Don't use 2 characters for surrogates in identifiers because there is
support for the \u{} syntax when escaped identifiers were introduced.
Also catch eof errors while reading identifier names
Introduce ascii_identifiers:
By setting ascii_identifiers to undefined (default value),
ascii_identifiers will print identifiers using the same setting as
ascii_only within the limits of the ecmascript 6 grammar.
ascii_identifiers accept true and false, allowing identifiers to be
printed under different settings than strings with the ascii_only setting.
* Implement getter/setter with computed value
* Fix parsing getter/setter after static or generator token
* Allow storing expressions for computed expression in AST_SymbolMethod
* Allow get and set in shorthand properties in object literals
Fixes#1094, #1146 and #1221
* Fixes#1147: template strings not obeying -b ascii_only true
* Allow evaluation of template expressions by adding optimizers and
walkers
* Make sure tagged templates are never changed
* Remove template tokenizer in parser, add template tokenizer in
tokenizer. It is using a brace counter to track brace position of
templates
* Add tokens `template_head` and `template_substitution` but parsing
tokens stays mostly the same
* Do not output strings anymore in AST_TemplateString, instead use
AST_TemplateSegment
* Fix parsing tagged templates, allowing multiple templates behind
as spec allows this
These changes don't influence tagged templates because raw content
may influence code execution, however they are safe to do in normal
templates:
* Allow basic string concatenation of templates where possible
* Allow custom character escape style similar to strings, except in
tagged templates
Note that expressions are still compressed in tagged templates.
Optional things that may be improved later:
* Custom quote style for templates if it doesn't have expressions.
Making it obey the quote_style option if this is the case.
* Escaped newlines should also produce SyntaxError
* Fix multiline comment parsing and add tests
* Adapt makePredicate to handle \u2028 and \u2029
* Move up nlb check in regex so it's checked before any escape handling
* Change error messages to conform ecma standard
* Find_eol not recornizing \u2028 and \u2029 as line terminator
* Remove \u180e as it is removed in unicode 6.3.0 from the category zs
This should simplify and improve implementation, make it easier to
implement template strings, and keep master a bit more in sync with
harmony.
Previous implementation wasn't broken, though the loop gave me the
impression it could read infinite numbers and annoyed me a bit. It was
also slightly unnecessary because the lookup involved only 3 characters.
* Don't interpret strings with escaped content as directive
* Don't interpret strings after empty statement as directive
* Adapt output to prevent strings being represent as directive
* Introduce UGLIFY_DEBUG to allow internal testing like EXPECT_DIRECTIVE
- Partially reverting 91cdb93e57 and eaf3911c31 and reimplement
- Add generators support for objects and classes
- Only classes can have static methods so restrict use of it
Special thanks to @rvanvelzen and @kzc for reviewing this patch and
providing constructive feedback over and over again.
- Make let, const, and class symbols be declared in a block scope.
- Piggy back on existing catch symbol implementation to get block-aware mangling working
- Make sure unused block-scoped declarations can be dropped
- Don't eliminate a block if it has a block-scoped declaration
- Remove silly empty anonymous blocks left over from drop_unused
- AST_Toplevel now gets to call drop_unused too, since block-scoped variables aren't global!
- Don't consider block declarations global
It's now available during tree walking, i.e. walker.has_directive("use
asm"), rather than as part of the scope. It's thus no longer necessary
to call `figure_out_scope` before codegen. Added special bits in the
code generator to overcome the fact that it doesn't inherit from
TreeWalker.
Fix#861
