xml の読込み – 2

昨日の xml 読込み で，データが大きい場合に比例ではなく極端に遅くなる？
それで，次の様なコードで計測してみた．

bool Test (c_tstring& str)
{
    for (size_t index=0 ; index<str.size() ; index++) {
        ElapseTime et ;
        tstring s_str = ::Parse_xml_quot(str,_T('|'), &index) ;
        std::cerr << ::To_tstring_Ki(long(s_str.size())) << _T("\t") << et.GetElapse() << std::endl ;
        }
    return true ;
    }

bool Test (void)
{
    tstring outPath = ::Get_i_Tools_tmp_date() ;
    tstring now_hms = ::Now_Format(_T("%H%M%S")) ;
    tstring out_txt = ::Path_AddLastSP(outPath) + now_hms + _T(".txt") ;
    tstring outrtxt = ::Path_AddLastSP(outPath) + now_hms + _T("_.txt") ;
    tstring str ;
    {
        v_tstring strAry ;
        {
            tstring str_1000 ;
            {
                tstring str_10 = _T("0123456789") ;
                for (long indexS=0 ; indexS<100 ; indexS++) {
                    str_1000 += str_10 ;
                    }
                }
            for (long index=0 ; index<2000 ; index++) {
                strAry.push_back(str_1000) ;
                }
            }
        str = ::String_Join_Line(strAry) ;
        }
    str[0] = _T('|') ;
    str += _T('|') ;
    size_t len = str.length() ;
    while (len > 100) {
        len = len/2 ;
        str[len-0] = _T('|') ;
        str[len-1] = _T('|') ;
        }
    {
        tstring tmp = str ;
        v_tstring tmpAry = ::String_Split(tmp.c_str(),false) ;
        tmpAry = ::SkipEmpty(tmpAry) ;
        str = ::String_Join_Line(tmpAry) ;
        }
    {
        ::SaveUTF8(out_txt.c_str(),str) ;
        ::Test(str) ;
        }
    {
        tstring rev = str ;
        {
            std::reverse(rev.begin(),rev.end()) ;
            v_tstring revAry = ::String_Split(rev.c_str(),false) ;
            revAry = ::SkipEmpty(revAry) ;
            rev = ::String_Join_Line(revAry) ;
            }
        ::SaveUTF8(outrtxt.c_str(),rev) ;
        ::Test(rev) ;
        }
    return true ;
    }

DS116 では，比例と思われる．

login as: Iwao
Iwao@DS116's password:
Iwao@DS116:~$ source ./set_ds_inc.sh
Iwao@DS116:~$ cd gcc_test/Test/T_P_xm_q/
Iwao@DS116:~/gcc_test/Test/T_P_xm_q$ g++ T_P_xm_q.cpp
Iwao@DS116:~/gcc_test/Test/T_P_xm_q$ ll
total 204
drwxrwxrwx+  2 Iwao users   4096 Jan  8 14:10 .
drwxrwxrwx+ 10 Iwao users   4096 Jan  8 14:09 ..
-rwxrwxrwx   1 Iwao users 183528 Jan  8 14:10 a.out
-rwxrwxrwx+  1 Iwao users   3009 Jan  8 13:59 T_P_xm_q.cpp
Iwao@DS116:~/gcc_test/Test/T_P_xm_q$ ./a.out
     61         0.000384092
     61         6.10352e-05
    122         5.31673e-05
    245         5.19753e-05
    489         6.29425e-05
    979         8.70228e-05
  1.91 K        0.000140905
  3.82 K        0.000212908
  7.64 K        0.000381947
 15.29 K        0.000729799
 30.58 K        0.00145721
 61.16 K        0.0028069
122.32 K        0.00562811
244.63 K        0.011224
489.26 K        0.0223849
978.52 K        0.0462639
      0         5.38826e-05
      0         3.91006e-05
978.52 K        0.045109
489.26 K        0.022413
244.63 K        0.0112309
122.32 K        0.00556397
 61.16 K        0.00286198
 30.58 K        0.00143504
 15.29 K        0.000734806
  7.64 K        0.000383139
  3.82 K        0.000211
  1.91 K        0.000123024
    979         8.51154e-05
    489         6.19888e-05
    245         5.00679e-05
    122         4.3869e-05
     61         4.00543e-05
     61         3.91006e-05
      0         3.31402e-05
      0         3.31402e-05
Iwao@DS116:~/gcc_test/Test/T_P_xm_q$

VC 6 は比例でなく，もっとかかっている．VC 14 では問題ない．VC 8 以降？は大丈夫だと思う．

C:\Users\Iwao>C:\Users\Iwao\AppData\Local\Temp\MICS.tmp\Temp\Test.exe\T_P_xm_q.exe
     61         0
     61         0
    122         0
    245         0
    489         0
    979         0
  1.91 K        0
  3.82 K        0
  7.64 K        0
 15.29 K        0
 30.58 K        0.016
 61.16 K        0.016
122.32 K        0.109
244.63 K        0.531
489.26 K        2.433
978.52 K        14.68
      0         0
      0         0
978.52 K        14.664
489.26 K        1.342
244.63 K        0.374
122.32 K        0.125
 61.16 K        0.016
 30.58 K        0.015
 15.29 K        0
  7.64 K        0
  3.82 K        0
  1.91 K        0
    979         0
    489         0
    245         0
    122         0
     61         0
     61         0
      0         0
      0         0

C:\Users\Iwao>

修正のコードは，

tstring Parse_xml_quot (c_tstring& str,const TCHAR qm,size_t* index)
{
    tstring q_str ;
    if (*index < str.length()) {
        TCHAR c = str[*index] ;
        if (c == qm) {
            /*
            q_str += c ;
            for (size_t indexQ=*index+1 ; indexQ<str.size() ; indexQ++) {
                TCHAR cq = str[indexQ] ;
                q_str += cq ;
                if (cq == qm) {
                    *index = indexQ ;
                    break ;
                    }
                }
            */
            size_t pos = ::String__Find(str,qm,*index+1) ;
            if (pos == tstring::npos) {
                q_str = str.substr(*index) ;
                *index = str.length() ;
                }
            else {
                size_t q_len = pos+1 - *index ;
                q_str = str.substr(*index,q_len) ;
                *index = pos ;
                }
            }
        }
    return q_str ;
    }

parsexml.hxx

2019/06/06
2019/01 のコードは属性の値のネスト（attr='”a” “b”‘）に対応できてないバグあり．

https://jml.mish.work/index.php/cpp/xml.html