On binary files, it seems that testing the UTF-8 sequences in
pcresearch.c is faster than asking pcre_exec to do that (because
of the retry I assume); see attached patch. It actually checks
UTF-8 only if an invalid sequence was already found by pcre_exec,
assuming that pcre_exec can check the validity of a valid text
file in a faster way.
On some file similar to PDF (test 1):
Before: 1.77s
After: 1.38s
But now, the main problem is the many pcre_exec. Indeed, if I replace
the non-ASCII bytes by \n with:
LC_ALL=C tr \\200-\\377 \\n
(now, one has a valid file but with many short lines), the grep -P time
is 1.52s (test 2). And if I replace the non-ASCII bytes by null bytes
with:
LC_ALL=C tr \\200-\\377 \\000
the grep -P time is 0.30s (test 3), thus it is much faster.
Note also that libpcre is much slower than normal grep on simple words,
but on "a[0-9]b", it can be faster:
grep PCRE PCRE+patch
test 1 4.31 1.90 1.53
test 2 0.18 1.61 1.63
test 3 3.28 0.39 0.39
With grep, I wonder why test 2 is much faster.
--
Vincent Lefèvre <[email protected]> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)
diff --git a/src/pcresearch.c b/src/pcresearch.c
index 5451029..6bff1e4 100644
--- a/src/pcresearch.c
+++ b/src/pcresearch.c
@@ -38,6 +38,8 @@ static pcre_extra *extra;
# endif
#endif
+#define INVALID(C) (to_uchar (C) < 0x80 || to_uchar (C) > 0xbf)
+
/* Table, indexed by ! (flag & PCRE_NOTBOL), of whether the empty
string matches when that flag is used. */
static int empty_match[2];
@@ -156,6 +158,7 @@ Pexecute (char const *buf, size_t size, size_t *match_size,
char const *line_start = buf;
int e = PCRE_ERROR_NOMATCH;
char const *line_end;
+ int invalid = 0;
/* If the input type is unknown, the caller is still testing the
input, which means the current buffer cannot contain encoding
@@ -212,25 +215,54 @@ Pexecute (char const *buf, size_t size, size_t
*match_size,
if (multiline)
options |= PCRE_NO_UTF8_CHECK;
- e = pcre_exec (cre, extra, p, search_bytes, 0,
- options, sub, NSUB);
- if (e != PCRE_ERROR_BADUTF8)
+ int valid_bytes = search_bytes;
+ if (invalid)
{
- if (0 < e && multiline && sub[1] - sub[0] != 0)
+ /* At least an encoding error was found. Other such errors
+ are likely to occur, and detecting them here is faster
+ in average than relying on pcre. */
+ options |= PCRE_NO_UTF8_CHECK;
+ char const *p2 = p;
+ while (p2 != line_end)
{
- char const *nl = memchr (p + sub[0], eolbyte,
- sub[1] - sub[0]);
- if (nl)
+ unsigned char c = p2[0];
+ size_t len =
+ c < 0x80 ? 1 :
+ c < 0xc2 || c > 0xf7 || INVALID(p2[1]) ? 0 :
+ c < 0xe0 ? 2 : INVALID(p2[2]) ? 0 :
+ c < 0xf0 ? 3 : INVALID(p2[3]) ? 0 : 4;
+ if (len == 0)
{
- /* This match crosses a line boundary; reject it. */
- p += sub[0];
- line_end = nl;
- continue;
+ valid_bytes = p2 - p;
+ break;
}
+ p2 += len;
}
- break;
}
- int valid_bytes = sub[0];
+
+ if (valid_bytes == search_bytes)
+ {
+ e = pcre_exec (cre, extra, p, search_bytes, 0,
+ options, sub, NSUB);
+ if (e != PCRE_ERROR_BADUTF8)
+ {
+ if (0 < e && multiline && sub[1] - sub[0] != 0)
+ {
+ char const *nl = memchr (p + sub[0], eolbyte,
+ sub[1] - sub[0]);
+ if (nl)
+ {
+ /* This match crosses a line boundary; reject it. */
+ p += sub[0];
+ line_end = nl;
+ continue;
+ }
+ }
+ break;
+ }
+ invalid = 1;
+ valid_bytes = sub[0];
+ }
/* Try to match the string before the encoding error.
Again, handle the empty-match case specially, for speed. */
