lib: fix xml_escape(), again

[JuhaSointusalo]

Commit ff0b8d0 (fix CDATA escaping) introduced two bugs to xml_escape():

• it eats ']' characters that are not part of ']]>'
• it eats the next character after ']]>'

Fixes #2809

[JuhaSointusalo]

Test code:

// cl /MDd /W4 /EHsc test-xml_escape.cpp /I \dev\boinc\src\lib /I /dev/boinc/src/win_build /link /libpath:\dev\boinc\src\win_build\Build\x64\Debug libboinc.lib shell32.lib advapi32.lib
#include <iostream>
#include <string>
#include "parse.h"

using namespace std;

void test(string str) {
    char out[2000] = "";
    cout << "test: '" << str << "'\n";
    xml_escape(str.c_str(), out, sizeof(out));
    cout << "escaped: '" << out << "'\n";
    xml_unescape(out);
    cout << "unescaped: '" << out << "'\n";
    cout << "test==unescaped: " << (str == out) << "\n\n";
}

int main() {
    test("");
    test("abc");
    test("abc<def");
    test("abc&def");
    test("abc]]>def");
    test("<");
    test("&");
    test("]]>");
    test("]");
    test("abc]def");
    test("abc]]]>def");
    test("abc&lt;def");
    test("abc&&def");
    test("abc]]>]]>def");
    string str = "\x9\xa\xd";
    for (int i = 32; i < 256; ++i) {
        str = str + (char)i;
    }
    test(str);
    test("abc]\30]\30>def");
    return 0;
}

Results:

test: ''
escaped: ''
unescaped: ''
test==unescaped: 1

test: 'abc'
escaped: 'abc'
unescaped: 'abc'
test==unescaped: 1

test: 'abc<def'
escaped: 'abc&lt;def'
unescaped: 'abc<def'
test==unescaped: 1

test: 'abc&def'
escaped: 'abc&amp;def'
unescaped: 'abc&def'
test==unescaped: 1

test: 'abc]]>def'
escaped: 'abc]]&gt;def'
unescaped: 'abc]]>def'
test==unescaped: 1

test: '<'
escaped: '&lt;'
unescaped: '<'
test==unescaped: 1

test: '&'
escaped: '&amp;'
unescaped: '&'
test==unescaped: 1

test: ']]>'
escaped: ']]&gt;'
unescaped: ']]>'
test==unescaped: 1

test: ']'
escaped: ']'
unescaped: ']'
test==unescaped: 1

test: 'abc]def'
escaped: 'abc]def'
unescaped: 'abc]def'
test==unescaped: 1

test: 'abc]]]>def'
escaped: 'abc]]]&gt;def'
unescaped: 'abc]]]>def'
test==unescaped: 1

test: 'abc&lt;def'
escaped: 'abc&amp;lt;def'
unescaped: 'abc&lt;def'
test==unescaped: 1

test: 'abc&&def'
escaped: 'abc&amp;&amp;def'
unescaped: 'abc&&def'
test==unescaped: 1

test: 'abc]]>]]>def'
escaped: 'abc]]&gt;]]&gt;def'
unescaped: 'abc]]>]]>def'
test==unescaped: 1

test: '	

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~�ÇüéâäàåçêëèïîìÄÅÉæÆôöòûùÿÖÜø£Ø׃áíóúñѪº¿®¬½¼¡«»░▒▓│┤ÁÂÀ©╣║╗╝¢¥┐└┴┬├─┼ãÃ╚╔╩╦╠═╬¤ðÐÊËÈıÍÎÏ┘┌█▄¦Ì▀ÓßÔÒõÕµþÞÚÛÙýݯ´­±‗¾¶§÷¸°¨·¹³²■ '
escaped: '&#9;&#10;&#13; !"#$%&amp;'()*+,-./0123456789:;&lt;=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~�&#128;&#129;&#130;&#131;&#132;&#133;&#134;&#135;&#136;&#137;&#138;&#139;&#140;&#141;&#142;&#143;&#144;&#145;&#146;&#147;&#148;&#149;&#150;&#151;&#152;&#153;&#154;&#155;&#156;&#157;&#158;&#159;&#160;&#161;&#162;&#163;&#164;&#165;&#166;&#167;&#168;&#169;&#170;&#171;&#172;&#173;&#174;&#175;&#176;&#177;&#178;&#179;&#180;&#181;&#182;&#183;&#184;&#185;&#186;&#187;&#188;&#189;&#190;&#191;&#192;&#193;&#194;&#195;&#196;&#197;&#198;&#199;&#200;&#201;&#202;&#203;&#204;&#205;&#206;&#207;&#208;&#209;&#210;&#211;&#212;&#213;&#214;&#215;&#216;&#217;&#218;&#219;&#220;&#221;&#222;&#223;&#224;&#225;&#226;&#227;&#228;&#229;&#230;&#231;&#232;&#233;&#234;&#235;&#236;&#237;&#238;&#239;&#240;&#241;&#242;&#243;&#244;&#245;&#246;&#247;&#248;&#249;&#250;&#251;&#252;&#253;&#254;&#255;'
unescaped: '	

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~�ÇüéâäàåçêëèïîìÄÅÉæÆôöòûùÿÖÜø£Ø׃áíóúñѪº¿®¬½¼¡«»░▒▓│┤ÁÂÀ©╣║╗╝¢¥┐└┴┬├─┼ãÃ╚╔╩╦╠═╬¤ðÐÊËÈıÍÎÏ┘┌█▄¦Ì▀ÓßÔÒõÕµþÞÚÛÙýݯ´­±‗¾¶§÷¸°¨·¹³²■ '
test==unescaped: 1

test: 'abc]�]�>def'
escaped: 'abc]]>def'
unescaped: 'abc]]>def'
test==unescaped: 0

[JuhaSointusalo]

The code still has (at least :/) one bug. For some reason xml_escape() eats characters less than 32. If it's given a string like ]\xf]\xf> it munches the \xfs and leaves ]]> behind, putting us back to where we started from.

I think there's two ways to fix it. Either escape not ]]> but all > characters or don't eat characters. The first one is allowed by XML standard. The second one would require figuring out why such a low level function gets to decide which characters are good and which are bad. Both would require full testing of entire BOINC and I'm not too keen on doing that right now.

Quite frankly, I'm beginning to think that the next time someone finds a bug in BOINC's XML code it's time to start porting the code to some XML library.

[lfield]

Looks ok to me. I would suggest that if there is any other issue we back out these changes, keep the original issue for this release and then move to an XML library afterwards.