correctly parse enum values #104
Conversation
enum values can be expressions including - unary operators - binary operators - string and number literals - parenthesis - identifiers See: https://github.com/microsoft/TypeScript/blob/master/doc/spec.md#92-enum-members https://github.com/microsoft/TypeScript/blob/master/doc/spec.md#419-binary-operators Note: this implementation performs a flat parse that neither handles expression-tree-building nor operator precedences. Note: This commit removes the minus from the NumericLit-token and parses it explicitly in TSDefParser.numberLiteral
SrTobi
force-pushed
the
value-expressions-pr
branch
from
91adb75 to
1ed9dbd
Compare
|
The CI fails because it uses typescript |
|
TypeScript 2.6.2 was added just because it was latest when #56 was created. |
|
ah ok. I did that. thx |
| B = 1 << 3, | ||
| C = 10**2+1*8- - -3/3>>2<<3>>>+19%~~~3+ + +9 | ||
| } | ||
| } |
There was a problem hiding this comment.
Please make sure there is a (single) new line at the end of the file, so that git and GitHub don't have that mark.
| @@ -105,7 +117,20 @@ class TSDefParser extends StdTokenParsers with ImplicitConversions { | |||
| "enum" ~> typeName ~ ("{" ~> ambientEnumBody <~ "}") ^^ EnumDecl | |||
|
|
|||
| lazy val ambientEnumBody: Parser[List[Ident]] = | |||
| repsep(identifier <~ opt("=" ~ (numericLit | stringLit) ), ",") <~ opt(",") | |||
| repsep(identifier <~ opt("=" ~! valueExpression), ",") <~ opt(",") | |||
There was a problem hiding this comment.
they are non backtracking operators. basically, after the = is parsed here the parser will not go back and try other alternatives prior to = when an error comes afterwards. this is very important to get better error locations when parsing fails, because we know that when we arrived after the = there is no other rule that could parse this. so there is no point in going back and try to parse a class declaration instead of the enum declaration, for example. Instead we just stop and show the much more accurate error location.
I was already about to make a pr which added many more of these operators but it's not always totally clear where to introduce these cuts while preserving the semantics and I want to make sure that I am correct. I will make the pr in a few days
| // for value expressions | ||
| val binaryValueOperators = Set( | ||
| "+", "-", "*", "/", "**", "%", | ||
| "<<", ">>>", ">>", "&", "|", "^" |
There was a problem hiding this comment.
These 2 lines should be indented with 4 spaces.
There was a problem hiding this comment.
And given my comments below, consider just including those directly inside lexical.delimiters ++= above.
| lexical.delimiters ++= binaryValueOperators | ||
|
|
||
| val unaryValueOperators = Set( | ||
| "+", "-", "~" |
There was a problem hiding this comment.
This line should be indented with 4 spaces.
| "(" ~! valueExpression ~ ")" | ||
|
|
||
| lazy val binaryValueOperator = | ||
| elem("binary operator", tok => binaryValueOperators.contains(tok.chars)) |
There was a problem hiding this comment.
I'd rather you use a more direct style, like
| elem("binary operator", tok => binaryValueOperators.contains(tok.chars)) | |
| "+" | "-" | "*" | "/" | "**" | "%" | "<<" | ">>>" | ">>" | "&" | "|" | "^" |
There was a problem hiding this comment.
may I ask why? in the current version the operators occur only once instead of twice at different locations
| elem("binary operator", tok => binaryValueOperators.contains(tok.chars)) | ||
|
|
||
| lazy val unaryValueOperator = | ||
| elem("unary operator", tok => unaryValueOperators.contains(tok.chars)) |
There was a problem hiding this comment.
Same here:
| elem("unary operator", tok => unaryValueOperators.contains(tok.chars)) | |
| "+" | "-" | "~" |
| repsep(identifier <~ opt("=" ~! valueExpression), ",") <~ opt(",") | ||
|
|
||
| lazy val valueExpression: Parser[_] = | ||
| rep1sep(success() ~>! rep(unaryValueOperator) ~ |
There was a problem hiding this comment.
I don't understand the purpose of success() ~>! . Could you enlighten me?
this pr fixes #103
enum values can be expressions including
See:
https://github.com/microsoft/TypeScript/blob/master/doc/spec.md#92-enum-members
https://github.com/microsoft/TypeScript/blob/master/doc/spec.md#419-binary-operators
Note:
this implementation performs a flat parse that neither
handles expression-tree-building nor operator precedences.
Note:
This commit removes the minus from the NumericLit-token and
parses it explicitly in TSDefParser.numberLiteral