-02-13

Added

  • @kddeisz - Automatically convert arrays of all string literals to %w arrays.
  • @kddeisz - Automatically convert arrays of all symbol literals to %i arrays.

Changed

  • @kddeisz - Move the args_add and args_new handling into the parser.
  • @uri, @kddeisz - Change command_call nodes to properly indent when broken and to not add a trailing comma.
  • @kddeisz - Rename the trailingComma option to addTrailingCommas to not conflict with the JS option.

0.4.1 - 2019-02-12

Changed

  • @kddeisz - Provide the makeList utility for the nodes that are lists from within ripper.
  • @awinograd, @kddeisz - Again, this time for real, properly escape strings.
  • @kddeisz - Fix up trailing commas on command calls.

0.4.0 - 2019-02-12

Added

Changed

0.3.7 - 2019-02-11

Changed

  • @kddeisz - Split up statements even if they started on the same line with ;s unless they are within an embedded expression.
  • @kddeisz - Properly handle escaped quotes within strings.

0.3.6 - 2019-02-10

Changed

0.3.5 - 2019-02-09

Changed

  • @kddeisz - Handle lonely operators in Ruby 2.5.

0.3.4 - 2019-02-09

Changed

  • @kddeisz - Comments are now properly attached inside defs nodes.
  • @kddeisz - Support multiple inline comments on nodes.
  • @kddeisz - Support inline comments from within the EXPR_END|EXPR_LABEL lexer state.
  • @cbothner - Stop transforming multistatement blocks with to_proc.
  • @kddeisz - do blocks necessarily need to break their parent nodes.
  • @eins78, @kddeisz - Handle next node edge case with args_add as the body.

0.3.3 - 2019-02-09

Changed

  • @bugthing, @kddeisz - Command nodes within conditionals now break parents to disallow them from being turned into ternary expressions.
  • @awinograd, @kddeisz - Properly escape double quotes when using preferSingleQuotes: false.

0.3.2 - 2019-02-09

Changed

  • @kddeisz - Don't define duplicated methods in the parser.
  • @kddeisz - Let prettier know about .rb and .rake files so you don't have to specify the parser when running.
  • @kddeisz - Renamed the package to @prettier/plugin-ruby.

0.3.1 - 2019-02-07

Changed

  • @kddeisz - Automatically add parens to method declarations.
  • @kddeisz - Handle comments on bare hash assocs.
  • @kddeisz - Handle method_add_block nodes where the statements may be nested one more level.
  • @kddeisz - Handle heredocs nested no matter how many levels deep.

0.3.0 - 2019-02-07

Added

Changed

  • @kddeisz - Ignore current indentation when creating embdocs so that =begin is always at the beginning of the line.
  • @kddeisz - Move regexp_add and regexp_new handling into the parser.
  • @kddeisz - Move xstring_add and xstring_new handling into the parser.
  • @kddeisz - Move string_add and string_content handling into the parser.
  • @kddeisz - Move mrhs_add and mrhs_new handling into the parser.
  • @kddeisz - Move mlhs_add and mlhs_new handling into the parser.

0.2.1 - 2019-02-06

Changed

  • @kddeisz - Handle brace blocks on commands properly.
  • @kddeisz - Break parent and return do blocks when called from a command node.
  • @kddeisz - Handle edge cases with if statements where there is no body of the if (so it can't be converted to a ternary).

0.2.0 - 2019-02-06

Added

  • @kddeisz - Handle methref nodes from Ruby 2.7.
  • @kddeisz - Allow module nodes to shorten using ; when the block is empty.

Changed

  • @kddeisz - Handle splat within an array, as in [1, 2, *foo].
  • @kddeisz - Disallow comments from being attached to intermediary regex nodes.
  • @kddeisz - Fix to_proc transforms to reference the method called as opposed to the parameter name.
  • @kddeisz - Change statement lists to be generated within the parser instead of the printer, thereby allowing finer control over comments.
  • @kddeisz - Completely revamp comment parsing by switching off the internal lexer state from ripper. This should drastically increase accuracy of comment parsing in general, and set us up for success in the future.
  • @kddeisz - Allow comments to be attached to CHAR nodes.
  • @kddeisz - Disallow comments from being attached to args_new nodes.
  • @kddeisz - Track start and end lines so we can better insert block comments.
  • @kddeisz - Handle intermediary array nodes in the parse for better comment handling.

0.1.2 - 2019-02-05

Changed

  • @kddeisz - Handle guard clauses that return with no parens.

0.1.1 - 2019-02-05

Changed

  • @kddeisz - Handle class method calls with the :: operator.
  • @kddeisz - Handle strings with apostrophes when using preferSingleQuote.
  • @kddeisz - Have travis run multiple ruby versions.
  • @kddeisz - Explicitly fail if ruby version is < 2.5.
  • @kddeisz - Disallow comments from being attached to intermediary string nodes.

0.1.0 - 2019-02-04

Added

  • Initial release 🎉