// http://www.nongnu.org/chmspec/latest/Internal.html#SYSTEM
bool ChmDoc::ParseSystemData()
{
size_t dataLen;
ScopedMem<unsigned char> data(GetData("/#SYSTEM", &dataLen));
if (!data)
return false;
ByteReader r(data, dataLen);
DWORD len = 0;
// Note: skipping DWORD version at offset 0. It's supposed to be 2 or 3.
for (size_t off = 4; off + 4 < dataLen; off += len + 4) {
// Note: at some point we seem to get off-sync i.e. I'm seeing
// many entries with type == 0 and len == 0. Seems harmless.
len = r.WordLE(off + 2);
if (len == 0)
continue;
WORD type = r.WordLE(off);
switch (type) {
case 0:
if (!tocPath)
tocPath.Set(GetCharZ(data, dataLen, off + 4));
break;
case 1:
if (!indexPath)
indexPath.Set(GetCharZ(data, dataLen, off + 4));
break;
case 2:
if (!homePath)
homePath.Set(GetCharZ(data, dataLen, off + 4));
break;
case 3:
if (!title)
title.Set(GetCharZ(data, dataLen, off + 4));
break;
case 4:
if (!codepage && len >= 4)
codepage = LcidToCodepage(r.DWordLE(off + 4));
break;
case 6:
// compiled file - ignore
break;
case 9:
if (!creator)
creator.Set(GetCharZ(data, dataLen, off + 4));
break;
case 16:
// default font - ignore
break;
}
}
return true;
}
bool ChmDoc::Load(const WCHAR *fileName)
{
chmHandle = chm_open((WCHAR *)fileName);
if (!chmHandle)
return false;
ParseWindowsData();
if (!ParseSystemData())
return false;
UINT fileCodepage = codepage;
char header[24] = { 0 };
if (file::ReadN(fileName, header, sizeof(header))) {
DWORD lcid = ByteReader(header, sizeof(header)).DWordLE(20);
fileCodepage = LcidToCodepage(lcid);
}
if (!codepage)
codepage = fileCodepage;
// if file and #SYSTEM codepage disagree, prefer #SYSTEM's (unless it leads to wrong paths)
FixPathCodepage(homePath, fileCodepage);
FixPathCodepage(tocPath, fileCodepage);
FixPathCodepage(indexPath, fileCodepage);
if (GetACP() == codepage)
codepage = CP_ACP;
if (!HasData(homePath)) {
const char *pathsToTest[] = {
"/index.htm", "/index.html", "/default.htm", "/default.html"
};
for (int i = 0; i < dimof(pathsToTest); i++) {
if (HasData(pathsToTest[i])) {
homePath.SetCopy(pathsToTest[i]);
}
}
if (!HasData(homePath))
return false;
}
return true;
}
bool ChmDoc::Load(const WCHAR *fileName)
{
chmHandle = chm_open((WCHAR *)fileName);
if (!chmHandle)
return false;
ParseWindowsData();
if (!ParseSystemData())
return false;
if (!HasData(homePath)) {
const char *pathsToTest[] = {
"/index.htm", "/index.html", "/default.htm", "/default.html"
};
for (int i = 0; i < dimof(pathsToTest); i++) {
if (HasData(pathsToTest[i]))
homePath.Set(str::Dup(pathsToTest[i]));
}
}
if (!HasData(homePath))
return false;
if (!codepage) {
char header[24];
if (file::ReadAll(fileName, header, sizeof(header))) {
DWORD lcid = ByteReader(header, sizeof(header)).DWordLE(20);
codepage = LcidToCodepage(lcid);
}
else
codepage = CP_CHM_DEFAULT;
}
if (GetACP() == codepage)
codepage = CP_ACP;
return true;
}