void
CKeyMap::addKeyAliasEntry(KeyID targetID, SInt32 group,
KeyModifierMask targetRequired,
KeyModifierMask targetSensitive,
KeyID sourceID,
KeyModifierMask sourceRequired,
KeyModifierMask sourceSensitive)
{
// if we can already generate the target as desired then we're done.
if (findCompatibleKey(targetID, group, targetRequired,
targetSensitive) != NULL) {
return;
}
// find a compatible source, preferably in the same group
for (SInt32 gd = 0, n = getNumGroups(); gd < n; ++gd) {
SInt32 eg = getEffectiveGroup(group, gd);
const KeyItemList* sourceEntry =
findCompatibleKey(sourceID, eg,
sourceRequired, sourceSensitive);
if (sourceEntry != NULL && sourceEntry->size() == 1) {
CKeyMap::KeyItem targetItem = sourceEntry->back();
targetItem.m_id = targetID;
targetItem.m_group = eg;
addKeyEntry(targetItem);
break;
}
}
}
void
CMSWindowsKeyState::getKeyMap(CKeyMap& keyMap)
{
// update keyboard groups
if (getGroups(m_groups)) {
m_groupMap.clear();
SInt32 numGroups = (SInt32)m_groups.size();
for (SInt32 g = 0; g < numGroups; ++g) {
m_groupMap[m_groups[g]] = g;
}
}
HKL activeLayout = GetKeyboardLayout(0);
// clear table
memset(m_virtualKeyToButton, 0, sizeof(m_virtualKeyToButton));
m_keyToVKMap.clear();
CKeyMap::KeyItem item;
SInt32 numGroups = (SInt32)m_groups.size();
for (SInt32 g = 0; g < numGroups; ++g) {
item.m_group = g;
ActivateKeyboardLayout(m_groups[g], 0);
// clear tables
memset(m_buttonToVK, 0, sizeof(m_buttonToVK));
memset(m_buttonToNumpadVK, 0, sizeof(m_buttonToNumpadVK));
// map buttons (scancodes) to virtual keys
for (KeyButton i = 1; i < 256; ++i) {
UINT vk = MapVirtualKey(i, 1);
if (vk == 0) {
// unmapped
continue;
}
// deal with certain virtual keys specially
switch (vk) {
case VK_SHIFT:
vk = VK_LSHIFT;
break;
case VK_CONTROL:
vk = VK_LCONTROL;
break;
case VK_MENU:
vk = VK_LMENU;
break;
case VK_NUMLOCK:
vk = VK_PAUSE;
break;
case VK_NUMPAD0:
case VK_NUMPAD1:
case VK_NUMPAD2:
case VK_NUMPAD3:
case VK_NUMPAD4:
case VK_NUMPAD5:
case VK_NUMPAD6:
case VK_NUMPAD7:
case VK_NUMPAD8:
case VK_NUMPAD9:
case VK_DECIMAL:
// numpad keys are saved in their own table
m_buttonToNumpadVK[i] = vk;
continue;
case VK_LWIN:
case VK_RWIN:
// add extended key only for these on 95 family
if (m_is95Family) {
m_buttonToVK[i | 0x100u] = vk;
continue;
}
break;
case VK_RETURN:
case VK_PRIOR:
case VK_NEXT:
case VK_END:
case VK_HOME:
case VK_LEFT:
case VK_UP:
case VK_RIGHT:
case VK_DOWN:
case VK_INSERT:
case VK_DELETE:
// also add extended key for these
m_buttonToVK[i | 0x100u] = vk;
break;
}
if (m_buttonToVK[i] == 0) {
m_buttonToVK[i] = vk;
}
}
// now map virtual keys to buttons. multiple virtual keys may map
// to a single button. if the virtual key matches the one in
// m_buttonToVK then we use the button as is. if not then it's
// either a numpad key and we use the button as is or it's an
// extended button.
