Matrix URIs and resolving them

CMakeLists.txt

@@ -149,6 +149,8 @@ set(lib_SRCS
     lib/room.cpp
     lib/user.cpp
     lib/avatar.cpp
+    lib/uri.cpp
+    lib/uriresolver.cpp
     lib/syncdata.cpp
     lib/settings.cpp
     lib/networksettings.cpp

lib/quotient_common.h

@@ -14,15 +14,15 @@ enum RunningPolicy { ForegroundRequest = 0x0, BackgroundRequest = 0x1 };
 
 Q_ENUM_NS(RunningPolicy)
 
-enum ResourceResolveResult : short {
+enum UriResolveResult : short {
     StillResolving = -1,
-    Resolved = 0,
-    UnknownMatrixId,
-    MalformedMatrixId,
-    NoAccount,
-    EmptyMatrixId
+    UriResolved = 0,
+    CouldNotResolve,
+    IncorrectAction,
+    InvalidUri,
+    NoAccount
 };
-Q_ENUM_NS(ResourceResolveResult)
+Q_ENUM_NS(UriResolveResult)
 
 } // namespace Quotient
 /// \deprecated Use namespace Quotient instead

lib/uri.cpp

@@ -0,0 +1,191 @@
+#include "uri.h"
+
+#include "logging.h"
+
+#include <QtCore/QRegularExpression>
+
+using namespace Quotient;
+
+struct ReplacePair { QByteArray uriString; char sigil; };
+/// Defines bi-directional mapping of path prefixes and sigils
+static const auto replacePairs = {
+    ReplacePair { "user/", '@' },
+    { "roomid/", '!' },
+    { "room/", '#' },
+    // The notation for bare event ids is not proposed in MSC2312 but there's
+    // https://github.com/matrix-org/matrix-doc/pull/2644
+    { "event/", '$' }
+};
+
+Uri::Uri(QByteArray primaryId, QByteArray secondaryId, QString query)
+{
+    if (primaryId.isEmpty())
+        primaryType_ = Empty;
+    else {
+        setScheme("matrix");
+        QString pathToBe;
+        primaryType_ = Invalid;
+        if (primaryId.size() < 2) // There should be something after sigil
+            return;
+        for (const auto& p: replacePairs)
+            if (primaryId[0] == p.sigil) {
+                primaryType_ = Type(p.sigil);
+                pathToBe = p.uriString + primaryId.mid(1);
+                break;
+            }
+        if (!secondaryId.isEmpty()) {
+            if (secondaryId.size() < 2) {
+                primaryType_ = Invalid;
+                return;
+            }
+            pathToBe += "/event/" + secondaryId.mid(1);
+        }
+        setPath(pathToBe);
+    }
+    setQuery(std::move(query));
+}
+
+Uri::Uri(QUrl url) : QUrl(std::move(url))
+{
+    // NB: don't try to use `url` from here on, it's moved-from and empty
+    if (isEmpty())
+        return; // primaryType_ == Empty
+
+    if (!QUrl::isValid()) { // MatrixUri::isValid() checks primaryType_
+        primaryType_ = Invalid;
+        return;
+    }
+
+    if (scheme() == "matrix") {
+        // Check sanity as per https://github.com/matrix-org/matrix-doc/pull/2312
+        const auto& urlPath = path();
+        const auto& splitPath = urlPath.splitRef('/');
+        switch (splitPath.size()) {
+        case 2:
+            break;
+        case 4:
+            if (splitPath[2] == "event")
+                break;
+            [[fallthrough]];
+        default:
+            return; // Invalid
+        }
+
+        for (const auto& p: replacePairs)
+            if (urlPath.startsWith(p.uriString)) {
+                primaryType_ = Type(p.sigil);
+                return; // The only valid return path for matrix: URIs
+            }
+        qCDebug(MAIN) << "The matrix: URI is not recognised:"
+                      << toDisplayString();
+        return;
+    }
+
+    primaryType_ = NonMatrix; // Default, unless overridden by the code below
+    if (scheme() == "https" && authority() == "matrix.to") {
+        // See https://matrix.org/docs/spec/appendices#matrix-to-navigation
+        static const QRegularExpression MatrixToUrlRE {
+            R"(^/(?<main>[^/?]+)(/(?<sec>[^?]+))?(\?(?<query>.+))?$)"
+        };
+        // matrix.to accepts both literal sigils (as well as & and ? used in
+        // its "query" substitute) and their %-encoded forms;
+        // so force QUrl to decode everything.
+        auto f = fragment(QUrl::FullyDecoded);
+        if (auto&& m = MatrixToUrlRE.match(f); m.hasMatch())
+            *this = Uri { m.captured("main").toUtf8(),
+                          m.captured("sec").toUtf8(), m.captured("query") };
+    }
+}
+
+Uri::Uri(const QString& uriOrId) : Uri(fromUserInput(uriOrId)) {}
+
+Uri Uri::fromUserInput(const QString& uriOrId)
+{
+    if (uriOrId.isEmpty())
+        return {}; // type() == None
+
+    // A quick check if uriOrId is a plain Matrix id
+    // Bare event ids cannot be resolved without a room scope as per the current
+    // spec but there's a movement towards making them navigable (see, e.g.,
+    // https://github.com/matrix-org/matrix-doc/pull/2644) - so treat them
+    // as valid
+    if (QStringLiteral("!@#+$").contains(uriOrId[0]))
+        return Uri { uriOrId.toUtf8() };
+
+    return Uri { QUrl::fromUserInput(uriOrId) };
+}
+
+Uri::Type Uri::type() const { return primaryType_; }
+
+Uri::SecondaryType Uri::secondaryType() const
+{
+    return path().section('/', 2, 2) == "event" ? EventId : NoSecondaryId;
+}
+
+QUrl Uri::toUrl(UriForm form) const
+{
+    if (!isValid())
+        return {};
+
+    if (form == CanonicalUri || type() == NonMatrix)
+        return *this;
+
+    QUrl url;
+    url.setScheme("https");
+    url.setHost("matrix.to");
+    url.setPath("/");
+    auto fragment = primaryId();
+    if (const auto& secId = secondaryId(); !secId.isEmpty())
+        fragment += '/' + secId;
+    if (const auto& q = query(); !q.isEmpty())
+        fragment += '?' + q;
+    url.setFragment(fragment);
+    return url;
+}
+
+QString Uri::primaryId() const
+{
+    if (primaryType_ == Empty || primaryType_ == Invalid)
+        return {};
+
+    const auto& idStem = path().section('/', 1, 1);
+    return idStem.isEmpty() ? idStem : primaryType_ + idStem;
+}
+
+QString Uri::secondaryId() const
+{
+    const auto& idStem = path().section('/', 3);
+    return idStem.isEmpty() ? idStem : secondaryType() + idStem;
+}
+
+static const auto ActionKey = QStringLiteral("action");
+
+QString Uri::action() const
+{
+    return type() == NonMatrix || !isValid()
+               ? QString()
+               : QUrlQuery { query() }.queryItemValue(ActionKey);
+}
+
+void Uri::setAction(const QString& newAction)
+{
+    if (!isValid()) {
+        qCWarning(MAIN) << "Cannot set an action on an invalid Quotient::Uri";
+        return;
+    }
+    QUrlQuery q { query() };
+    q.removeQueryItem(ActionKey);
+    q.addQueryItem(ActionKey, newAction);
+    setQuery(q);
+}
+
+QStringList Uri::viaServers() const
+{
+    return QUrlQuery { query() }.allQueryItemValues(QStringLiteral("via"),
+                                                    QUrl::EncodeReserved);
+}
+
+bool Uri::isValid() const
+{
+    return primaryType_ != Empty && primaryType_ != Invalid;
+}

