string HtmlParser::buscarTextoInTag(char *html, Tags *tag) {
GumboOutput* output = gumbo_parse(html);
string res = search_text(output->root, tag);
gumbo_destroy_output(&kGumboDefaultOptions, output);
return res;
}
/* Exported function documented in search.h */
void search_step(struct search_context *context, search_flags_t flags,
const char *string)
{
int string_len;
int i = 0;
if (context == NULL) {
warn_user("SearchError", 0);
return;
}
guit->search->add_recent(string, context->gui_p);
string_len = strlen(string);
for (i = 0; i < string_len; i++)
if (string[i] != '#' && string[i] != '*')
break;
if (i >= string_len) {
union content_msg_data msg_data;
free_matches(context);
guit->search->status(true, context->gui_p);
guit->search->back_state(false, context->gui_p);
guit->search->forward_state(false, context->gui_p);
msg_data.scroll.area = false;
msg_data.scroll.x0 = 0;
msg_data.scroll.y0 = 0;
content_broadcast(context->c, CONTENT_MSG_SCROLL, msg_data);
return;
}
search_text(string, string_len, context, flags);
}
void HtmlParser::search_for_links(GumboNode* node, Tags *tag) {
if (node->type != GUMBO_NODE_ELEMENT) {
return;
}
GumboAttribute* attribute;
if (node->v.element.tag == tag->htmlTag){
map<string, string> attrList;
attrValue attrElement;
if ( (attribute = gumbo_get_attribute(&node->v.element.attributes, tag->attr.c_str()))) {
attrElement.attrList.insert( make_pair(attribute->name, attribute->value));
} else {
//Recogemos el resto de atributos y los incluimos en la variable tag
for (unsigned int i=0; i<node->v.element.attributes.length; i++){
attribute = ((GumboAttribute *)node->v.element.attributes.data[i]);
attrElement.attrList.insert( make_pair(attribute->name, attribute->value));
}
}
attrElement.content = isObtainContentTag() ? search_text(node, tag) : "";
tag->tagElement.push_back(attrElement);
}
GumboVector* children = &node->v.element.children;
// std::cout << "****** elementos: " << children->length << endl;
for (unsigned int i = 0; i < children->length; ++i) {
search_for_links(static_cast<GumboNode*>(children->data[i]), tag);
}
}
string HtmlParser::search_text(GumboNode* node, Tags *tag) {
if (node->type == GUMBO_NODE_TEXT) {
return std::string(node->v.text.text);
} else if (node->type == GUMBO_NODE_ELEMENT &&
node->v.element.tag != GUMBO_TAG_STYLE) {
std::string contents = "";
GumboVector* children = &node->v.element.children;
for (int i = 0; i < children->length; ++i) {
const std::string text = search_text((GumboNode*) children->data[i], tag);
if (i != 0 && !text.empty()) {
contents.append(" ");
}
contents.append(text);
}
return contents;
} else {
return "";
}
}
DataStore Catalog::text_process(std::string input_text){
int flag = 0;
int flag_case = 1; //used to distinguish call from filemanager and text_process
std::string search_txt, sub1, sub2;
std::size_t pos;
std::vector<std::string>::iterator itr_patt;
if (input_text.find("\"") == 0){ //Search text is saperated if given in "quotes"
int i = input_text.find("\"", 1);
sub1 = input_text.substr(1, i - 1);
int len = input_text.length();
sub2 = input_text.substr(i + 1, len);
search_txt = sub1;
flag = 1;
}
else{ sub2 = input_text; }
std::istringstream iss(sub2);
do{
std::string sub; //search text is filtered if the text is only one word
iss >> sub;
if (flag == 0){
search_txt = sub;
flag = 1;
}
else{
if (sub.find("*") == 0){ //storing valit extensions
if (sub.size() != 0){
pos = sub.find(".");
sub = sub.substr(pos + 1);
if (sub == "*")flag_case = 0;
patt.push_back(sub);
}
}
else{
if (int(sub.find(".")) > 0) //To store filename that may contain text
patt.push_back(sub);
}
}
} while (iss);
if (patt.size() == 0)
flag_case = 0;
return search_text(search_txt, flag_case);
}
bool wildcard_samples()
{
// wildcard search examples
strref search_text("in 2 out of 5 cases the trees will outnumber the carrots, willoutnumber");
// basic wildcard find substring (*)
strref search("the*will");
strref substr = search_text.find_wildcard(search);
printf("found \"" STRREF_FMT "\" matching \"" STRREF_FMT "\"\n", STRREF_ARG(substr), STRREF_ARG(search));
if (!substr.same_str_case("the trees will"))
return false;
// find word beginning with c
substr = search_text.find_wildcard("<c*>");
if (!substr.same_str_case("cases"))
return false;
// find word ending with e
substr = search_text.find_wildcard("<*%e>");
