const CssSelector *CssPullParser::NextSelector()
{
if (!currSel)
return nullptr;
SkipWsAndComments(currSel, selEnd);
if (currSel == selEnd)
return nullptr;
sel.s = currSel;
// skip single selector
const char *sEnd = currSel;
while (currSel < selEnd && *currSel != ',') {
if (*currSel == '"' || *currSel == '\'') {
bool ok = SkipQuotedString(currSel, selEnd);
CrashIf(!ok);
sEnd = currSel;
}
else if (*currSel == '\\' && currSel < selEnd - 1) {
currSel += 2;
sEnd = currSel;
}
else if (!SkipWsAndComments(currSel, selEnd)) {
sEnd = ++currSel;
}
}
if (currSel < selEnd)
currSel++;
sel.sLen = sEnd - sel.s;
sel.tag = Tag_NotFound;
sel.clazz = nullptr;
sel.clazzLen = 0;
// parse "*", "el", ".class" and "el.class"
const char *c = sEnd;
for (; c > sel.s && (isalnum((unsigned char)*(c - 1)) || *(c - 1) == '-'); c--);
if (c > sel.s && *(c - 1) == '.') {
sel.clazz = c;
sel.clazzLen = sEnd - c;
c--;
}
for (; c > sel.s && (isalnum((unsigned char)*(c - 1)) || *(c - 1) == '-'); c--);
if (sel.clazz - 1 == sel.s) {
sel.tag = Tag_Any;
}
else if (c == (sel.clazz ? sel.clazz - 1 : sEnd) && c == sel.s + 1 && *sel.s == '*') {
sel.tag = Tag_Any;
}
else if (c == sel.s) {
sel.tag = FindHtmlTag(sel.s, sel.clazz ? sel.clazz - sel.s - 1 : sel.sLen);
}
return &sel;
}
bool CssPullParser::NextRule()
{
if (inProps)
while (NextProperty());
CrashIf(inProps && currPos < end);
if (inlineStyle || currPos == end)
return false;
if (currPos == s) {
SkipWsAndComments(currPos, end);
if (currPos + 4 < end && str::StartsWith(currPos, "<!--"))
currPos += 4;
}
SkipWsAndComments(currPos, end);
currSel = currPos;
// skip selectors
while (currPos < end && *currPos != '{') {
if (*currPos == '"' || *currPos == '\'') {
if (!SkipQuotedString(currPos, end))
break;
}
else if (*currPos == ';') {
currPos++;
SkipWsAndComments(currPos, end);
currSel = currPos;
}
else if (!SkipWsAndComments(currPos, end)) {
currPos++;
}
}
if (currPos == end) {
currSel = nullptr;
return false;
}
selEnd = currPos++;
inProps = true;
return true;
}
static bool SkipBlock(const char*& s, const char *end)
{
CrashIf(s >= end || *s != '{');
s++;
while (s < end && *s != '}') {
if (*s == '"' || *s == '\'') {
if (!SkipQuotedString(s, end))
return false;
}
else if (*s == '{') {
if (!SkipBlock(s, end))
return false;
}
else if (*s == '\\' && s < end - 1)
s += 2;
else if (!SkipWsAndComments(s, end))
s++;
}
if (s == end)
return false;
s++;
return true;
}
static SquareTreeNode *ParseSquareTreeRec(char *& data, bool isTopLevel=false)
{
SquareTreeNode *node = new SquareTreeNode();
while (*(data = SkipWsAndComments(data))) {
// all non-empty non-comment lines contain a key-value pair
// where the value is either a string (separated by '=' or ':')
// or a list of child nodes (if the key is followed by '[' alone)
char *key = data;
for (data = key; *data && *data != '=' && *data != ':' && *data != '[' && *data != ']' && *data != '\n'; data++);
if (!*data || '\n' == *data) {
// use first whitespace as a fallback separator
for (data = key; *data && !str::IsWs(*data); data++);
}
char *separator = data;
if (*data && *data != '\n') {
// skip to the first non-whitespace character on the same line (value)
data = SkipWs(data + 1, true);
