[fixed] Bug in gettokenindex()

[PeterH]

gettokenindex() (at least with sep=<crlf>) will return the wrong token-(= line-)number, if a "string" (it's *not* XY-syntax!) before the searched token is continued over 2 lines.
Example:

Code: Select all

//§ Test GetTokenIndex()
"Test"
   $lines = <<<#>>>
gd
eins1   <- line 2: OK
gfhjh
lines "with
continued" string
zwei2   <- line 6: NOK
jhgh
xfhg
drei3   <- line 9: NOK
#>>>;

   $x1 = gettokenindex("eins*", $lines, <crlf>, 'iw');
   $x2 = gettokenindex("zwei*", $lines, <crlf>, 'iw');
   $x3 = gettokenindex("drei*", $lines, <crlf>, 'iw');

   echo "x1 | 2 | 3 = $x1 | $x2 | $x3<crlf>Should be 2 | 6 | 9";

And a related wish: gettokenindex() should have a starting-position, like e.g. strpos() has.
If I scan a (much larger!) file like above e.g. for section numbers, I should be able to start scan n+1 with the found position of scan n. But now all scans will/must restart at pos 1.
Very bad: if there's a table of contents - all section numbers will be found there - I can't scan for the real section.
(What I'm going to code won't work without that!)

[admin]

Debatable. SC gettokenindex() ignores separators that are within quotes.

[PeterH]

OK - then there would be a bug in gettoken()
As, when you find a token with gettokenindex(), and then get that token with gettoken(), you get the wrong one.
And from the meaning of the scanned text it even would not meet syntax of .xys - as this is not an allowed situation.

So you should either meet the meaning of token: the text between 2 separators, or add an operand to define the syntax of the scanned text. And take care that the logic of gettokenindex() and gettoken() fit together.
Especially, if the text is text (pun intended), and the goal is to get a line.
(A line is a line is a line - pedantry again, and with reason, again.)

Debated enough?

[admin]

Yep, convinced, I'll fix gettokenindex().

[PeterH]

Yep, convinced, I'll fix gettokenindex().

Fine

A word to "related wish: gettokenindex() should have a starting-position" ?
It's missing badly!

[admin]

It's coming.

[PeterH]

[PeterH]

Oooh - *both* in avail beta

1) have a problem with smilies
2) must go to test

[PeterH]

OK: what I named a bug is fiixed!
(Easy to test, as the script stumbled here.)

Now *I* have a problem with an anomaly in the scanned fiile. Must recognize and handle...

And then, for using//testing the new start=, further programming has to be done...

Much work to do, so:

Thanks!