feat: strip prefixes and suffixes in links (#5486)
This commit is contained in:
nerix
+59
-21
@@ -113,19 +113,58 @@ bool startsWithPort(QStringView string)
|
||||
return true;
|
||||
}
|
||||
|
||||
/// @brief Strips ignored characters off @a source
|
||||
///
|
||||
/// As per https://github.github.com/gfm/#autolinks-extension-:
|
||||
///
|
||||
/// '<', '*', '_', '~', and '(' are ignored at the beginning
|
||||
/// '>', '?', '!', '.', ',', ':', '*', '~', and ')' are ignored at the end
|
||||
///
|
||||
/// A difference to GFM is that the source isn't scanned for parentheses and '_'
|
||||
/// isn't a valid suffix.
|
||||
void strip(QStringView &source)
|
||||
{
|
||||
while (!source.isEmpty())
|
||||
{
|
||||
auto c = source.first();
|
||||
if (c == u'<' || c == u'*' || c == u'_' || c == u'~' || c == u'(')
|
||||
{
|
||||
source = source.mid(1);
|
||||
continue;
|
||||
}
|
||||
break;
|
||||
}
|
||||
|
||||
while (!source.isEmpty())
|
||||
{
|
||||
auto c = source.last();
|
||||
if (c == u'>' || c == u'?' || c == u'!' || c == u'.' || c == u',' ||
|
||||
c == u':' || c == u'*' || c == u'~' || c == u')')
|
||||
{
|
||||
source.chop(1);
|
||||
continue;
|
||||
}
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
namespace chatterino {
|
||||
namespace chatterino::linkparser {
|
||||
|
||||
LinkParser::LinkParser(const QString &unparsedString)
|
||||
std::optional<Parsed> parse(const QString &source) noexcept
|
||||
{
|
||||
ParsedLink result;
|
||||
std::optional<Parsed> result;
|
||||
// This is not implemented with a regex to increase performance.
|
||||
QStringView remaining(unparsedString);
|
||||
QStringView protocol(remaining);
|
||||
|
||||
QStringView link{source};
|
||||
strip(link);
|
||||
|
||||
QStringView remaining = link;
|
||||
QStringView protocol;
|
||||
|
||||
// Check protocol for https?://
|
||||
if (remaining.startsWith(QStringLiteral("http"), Qt::CaseInsensitive) &&
|
||||
if (remaining.startsWith(u"http", Qt::CaseInsensitive) &&
|
||||
remaining.length() >= 4 + 3 + 1) // 'http' + '://' + [any]
|
||||
{
|
||||
// optimistic view assuming there's a protocol (http or https)
|
||||
@@ -136,11 +175,11 @@ LinkParser::LinkParser(const QString &unparsedString)
|
||||
withProto = withProto.mid(1);
|
||||
}
|
||||
|
||||
if (withProto.startsWith(QStringLiteral("://")))
|
||||
if (withProto.startsWith(u"://"))
|
||||
{
|
||||
// there's really a protocol => consume it
|
||||
remaining = withProto.mid(3);
|
||||
result.protocol = {protocol.begin(), remaining.begin()};
|
||||
protocol = {link.begin(), remaining.begin()};
|
||||
}
|
||||
}
|
||||
|
||||
@@ -161,7 +200,7 @@ LinkParser::LinkParser(const QString &unparsedString)
|
||||
{
|
||||
if (lastWasDot) // no double dots ..
