From 5df9ff919db13d7edeefd1d48ce942ba0d8cc0dc Mon Sep 17 00:00:00 2001 From: Bruno Haible Date: Fri, 21 Jan 2011 14:06:47 +0100 Subject: [PATCH] Make uN_strstr functions O(n) worst-case. * lib/unistr/u-strstr.h (FUNC): In the 8-bit case, use strstr. In the 16-bit and 32-bit unit cases, use the unibyte algorithm from lib/mbsstr.c. * lib/unistr/u8-strstr.c: Include . (UNIT_IS_UINT8_T): New macro. * lib/unistr/u16-strstr.c: Include malloca.h and str-kmp.h. (U_STRLEN, U_STRNLEN): New macros. * lib/unistr/u32-strstr.c: Include malloca.h and str-kmp.h. (U_STRLEN, U_STRNLEN): New macros. * modules/unistr/u8-strstr (Depends-on): Add strstr. (configure.ac): Update required libunistring version. * modules/unistr/u16-strstr (Files): Add lib/str-kmp.h. (Depends-on): Add unistr/u16-strlen, unistr/u16-strnlen, stdbool, malloca. (configure.ac): Update required libunistring version. * modules/unistr/u32-strstr (Files): Add lib/str-kmp.h. (Depends-on): Add unistr/u32-strlen, unistr/u32-strnlen, stdbool, malloca. (configure.ac): Update required libunistring version. --- ChangeLog | 24 +++++++++++ lib/unistr/u-strstr.h | 102 +++++++++++++++++++++++++++++++++++++++------- lib/unistr/u16-strstr.c | 10 ++++- lib/unistr/u32-strstr.c | 10 ++++- lib/unistr/u8-strstr.c | 3 ++ modules/unistr/u16-strstr | 7 +++- modules/unistr/u32-strstr | 7 +++- modules/unistr/u8-strstr | 3 +- 8 files changed, 146 insertions(+), 20 deletions(-) diff --git a/ChangeLog b/ChangeLog index 468e317f5..751447491 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,6 +1,30 @@ 2011-01-21 Pádraig Brady Bruno Haible + Make uN_strstr functions O(n) worst-case. + * lib/unistr/u-strstr.h (FUNC): In the 8-bit case, use strstr. In the + 16-bit and 32-bit unit cases, use the unibyte algorithm from + lib/mbsstr.c. + * lib/unistr/u8-strstr.c: Include . + (UNIT_IS_UINT8_T): New macro. + * lib/unistr/u16-strstr.c: Include malloca.h and str-kmp.h. + (U_STRLEN, U_STRNLEN): New macros. + * lib/unistr/u32-strstr.c: Include malloca.h and str-kmp.h. + (U_STRLEN, U_STRNLEN): New macros. + * modules/unistr/u8-strstr (Depends-on): Add strstr. + (configure.ac): Update required libunistring version. + * modules/unistr/u16-strstr (Files): Add lib/str-kmp.h. + (Depends-on): Add unistr/u16-strlen, unistr/u16-strnlen, stdbool, + malloca. + (configure.ac): Update required libunistring version. + * modules/unistr/u32-strstr (Files): Add lib/str-kmp.h. + (Depends-on): Add unistr/u32-strlen, unistr/u32-strnlen, stdbool, + malloca. + (configure.ac): Update required libunistring version. + +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 diff --git a/lib/unistr/u-strstr.h b/lib/unistr/u-strstr.h index dee538deb..336ce8bfa 100644 --- a/lib/unistr/u-strstr.h +++ b/lib/unistr/u-strstr.h @@ -1,6 +1,6 @@ /* Substring test for UTF-8/UTF-16/UTF-32 strings. Copyright (C) 1999, 2002, 2006, 2010-2011 Free Software Foundation, Inc. - Written by Bruno Haible , 2002. + Written by Bruno Haible , 2002, 2005. This program is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published @@ -38,22 +38,94 @@ FUNC (const UNIT *haystack, const UNIT *needle) } #endif - /* Search for needle's first unit. */ - for (; *haystack != 0; haystack++) - if (*haystack == first) +#if UNIT_IS_UINT8_T + return (uint8_t *) strstr ((const char *) haystack, (const char *) needle); +#else + { + /* Minimizing the worst-case complexity: + Let n = U_STRLEN(haystack), m = U_STRLEN(needle). + The naïve algorithm is O(n*m) worst-case. + The Knuth-Morris-Pratt algorithm is O(n) worst-case but it needs a + memory allocation. + To achieve linear complexity and yet amortize the cost of the + memory allocation, we activate the Knuth-Morris-Pratt algorithm + only once the naïve algorithm has already run for some time; more + precisely, when + - the outer loop count is >= 10, + - the average number of comparisons per outer loop is >= 5, + - the total number of comparisons is >= m. + But we try it only once. If the memory allocation attempt failed, + we don't retry it. */ + bool try_kmp = true; + size_t outer_loop_count = 0; + size_t comparison_count = 0; + size_t last_ccount = 0; /* last comparison count */ + const UNIT *needle_last_ccount = needle; /* = needle + last_ccount */ + + /* Speed up the following searches of needle by caching its first + character. */ + UNIT b = *needle++; + + for (;; haystack++) { - /* Compare with needle's remaining units. */ - const UNIT *hptr = haystack + 1; - const UNIT *nptr = needle + 1; - for (;;) + if (*haystack == 0) + /* No match. */ + return NULL; + + /* See whether it's advisable to use an asymptotically faster + algorithm. */ + if (try_kmp + && outer_loop_count >= 10 + && comparison_count >= 5 * outer_loop_count) { - if (*hptr != *nptr) - break; - hptr++; nptr++; - if (*nptr == 0) - return (UNIT *) haystack; + /* See if needle + comparison_count now reaches the end of + needle. */ + if (needle_last_ccount != NULL) + { + needle_last_ccount += + U_STRNLEN (needle_last_ccount, + comparison_count - last_ccount); + if (*needle_last_ccount == 0) + needle_last_ccount = NULL; + last_ccount = comparison_count; + } + if (needle_last_ccount == NULL) + { + /* Try the Knuth-Morris-Pratt algorithm. */ + const UNIT *result; + bool success = + knuth_morris_pratt (haystack, + needle - 1, U_STRLEN (needle - 1), + &result); + if (success) + return (UNIT *) result; + try_kmp = false; + } } - } - return NULL; + outer_loop_count++; + comparison_count++; + if (*haystack == b) + /* The first character matches. */ + { + const UNIT *rhaystack = haystack + 1; + const UNIT *rneedle = needle; + + for (;; rhaystack++, rneedle++) + { + if (*rneedle == 0) + /* Found a match. */ + return (UNIT *) haystack; + if (*rhaystack == 0) + /* No match. */ + return NULL; + comparison_count++; + if (*rhaystack != *rneedle) + /* Nothing in this round. */ + break; + } + } + } + } +#endif } diff --git a/lib/unistr/u16-strstr.c b/lib/unistr/u16-strstr.c index 06f17b5af..934a4fccf 100644 --- a/lib/unistr/u16-strstr.c +++ b/lib/unistr/u16-strstr.c @@ -20,10 +20,18 @@ /* Specification. */ #include "unistr.h" +#include "malloca.h" + /* FIXME: Maybe walking the string via u16_mblen is a win? */ -#define FUNC u16_strstr #define UNIT uint16_t + +#define CANON_ELEMENT(c) c +#include "str-kmp.h" + +#define FUNC u16_strstr #define U_STRCHR u16_strchr #define U_STRMBTOUC u16_strmbtouc +#define U_STRLEN u16_strlen +#define U_STRNLEN u16_strnlen #include "u-strstr.h" diff --git a/lib/unistr/u32-strstr.c b/lib/unistr/u32-strstr.c index 99ccef33a..00a7f3e22 100644 --- a/lib/unistr/u32-strstr.c +++ b/lib/unistr/u32-strstr.c @@ -20,7 +20,15 @@ /* Specification. */ #include "unistr.h" -#define FUNC u32_strstr +#include "malloca.h" + #define UNIT uint32_t + +#define CANON_ELEMENT(c) c +#include "str-kmp.h" + +#define FUNC u32_strstr #define U_STRCHR u32_strchr +#define U_STRLEN u32_strlen +#define U_STRNLEN u32_strnlen #include "u-strstr.h" diff --git a/lib/unistr/u8-strstr.c b/lib/unistr/u8-strstr.c index 0638baa2d..3f725cb54 100644 --- a/lib/unistr/u8-strstr.c +++ b/lib/unistr/u8-strstr.c @@ -20,10 +20,13 @@ /* Specification. */ #include "unistr.h" +#include + /* FIXME: Maybe walking the string via u8_mblen is a win? */ #define FUNC u8_strstr #define UNIT uint8_t #define U_STRCHR u8_strchr #define U_STRMBTOUC u8_strmbtouc +#define UNIT_IS_UINT8_T 1 #include "u-strstr.h" diff --git a/modules/unistr/u16-strstr b/modules/unistr/u16-strstr index 014ae113b..af79f0b36 100644 --- a/modules/unistr/u16-strstr +++ b/modules/unistr/u16-strstr @@ -4,14 +4,19 @@ Substring test for UTF-16 strings. Files: lib/unistr/u16-strstr.c lib/unistr/u-strstr.h +lib/str-kmp.h Depends-on: unistr/base unistr/u16-strchr unistr/u16-strmbtouc +unistr/u16-strlen +unistr/u16-strnlen +stdbool +malloca configure.ac: -gl_LIBUNISTRING_MODULE([0.9], [unistr/u16-strstr]) +gl_LIBUNISTRING_MODULE([0.9.4], [unistr/u16-strstr]) Makefile.am: if LIBUNISTRING_COMPILE_UNISTR_U16_STRSTR diff --git a/modules/unistr/u32-strstr b/modules/unistr/u32-strstr index 45533c243..c8da9f71f 100644 --- a/modules/unistr/u32-strstr +++ b/modules/unistr/u32-strstr @@ -4,13 +4,18 @@ Substring test for UTF-32 strings. Files: lib/unistr/u32-strstr.c lib/unistr/u-strstr.h +lib/str-kmp.h Depends-on: unistr/base unistr/u32-strchr +unistr/u32-strlen +unistr/u32-strnlen +stdbool +malloca configure.ac: -gl_LIBUNISTRING_MODULE([0.9], [unistr/u32-strstr]) +gl_LIBUNISTRING_MODULE([0.9.4], [unistr/u32-strstr]) Makefile.am: if LIBUNISTRING_COMPILE_UNISTR_U32_STRSTR diff --git a/modules/unistr/u8-strstr b/modules/unistr/u8-strstr index 30d99a459..b4ec6a689 100644 --- a/modules/unistr/u8-strstr +++ b/modules/unistr/u8-strstr @@ -9,9 +9,10 @@ Depends-on: unistr/base unistr/u8-strchr unistr/u8-strmbtouc +strstr configure.ac: -gl_LIBUNISTRING_MODULE([0.9], [unistr/u8-strstr]) +gl_LIBUNISTRING_MODULE([0.9.4], [unistr/u8-strstr]) Makefile.am: if LIBUNISTRING_COMPILE_UNISTR_U8_STRSTR -- 2.11.0