// Main function
//
void MAppMain()
{
MString str; // Program's output
MVector<UInt32> L0, L1, L2; // Database lists
// Encode List 1 (2-3-letter combinations from "exclude.txt")
MStrings exc;
MTextFile(dir+"/exclude.txt").readAll().parseTo(exc,"\r\n");
for (MStrings::CIter it(exc); it(); ++it)
{
const MString& s = (*it).convToLower();
if (s.cnt() == 2)
L1.append(ch2int1(s[0]) + ch2int1(s[1])*28);
else if (s.cnt() == 3)
L1.append(ch2int1(s[0]) + ch2int1(s[1])*28 + ch2int1(s[2])*784);
}
// Encode List 0 and List 2 (converted to lowercase in the process)
MStrings words;
MTextFile(dir+"/words.txt").readAll().parseTo(words,"\r\n"); // Load dictionary
for (MStrings::CIter it(words); it(); ++it)
{
const MString s = (*it).convToLower();
// Encode 1-4-length words to List 2 (as is, no signatures)
if (s.cnt() >= 1 && s.cnt() <= 4)
{
if (s.cnt() == 1)
L2.append(ch2int2(s[0]));
else if (s.cnt() == 2)
L2.append(ch2int2(s[0]) + ch2int2(s[1])*28);
else if (s.cnt() == 3)
L2.append(ch2int2(s[0]) + ch2int2(s[1])*28 + ch2int2(s[2])*784);
else if (s.cnt() == 4)
L2.append(ch2int2(s[0]) + ch2int2(s[1])*28 +
ch2int2(s[2])*784 + ch2int2(s[3])*21952);
}
// Skip words with "wrong" lengths (<5 - handled above, >15 - too few to care)
if (s.cnt() < 5 || s.cnt() > 15) continue;
// Encode 5-15-length words to List 0 as signatures
UInt32 sig = (1-s.cnt()%2)<<ch2int0('*'); // Encode length bit (Note: produces a slightly smaller database than "s.cnt()%2")
for (MNum i = 0; i < s.cnt(); ++i) // Encode 19 letter bits
sig |= (1 << ch2int0(s[i]));
L0.append(sig);
}
// Sort and merge the lists
unique(L0);
unique(L1);
unique(L2); L2.setCnt(c2,0);
MVector<UInt32> L = L0; // Merged list
L.append(L1);
L.append(L2);
// Calculate element differences (at lists' edges add initial element)
MVector<UInt32> diffs;
for (MNum i = 1; i < L.cnt(); ++i)
if (L.get(i) > L.get(i-1))
diffs.append(L.get(i) - L.get(i-1));
// Encode 'diffs' as optimized bitmaps (in a string form for convenience)
// Each diff value is encoded as "0..01n..n": 'z' zeroes, followed by 1,
// followed by bitmap of the 'x-a' number, where 'x' is the diff and 'a' is
// the nearest (smaller or equal) to 'x' power of 2
MString bits;
for (MNum i = 0; i < diffs.cnt(); ++i)
{
UInt32 d = diffs.get(i);
for (UInt32 z = 0; ; ++z)
{
UInt32 a = (1<<z), b = a*2-1; // [a,b] - suitable range of 'd' to process
if (d<a || d>b) continue; // Outside of proper range - keep looking
for (UInt32 k = 0; k < z; ++k)
bits << '0';
bits << '1';
for (UInt32 k = 0; k < z; ++k)
bits << (GetBits(d-a,UInt8(k),UInt8(k))?'1':'0');
break;
}
}
while (bits.cnt()%8) bits << '0'; // Pad with zeroes to complete the last byte
// Encode the intermediary 'bits' string to actual bit vector
MVector<UInt8> bytes;
for (MNum i = 0; i < bits.cnt(); i+=8)
{
UInt8 b = 0;
for (MNum j = 0; j < 8; ++j) if (bits[i+j]=='1') b |= (1<<j);
bytes.append(b);
}
// Save the data to the file and compress it, if 7z is available at the sprcified location
MFile(dir+"/data",fmOverwrite) << bytes;
if (MSys::getFileInfo(GZipExe).name.cnt())
{
MSys::deleteFile(dir+"/data.gz");
MSys::open(GZipExe, (MString)"a -tgzip -mx=9 -sdel \""+dir+"/data.gz\" \""+dir+"/data\"");
}
// Compile some useful information (needed for hardcoding in the Node.js file)
// Prepare sizes
int size1 = 0;
while (!size1)
{
MApp::sleep(500);
size1 = (UInt32)MSys::getFileInfo(dir+"/data.gz").size;
}
int size2 = (int)MSys::getFileInfo(dir+"/solution.js").size;
// Initial elements and sizes
str <<"w=["<<L0.get(0)<<","<<L1.get(0)<<","<<L2.get(0)<<","
<<L0.cnt()<<","<<L1.cnt()<<","<<L2.cnt()<<"]\n";
// Letter->index conversion string for L0
str<<"C0: \" ";
for (int j='a'; j<='z'; ++j) str<<MChar(int('a'+ch2int0(j)-1));
str<<MChar(int('a'+ch2int0('\'')-1))<<"\"\n";
// Letter->index conversion string for L1
MString c1 = " ...........................";
for (int j='a'; j<='z'; ++j) c1.setChar(ch2int1(j),j);
c1.setChar(ch2int1('\''),'\'');
str<<"C1: \""<<c1<<"\"\n";
// Letter->index conversion string for L2
str<<"C2: \" ";
for (int j='a'; j<='z'; ++j) str<<MChar(int('a'+ch2int2(j)-1));
str<<MChar(int('a'+ch2int2('\'')-1))<<"\"\n";
// Size report
str <<"\n"<<words.cnt()<<" words => "<<L.cnt()<<" signatures => "
<<bytes.cnt()<<" bytes =>\n"<<size1<<"+"<<size2<<" final bytes ("
<<(65536-size1-size2)<<" left)\n\n";
// Display the information
MMainFrame wnd; // Create app's mainframe window
wnd.plugBar(MEditBoxPtr(wnd,AutoRect,None,true,str)); // Create a text box
wnd.setCaption("Data file written");
wnd.runMessageLoop(); // Run the app
}
