Package: grep Tags: patch I ran following test, which used the regex enging in non-UTF8 locale.
$ yes abcd.abc | head -10000 > m $ time -p env LC_ALL=ja_JP.eucJP src/grep abcd.abd m real 7.28 user 6.36 sys 0.57 It's extremally slow. When regex engine is used in grep, a text is splitted by line. However all of buffer is passed to re_search and re_match. I seem that it's wrong. Norihiro
>From 7187092186b982b95e94df81393e8fa72060985c Mon Sep 17 00:00:00 2001 From: Norihiro Tanaka <[email protected]> Date: Mon, 17 Mar 2014 23:46:31 +0900 Subject: [PATCH] grep: matching line-by-line with regex * src/dfasearch.c (EGexecute): matching line-by-line with regex. --- src/dfasearch.c | 34 ++++++++++++++++++---------------- 1 file changed, 18 insertions(+), 16 deletions(-) diff --git a/src/dfasearch.c b/src/dfasearch.c index 0b56960..8697383 100644 --- a/src/dfasearch.c +++ b/src/dfasearch.c @@ -204,7 +204,7 @@ size_t EGexecute (char const *buf, size_t size, size_t *match_size, char const *start_ptr) { - char const *buflim, *beg, *end, *match, *best_match, *mb_start; + char const *buflim, *beg, *end, *ptr, *match, *best_match, *mb_start; char eol = eolbyte; int backref; regoff_t start; @@ -272,18 +272,20 @@ EGexecute (char const *buf, size_t size, size_t *match_size, /* Successful, no backreferences encountered! */ if (!backref) goto success; + ptr = beg; } else { /* We are looking for the leftmost (then longest) exact match. We will go through the outer loop only once. */ - beg = start_ptr; + beg = buf; end = buflim; + ptr = start_ptr; } /* If the "line" is longer than the maximum regexp offset, die as if we've run out of memory. */ - if (TYPE_MAXIMUM (regoff_t) < end - buf - 1) + if (TYPE_MAXIMUM (regoff_t) < end - beg - 1) xalloc_die (); /* If we've made it to this point, this means DFA has seen @@ -294,24 +296,24 @@ EGexecute (char const *buf, size_t size, size_t *match_size, { patterns[i].regexbuf.not_eol = 0; start = re_search (&(patterns[i].regexbuf), - buf, end - buf - 1, - beg - buf, end - beg - 1, + beg, end - beg - 1, + ptr - beg, end - ptr - 1, &(patterns[i].regs)); if (start < -1) xalloc_die (); else if (0 <= start) { len = patterns[i].regs.end[0] - start; - match = buf + start; + match = beg + start; if (match > best_match) continue; if (start_ptr && !match_words) goto assess_pattern_match; if ((!match_lines && !match_words) - || (match_lines && len == end - beg - 1)) + || (match_lines && len == end - ptr - 1)) { - match = beg; - len = end - beg; + match = ptr; + len = end - ptr; goto assess_pattern_match; } /* If -w, check if the match aligns with word boundaries. @@ -325,8 +327,8 @@ EGexecute (char const *buf, size_t size, size_t *match_size, while (match <= best_match) { regoff_t shorter_len = 0; - if ((match == buf || !WCHAR (to_uchar (match[-1]))) - && (start + len == end - buf - 1 + if ((match == beg || !WCHAR (to_uchar (match[-1]))) + && (start + len == end - beg - 1 || !WCHAR (to_uchar (match[len])))) goto assess_pattern_match; if (len > 0) @@ -335,8 +337,8 @@ EGexecute (char const *buf, size_t size, size_t *match_size, --len; patterns[i].regexbuf.not_eol = 1; shorter_len = re_match (&(patterns[i].regexbuf), - buf, match + len - beg, - match - buf, + beg, match + len - ptr, + match - beg, &(patterns[i].regs)); if (shorter_len < -1) xalloc_die (); @@ -351,8 +353,8 @@ EGexecute (char const *buf, size_t size, size_t *match_size, match++; patterns[i].regexbuf.not_eol = 0; start = re_search (&(patterns[i].regexbuf), - buf, end - buf - 1, - match - buf, end - match - 1, + beg, end - beg - 1, + match - beg, end - match - 1, &(patterns[i].regs)); if (start < 0) { @@ -361,7 +363,7 @@ EGexecute (char const *buf, size_t size, size_t *match_size, break; } len = patterns[i].regs.end[0] - start; - match = buf + start; + match = beg + start; } } /* while (match <= best_match) */ continue; -- 1.9.0