for (UINT i = 1; i < 255; ++i) {
// skip virtual keys we don't want
switch (i) {
case VK_LBUTTON:
case VK_RBUTTON:
case VK_MBUTTON:
case VK_XBUTTON1:
case VK_XBUTTON2:
case VK_SHIFT:
case VK_CONTROL:
case VK_MENU:
continue;
}
// get the button
KeyButton button = static_cast<KeyButton>(MapVirtualKey(i, 0));
if (button == 0) {
continue;
}
// deal with certain virtual keys specially
switch (i) {
case VK_NUMPAD0:
case VK_NUMPAD1:
case VK_NUMPAD2:
case VK_NUMPAD3:
case VK_NUMPAD4:
case VK_NUMPAD5:
case VK_NUMPAD6:
case VK_NUMPAD7:
case VK_NUMPAD8:
case VK_NUMPAD9:
case VK_DECIMAL:
m_buttonToNumpadVK[button] = i;
break;
default:
// add extended key if virtual keys don't match
if (m_buttonToVK[button] != i) {
m_buttonToVK[button | 0x100u] = i;
}
break;
}
}
// add alt+printscreen
if (m_buttonToVK[0x54u] == 0) {
m_buttonToVK[0x54u] = VK_SNAPSHOT;
}
// set virtual key to button table
if (GetKeyboardLayout(0) == m_groups[g]) {
for (KeyButton i = 0; i < 512; ++i) {
if (m_buttonToVK[i] != 0) {
if (m_virtualKeyToButton[m_buttonToVK[i]] == 0) {
m_virtualKeyToButton[m_buttonToVK[i]] = i;
}
}
if (m_buttonToNumpadVK[i] != 0) {
if (m_virtualKeyToButton[m_buttonToNumpadVK[i]] == 0) {
m_virtualKeyToButton[m_buttonToNumpadVK[i]] = i;
}
}
}
}
// add numpad keys
for (KeyButton i = 0; i < 512; ++i) {
if (m_buttonToNumpadVK[i] != 0) {
item.m_id = getKeyID(m_buttonToNumpadVK[i], i);
item.m_button = i;
item.m_required = KeyModifierNumLock;
item.m_sensitive = KeyModifierNumLock | KeyModifierShift;
item.m_generates = 0;
item.m_client = m_buttonToNumpadVK[i];
addKeyEntry(keyMap, item);
}
}
// add other keys
BYTE keys[256];
memset(keys, 0, sizeof(keys));
for (KeyButton i = 0; i < 512; ++i) {
if (m_buttonToVK[i] != 0) {
// initialize item
item.m_id = getKeyID(m_buttonToVK[i], i);
item.m_button = i;
item.m_required = 0;
item.m_sensitive = 0;
item.m_client = m_buttonToVK[i];
// get flags for modifier keys
CKeyMap::initModifierKey(item);
if (item.m_id == 0) {
// translate virtual key to a character with and without
// shift, caps lock, and AltGr.
struct Modifier {
UINT m_vk1;
UINT m_vk2;
BYTE m_state;
KeyModifierMask m_mask;
};
static const Modifier modifiers[] = {
{ VK_SHIFT, VK_SHIFT, 0x80u, KeyModifierShift },
{ VK_CAPITAL, VK_CAPITAL, 0x01u, KeyModifierCapsLock },
{ VK_CONTROL, VK_MENU, 0x80u, KeyModifierControl |
KeyModifierAlt }
};
static const size_t s_numModifiers =
sizeof(modifiers) / sizeof(modifiers[0]);
static const size_t s_numCombinations = 1 << s_numModifiers;
KeyID id[s_numCombinations];
bool anyFound = false;
KeyButton button = static_cast<KeyButton>(i & 0xffu);
for (size_t j = 0; j < s_numCombinations; ++j) {
for (size_t k = 0; k < s_numModifiers; ++k) {
//if ((j & (1 << k)) != 0) {
// http://msdn.microsoft.com/en-us/library/ke55d167.aspx
if ((j & (1i64 << k)) != 0) {
keys[modifiers[k].m_vk1] = modifiers[k].m_state;
keys[modifiers[k].m_vk2] = modifiers[k].m_state;
}
else {
keys[modifiers[k].m_vk1] = 0;
keys[modifiers[k].m_vk2] = 0;
}
}
id[j] = getIDForKey(item, button,
m_buttonToVK[i], keys, m_groups[g]);
if (id[j] != 0) {
anyFound = true;
}
}
if (anyFound) {
// determine what modifiers we're sensitive to.
