On Sat, Mar 19, 2005 at 09:38:14PM +0100, Marc Haber wrote: || tags #280178 patch moreinfo || thanks || || On Sun, Nov 07, 2004 at 09:38:29PM +0100, Vincent Zweije wrote: || > There are bugs in how ippl reads its configuration file, which may cause || > valid configurations to be rejected or interpreted incorrectly. || || Can you please give examples which can be used to do pre/post tests of || your patch?
It's a long time ago.
I'll go through the patch and annotate it, in the hope that you can
figure out some test cases.
General note: I've added () for old lex compatibility.
If you need working test cases, I'll need more time.
|| > A patch is attached which thoroughly cleans up the tokeniser
|| > specification.
||
|| I intend to apply that patch and upload to experimental after I have
|| received test cases.
|| diff -ru ippl-1.4.14-3/Source/ippl.l ippl-1.4.14-3.1/Source/ippl.l
|| --- ippl-1.4.14-3/Source/ippl.l 2004-11-07 20:56:41.000000000 +0100
|| +++ ippl-1.4.14-3.1/Source/ippl.l 2004-10-23 18:49:22.000000000 +0200
|| @@ -33,39 +33,35 @@
||
|| %}
||
|| -white [ \t]+
|| +cr [\r]
|| +lf [\n]
|| +white [ \a\b\t\v\f]+
|| +digit [0-9]
|| +letter [A-Za-z]
|| +alpha [A-Za-z0-9]
|| +hyp [-_]
|| +filechar [-_A-Za-z0-9.\\]
|| +labelchar [-_A-Za-z0-9]
The following characters are now also accepted as white space:
\a - ascii 7 - audible bell
\b - ascii 8 - backspace
\t - ascii 9 - tab
\v - ascii 11 - vertical tab
\f - ascii 12 - form feed
Pre-patch should report an unexpected character on each of these in the
config file.
|| -octdigit [0-7]
|| -hexdigit [0-9A-Fa-f]
|| -decdigit [0-9]
|| +eol ({cr}|{lf}|{cr}{lf}|{lf}{cr})
eol:
The following character sequence are now interpreted as end-of-line:
\n - ascii 10 - line feed (already pre-patch)
\r - ascii 13 - carriage return (new)
\n\r - ascii 10,13 - line feed, carriage return (new)
\r\n - ascii 13,10 - carriage return, line feed (new)
Pre-patch should complain about \r being unexpected.
|| -octchar {octdigit}+
|| -hexchar {hexdigit}+
|| -decchar {decdigit}+
|| +integer ({digit}+)
||
|| -char {octchar}|{hexchar}|{decchar}
|| -integer {decdigit}+
integer:
No change.
|| +byte (0*(1[0-9][0-9]|2[0-4][0-9]|25[0-5]|[1-9][0-9]|[0-9]))
|| +ip ({byte}\.{byte}\.{byte}\.{byte})
||
|| -ip {char}"."{char}"."{char}"."{char}
ip:
Hexadecimal digits are no longer accepted in IP address tokens.
IP address tokens are limited to byte-range values (0-255):
byte-overflow is no longer possible when decoding.
Try a config pattern such as:
1.2.3.256 (possibly accepted as 1.2.3.0 pre-patch)
a.b.c.de (pre-patch accepted as IP address, but still looked up by
forgiving address decoding function. Handled properly as
hostmask post-patch)
a.b.c.de/32 (pre-patch accepted, post-patch rejected)
|| +labelmask ({labelchar}+|"*")
|| +hostmask ({labelmask}("."{labelmask})*"."?)
||
|| -letter [A-Za-z]
|| -digit [0-9]
|| -letdig {letter}|{digit}
|| -hyp "-"|"_"
|| -letdighyp {letdig}|{hyp}
|| -label {letdig}+({hyp}{letdig}+)*
|| +ident ({letter}{alpha}*(-{alpha}+)*)
||
|| -hostmask ({label}|"*")("."({label}|"*"))*"."?
hostmask:
Factorized labelmask. No effective change.
|| -ident {letter}("-"?{letdig}+)*
ident:
Disambiguated. No effective change.
|| -letfile {letdighyp}|"\\"|"."
|| -file ("/"{letfile}+)("/"{letfile}+)+
|| +file (("/"{filechar}+)("/"{filechar}+)+)
file:
Sanely renamed letfile->filechar. No effective change.
|| %%
||
|| {white} { /* Ignore */ }
|| -"#".*$ { /* Ignore */ }
|| -"\n" { line_number++; return EOL; }
|| +#.* { /* Ignore */ }
|| +{eol} { line_number++; return EOL; }
comment:
Removed unnecessary "".
Removed unnecessary $ anchor.
No effective change.
eol:
As said, accepts extra sequences. However, a cr in or after a comment will
be gobbled by the comment token. This was already the case pre-patch,
and is harmless unless lines are ended by lone cr characters (as in
macintosh style line endings).
|| [lL][oO][gG][fF][oO][rR][mM][aA][tT] return LOGFORMAT;
|| [dD][eE][tT][aA][iI][lL][eE][dD] return DETAILED;
|| @@ -132,14 +128,14 @@
|| "addr_mask_req" { yylval.longval = 17; return NUMBER; }
|| "addr_mask_reply" { yylval.longval = 18; return NUMBER; }
||
|| -{ip} {
|| +{ident} {
|| yylval.stringval = strdup(yytext);
|| - return IP;
|| + return IDENTIFIER;
|| }
||
|| -{ident} {
|| +{ip} {
|| yylval.stringval = strdup(yytext);
|| - return IDENTIFIER;
|| + return IP;
|| }
||
|| {hostmask} {
I appear to have swapped the order of {ip} and {ident} to give {ident}
priority over {ip}, but in rear view I think this is not necessary because
they never recognize the same character sequences; {ip} always contains
3 dots, {ident} always contains 0 dots.
So much for now. Vincent.
--
Vincent Zweije <[EMAIL PROTECTED]> | "If you're flamed in a group you
<http://www.xs4all.nl/~zweije/> | don't read, does anybody get burnt?"
[Xhost should be taken out and shot] | -- Paul Tomblin on a.s.r.
signature.asc
Description: Digital signature
