regexp newline

[GeN1e]

AR0530 
DEMPRI    Pitre                         xxxxxxxxxxxxxxxxxxxxxxxx
DEMFIG01  Falahar                       xxxxxxxxxxxxxxxxxxxxxxxx
DEMFIG02  Valeria                       xxxxxxxxxxxxxxxxxxxxxxxx
DEMMAG    Dracandros                    xxxxxxxxxxxxxxxxxxxxxxxx
AR2014 
TRGENI01  Khan Zahraa                   xxxxxxxxxxxxxxxxxxxxxxxx
TRGENI03  Faafirah                      xxxxxxxxxxxxxxxxxxxxxxxx

Code: [Select]

<<<<<<<<dummy
>>>>>>>>
COPY dummy dummy.txt
COPY - area_actors.txt area_actors.txt
  REPLACE_EVALUATE ~\(.+\)~ BEGIN
    PATCH_IF STRING_LENGTH ~%MATCH1%~ < 11 BEGIN

      SPRINT areafile ~%MATCH1%~

      INNER_ACTION BEGIN
        APPEND_OUTER dummy.txt ~~~~~SET $"%areafile%"("%actor%")= ~%schedule%~ ~~~~~
      END
    END
  END ~~Yields this -

SET $"AR0530 
"("%actor%")= ~%schedule%~
SET $"AR2014 
"("%actor%")= ~%schedule%~ 

Bad. Using this instead of SPRINT areafile ~%MATCH1%~

Code: [Select]

      INNER_PATCH ~%MATCH1%~ BEGIN
        FOR (i=0;i<STRING_LENGTH ~%MATCH1%~;i+=1) BEGIN
          READ_ASCII i char (1)
          PATCH_IF (~%char%~ STRING_COMPARE_REGEXP ~[a-zA-Z0-9]~) BEGIN
            WRITE_BYTE i 0
          END
        END
        READ_ASCII 0 areafile (8) NULL
      ENDyields correct thing

SET $"AR0530"("%actor%")= ~%schedule%~
SET $"AR2014"("%actor%")= ~%schedule%~ 

I would assume that for whatever reason ~\(.+\)~ catches newline character. But I thought a dot is anything except newline?

Also, doesn't readme mention that ~~~~~ allows to store any char, including % in the string? I'm not complaining, quite the opposite in fact, as otherwise it would have taken more labor to do it right, but still it's strange.

[the bigg]

Maybe it's catching the \r? Either way, OCaml regexp are way beyond broken, and I can't change how they work.

The readme actually mentions ~~~~~ (it has been in since 168 or so) - just search it for ~~~~~. Of course, it isn't easy to find if you don't already know what to look for, but then what is?

[Mike1072]

Maybe it's catching the \r?

Definitely looks like it.

I did a straight-up REPLACE_TEXTUALLY ~\(.+\)~ ~@\1!~ on

Code: [Select]

A
B
Cwith the 3 styles of newlines.

When Linux formatted, it gave the expected

Code: [Select]

@A!\n
@B!\n
@C!\n
When Mac formatted, it gave

Code: [Select]

@A\r
B\r
C\r
!
And when Windows formatted, it gave

Code: [Select]

@A\r!\n
@B\r!\n
@C\r!\n
Is it possible to do any pre-processing on regular expressions before they get sent to OCaml?  Turn dots that aren't escaped or in character sets into [^%mnl%%lnl%]?  What I'd really like are some \t, \r, \n shortcuts, so we don't have to use variables....

[the bigg]

Regexp preprocessing is a compatibility bomb waiting to explode.

[Mike1072]

Sounds exciting!

[the bigg]

There you go.

[cmorgan]

Check CamDawg's stuff in Fixpack and bg1noc and some other places - he adss


D:\ie_modding\BG2_Fixpack-v8\bg2fixpack\lib\extra_regexp_vars.tph

Code: [Select]

OUTER_INNER_PATCH ~12~ BEGIN
  WRITE_BYTE 1 0x09
  READ_ASCII 1 tab (1)  // 0x09, tab
  WRITE_BYTE 1 0x0a
  READ_ASCII 1 lnl (1)  // 0x0a, Linux
  WRITE_BYTE 0 0x0d
  READ_ASCII 0 mnl (1)  // 0x0d, Mac
  READ_ASCII 0 wnl (2)  // 0x0d0a, Windows
END

Will this help? (at least you will get a good laugh at my misunderstanding )

[the bigg]

Mike is asking for that to be performed automatically, actually.

[erik]

Definitely going to explode hairily, if changed.