[petermrg]

I'm working on an opensource emule server. The alpha version is almost done but I have some problems understanding the emule's OP_SEARCHREQUEST.

Do any emule developers have documentation about this operation?

I know it's encoded as a binary expression tree but I don't understand how are the contraints encoded. Also expressions for identical searches are are different in emule, amule and other clients.
All the documentation I found about the ed2k protocol is outdated or wrong.



How is the "minsize" constraint encoded?

PS: please, don't answer "look at the code". I'm already doing that (and getting crazy).

This post has been edited by petermrg: 30 January 2013 - 12:11 PM

[tHeWiZaRdOfDoS]

Even if you don't like it, the source code is the way to go. See CSearchResultsWnd::DoNewEd2kSearch. Within there's a call to GetSearchPacket (same source file) which pretty much contains everything you need.
Note the calls to AddAndAttr and their corresponding parameters, e.g.

Quote

if (pParams->ullMinSize > 0)
AddAndAttr(FT_FILESIZE, ED2K_SEARCH_OP_GREATER_EQUAL, pParams->ullMinSize);

After that, check the loop below

Quote

for (int j = 0; j < s_SearchExpr.m_aExpr.GetCount(); j++)

which is used to write the contraints (note that FT_FILESIZE may be added several times).

[DavidXanatos]

Look inside class CSearchExprTarget it assembles the request into its binary form.

[petermrg]

Thanks Wizard and David!

Emule seems to encode constraints this way:

• MinSize: (input * 1024 * 1024) - 1;
  
• MaxSize: (input * 1024 * 1024) + 1;
  
• Availability: input - 1;
  
• Completed Sources: input - 1;
  
• Bitrate: input - 1;
  
• Duration: toSeconds(input) - 1;

This is my function to decode search expressions:

var searchExpr = function(token, type, value) {
    switch (type) {
        case 0xff: return 'str='+value;
        case 0x00: return ' AND ';
        case 0x01: return ' OR ';
        case 0x02: return ' NOT ';
        case 0x030001: return 'type='+value;
        case 0x040001: return 'ext='+value;
        case 0xd50001: return 'codec='+value;
        case 0x02000101: return 'size>'+value;
        case 0x02000102: return 'size<'+value;
        case 0x15000101: return 'sources>'+value;
        case 0xd4000101: return 'bitrate>'+value;
        case 0xd3000101: return 'duration>'+value;
        case 0x30000101: return 'sourcescompleted='+value;
        default: log.warn('searchExpr: unknown type: 0x'+type.toString(16)+'; token: 0x'+token.toString(16)+'; value: '+value);
    }
}

Buffer.prototype.getSearchString = function() {
    var token = this.getUInt8();
    switch (token) {
        case 0x01: // text match
            return searchExpr(token, 0xff, this.getString());
        case 0x02: // string
            var value = this.getString();
            var type = this.getUInt8();
            type+= this.getUInt16LE() * 0x100;
            return searchExpr(token, type, value);
        case 0x03: // 32bit value
            var value = this.getUInt32LE();
            var type = this.getUInt32LE();
            return searchExpr(token, type, value);
        case 0x08: // 64bit value
            var value = this.getUInt32LE();
            value+= this.getUInt32LE() * 0x10000 * 0x10000;
            var type = this.getUInt32LE();
            return searchExpr(token, type, value);
        case 0x00: // tree node
            var type = this.getUInt8();
            var s = '(' + this.getSearchString();
            s+= searchExpr(token, type, null);
            return s + this.getSearchString() + ')';
        default:
            log.warn('Buffer.getSearchString: unknown token: 0x'+token.toString(16));
    }
    return '';
};

Sample input search of text "aaa AND bbb OR ccc NOT ddd" and all possible constraints:

00000000: 0000 0000 0103 0061 6161 0001 0103 0062  .......aaa.....b
00000010: 6262 0002 0103 0063 6363 0103 0064 6464  bb.....ccc...ddd
00000020: 0000 0205 004d 5045 4734 0100 d500 0003  .....MPEG4..�...
00000030: 0f0e 0000 0101 00d3 0000 0399 0200 0001  .......�...�....
00000040: 0100 d400 0003 e603 0000 0101 0030 0000  ..�...�......0..
00000050: 03ff ff0f 1401 0100 0200 0008 0100 1043  .��............C
00000060: 0d00 0000 0201 0002 0000 0377 0300 0001  ...........w....
00000070: 0100 1502 0300 7171 7101 0004            ......qqq...

Sample output:

((str=aaa AND (str=bbb OR (str=ccc NOT str=ddd))) AND (codec=MPEG4 AND (duration>3599 AND (bitrate>665 AND (sourcescompleted=998 AND (size>336592895 AND (size<56959696897 AND (sources>887 AND ext=qqq))))))))

This post has been edited by petermrg: 31 January 2013 - 01:05 PM