// we're sensitive if the KeyID changes when the
// modifier does.
item.m_sensitive = 0;
for (size_t k = 0; k < s_numModifiers; ++k) {
for (size_t j = 0; j < s_numCombinations; ++j) {
//if (id[j] != id[j ^ (1u << k)]) {
// http://msdn.microsoft.com/en-us/library/ke55d167.aspx
if (id[j] != id[j ^ (1ui64 << k)]) {
item.m_sensitive |= modifiers[k].m_mask;
break;
}
}
}
// save each key. the map will automatically discard
// duplicates, like an unshift and shifted version of
// a key that's insensitive to shift.
for (size_t j = 0; j < s_numCombinations; ++j) {
item.m_id = id[j];
item.m_required = 0;
for (size_t k = 0; k < s_numModifiers; ++k) {
if ((j & (1i64 << k)) != 0) {
item.m_required |= modifiers[k].m_mask;
}
}
addKeyEntry(keyMap, item);
}
}
}
else {
// found in table
switch (m_buttonToVK[i]) {
case VK_TAB:
// add kKeyLeftTab, too
item.m_id = kKeyLeftTab;
item.m_required |= KeyModifierShift;
item.m_sensitive |= KeyModifierShift;
addKeyEntry(keyMap, item);
item.m_id = kKeyTab;
item.m_required &= ~KeyModifierShift;
break;
case VK_CANCEL:
item.m_required |= KeyModifierControl;
item.m_sensitive |= KeyModifierControl;
break;
case VK_SNAPSHOT:
item.m_sensitive |= KeyModifierAlt;
if ((i & 0x100u) == 0) {
// non-extended snapshot key requires alt
item.m_required |= KeyModifierAlt;
}
break;
}
addKeyEntry(keyMap, item);
}
}
}
}
// restore keyboard layout
ActivateKeyboardLayout(activeLayout, 0);
}
void ConfusabledataBuilder::build(const char * confusables, int32_t confusablesLen,
UErrorCode &status) {
// Convert the user input data from UTF-8 to UChar (UTF-16)
int32_t inputLen = 0;
if (U_FAILURE(status)) {
return;
}
u_strFromUTF8(NULL, 0, &inputLen, confusables, confusablesLen, &status);
if (status != U_BUFFER_OVERFLOW_ERROR) {
return;
}
status = U_ZERO_ERROR;
fInput = static_cast<UChar *>(uprv_malloc((inputLen+1) * sizeof(UChar)));
if (fInput == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
}
u_strFromUTF8(fInput, inputLen+1, NULL, confusables, confusablesLen, &status);
// Regular Expression to parse a line from Confusables.txt. The expression will match
// any line. What was matched is determined by examining which capture groups have a match.
// Capture Group 1: the source char
// Capture Group 2: the replacement chars
// Capture Group 3-6 the table type, SL, SA, ML, or MA
// Capture Group 7: A blank or comment only line.
// Capture Group 8: A syntactically invalid line. Anything that didn't match before.
// Example Line from the confusables.txt source file:
// "1D702 ; 006E 0329 ; SL # MATHEMATICAL ITALIC SMALL ETA ... "
fParseLine = uregex_openC(
"(?m)^[ \\t]*([0-9A-Fa-f]+)[ \\t]+;" // Match the source char
"[ \\t]*([0-9A-Fa-f]+" // Match the replacement char(s)
"(?:[ \\t]+[0-9A-Fa-f]+)*)[ \\t]*;" // (continued)
"\\s*(?:(SL)|(SA)|(ML)|(MA))" // Match the table type
"[ \\t]*(?:#.*?)?$" // Match any trailing #comment
"|^([ \\t]*(?:#.*?)?)$" // OR match empty lines or lines with only a #comment
"|^(.*?)$", // OR match any line, which catches illegal lines.
0, NULL, &status);
// Regular expression for parsing a hex number out of a space-separated list of them.
// Capture group 1 gets the number, with spaces removed.
fParseHexNum = uregex_openC("\\s*([0-9A-F]+)", 0, NULL, &status);
// Zap any Byte Order Mark at the start of input. Changing it to a space is benign
// given the syntax of the input.
if (*fInput == 0xfeff) {
*fInput = 0x20;
}
// Parse the input, one line per iteration of this loop.
uregex_setText(fParseLine, fInput, inputLen, &status);
while (uregex_findNext(fParseLine, &status)) {
fLineNum++;
if (uregex_start(fParseLine, 7, &status) >= 0) {
// this was a blank or comment line.
continue;
}
if (uregex_start(fParseLine, 8, &status) >= 0) {
// input file syntax error.
status = U_PARSE_ERROR;
return;
}
// We have a good input line. Extract the key character and mapping string, and
// put them into the appropriate mapping table.
