From 1b1e945287b0e30a238102974f7a51f24def2a79 Mon Sep 17 00:00:00 2001 From: =?utf8?q?P=C3=A1draig=20Brady?= Date: Fri, 21 Jan 2011 13:40:28 +0100 Subject: [PATCH] Prepare for faster uN_strstr functions. * lib/str-kmp.h: Support definable UNITs. (knuth_morris_pratt): Renamed from knuth_morris_pratt_unibyte. Add needle_len argument. * lib/mbsstr.c (mbsstr): Adjust for the changed str-kmp.h. * lib/mbscasestr.c (mbscasestr): Likewise. --- ChangeLog | 10 ++++++++++ lib/mbscasestr.c | 9 ++++++--- lib/mbsstr.c | 9 ++++++--- lib/str-kmp.h | 32 ++++++++++++++++++-------------- 4 files changed, 40 insertions(+), 20 deletions(-) diff --git a/ChangeLog b/ChangeLog index 9bb96b85b..468e317f5 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,13 @@ +2011-01-21 Pádraig Brady + Bruno Haible + + Prepare for faster uN_strstr functions. + * lib/str-kmp.h: Support definable UNITs. + (knuth_morris_pratt): Renamed from knuth_morris_pratt_unibyte. Add + needle_len argument. + * lib/mbsstr.c (mbsstr): Adjust for the changed str-kmp.h. + * lib/mbscasestr.c (mbscasestr): Likewise. + 2011-01-21 Pádraig Brady malloca-tests: make faster by unsetting MALLOC_PERTURB_ diff --git a/lib/mbscasestr.c b/lib/mbscasestr.c index 379b2fb03..f013808d5 100644 --- a/lib/mbscasestr.c +++ b/lib/mbscasestr.c @@ -30,6 +30,7 @@ #define TOLOWER(Ch) (isupper (Ch) ? tolower (Ch) : (Ch)) /* Knuth-Morris-Pratt algorithm. */ +#define UNIT unsigned char #define CANON_ELEMENT(c) TOLOWER (c) #include "str-kmp.h" @@ -368,10 +369,12 @@ mbscasestr (const char *haystack, const char *needle) if (needle_last_ccount == NULL) { /* Try the Knuth-Morris-Pratt algorithm. */ - const char *result; + const unsigned char *result; bool success = - knuth_morris_pratt_unibyte (haystack, needle - 1, - &result); + knuth_morris_pratt ((const unsigned char *) haystack, + (const unsigned char *) (needle - 1), + strlen (needle - 1), + &result); if (success) return (char *) result; try_kmp = false; diff --git a/lib/mbsstr.c b/lib/mbsstr.c index 513dde1ab..611000e6d 100644 --- a/lib/mbsstr.c +++ b/lib/mbsstr.c @@ -27,6 +27,7 @@ #include "mbuiter.h" /* Knuth-Morris-Pratt algorithm. */ +#define UNIT unsigned char #define CANON_ELEMENT(c) c #include "str-kmp.h" @@ -339,10 +340,12 @@ mbsstr (const char *haystack, const char *needle) if (needle_last_ccount == NULL) { /* Try the Knuth-Morris-Pratt algorithm. */ - const char *result; + const unsigned char *result; bool success = - knuth_morris_pratt_unibyte (haystack, needle - 1, - &result); + knuth_morris_pratt ((const unsigned char *) haystack, + (const unsigned char *) (needle - 1), + strlen (needle - 1), + &result); if (success) return (char *) result; try_kmp = false; diff --git a/lib/str-kmp.h b/lib/str-kmp.h index 95a73f571..d7e0a7454 100644 --- a/lib/str-kmp.h +++ b/lib/str-kmp.h @@ -1,4 +1,4 @@ -/* Substring search in a NUL terminated string of 'char' elements, +/* Substring search in a NUL terminated string of UNIT elements, using the Knuth-Morris-Pratt algorithm. Copyright (C) 2005-2011 Free Software Foundation, Inc. Written by Bruno Haible , 2005. @@ -18,21 +18,26 @@ Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ /* Before including this file, you need to define: + UNIT The element type of the needle and haystack. CANON_ELEMENT(c) A macro that canonicalizes an element right after - it has been fetched from one of the two strings. - The argument is an 'unsigned char'; the result - must be an 'unsigned char' as well. */ + it has been fetched from needle or haystack. + The argument is of type UNIT; the result must be + of type UNIT as well. */ /* Knuth-Morris-Pratt algorithm. See http://en.wikipedia.org/wiki/Knuth-Morris-Pratt_algorithm + HAYSTACK is the NUL terminated string in which to search for. + NEEDLE is the string to search for in HAYSTACK, consisting of NEEDLE_LEN + units. Return a boolean indicating success: Return true and set *RESULTP if the search was completed. Return false if it was aborted because not enough memory was available. */ static bool -knuth_morris_pratt_unibyte (const char *haystack, const char *needle, - const char **resultp) +knuth_morris_pratt (const UNIT *haystack, + const UNIT *needle, size_t needle_len, + const UNIT **resultp) { - size_t m = strlen (needle); + size_t m = needle_len; /* Allocate the table. */ size_t *table = (size_t *) nmalloca (m, sizeof (size_t)); @@ -66,14 +71,14 @@ knuth_morris_pratt_unibyte (const char *haystack, const char *needle, The inequality needle[x..i-1] != needle[0..i-1-x] is known to hold for x < table[i-1], by induction. Furthermore, if j>0: needle[i-1-j..i-2] = needle[0..j-1]. */ - unsigned char b = CANON_ELEMENT ((unsigned char) needle[i - 1]); + UNIT b = CANON_ELEMENT (needle[i - 1]); for (;;) { /* Invariants: The inequality needle[x..i-1] != needle[0..i-1-x] is known to hold for x < i-1-j. Furthermore, if j>0: needle[i-1-j..i-2] = needle[0..j-1]. */ - if (b == CANON_ELEMENT ((unsigned char) needle[j])) + if (b == CANON_ELEMENT (needle[j])) { /* Set table[i] := i-1-j. */ table[i] = i - ++j; @@ -108,17 +113,16 @@ knuth_morris_pratt_unibyte (const char *haystack, const char *needle, /* Search, using the table to accelerate the processing. */ { size_t j; - const char *rhaystack; - const char *phaystack; + const UNIT *rhaystack; + const UNIT *phaystack; *resultp = NULL; j = 0; rhaystack = haystack; phaystack = haystack; /* Invariant: phaystack = rhaystack + j. */ - while (*phaystack != '\0') - if (CANON_ELEMENT ((unsigned char) needle[j]) - == CANON_ELEMENT ((unsigned char) *phaystack)) + while (*phaystack != 0) + if (CANON_ELEMENT (needle[j]) == CANON_ELEMENT (*phaystack)) { j++; phaystack++; -- 2.11.0