void gumbo_print_caret_diagnostic(
GumboParser* parser, const GumboError* error, const char* source_text) {
GumboStringBuffer text;
gumbo_string_buffer_init(parser, &text);
gumbo_caret_diagnostic_to_string(parser, error, source_text, &text);
printf("%.*s", (int) text.length, text.data);
gumbo_string_buffer_destroy(parser, &text);
}
QList<GumboWellFormedError> GumboInterface::error_check()
{
QList<GumboWellFormedError> errlist;
int line_offset = 0;
// In case we ever have to revert to earlier versions, please note the following
// additional initialization is needed because Microsoft Visual Studio 2013 (and earlier?)
// do not properly initialize myoptions from the static const kGumboDefaultOptions defined
// in the gumbo library. Instead whatever was in memory at the time is used causing random
// issues later on so if reverting remember to keep these specific changes as the bug
// they work around took a long long time to track down
GumboOptions myoptions = kGumboDefaultOptions;
myoptions.tab_stop = 4;
myoptions.use_xhtml_rules = true;
myoptions.stop_on_first_error = false;
myoptions.max_errors = -1;
if (!m_source.isEmpty() && (m_output == NULL)) {
// fix any non html valid self-closing tags
m_source = fix_self_closing_tags(m_source);
m_utf8src = m_source.toStdString();
// remove any xml header line and trailing whitespace
if (m_utf8src.compare(0,5,"<?xml") == 0) {
size_t end = m_utf8src.find_first_of('>', 0);
end = m_utf8src.find_first_not_of("\n\r\t\v\f ",end+1);
m_utf8src.erase(0,end);
line_offset++;
}
// add in doctype if missing
if ((m_utf8src.compare(0,9,"<!DOCTYPE") != 0) && (m_utf8src.compare(0,9,"<!doctype") != 0)) {
m_utf8src.insert(0,"<!DOCTYPE html>\n");
line_offset--;
}
// GumboInterface::m_mutex.lock();
m_output = gumbo_parse_with_options(&myoptions, m_utf8src.data(), m_utf8src.length());
// GumboInterface::m_mutex.unlock();
}
const GumboVector* errors = &m_output->errors;
for (unsigned int i=0; i< errors->length; ++i) {
GumboError* er = static_cast<GumboError*>(errors->data[i]);
GumboWellFormedError gperror;
gperror.line = er->position.line + line_offset;;
gperror.column = er->position.column;
// unsigned int typenum = er->type;
GumboStringBuffer text;
gumbo_string_buffer_init(&text);
gumbo_error_to_string(er, &text);
std::string errmsg(text.data, text.length);
gperror.message = QString::fromStdString(errmsg);
gumbo_string_buffer_destroy(&text);
errlist.append(gperror);
}
return errlist;
}
void gumbo_string_buffer_clear(GumboStringBuffer* input) {
input->length = 0;
if (input->capacity > kDefaultStringBufferSize * 8) {
// This approach to clearing means that the buffer can grow unbounded and
// tie up memory that may be needed for parsing the rest of the document, so
// we free and reinitialize the buffer if its grown more than 3 doublings.
gumbo_string_buffer_destroy(input);
gumbo_string_buffer_init(input);
}
}
QList<GumboWellFormedError> GumboInterface::error_check()
{
QList<GumboWellFormedError> errlist;
int line_offset = 0;
GumboOptions myoptions = kGumboDefaultOptions;
myoptions.use_xhtml_rules = true;
myoptions.tab_stop = 4;
// leave this as false to prevent pre-mature stopping when no error exists
myoptions.stop_on_first_error = false;
if (!m_source.isEmpty() && (m_output == NULL)) {
m_utf8src = m_source.toStdString();
// remove any xml header line and trailing whitespace
if (m_utf8src.compare(0,5,"<?xml") == 0) {
size_t end = m_utf8src.find_first_of('>', 0);
end = m_utf8src.find_first_not_of("\n\r\t ",end+1);
m_utf8src.erase(0,end);
line_offset++;
}
// add in doctype if missing
if ((m_utf8src.compare(0,9,"<!DOCTYPE") != 0) && (m_utf8src.compare(0,9,"<!doctype") != 0)) {
m_utf8src.insert(0,"<!DOCTYPE html>\n");
line_offset--;
}
m_output = gumbo_parse_with_options(&myoptions, m_utf8src.data(), m_utf8src.length());
}
const GumboVector* errors = &m_output->errors;
for (int i=0; i< errors->length; ++i) {
GumboError* er = static_cast<GumboError*>(errors->data[i]);
GumboWellFormedError gperror;
gperror.line = er->position.line + line_offset;;
gperror.column = er->position.column;
unsigned int typenum = er->type;
GumboStringBuffer text;
gumbo_string_buffer_init(&text);
gumbo_error_to_string(er, &text);
std::string errmsg(text.data, text.length);
gperror.message = QString::fromStdString(errmsg);
gumbo_string_buffer_destroy(&text);
errlist.append(gperror);
}
return errlist;
}
