#include "myhtmlparse.h"
#include "indextext.h" // for lowercase_term()
#include "mimeparse.h"
// The original version for this compresses whitespace and suppresses newlines
// I can see no good reason to do this, and it actually helps preview to keep
// whitespace, especially if the html comes from a filter that generated it
// from text (ie: inside ' tags)
void
MyHtmlParser::process_text(const string &text)
{
if (!in_script_tag && !in_style_tag) {
#if 0
string::size_type b = 0;
while ((b = text.find_first_not_of(WHITESPACE, b)) != string::npos) {
if (pending_space || b != 0)
if (!dump.empty()) dump += ' ';
pending_space = true;
string::size_type e = text.find_first_of(WHITESPACE, b);
if (e == string::npos) {
dump += text.substr(b);
pending_space = false;
break;
}
dump += text.substr(b, e - b);
b = e + 1;
}
#else
if (pending_space)
dump += ' ';
dump += text;
#endif
}
}
void
MyHtmlParser::opening_tag(const string &tag, const map &p)
{
#if 0
cout << "TAG: " << tag << ": " << endl;
map::const_iterator x;
for (x = p.begin(); x != p.end(); x++) {
cout << " " << x->first << " -> '" << x->second << "'" << endl;
}
#endif
if (tag.empty()) return;
switch (tag[0]) {
case 'a':
if (tag == "address") pending_space = true;
break;
case 'b':
if (tag == "body") {
dump = "";
break;
}
if (tag == "blockquote" || tag == "br") {
dump += '\n';
pending_space = true;
}
break;
case 'c':
if (tag == "center") pending_space = true;
break;
case 'd':
if (tag == "dd" || tag == "dir" || tag == "div" || tag == "dl" ||
tag == "dt") pending_space = true;
if (tag == "dt")
dump += '\n';
break;
case 'e':
if (tag == "embed") pending_space = true;
break;
case 'f':
if (tag == "fieldset" || tag == "form") pending_space = true;
break;
case 'h':
// hr, and h1, ..., h6
if (tag.length() == 2 && strchr("r123456", tag[1])) {
dump += '\n';
pending_space = true;
}
break;
case 'i':
if (tag == "iframe" || tag == "img" || tag == "isindex" ||
tag == "input") pending_space = true;
break;
case 'k':
if (tag == "keygen") pending_space = true;
break;
case 'l':
if (tag == "legend" || tag == "li" || tag == "listing") {
dump += '\n';
pending_space = true;
}
break;
case 'm':
if (tag == "meta") {
map::const_iterator i, j;
if ((i = p.find("content")) != p.end()) {
if ((j = p.find("name")) != p.end()) {
string name = j->second;
lowercase_term(name);
if (name == "description") {
if (sample.empty()) {
sample = i->second;
decode_entities(sample);
}
} else if (name == "keywords") {
if (!keywords.empty()) keywords += ' ';
string tmp = i->second;
decode_entities(tmp);
keywords += tmp;
} else if (name == "date") {
// Yes this doesnt exist. It's output by filters
// And the format isn't even standard http/html
// FIXME
string tmp = i->second;
decode_entities(tmp);
struct tm tm;
if (strptime(tmp.c_str(),
" %Y-%m-%d %H:%M:%S ", &tm)) {
char ascuxtime[100];
sprintf(ascuxtime, "%ld", (long)mktime(&tm));
dmtime = ascuxtime;
}
} else if (name == "robots") {
string val = i->second;
decode_entities(val);
lowercase_term(val);
if (val.find("none") != string::npos ||
val.find("noindex") != string::npos) {
indexing_allowed = false;
throw true;
}
}
} else if ((j = p.find("http-equiv")) != p.end()) {
string hequiv = j->second;
lowercase_term(hequiv);
if (hequiv == "content-type") {
string value = i->second;
MimeHeaderValue p;
parseMimeHeaderValue(value, p);
map::const_iterator k;
if ((k = p.params.find("charset")) !=
p.params.end()) {
doccharset = k->second;
if (doccharset != ocharset) {
LOGDEB1(("Doc specified charset '%s' "
"differs from announced '%s'\n",
doccharset.c_str(),
ocharset.c_str()));
throw true;
}
}
}
}
}
break;
}
if (tag == "marquee" || tag == "menu" || tag == "multicol")
pending_space = true;
break;
case 'o':
if (tag == "ol" || tag == "option") pending_space = true;
break;
case 'p':
if (tag == "p" || tag == "pre" || tag == "plaintext") {
dump += '\n';
pending_space = true;
}
break;
case 'q':
if (tag == "q") pending_space = true;
break;
case 's':
if (tag == "style") {
in_style_tag = true;
break;
}
if (tag == "script") {
in_script_tag = true;
break;
}
if (tag == "select") pending_space = true;
break;
case 't':
if (tag == "table" || tag == "td" || tag == "textarea" ||
tag == "th") pending_space = true;
break;
case 'u':
if (tag == "ul") pending_space = true;
break;
case 'x':
if (tag == "xmp") pending_space = true;
break;
}
}
void
MyHtmlParser::closing_tag(const string &tag)
{
if (tag.empty()) return;
switch (tag[0]) {
case 'a':
if (tag == "address") pending_space = true;
break;
case 'b':
if (tag == "body") {
throw true;
}
if (tag == "blockquote" || tag == "br") pending_space = true;
break;
case 'c':
if (tag == "center") pending_space = true;
break;
case 'd':
if (tag == "dd" || tag == "dir" || tag == "div" || tag == "dl" ||
tag == "dt") pending_space = true;
break;
case 'f':
if (tag == "fieldset" || tag == "form") pending_space = true;
break;
case 'h':
// hr, and h1, ..., h6
if (tag.length() == 2 && strchr("r123456", tag[1]))
pending_space = true;
break;
case 'i':
if (tag == "iframe") pending_space = true;
break;
case 'l':
if (tag == "legend" || tag == "li" || tag == "listing")
pending_space = true;
break;
case 'm':
if (tag == "marquee" || tag == "menu") pending_space = true;
break;
case 'o':
if (tag == "ol" || tag == "option") pending_space = true;
break;
case 'p':
if (tag == "p" || tag == "pre") pending_space = true;
break;
case 'q':
if (tag == "q") pending_space = true;
break;
case 's':
if (tag == "style") {
in_style_tag = false;
break;
}
if (tag == "script") {
in_script_tag = false;
break;
}
if (tag == "select") pending_space = true;
break;
case 't':
if (tag == "title") {
if (title.empty()) {
title = dump;
dump = "";
}
break;
}
if (tag == "table" || tag == "td" || tag == "textarea" ||
tag == "th") pending_space = true;
break;
case 'u':
if (tag == "ul") pending_space = true;
break;
case 'x':
if (tag == "xmp") pending_space = true;
break;
}
}