|
||||
{
|
||||
return;
|
||||
return result;
|
||||
}
|
||||
lastDotPos = i;
|
||||
lastWasDot = true;
|
||||
@@ -181,7 +220,7 @@ LinkParser::LinkParser(const QString &unparsedString)
|
||||
|
||||
if (!startsWithPort(remaining))
|
||||
{
|
||||
return;
|
||||
return result;
|
||||
}
|
||||
|
||||
break;
|
||||
@@ -198,23 +237,22 @@ LinkParser::LinkParser(const QString &unparsedString)
|
||||
|
||||
if (lastWasDot || lastDotPos <= 0)
|
||||
{
|
||||
return;
|
||||
return result;
|
||||
}
|
||||
|
||||
// check host/tld
|
||||
if ((nDots == 3 && isValidIpv4(host)) ||
|
||||
isValidTld(host.mid(lastDotPos + 1)))
|
||||
{
|
||||
result.host = host;
|
||||
result.rest = rest;
|
||||
result.source = unparsedString;
|
||||
this->result_ = std::move(result);
|
||||
result = Parsed{
|
||||
.protocol = protocol,
|
||||
.host = host,
|
||||
.rest = rest,
|
||||
.link = link,
|
||||
};
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
const std::optional<ParsedLink> &LinkParser::result() const
|
||||
{
|
||||
return this->result_;
|
||||
}
|
||||
|
||||
} // namespace chatterino
|
||||
} // namespace chatterino::linkparser
|
||||
|
||||
+102
-19
@@ -4,43 +4,126 @@
|
||||
|
||||
#include <optional>
|
||||
|
||||
namespace chatterino {
|
||||
namespace chatterino::linkparser {
|
||||
|
||||
struct ParsedLink {
|
||||
/// @brief Represents a parsed link
|
||||
///
|
||||
/// A parsed link is represented as views over the source string for its
|
||||
/// different segments. In this simplified model, a link consists of an optional
|
||||
/// @a protocol, a mandatory @a host and an optional @a rest. These segments are
|
||||
/// always next to eachother in the input string, however together, they don't
|
||||
/// span the whole input as it could contain prefixes or suffixes.
|
||||
///
|
||||
/// Prefixes and suffixes are almost identical to the ones in GitHub Flavored
|
||||
/// Markdown (GFM - https://github.github.com/gfm/#autolinks-extension-).
|
||||
/// The main differences are that '_' isn't a valid suffix and parentheses
|
||||
/// aren't counted (e.g. "(a.com/(foo)! would result in "a.com/(foo").
|
||||
/// Matching is done case insensitive (e.g. "HTTp://a.com" would be valid).
|
||||
///
|
||||
/// A @a protocol can either be empty, "http://", or "https://".
|
||||
/// A @a host can either be an IPv4 address or a hostname. The hostname must end
|
||||
/// in a valid top level domain. Otherwise, there are no restrictions on it.
|
||||
/// The @a rest can start with an optional port followed by either a '/', '?',
|
||||
/// or '#'.
|
||||
///
|
||||
/// @b Example
|
||||
///
|
||||
/// ```text
|
||||
/// (https://wiki.chatterino.com/Help/#overview)
|
||||
/// ▏▏proto ▕ host ▏ rest ▏▏
|
||||
/// ▏▏ link ▏▏
|
||||
/// ▏ source ▏
|
||||
/// ```
|
||||
struct Parsed {
|
||||
/// The parsed protocol of the link. Can be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// https://www.forsen.tv/commands
|
||||
/// ^------^
|
||||
/// ▏╌╌╌╌╌╌▕
|
||||
///
|
||||
/// www.forsen.tv/commands
|
||||
/// (empty)
|
||||
/// ```
|
||||
QStringView protocol;
|
||||
|
||||
/// The parsed host of the link. Can not be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// https://www.forsen.tv/commands
|
||||
/// ^-----------^
|
||||
/// ▏╌╌╌╌╌╌╌╌╌╌╌▕
|
||||
/// ```
|
||||
QStringView host;
|
||||
|
||||
/// The remainder of the link. Can be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// https://www.forsen.tv/commands
|
||||
/// ^-------^
|
||||
/// ▏╌╌╌╌╌╌╌▕
|
||||
///
|
||||
/// https://www.forsen.tv
|
||||
/// (empty)
|
||||
/// ```
|
||||
QStringView rest;
|
||||
|
||||
/// The original unparsed link.
|
||||
/// The matched link. Can not be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// (https://www.forsen.tv/commands)
|
||||
/// ▏╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌▕
|
||||
/// ```
|
||||
QStringView link;
|
||||
|
||||
/// Checks if the parsed link contains a prefix
|
||||
bool hasPrefix(const QString &source) const noexcept
|
||||
{
|
||||
return this->link.begin() != source.begin();
|
||||
}
|
||||
|
||||
/// The prefix before the parsed link inside @a source. May be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// (https://www.forsen.tv/commands)
|
||||
/// ^
|
||||
///
|
||||
/// https://www.forsen.tv/commands
|
||||
/// ^----------------------------^
|
||||
QString source;
|
||||
/// (empty)
|
||||
/// ```
|
||||
QStringView prefix(const QString &source) const noexcept
|
||||
{
|
||||
return {source.data(), this->link.begin()};
|
||||
}
|
||||
|
||||
/// Checks if the parsed link contains a suffix
|
||||
bool hasSuffix(const QString &source) const noexcept
|
||||
{
|
||||
return this->link.end() != source.end();
|
||||
}
|
||||
|
||||
/// The suffix after the parsed link inside @a source. May be empty.