if (!substr.same_str_case("the"))
return false;
// find whole line
substr = search_text.find_wildcard("@*^");
if (!substr.same_str_case(search_text))
return false;
// using ranges and single character number
substr = search_text.find_wildcard(strref("[0-9] out of #"));
if (!substr.same_str_case("2 out of 5"))
return false;
// searching any substring between two ranged characters
substr = search_text.find_wildcard(strref("#*[0-9]"));
if (!substr.same_str_case("2 out of 5"))
return false;
// searching for substring but spaces are not allowed
substr = search_text.find_wildcard("will*{! }number");
printf("substr = " STRREF_FMT "\n", STRREF_ARG(substr));
if (!substr.same_str_case("willoutnumber"))
return false;
// find r folloed by numbers without another substring to find
substr = strref("no k12 r13 [99]").find_wildcard("r*{0-9}");
if (!substr.same_str_case("r13"))
return false;
// search wildcard iteratively
search = "radio, gorilla, zebra, monkey, human, rat, car, ocelot, conrad, butler";
substr.clear();
printf("Words with 'r': ");
while ((substr = search.wildcard_after("<*$r*$>", substr)))
printf("\"" STRREF_FMT "\" ", STRREF_ARG(substr));
printf("\n");
// search for a more complex expression
search = strref(aSomeFiles, sizeof(aSomeFiles)-1);
substr.clear();
printf("Lines with two subsequent characters d-e (dd, de, ed, ee) and only alphanumeric or dot after:\n");
while ((substr = search.wildcard_after("@*@[d-e][d-e]*{0-9A-Za-z.}^", substr))) {
printf("\"" STRREF_FMT "\"\n", STRREF_ARG(substr));
}
substr.clear();
printf("\n.json files:\n\n");
while ((substr = search.wildcard_after("<*{!/}.json^", substr))) {
printf("\"" STRREF_FMT "\"\n", STRREF_ARG(substr));
}
printf("\nsome words with a character a-f followed by a character that is not a-f:\n\n");
search = strref(aTextSample, sizeof(aTextSample)-1);
substr.clear();
while ((substr = search.wildcard_after("<*$[a-f][!a-f\\1-\\x40]*>", substr))) {
printf("\"" STRREF_FMT "\" ", STRREF_ARG(substr));
}
printf("\n");
return true;
}
static void search_button_clicked(GtkWidget *widget, viewer_t *viewer)
{
search_text(viewer, 1);
}
static void search_entry_validated(GtkWidget *widget, viewer_t *viewer)
{
search_text(viewer, 0);
}
static void read_commands(struct client *client)
{
char buf[1024];
int pos = 0;
if (!client->authenticated)
client->authenticated = addr.sa.sa_family == AF_UNIX;
while (1) {
int rc, s, i;
rc = read(client->fd, buf + pos, sizeof(buf) - pos);
if (rc == -1) {
if (errno == EINTR)
continue;
if (errno == EAGAIN)
return;
goto close;
}
if (rc == 0)
goto close;
pos += rc;
s = 0;
for (i = 0; i < pos; i++) {
const char *line, *msg;
char *cmd, *arg;
int ret;
if (buf[i] != '\n')
continue;
buf[i] = 0;
line = buf + s;
s = i + 1;
if (!client->authenticated) {
if (!server_password) {
msg = "password is unset, tcp/ip disabled";
d_print("%s\n", msg);
ret = send_answer(client->fd, "%s\n\n", msg);
goto close;
}
if (strncmp(line, "passwd ", 7) == 0)
line += 7;
client->authenticated = !strcmp(line, server_password);
if (!client->authenticated) {
msg = "authentication failed";
d_print("%s\n", msg);
ret = send_answer(client->fd, "%s\n\n", msg);
goto close;
}
ret = write_all(client->fd, "\n", 1);
continue;
}
while (isspace((unsigned char)*line))
line++;
if (*line == '/') {
int restricted = 0;
line++;
search_direction = SEARCH_FORWARD;
if (*line == '/') {
line++;
restricted = 1;
}
search_text(line, restricted, 1);
ret = write_all(client->fd, "\n", 1);
} else if (*line == '?') {
int restricted = 0;
line++;
search_direction = SEARCH_BACKWARD;
if (*line == '?') {
line++;
restricted = 1;
}
search_text(line, restricted, 1);
ret = write_all(client->fd, "\n", 1);
} else if (parse_command(line, &cmd, &arg)) {
if (!strcmp(cmd, "status")) {
ret = cmd_status(client);
} else {
if (strcmp(cmd, "passwd") != 0) {
set_client_fd(client->fd);
run_parsed_command(cmd, arg);
set_client_fd(-1);
}
ret = write_all(client->fd, "\n", 1);
}
free(cmd);
free(arg);
} else {
// don't hang cmus-remote
ret = write_all(client->fd, "\n", 1);
}
if (ret < 0) {
d_print("write: %s\n", strerror(errno));
goto close;
}
}
memmove(buf, buf + s, pos - s);
pos -= s;
}
return;
close:
close(client->fd);
list_del(&client->node);
free(client);
}
int wmain(int argc, wchar_t* argv[]) {
std::string search_text(argv[2], argv[2] + wcslen(argv[2]));
Test(argv[1], search_text);
return 0;
}