UChar32 keyChar = SpoofImpl::ScanHex(fInput, uregex_start(fParseLine, 1, &status),
uregex_end(fParseLine, 1, &status), status);
int32_t mapStringStart = uregex_start(fParseLine, 2, &status);
int32_t mapStringLength = uregex_end(fParseLine, 2, &status) - mapStringStart;
uregex_setText(fParseHexNum, &fInput[mapStringStart], mapStringLength, &status);
UnicodeString *mapString = new UnicodeString();
if (mapString == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
while (uregex_findNext(fParseHexNum, &status)) {
UChar32 c = SpoofImpl::ScanHex(&fInput[mapStringStart], uregex_start(fParseHexNum, 1, &status),
uregex_end(fParseHexNum, 1, &status), status);
mapString->append(c);
}
U_ASSERT(mapString->length() >= 1);
// Put the map (value) string into the string pool
// This a little like a Java intern() - any duplicates will be eliminated.
SPUString *smapString = stringPool->addString(mapString, status);
// Add the UChar32 -> string mapping to the appropriate table.
UHashtable *table = uregex_start(fParseLine, 3, &status) >= 0 ? fSLTable :
uregex_start(fParseLine, 4, &status) >= 0 ? fSATable :
uregex_start(fParseLine, 5, &status) >= 0 ? fMLTable :
uregex_start(fParseLine, 6, &status) >= 0 ? fMATable :
NULL;
U_ASSERT(table != NULL);
uhash_iput(table, keyChar, smapString, &status);
fKeySet->add(keyChar);
if (U_FAILURE(status)) {
return;
}
}
// Input data is now all parsed and collected.
// Now create the run-time binary form of the data.
//
// This is done in two steps. First the data is assembled into vectors and strings,
// for ease of construction, then the contents of these collections are dumped
// into the actual raw-bytes data storage.
// Build up the string array, and record the index of each string therein
// in the (build time only) string pool.
// Strings of length one are not entered into the strings array.
// At the same time, build up the string lengths table, which records the
// position in the string table of the first string of each length >= 4.
// (Strings in the table are sorted by length)
stringPool->sort(status);
fStringTable = new UnicodeString();
fStringLengthsTable = new UVector(status);
int32_t previousStringLength = 0;
int32_t previousStringIndex = 0;
int32_t poolSize = stringPool->size();
int32_t i;
for (i=0; i<poolSize; i++) {
SPUString *s = stringPool->getByIndex(i);
int32_t strLen = s->fStr->length();
int32_t strIndex = fStringTable->length();
U_ASSERT(strLen >= previousStringLength);
if (strLen == 1) {
// strings of length one do not get an entry in the string table.
// Keep the single string character itself here, which is the same
// convention that is used in the final run-time string table index.
s->fStrTableIndex = s->fStr->charAt(0);
} else {
if ((strLen > previousStringLength) && (previousStringLength >= 4)) {
fStringLengthsTable->addElement(previousStringIndex, status);
fStringLengthsTable->addElement(previousStringLength, status);
}
s->fStrTableIndex = strIndex;
fStringTable->append(*(s->fStr));
}
previousStringLength = strLen;
previousStringIndex = strIndex;
}
// Make the final entry to the string lengths table.
// (it holds an entry for the _last_ string of each length, so adding the
// final one doesn't happen in the main loop because no longer string was encountered.)
if (previousStringLength >= 4) {
fStringLengthsTable->addElement(previousStringIndex, status);
fStringLengthsTable->addElement(previousStringLength, status);
}
// Construct the compile-time Key and Value tables
//
// For each key code point, check which mapping tables it applies to,
// and create the final data for the key & value structures.
//
// The four logical mapping tables are conflated into one combined table.
// If multiple logical tables have the same mapping for some key, they
// share a single entry in the combined table.
// If more than one mapping exists for the same key code point, multiple
// entries will be created in the table
for (int32_t range=0; range<fKeySet->getRangeCount(); range++) {
// It is an oddity of the UnicodeSet API that simply enumerating the contained
// code points requires a nested loop.
for (UChar32 keyChar=fKeySet->getRangeStart(range);
keyChar <= fKeySet->getRangeEnd(range); keyChar++) {
addKeyEntry(keyChar, fSLTable, USPOOF_SL_TABLE_FLAG, status);
addKeyEntry(keyChar, fSATable, USPOOF_SA_TABLE_FLAG, status);
addKeyEntry(keyChar, fMLTable, USPOOF_ML_TABLE_FLAG, status);
addKeyEntry(keyChar, fMATable, USPOOF_MA_TABLE_FLAG, status);
}
}
// Put the assembled data into the flat runtime array
outputData(status);
// All of the intermediate allocated data belongs to the ConfusabledataBuilder
// object (this), and is deleted in the destructor.
return;
}