|
||||
///
|
||||
/// ```text
|
||||
/// (https://www.forsen.tv/commands)
|
||||
/// ^
|
||||
///
|
||||
/// https://www.forsen.tv/commands
|
||||
/// (empty)
|
||||
/// ```
|
||||
QStringView suffix(const QString &source) const noexcept
|
||||
{
|
||||
return {
|
||||
this->link.begin() + this->link.size(),
|
||||
source.data() + source.length(),
|
||||
};
|
||||
}
|
||||
};
|
||||
|
||||
class LinkParser
|
||||
{
|
||||
public:
|
||||
explicit LinkParser(const QString &unparsedString);
|
||||
/// @brief Parses a link from @a source into its segments
|
||||
///
|
||||
/// If no link is contained in @a source, `std::nullopt` will be returned.
|
||||
/// The returned value is valid as long as @a source exists, as it contains
|
||||
/// views into @a source.
|
||||
///
|
||||
/// For the accepted links, see Parsed.
|
||||
std::optional<Parsed> parse(const QString &source) noexcept;
|
||||
|
||||
const std::optional<ParsedLink> &result() const;
|
||||
|
||||
private:
|
||||
std::optional<ParsedLink> result_{};
|
||||
};
|
||||
|
||||
} // namespace chatterino
|
||||
} // namespace chatterino::linkparser
|
||||
|
||||
@@ -152,10 +152,10 @@ bool appendWhisperMessageWordsLocally(const QStringList &words)
|
||||
void operator()(const QString &string,
|
||||
MessageBuilder &b) const
|
||||
{
|
||||
LinkParser parser(string);
|
||||
if (parser.result())
|
||||
auto link = linkparser::parse(string);
|
||||
if (link)
|
||||
{
|
||||
b.addLink(*parser.result());
|
||||
b.addLink(*link, string);
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
@@ -94,10 +94,10 @@ MessageBuilder::MessageBuilder(SystemMessageTag, const QString &text,
|
||||
text.split(QRegularExpression("\\s"), Qt::SkipEmptyParts);
|
||||
for (const auto &word : textFragments)
|
||||
{
|
||||
LinkParser parser(word);
|
||||
if (parser.result())
|
||||
auto link = linkparser::parse(word);
|
||||
if (link)
|
||||
{
|
||||
this->addLink(*parser.result());
|
||||
this->addLink(*link, word);
|
||||
continue;
|
||||
}
|
||||
|
||||
@@ -637,30 +637,45 @@ std::unique_ptr<MessageElement> MessageBuilder::releaseBack()
|
||||
return ptr;
|
||||
}
|
||||
|
||||
void MessageBuilder::addLink(const ParsedLink &parsedLink)
|
||||
void MessageBuilder::addLink(const linkparser::Parsed &parsedLink,
|
||||
const QString &source)
|
||||
{
|
||||
QString lowercaseLinkString;
|
||||
QString origLink = parsedLink.source;
|
||||
QString origLink = parsedLink.link.toString();
|
||||
QString fullUrl;
|
||||
|
||||
if (parsedLink.protocol.isNull())
|
||||
{
|
||||
fullUrl = QStringLiteral("http://") + parsedLink.source;
|
||||
fullUrl = QStringLiteral("http://") + origLink;
|
||||
}
|
||||
else
|
||||
{
|
||||
lowercaseLinkString += parsedLink.protocol;
|
||||
fullUrl = parsedLink.source;
|
||||
fullUrl = origLink;
|
||||
}
|
||||
|
||||
lowercaseLinkString += parsedLink.host.toString().toLower();
|
||||
lowercaseLinkString += parsedLink.rest;
|
||||
|
||||
auto textColor = MessageColor(MessageColor::Link);
|
||||
|
||||
if (parsedLink.hasPrefix(source))
|
||||
{
|
||||
this->emplace<TextElement>(parsedLink.prefix(source).toString(),
|
||||
MessageElementFlag::Text, this->textColor_)
|
||||
->setTrailingSpace(false);
|
||||
}
|