}
char *value = data;
// skip to the end of the line
for (; *data && *data != '\n'; data++);
if (IsBracketLine(separator) ||
// also tolerate "key \n [ \n ... \n ]" (else the key
// gets an empty value and the child node an empty key)
str::IsWs(*separator) && '\n' == *value && IsBracketLine(SkipWsAndComments(data))) {
// parse child node(s)
data = SkipWsAndComments(separator) + 1;
*SkipWsRev(key, separator) = '\0';
node->data.Append(SquareTreeNode::DataItem(key, ParseSquareTreeRec(data)));
// arrays are created by either reusing the same key for a different child
// or by concatenating multiple children ("[ \n ] [ \n ] [ \n ]")
while (IsBracketLine((data = SkipWsAndComments(data)))) {
data++;
node->data.Append(SquareTreeNode::DataItem(key, ParseSquareTreeRec(data)));
}
}
else if (']' == *key) {
// finish parsing child node
data = key + 1;
if (!isTopLevel)
return node;
// ignore superfluous closing square brackets instead of
// ignoring all content following them
}
else if ('[' == *key && ']' == SkipWsRev(value, data)[-1]) {
// treat INI section headers as top-level node names
// (else "[Section]" would be ignored)
if (!isTopLevel) {
data = key;
return node;
}
// trim whitespace around section name (for consistency with GetPrivateProfileString)
key = SkipWs(key + 1);
*SkipWsRev(key, SkipWsRev(value, data) - 1) = '\0';
node->data.Append(SquareTreeNode::DataItem(key, ParseSquareTreeRec(data)));
}
else if ('[' == *separator || ']' == *separator) {
// invalid line (ignored)
}
else {
// string value (decoding is left to the consumer)
bool hasMoreLines = '\n' == *data;
*SkipWsRev(key, separator) = '\0';
*SkipWsRev(value, data) = '\0';
node->data.Append(SquareTreeNode::DataItem(key, value));
if (hasMoreLines)
data++;
}
}
// assume that all square brackets have been properly balanced
return node;
}
const CssProperty *CssPullParser::NextProperty()
{
if (currPos == s)
inlineStyle = inProps = true;
else if (!inProps)
return nullptr;
GetNextProperty:
SkipWsAndComments(currPos, end);
if (currPos == end)
return nullptr;
if (*currPos == '}') {
currPos++;
inProps = false;
return nullptr;
}
if (*currPos == '{') {
if (!SkipBlock(currPos, end))
return nullptr;
goto GetNextProperty;
}
if (*currPos == ';') {
currPos++;
goto GetNextProperty;
}
const char *name = currPos;
// skip identifier
while (currPos < end && !str::IsWs(*currPos) && *currPos != ':' &&
*currPos != ';' && *currPos != '{' && *currPos != '}') {
currPos++;
}
SkipWsAndComments(currPos, end);
if (currPos == end || *currPos != ':')
goto GetNextProperty;
prop.type = FindCssProp(name, currPos - name);
currPos++;
SkipWsAndComments(currPos, end);
prop.s = currPos;
// skip value
const char *valEnd = currPos;
while (currPos < end && *currPos != ';' && *currPos != '}') {
if (*currPos == '"' || *currPos == '\'') {
if (!SkipQuotedString(currPos, end))
return nullptr;
valEnd = currPos;
}
else if (*currPos == '{') {
if (!SkipBlock(currPos, end))
return nullptr;
valEnd = currPos;
}
else if (*currPos == '\\' && currPos < end - 1) {
currPos += 2;
valEnd = currPos;
}
else if (!SkipWsAndComments(currPos, end)) {
valEnd = ++currPos;
}
}
prop.sLen = valEnd - prop.s;
return ∝
}