lib/uri.h

@@ -0,0 +1,85 @@
+#pragma once
+
+#include "quotient_common.h"
+
+#include <QtCore/QUrl>
+#include <QtCore/QUrlQuery>
+
+namespace Quotient {
+
+/*! \brief A wrapper around a Matrix URI or identifier
+ *
+ * This class encapsulates a Matrix resource identifier, passed in either of
+ * 3 forms: a plain Matrix identifier (sigil, localpart, serverpart or, for
+ * modern event ids, sigil and base64 hash); an MSC2312 URI (aka matrix: URI);
+ * or a matrix.to URL. The input can be either encoded (serverparts with
+ * punycode, the rest with percent-encoding) or unencoded (in this case it is
+ * the caller's responsibility to resolve all possible ambiguities).
+ *
+ * The class provides functions to check the validity of the identifier,
+ * its type, and obtain components, also in either unencoded (for displaying)
+ * or encoded (for APIs) form.
+ */
+class Uri : private QUrl {
+    Q_GADGET
+public:
+    enum Type : char {
+        Invalid = char(-1),
+        Empty = 0x0,
+        UserId = '@',
+        RoomId = '!',
+        RoomAlias = '#',
+        Group = '+',
+        BareEventId = '$', // https://github.com/matrix-org/matrix-doc/pull/2644
+        NonMatrix = ':'
+    };
+    Q_ENUM(Type)
+    enum SecondaryType : char { NoSecondaryId = 0x0, EventId = '$' };
+    Q_ENUM(SecondaryType)
+
+    enum UriForm : short { CanonicalUri, MatrixToUri };
+    Q_ENUM(UriForm)
+
+    /// Construct an empty Matrix URI
+    Uri() = default;
+    /*! \brief Decode a user input string to a Matrix identifier
+     *
+     * Accepts plain Matrix ids, MSC2312 URIs (aka matrix: URIs) and
+     * matrix.to URLs. In case of URIs/URLs, it uses QUrl's TolerantMode
+     * parser to decode common mistakes/irregularities (see QUrl documentation
+     * for more details).
+     */
+    Uri(const QString& uriOrId);
+
+    /// Construct a Matrix URI from components
+    explicit Uri(QByteArray primaryId, QByteArray secondaryId = {},
+                 QString query = {});
+    /// Construct a Matrix URI from matrix.to or MSC2312 (matrix:) URI
+    explicit Uri(QUrl url);
+
+    static Uri fromUserInput(const QString& uriOrId);
+    static Uri fromUrl(QUrl url);
+
+    /// Get the primary type of the Matrix URI (user id, room id or alias)
+    /*! Note that this does not include an event as a separate type, since
+     * events can only be addressed inside of rooms, which, in turn, are
+     * addressed either by id or alias. If you need to check whether the URI
+     * is specifically an event URI, use secondaryType() instead.
+     */
+    Q_INVOKABLE Type type() const;
+    Q_INVOKABLE SecondaryType secondaryType() const;
+    Q_INVOKABLE QUrl toUrl(UriForm form = CanonicalUri) const;
+    Q_INVOKABLE QString primaryId() const;
+    Q_INVOKABLE QString secondaryId() const;
+    Q_INVOKABLE QString action() const;
+    Q_INVOKABLE void setAction(const QString& newAction);
+    Q_INVOKABLE QStringList viaServers() const;
+    Q_INVOKABLE bool isValid() const;
+    using QUrl::path, QUrl::query, QUrl::fragment;
+    using QUrl::isEmpty, QUrl::toDisplayString;
+
+private:
+    Type primaryType_ = Empty;
+};
+
+}