On Fri, 2 Sep 2016 15:35:22 -0700
Paul Eggert <egg...@cs.ucla.edu> wrote:
> Norihiro Tanaka wrote:
> > However, the patch adds an argument to dfasyntax(). To synchronize
> > between grep and dfa easily, I expect it is applied before dfa is moved
> > to gnulib.
>
> Since we're already changing the DFA API already, how about if we have a
> flags arg that combines all these little Boolean arguments? That will make
> future changes less disruptive. Although the existing code supports any
> newline terminator, in practice only '\0' and '\n' are useful, so it's really
> just a boolean.
>
> Most of the changes in your proposed patch are subsumed by the changes for
> multithreading, so the patch can be simplified now. I installed the attached,
> and plan to follow up shortly about the corresponding Gawk changes that I'll
> propose.
Thanks for reviewing. You say we can simplified by the changes for
multithreading, but two changes in the patch are needed.
From fb1fe06e6f912039d524d42a6fff274550f55463 Mon Sep 17 00:00:00 2001
From: Norihiro Tanaka <nori...@kcn.ne.jp>
Date: Sat, 3 Sep 2016 12:02:37 +0900
Subject: [PATCH] dfa: additional change new option for anchored searches
* src/dfa.c (dfaexec_main): Do it.
---
src/dfa.c | 10 +++++++++-
1 files changed, 9 insertions(+), 1 deletions(-)
diff --git a/src/dfa.c b/src/dfa.c
index e009bc6..41b0ffb 100644
--- a/src/dfa.c
+++ b/src/dfa.c
@@ -3172,10 +3172,18 @@ dfaexec_main (struct dfa *d, char const *begin, char
*end, bool allow_nl,
mbp = p;
s = allow_nl ? d->newlines[s1] : 0;
+
+ s = (allow_nl ? d->newlines[s1]
+ : (d->syntax.sbit[eol] == CTX_NEWLINE ? 0
+ : (d->syntax.sbit[eol] == CTX_LETTER ? d->min_trcount - 1
+ : d->initstate_notbol)));
}
else if (d->fails[s])
{
- if (d->success[s] & d->syntax.sbit[*p])
+ if ((d->success[s] & d->syntax.sbit[*p])
+ || ((char *) p == end
+ && ACCEPTS_IN_CONTEXT (d->states[s].context, CTX_NEWLINE, s,
+ *d)))
goto done;
if (multibyte && s < d->min_trcount)
--
1.7.1
