static MessageBody* get_body(CollectedPart *body_part, gboolean sanitize_body, GPtrArray *inlines) {
g_return_val_if_fail(body_part != NULL, NULL);
MessageBody *mb = new_message_body();
// We keep the raw size intentionally
mb->size = body_part->content->len;
mb->content_type = g_strdup(body_part->content_type);
if (sanitize_body) {
// Parse any HTML tags
GString *raw_content = g_string_new_len((const gchar*) body_part->content->data, body_part->content->len);
GumboOutput* output = gumbo_parse_with_options(&kGumboDefaultOptions, raw_content->str, raw_content->len);
// Remove unallowed HTML tags (like scripts, bad href etc..)
GString *sanitized_content = sanitize(output->document, inlines);
mb->content = sanitized_content;
gumbo_destroy_output(&kGumboDefaultOptions, output);
g_string_free(raw_content, TRUE);
} else {
mb->content = g_string_new_len((const gchar*) body_part->content->data, body_part->content->len);
}
return mb;
}
void GumboInterface::parse()
{
if (!m_source.isEmpty() && (m_output == NULL)) {
if (!m_version.startsWith('3')) {
m_hasnbsp = m_source.contains(" ");
}
m_utf8src = m_source.toStdString();
// remove any xml header line and any trailing whitespace
if (m_utf8src.compare(0,5,"<?xml") == 0) {
size_t end = m_utf8src.find_first_of('>', 5);
end = m_utf8src.find_first_not_of("\n\r\t\v\f ",end+1);
m_utf8src.erase(0,end);
}
// 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;
// GumboInterface::m_mutex.lock();
m_output = gumbo_parse_with_options(&myoptions, m_utf8src.data(), m_utf8src.length());
// GumboInterface::m_mutex.unlock();
}
}
int main(int argc, const char** argv){
if(argc != 2){
printf("得到html文件的各个节点（请加上html文件名作为执行参数）");
exit(1);
}
const char* filename = argv[1];
//
FILE *fp;
if(!(fp = fopen(filename, "r"))){
printf("file open error!");
exit(1);
}
//
char* input;
int length;
readFile(fp, &input, &length);
//
GumboOutput* output = gumbo_parse_with_options(
&kGumboDefaultOptions, input, length);
findTag(output->root, 0);
//
gumbo_destroy_output(&kGumboDefaultOptions, output);
fclose(fp);
free(input);
}
static int
es_gumbo_parse(duk_context *ctx)
{
duk_size_t len;
const char *str = duk_to_lstring(ctx, 0, &len);
es_gumbo_output_t *ego = calloc(1, sizeof(es_gumbo_output_t));
atomic_set(&ego->ego_refcount, 1);
ego->ego_output =
gumbo_parse_with_options(&kGumboDefaultOptions, str, len);
/*
for(int i = 0; i < ego->ego_output->errors.length; i++) {
GumboError *ge = ego->ego_output->errors.data[i];
}
*/
duk_pop(ctx);
duk_push_object(ctx);
push_gumbo_node(ctx, ego->ego_output->document, ego);
duk_put_prop_string(ctx, -2, "document");
push_gumbo_node(ctx, ego->ego_output->root, ego);
duk_put_prop_string(ctx, -2, "root");
es_gumbo_output_release(ego);
return 1;
}
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 GumboInterface::parse()
{
if (!m_source.isEmpty() && (m_output == NULL)) {
m_utf8src = m_source.toStdString();
// remove any xml header line and any trailing whitespace
if (m_utf8src.compare(0,5,"<?xml") == 0) {
size_t end = m_utf8src.find_first_of('>', 5);
end = m_utf8src.find_first_not_of("\n\r\t ",end+1);
m_utf8src.erase(0,end);
}
GumboOptions myoptions = kGumboDefaultOptions;
myoptions.use_xhtml_rules = true;
myoptions.tab_stop = 4;
m_output = gumbo_parse_with_options(&myoptions, m_utf8src.data(), m_utf8src.length());
}
}
static int parse(lua_State *L) {
size_t length;
const char *input = luaL_checklstring(L, 1, &length);
GumboOptions options = kGumboDefaultOptions;
options.tab_stop = luaL_optint(L, 2, 8);
GumboOutput *output = gumbo_parse_with_options(&options, input, length);
if (output) {
push_node(L, output->document);
lua_rawgeti(L, -1, output->root->index_within_parent + 1);
lua_setfield(L, -2, "root");
gumbo_destroy_output(&options, output);
return 1;
} else {
lua_pushnil(L);
lua_pushliteral(L, "Failed to parse");
return 2;
}
}
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;
}
static int parse(lua_State *L) {
size_t input_len, tagname_len;
GumboOptions options = kGumboDefaultOptions;
options.max_errors = 0;
const char *input = luaL_checklstring(L, 1, &input_len);
options.tab_stop = (int)luaL_optinteger(L, 2, 8);
const char *tagname = luaL_optlstring(L, 3, NULL, &tagname_len);
if (tagname != NULL) {
options.fragment_context = gumbo_tagn_enum(tagname, tagname_len);
}
static const char *namespaces[] = {"html", "svg", "math", NULL};
options.fragment_namespace = luaL_checkoption(L, 4, "html", namespaces);
for (int i = 1; i <= nupvalues; i++) {
luaL_checktype(L, i + 4, LUA_TTABLE);
}
lua_pushcclosure(L, push_document, nupvalues);
GumboOutput *output = gumbo_parse_with_options(&options, input, input_len);
if (output == NULL) {
lua_pushnil(L);
lua_pushliteral(L, "gumbo_parse_with_options() returned NULL");
return 2;
}
GumboOutputStatus status = output->status;
if (status != GUMBO_STATUS_OK) {
gumbo_destroy_output(output);
lua_pushnil(L);
lua_pushstring(L, gumbo_status_to_string(status));
return 2;
}
lua_pushlightuserdata(L, &output->document->v.document);
int err = lua_pcall(L, 1, 1, 0);
gumbo_destroy_output(output);
if (err == 0) { // LUA_OK
return 1;
} else {
lua_pushnil(L);
lua_pushvalue(L, -2);
return 2;
}
}
