diff options
author | graehl <graehl@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-08-12 01:19:27 +0000 |
---|---|---|
committer | graehl <graehl@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-08-12 01:19:27 +0000 |
commit | 734d2f82f79c3bfa4ee8d57d0ab6c1eafa6b91cc (patch) | |
tree | b5b7a301e52927ed6bfd3c3e0b056bb4a9af920a | |
parent | aa306c4142789fc2462dc65d94321e026b722ed8 (diff) |
fidget
git-svn-id: https://ws10smt.googlecode.com/svn/trunk@532 ec762483-ff6d-05da-a07a-a48fb63a330f
-rwxr-xr-x | decoder/fast_lexical_cast.hpp | 1 | ||||
-rw-r--r-- | utils/stringlib.h | 14 | ||||
-rwxr-xr-x | utils/utoa.h | 24 |
3 files changed, 21 insertions, 18 deletions
diff --git a/decoder/fast_lexical_cast.hpp b/decoder/fast_lexical_cast.hpp index c1b15042..ae49c934 100755 --- a/decoder/fast_lexical_cast.hpp +++ b/decoder/fast_lexical_cast.hpp @@ -2,6 +2,7 @@ #define FAST_LEXICAL_CAST_HPP #define BOOST_LEXICAL_CAST_ASSUME_C_LOCALE +// This should make casting to/from string reasonably fast (see http://accu.org/index.php/journals/1375) #include <boost/lexical_cast.hpp> diff --git a/utils/stringlib.h b/utils/stringlib.h index 84e95d44..4f79eb31 100644 --- a/utils/stringlib.h +++ b/utils/stringlib.h @@ -20,22 +20,22 @@ #include <sstream> #include <algorithm> -inline std::size_t skip_ws(std::string const& s,std::size_t starting=0,char const* ws=" \t\n\r") { +namespace { +const char c_isspace[]=" \t\n\r\f\v"; // somewhat ridiculous, including characters nobody uses. +const char common_isspace[]=" \t\n\r"; // even \n\r is borderline, but maybe you pass multiline DOS format text. +} + +inline std::size_t skip_ws(std::string const& s,std::size_t starting=0,char const* ws=common_isspace) { return s.find_first_not_of(ws,starting); } // returns position of end of all non-ws chars before ending, i.e. string(s.begin()+skip_ws(s),s.begin()+trailing_ws(s)) strips both ends -inline std::size_t trailing_ws(std::string const& s,std::size_t ending=std::string::npos,char const* ws=" \t\n\r") { +inline std::size_t trailing_ws(std::string const& s,std::size_t ending=std::string::npos,char const* ws=common_isspace) { std::size_t n=s.find_last_not_of(ws,ending); if (n==std::string::npos) return n; else return n+1; } -//TEST: if string is all whitespace, make sure that string(a+npos,a+npos) can't segfault (i.e. won't access any memory because begin==end) -inline std::string strip_ws(std::string const& s) { - return std::string(s.begin()+skip_ws(s),s.begin()+trailing_ws(s)); -} - inline bool is_single_line(std::string const& line) { return std::count(line.begin(),line.end(),'\n')==0; // but we want to allow terminal newlines/blanks diff --git a/utils/utoa.h b/utils/utoa.h index 6c1cf2de..8d304f97 100755 --- a/utils/utoa.h +++ b/utils/utoa.h @@ -1,6 +1,7 @@ #ifndef UTOA_H #define UTOA_H +#include <stdint.h> #include <string> #include <cstring> @@ -8,7 +9,10 @@ # define DIGIT_LOOKUP_TABLE 0 #endif -const unsigned utoa_bufsize=21; +// The largest 32-bit integer is 4294967295, that is 10 chars +// 1 more for sign, and 1 for 0-termination of string +// generally: 2 + std::numeric_limits<T>::is_signed + std::numeric_limits<T>::digits10 +const unsigned utoa_bufsize=12; const unsigned utoa_bufsizem1=utoa_bufsize-1; #ifdef DIGIT_LOOKUP_TABLE @@ -27,11 +31,11 @@ inline char digit_to_char(int d) { } // returns n in string [return,num); *num=0 yourself before calling if you want a c_str -inline char *utoa(char *num,unsigned n) { +inline char *utoa(char *num,uint32_t n) { if ( !n ) { *--num='0'; } else { - unsigned rem; + uint32_t rem; // 3digit lookup table, divide by 1000 faster? while ( n ) { #if 1 @@ -48,24 +52,23 @@ inline char *utoa(char *num,unsigned n) { return num; } - -inline char *itoa(char *p,int n) { +inline char *itoa(char *p,int32_t n) { if (n<0) { - p=utoa(p,-n); // TODO: check that (unsigned)(-INT_MIN) == 0x1000000 in 2s complement and not == 0 + p=utoa(p,-n); // (unsigned)(-INT_MIN) == 0x1000000 in 2s complement and not == 0. *--p='-'; return p; } else return utoa(p,n); } -inline std::string utos(unsigned n) { +inline std::string utos(uint32_t n) { char buf[utoa_bufsize]; char *end=buf+utoa_bufsize; char *p=utoa(end,n); return std::string(p,end); } -inline std::string itos(int n) { +inline std::string itos(int32_t n) { char buf[utoa_bufsize]; char *end=buf+utoa_bufsize; char *p=itoa(end,n); @@ -73,7 +76,7 @@ inline std::string itos(int n) { } //returns position of '\0' terminating number written starting at to -inline char* append_utoa(char *to,unsigned n) { +inline char* append_utoa(char *to,uint32_t n) { char buf[utoa_bufsize]; char *end=buf+utoa_bufsize; char *s=utoa(end,n); @@ -85,7 +88,7 @@ inline char* append_utoa(char *to,unsigned n) { } //returns position of '\0' terminating number written starting at to -inline char* append_itoa(char *to,unsigned n) { +inline char* append_itoa(char *to,int32_t n) { char buf[utoa_bufsize]; char *end=buf+utoa_bufsize; char *s=itoa(end,n); @@ -96,5 +99,4 @@ inline char* append_itoa(char *to,unsigned n) { return to; } - #endif |