unofficial mirror of emacs-devel@gnu.org 
 help / color / mirror / code / Atom feed
From: Yuan Fu <casouri@gmail.com>
To: "Mattias Engdegård" <mattiase@acm.org>
Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel <emacs-devel@gnu.org>
Subject: Re: Matching regex case-sensitively in C strings?
Date: Tue, 8 Nov 2022 11:31:35 -0800	[thread overview]
Message-ID: <5711A9D3-7BCB-44AE-8911-5E039FF5FBB8@gmail.com> (raw)
In-Reply-To: <580E87E6-DCFD-42AE-807A-339BBB3878C2@acm.org>

[-- Attachment #1: Type: text/plain, Size: 1994 bytes --]



> On Nov 8, 2022, at 2:18 AM, Mattias Engdegård <mattiase@acm.org> wrote:
> 
> 7 nov. 2022 kl. 21.35 skrev Yuan Fu <casouri@gmail.com>:
> 
>> fast_c_string_match_ignore_case (Lisp_Object regexp,
>> 				 const char *string, ptrdiff_t len)
>> {
>> regexp = string_make_unibyte (regexp);
> 
> This is expensive and not obviously correct when it makes a difference. Ie, no longer "fast", and may hide bugs.
> Something should be done about that.
> 
>> // Why do we need to unwind stack?
>> specpdl_ref count = SPECPDL_INDEX ();
> 
> Because freeze_pattern pushes an unwind-protect on the specpdl.
> 
>> struct regexp_cache *cache_entry
>>   = compile_pattern (regexp, 0, Vascii_canon_table, 0, 0);
> 
> `Vascii_canon_table` is what makes it case-insensitive; you want to use Qnil (but you probably already know that now).
> Since this is the only thing that differs from your intended use, I suggest you generalise this subroutine with a boolean parameter.
> 
>> // What does freezing a pattern do?
>> freeze_pattern (cache_entry);
> 
> It locks the compiled pattern record to make the regexp engine reentrant (but here it also seems to be used for GC purposes; not sure about that).
> 
>> // What is re_match_object for? I see that it can be t, nil or a string.
>> re_match_object = Qt;
> 
> Described in regex-emacs.h:
> 
>> /* The string or buffer being matched.
>>   It is used for looking up syntax properties.
>> 
>>   If the value is a Lisp string object, match text in that string; if
>>   it's nil, match text in the current buffer; if it's t, match text
>>   in a C string.
>> 
>>   This value is effectively another parameter to re_search_2 and
>>   re_match_2.  No calls into Lisp or thread switches are allowed
>>   before setting re_match_object and calling into the regex search
>>   and match functions.  These functions capture the current value of
>>   re_match_object into gl_state on entry.
>> 

Thanks! How about:

Yuan


[-- Attachment #2: c_string_match.diff --]
[-- Type: application/octet-stream, Size: 2112 bytes --]

diff --git a/src/lisp.h b/src/lisp.h
index 1e41e2064c9..d0b3f8f05a5 100644
--- a/src/lisp.h
+++ b/src/lisp.h
@@ -4772,6 +4772,8 @@ fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string)
 
 extern ptrdiff_t fast_c_string_match_ignore_case (Lisp_Object, const char *,
 						  ptrdiff_t);
+extern ptrdiff_t fast_c_string_match (Lisp_Object, const char *, ptrdiff_t,
+				      bool);
 extern ptrdiff_t fast_looking_at (Lisp_Object, ptrdiff_t, ptrdiff_t,
                                   ptrdiff_t, ptrdiff_t, Lisp_Object);
 extern ptrdiff_t find_newline1 (ptrdiff_t, ptrdiff_t, ptrdiff_t, ptrdiff_t,
diff --git a/src/search.c b/src/search.c
index b5d6a442c0f..90856cf5c12 100644
--- a/src/search.c
+++ b/src/search.c
@@ -505,10 +505,29 @@ fast_string_match_internal (Lisp_Object regexp, Lisp_Object string,
 fast_c_string_match_ignore_case (Lisp_Object regexp,
 				 const char *string, ptrdiff_t len)
 {
+  return fast_c_string_match (regexp, string, len, true);
+}
+
+/* Match REGEXP against STRING, searching all of STRING and return the
+   index of the match, or negative on failure.  This does not clobber
+   the match data.  Ignore case when searching if IGNORE_CASE is true.
+
+   We assume that STRING contains single-byte characters.  */
+
+ptrdiff_t
+fast_c_string_match (Lisp_Object regepx,
+		     const char *string, ptrdiff_t len, bool ignore_case)
+{
+  /* FIXME: This is expensive and not obviously correct when it makes
+     a difference. I.e., no longer "fast", and may hide bugs.
+     Something should be done about this.  */
   regexp = string_make_unibyte (regexp);
+  /* Record specpdl index because freeze_pattern pushes an
+     unwind-protect on the specpdl.  */
   specpdl_ref count = SPECPDL_INDEX ();
+  Lisp_Object translate_table = ignore_case ? Vascii_canon_table : Qnil;
   struct regexp_cache *cache_entry
-    = compile_pattern (regexp, 0, Vascii_canon_table, 0, 0);
+    = compile_pattern (regexp, 0, translate_table, 0, 0);
   freeze_pattern (cache_entry);
   re_match_object = Qt;
   ptrdiff_t val = re_search (&cache_entry->buf, string, len, 0, len, 0);

[-- Attachment #3: Type: text/plain, Size: 2 bytes --]




  parent reply	other threads:[~2022-11-08 19:31 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-07  5:41 Matching regex case-sensitively in C strings? Yuan Fu
2022-11-07 11:52 ` Eli Zaretskii
2022-11-07 20:35   ` Yuan Fu
2022-11-08 10:18     ` Mattias Engdegård
2022-11-08 12:29       ` Eli Zaretskii
2022-11-08 19:31       ` Yuan Fu [this message]
2022-11-08 19:37         ` Eli Zaretskii
2022-11-08 20:59           ` Yuan Fu
2022-11-09 10:53             ` Mattias Engdegård
2022-11-09 10:33           ` Mattias Engdegård
2022-11-09 13:06             ` Eli Zaretskii
2022-11-10  9:52               ` Mattias Engdegård
2022-11-10 11:18                 ` Eli Zaretskii
2022-11-10 11:35                   ` Robert Pluim
2022-11-10 14:20                   ` Mattias Engdegård
2022-11-10 22:25                     ` Yuan Fu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/emacs/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5711A9D3-7BCB-44AE-8911-5E039FF5FBB8@gmail.com \
    --to=casouri@gmail.com \
    --cc=eliz@gnu.org \
    --cc=emacs-devel@gnu.org \
    --cc=mattiase@acm.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/emacs.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).