Require superClass for Lexer in split mode for JavaScript target #1966
Conversation
…s available on Lexer prototype for use in lexer actions (this.SOME_MODE) like in Java actions, Utils.escapeWhiteSpace uses regex for global replace (was replacing only first occurnce of \n etc).
| if (escapeSpaces) { | ||
| s = s.replace(" ", "\u00B7"); | ||
| s = s.replace(/ /g, "\u00B7"); | ||
| } | ||
| return s; | ||
| } |
| @@ -160,6 +160,7 @@ Object.defineProperty(<parser.name>.prototype, "atn", { | |||
| }); | |||
|
|
|||
| <parser.name>.EOF = antlr4.Token.EOF; | |||
| <parser.name>.prototype.EOF = antlr4.Token.EOF; | |||
There was a problem hiding this comment.
sorry but this is superfluous
| @@ -802,6 +803,9 @@ var antlr4 = require('antlr4/index'); | |||
| >> | |||
|
|
|||
| Lexer(lexer, atn, actionFuncs, sempredFuncs, superClass) ::= << | |||
| <if(superClass)> | |||
| var <superClass> = require('./<superClass>').<superClass>; | |||
| <endif> | |||
| @@ -819,14 +823,18 @@ function <lexer.name>(input) { | |||
| <lexer.name>.prototype.constructor = <lexer.name>; | |||
|
|
|||
| <lexer.name>.EOF = antlr4.Token.EOF; | |||
| <lexer.name>.prototype.EOF = antlr4.Token.EOF; | |||
| <lexer.tokens:{k | <lexer.name>.<k> = <lexer.tokens.(k)>;}; separator="\n", wrap, anchor> | ||
| <lexer.tokens:{k | <lexer.name>.prototype.<k> = <lexer.tokens.(k)>;}; separator="\n", wrap, anchor> |
|
|
||
| <if(lexer.channels)> | ||
| <lexer.channels:{c| <lexer.name>.<c> = <lexer.channels.(c)>;}; separator="\n"> | ||
| <lexer.channels:{c| <lexer.name>.prototype.<c> = <lexer.channels.(c)>;}; separator="\n"> | ||
|
|
|
|
||
| <endif> | ||
| <if(rest(lexer.modes))> | ||
| <rest(lexer.modes):{m| <lexer.name>.<m> = <i>;}; separator="\n"> | ||
| <rest(lexer.modes):{m| <lexer.name>.prototype.<m> = <i>;}; separator="\n"> | ||
|
|
|
Hi, |
|
Hi |
|
I meant bless not bleed |
|
So the downside of ids on the class vs instance is that superclasses will need to reach into the subclass (usually the generated Lexer) to access definitions. e.g. if I have a token: From my lexer's super class, I need to do this: I guess its the same with Java currrently: Example: |
|
Reversed proto changes. |
|
@parrt blessed |
In split mode grammars, the generated lexer.js wasn't requiring the superclass.
Lexer ids such as token/mode/channel names were set on the generated lexer and not its prototype making them class/static variables. For grammar actions to be universal across Java and JavaScript, these names must be accessible on the lexer instance (via [this.]SOME_MODE in Java, or this.SOME_MODE in JavaScript since JS doesn't scope to the local class without this.).
Finally,
Utils.escapeWhiteSpacewas replacing only the first instance of\netc. Switched to a regex with global replace.