Re: [bug-gettext] [PATCH 4/4] xg-js: fix end-of string miss bug due to backslashes

[Daiki Ueno]

Andreas Stricker <address@hidden> writes:

> --- a/gettext-tools/src/x-javascript.c
> +++ b/gettext-tools/src/x-javascript.c
> @@ -1009,7 +1009,7 @@ phase7_getuc (int quote_char, unsigned int 
> *backslash_counter)
>            case '\n':
>              continue;
>            case '\\':
> -            ++*backslash_counter;
> +            *backslash_counter = 0;

Thanks for looking into this.  However, I doubt that backslash_counter
is even needed for JavaScript parser.  The original code in the Python
parser has a check like this:

  if (c == quote_char && (interpret_ansic || (*backslash_counter & 1) == 0))

In JavaScript parser, interpret_ansic is always true, and the latter
condition won't be checked (i.e. this seems to be a regression
introduced by my rewrite :)

By the way, ECMA 262 says:

   * The CV of DoubleStringCharacter :: \ EscapeSequence is the CV of the
     EscapeSequence.

   * The CV of CharacterEscapeSequence :: NonEscapeCharacter is the CV of
     the NonEscapeCharacter.

where CV stands for "character value".  That means, "\xxx" should be
treated as "xxx", while the current implementation treats it as "\\xxx".

Regards,
-- 
Daiki Ueno