||||
auto *el = this->emplace<LinkElement>(
|
||||
LinkElement::Parsed{.lowercase = lowercaseLinkString,
|
||||
.original = origLink},
|
||||
fullUrl, MessageElementFlag::Text, textColor);
|
||||
if (parsedLink.hasSuffix(source))
|
||||
{
|
||||
el->setTrailingSpace(false);
|
||||
this->emplace<TextElement>(parsedLink.suffix(source).toString(),
|
||||
MessageElementFlag::Text, this->textColor_);
|
||||
}
|
||||
|
||||
getIApp()->getLinkResolver()->resolve(el->linkInfo());
|
||||
}
|
||||
|
||||
@@ -676,20 +691,18 @@ void MessageBuilder::addIrcMessageText(const QString &text)
|
||||
int fg = -1;
|
||||
int bg = -1;
|
||||
|
||||
for (const auto &word : words)
|
||||
for (const auto &string : words)
|
||||
{
|
||||
if (word.isEmpty())
|
||||
if (string.isEmpty())
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
auto string = QString(word);
|
||||
|
||||
// Actually just text
|
||||
LinkParser parser(string);
|
||||
if (parser.result())
|
||||
auto link = linkparser::parse(string);
|
||||
if (link)
|
||||
{
|
||||
this->addLink(*parser.result());
|
||||
this->addLink(*link, string);
|
||||
continue;
|
||||
}
|
||||
|
||||
@@ -772,15 +785,13 @@ void MessageBuilder::addTextOrEmoji(EmotePtr emote)
|
||||
this->emplace<EmoteElement>(emote, MessageElementFlag::EmojiAll);
|
||||
}
|
||||
|
||||
void MessageBuilder::addTextOrEmoji(const QString &string_)
|
||||
void MessageBuilder::addTextOrEmoji(const QString &string)
|
||||
{
|
||||
auto string = QString(string_);
|
||||
|
||||
// Actually just text
|
||||
LinkParser linkParser(string);
|
||||
if (linkParser.result())
|
||||
auto link = linkparser::parse(string);
|
||||
if (link)
|
||||
{
|
||||
this->addLink(*linkParser.result());
|
||||
this->addLink(*link, string);
|
||||
return;
|
||||
}
|
||||
|
||||
|
||||
@@ -24,7 +24,9 @@ class TextElement;
|
||||
struct Emote;
|
||||
using EmotePtr = std::shared_ptr<const Emote>;
|
||||
|
||||
struct ParsedLink;
|
||||
namespace linkparser {
|
||||
struct Parsed;
|
||||
} // namespace linkparser
|
||||
|
||||
struct SystemMessageTag {
|
||||
};
|
||||
@@ -112,7 +114,7 @@ public:
|
||||
std::weak_ptr<Message> weakOf();
|
||||
|
||||
void append(std::unique_ptr<MessageElement> element);
|
||||
void addLink(const ParsedLink &parsedLink);
|
||||
void addLink(const linkparser::Parsed &parsedLink, const QString &source);
|
||||
|
||||
/**
|
||||
* Adds the text, applies irc colors, adds links,
|
||||
|
||||
@@ -14,7 +14,7 @@ bool LinkPredicate::appliesToImpl(const Message &message)
|
||||
{
|
||||
for (const auto &word : message.messageText.split(' ', Qt::SkipEmptyParts))
|
||||
{
|
||||
if (LinkParser(word).result())
|
||||
if (linkparser::parse(word).has_value())
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
@@ -733,12 +733,12 @@ void TwitchMessageBuilder::addTextOrEmoji(const QString &string_)
|
||||
}
|
||||
|
||||
// Actually just text
|
||||
LinkParser parsed(string);
|
||||
auto link = linkparser::parse(string);
|
||||
auto textColor = this->textColor_;
|
||||
|
||||
if (parsed.result())
|
||||
if (link)
|
||||
{
|
||||
this->addLink(*parsed.result());
|
||||
this->addLink(*link, string);
|
||||
return;
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user