void Page::parseTags()
{
// Check redirection
QUrl redir = m_replyTags->attribute(QNetworkRequest::RedirectionTargetAttribute).toUrl();
if (!redir.isEmpty())
{
m_urlRegex = m_site->fixUrl(redir.toString(), m_urlRegex);
loadTags();
return;
}
QString source = QString::fromUtf8(m_replyTags->readAll());
m_tags.clear();
if (m_site->contains("Regex/Tags"))
{
QRegExp rxtags(m_site->value("Regex/Tags"));
rxtags.setMinimal(true);
int p = 0;
QStringList order = m_site->value("Regex/TagsOrder").split('|', QString::SkipEmptyParts);
while ((p = rxtags.indexIn(source, p)) != -1)
{
p += rxtags.matchedLength();
QString type = "", tag = "";
int count = 1;
if (order.empty())
{
switch (rxtags.captureCount())
{
case 4: order << "type" << "" << "count" << "tag"; break;
case 3: order << "type" << "tag" << "count"; break;
case 2: order << "type" << "tag"; break;
case 1: order << "tag"; break;
}
}
for (int o = 0; o < order.size(); o++)
{
if (order.at(o) == "tag" && tag.isEmpty())
{ tag = rxtags.cap(o + 1).replace(" ", "_").replace("&", "&").trimmed(); }
else if (order.at(o) == "type" && type.isEmpty())
{
type = rxtags.cap(o + 1).toLower().trimmed();
if (type.contains(", "))
{ type = type.split(", ").at(0).trimmed(); }
if (type == "series")
{ type = "copyright"; }
else if (type == "mangaka")
{ type = "artist"; }
else if (type == "game")
{ type = "copyright"; }
else if (type == "studio")
{ type = "circle"; }
else if (type == "source")
{ type = "general"; }
else if (type == "character group")
{ type = "general"; }
else if (type.length() == 1)
{
int tpe = type.toInt();
if (tpe >= 0 && tpe <= 4)
{
QStringList types = QStringList() << "general" << "artist" << "unknown" << "copyright" << "character";
type = types[tpe];
}
}
}
else if (order.at(o) == "count" && count != 0)
{ count = rxtags.cap(o + 1).toLower().endsWith('k') ? rxtags.cap(3).left(rxtags.cap(3).length() - 1).toInt() * 1000 : rxtags.cap(3).toInt(); }
}
if (type.isEmpty())
{ type = "unknown"; }
m_tags.append(Tag(tag, type, count));
}
}
// Getting last page
if (m_site->contains("Regex/Count") && m_imagesCount < 1)
{
QRegExp rxlast(m_site->value("Regex/Count"));
rxlast.indexIn(source, 0);
m_imagesCount = rxlast.cap(1).remove(",").toInt();
}
if (m_imagesCount < 1)
{
for (Tag tag : m_tags)
{
if (tag.text() == m_search.join(" "))
{
m_imagesCount = tag.count();
if (m_pagesCount < 0)
m_pagesCount = (int)ceil((m_imagesCount * 1.) / m_imagesPerPage);
}
}
}
if (m_site->contains("Regex/LastPage") && (m_imagesCount < 1 || m_imagesCount % 1000 == 0))
{
QRegExp rxlast(m_site->value("Regex/LastPage"));
rxlast.indexIn(source, 0);
m_pagesCount = rxlast.cap(1).remove(",").toInt();
if (m_pagesCount != 0)
m_imagesCount = m_pagesCount * m_imagesPerPage;
}
// Wiki
m_wiki.clear();
if (m_site->contains("Regex/Wiki"))
{
QRegExp rxwiki(m_site->value("Regex/Wiki"));
rxwiki.setMinimal(true);
if (rxwiki.indexIn(source) != -1)
{
m_wiki = rxwiki.cap(1);
m_wiki.remove("/wiki/show?title=").remove(QRegExp("<p><a href=\"([^\"]+)\">Full entry »</a></p>")).replace("<h6>", "<span class=\"title\">").replace("</h6>", "</span>");
}
}
m_replyTags->deleteLater();
m_replyTagsExists = false;
emit finishedLoadingTags(this);
}
ParsedPage HtmlApi::parsePage(Page *parentPage, const QString &source, int first, int limit) const
{
ParsedPage ret;
// Getting tags
if (contains("Regex/Tags"))
{
QList<Tag> tgs = Tag::FromRegexp(value("Regex/Tags"), source);
if (!tgs.isEmpty())
{ ret.tags = tgs; }
}
// Getting images
QRegularExpression rxImages(value("Regex/Image"), QRegularExpression::DotMatchesEverythingOption);
auto matches = rxImages.globalMatch(source);
int id = 0;
while (matches.hasNext())
{
auto match = matches.next();
QMap<QString, QString> d = multiMatchToMap(match, rxImages.namedCaptureGroups());
// JSON elements
if (d.contains("json") && !d["json"].isEmpty())
{
QVariant src = Json::parse(d["json"]);
if (!src.isNull())
{
QMap<QString, QVariant> map = src.toMap();
for (auto it = map.begin(); it != map.end(); ++it)
{ d[it.key()] = it.value().toString(); }
}
}
QSharedPointer<Image> img = parseImage(parentPage, d, id + first);
if (!img.isNull())
{ ret.images.append(img); }
id++;
}
// Navigation
if (contains("Regex/NextPage"))
{
QRegularExpression rxNextPage(value("Regex/NextPage"));
auto match = rxNextPage.match(source);
if (match.hasMatch())
{ ret.urlNextPage = QUrl(match.captured(1)); }
}
if (contains("Regex/PrevPage"))
{
QRegularExpression rxPrevPage(value("Regex/PrevPage"));
auto match = rxPrevPage.match(source);
if (match.hasMatch())
{ ret.urlPrevPage = QUrl(match.captured(1)); }
}
// Last page
if (contains("LastPage"))
{ ret.pageCount = value("LastPage").toInt(); }
else if (contains("Regex/LastPage"))
{
QRegularExpression rxlast(value("Regex/LastPage"));
auto match = rxlast.match(source);
int cnt = match.hasMatch() ? match.captured(1).remove(",").toInt() : 0;
if (cnt > 0)
{
int pagesCount = cnt;
if (value("Urls/Tags").contains("{pid}") || (contains("Urls/PagePart") && value("Urls/PagePart").contains("{pid}")))
{
int forced = forcedLimit();
int ppid = forced > 0 ? forced : limit;
pagesCount = qFloor(static_cast<qreal>(pagesCount) / static_cast<qreal>(ppid)) + 1;
}
ret.pageCount = pagesCount;
}
}
// Count images
if (contains("Regex/Count"))
{
QRegularExpression rxlast(value("Regex/Count"));
auto match = rxlast.match(source);
int cnt = match.hasMatch() ? match.captured(1).remove(",").toInt() : 0;
if (cnt > 0)
{ ret.imageCount = cnt; }
}
// Wiki
if (contains("Regex/Wiki"))
{
QRegularExpression rxwiki(value("Regex/Wiki"), QRegularExpression::DotMatchesEverythingOption);
auto match = rxwiki.match(source);
if (match.hasMatch())
{
QString wiki = match.captured(1);
wiki.remove("/wiki/show?title=");
wiki.remove(QRegularExpression("<p><a href=\"([^\"]+)\">Full entry »</a></p>"));
wiki.replace("<h6>", "<span class=\"title\">").replace("</h6>", "</span>");
ret.wiki = wiki;
}
}
return ret;
}
