add greasemonkey support

src/browserwindow.cpp

@@ -3,6 +3,7 @@
 #include "adblock.h"
 #include "commandline.h"
 #include "completionview.h"
+#include "greasemonkey.h"
 #include "hints.h"
 #include "history.h"
 #include "statusbar.h"
@@ -61,6 +62,7 @@ BrowserWindow::BrowserWindow(const Session &session, const QUrl &startUrl, QWidg
     connect(m_commandLine, &QLineEdit::textEdited, this, &BrowserWindow::updateCompletion);
     connect(m_commandLine, &CommandLine::completionCycled, this, &BrowserWindow::cycleCompletion);
     connect(m_session.adBlocker, &AdBlocker::message, m_status, &StatusBar::showMessage);
+    connect(m_session.greasemonkey, &Greasemonkey::message, m_status, &StatusBar::showMessage);
 
     qApp->installEventFilter(this);
 
@@ -473,6 +475,8 @@ void BrowserWindow::runCommand(const QString &command)
         } else {
             m_status->showMessage(QStringLiteral("Could not clear history"), true);
         }
+    } else if (name == QLatin1String("greasemonkey-reload")) {
+        m_session.greasemonkey->reload();
     } else if (name == QLatin1String("adblock-update")) {
         m_session.adBlocker->update();
     } else if (name == QLatin1String("adblock-toggle")) {

src/browserwindow.h

@@ -8,6 +8,7 @@
 class AdBlocker;
 class CommandLine;
 class CompletionView;
+class Greasemonkey;
 class History;
 class StatusBar;
 class QWebEngineProfile;
@@ -17,6 +18,7 @@ class QWebEngineView;
 struct Session {
     QWebEngineProfile *profile;
     AdBlocker *adBlocker;
+    Greasemonkey *greasemonkey;
     History *history;
 };
 

src/greasemonkey.cpp

@@ -0,0 +1,425 @@
+#include "greasemonkey.h"
+
+#include <QCoreApplication>
+#include <QCryptographicHash>
+#include <QFile>
+#include <QJsonArray>
+#include <QJsonDocument>
+#include <QJsonObject>
+#include <QNetworkAccessManager>
+#include <QNetworkReply>
+#include <QRegularExpression>
+#include <QSaveFile>
+#include <QStandardPaths>
+#include <QWebEngineProfile>
+#include <QWebEngineScript>
+#include <QWebEngineScriptCollection>
+
+#include <memory>
+
+namespace {
+
+constexpr auto kScriptNamePrefix = "att-greasemonkey:";
+
+// Runs around every userscript, in the page's main world. The GM functions are
+// locals of the wrapper, so the page itself can't see or call them.
+//
+// GM_setValue & co. are backed by the page's localStorage under a per-script
+// prefix, as in qutebrowser: values are per site, not shared across sites as in
+// browser-extension userscript managers. GM_xmlhttpRequest is a plain XHR, so
+// it is subject to CORS.
+constexpr auto kWrapperHead = R"JS((function () {
+'use strict';
+const GM_info = @INFO@;
+const __attStorePrefix = '__att_gm__' + @KEY@ + '__';
+const __attMemoryStore = new Map();
+
+function __attStore() {
+    try {
+        const s = window.localStorage;
+        s.length; // throws on opaque origins
+        return s;
+    } catch (e) {
+        return {
+            getItem: k => (__attMemoryStore.has(k) ? __attMemoryStore.get(k) : null),
+            setItem: (k, v) => __attMemoryStore.set(k, String(v)),
+            removeItem: k => __attMemoryStore.delete(k),
+            get length() { return __attMemoryStore.size; },
+            key: i => Array.from(__attMemoryStore.keys())[i] ?? null,
+        };
+    }
+}
+
+function GM_setValue(key, value) {
+    __attStore().setItem(__attStorePrefix + key, JSON.stringify(value));
+}
+function GM_getValue(key, defaultValue) {
+    const raw = __attStore().getItem(__attStorePrefix + key);
+    if (raw === null)
+        return defaultValue;
+    try {
+        return JSON.parse(raw);
+    } catch (e) {
+        return defaultValue;
+    }
+}
+function GM_deleteValue(key) {
+    __attStore().removeItem(__attStorePrefix + key);
+}
+function GM_listValues() {
+    const store = __attStore();
+    const keys = [];
+    for (let i = 0; i < store.length; i++) {
+        const k = store.key(i);
+        if (k && k.startsWith(__attStorePrefix))
+            keys.push(k.slice(__attStorePrefix.length));
+    }
+    return keys;
+}
+
+function GM_addStyle(css) {
+    const style = document.createElement('style');
+    style.textContent = css;
+    const parent = document.head || document.documentElement;
+    if (parent) {
+        parent.appendChild(style);
+    } else {
+        // @run-at document-start runs before <html> exists, which is exactly
+        // when style scripts want to run, to avoid a flash of unstyled page.
+        new MutationObserver((_, observer) => {
+            const p = document.head || document.documentElement;
+            if (p) {
+                observer.disconnect();
+                p.appendChild(style);
+            }
+        }).observe(document, { childList: true, subtree: true });
+    }
+    return style;
+}
+
+function GM_addElement(parent, tag, attributes) {
+    if (typeof parent === 'string') {
+        attributes = tag;
+        tag = parent;
+        parent = document.head || document.documentElement;
+    }
+    const el = document.createElement(tag);
+    for (const [k, v] of Object.entries(attributes || {})) {
+        if (k === 'textContent')
+            el.textContent = v;
+        else
+            el.setAttribute(k, v);
+    }
+    parent.appendChild(el);
+    return el;
+}
+
+function GM_openInTab(url) {
+    return window.open(url, '_blank');
+}
+
+function GM_setClipboard(text) {
+    navigator.clipboard?.writeText(String(text)).catch(() => {});
+}
+
+function GM_log(...args) {
+    console.log(...args);
+}
+
+function GM_notification(details, title) {
+    const text = typeof details === 'string' ? details : details.text;
+    console.info('[' + (title || GM_info.script.name) + '] ' + text);
+}
+
+// Userscript managers add these to their extension menu; att has none, but
+// scripts commonly call them unconditionally.
+function GM_registerMenuCommand() { return 0; }
+function GM_unregisterMenuCommand() {}
+
+function GM_getResourceText() { return null; }
+function GM_getResourceURL() { return null; }
+
+function GM_xmlhttpRequest(details) {
+    const xhr = new XMLHttpRequest();
+    const response = () => ({
+        readyState: xhr.readyState,
+        responseHeaders: xhr.getAllResponseHeaders(),
+        responseText: xhr.responseType === '' || xhr.responseType === 'text' ? xhr.responseText : '',
+        response: xhr.response,
+        status: xhr.status,
+        statusText: xhr.statusText,
+        finalUrl: xhr.responseURL,
+        context: details.context,
+    });
+    const on = name => () => details[name] && details[name](response());
+    xhr.onload = on('onload');
+    xhr.onerror = on('onerror');
+    xhr.onabort = on('onabort');
+    xhr.ontimeout = on('ontimeout');
+    xhr.onprogress = on('onprogress');
+    xhr.onreadystatechange = on('onreadystatechange');
+    xhr.open(details.method || 'GET', details.url, true, details.user, details.password);
+    if (details.responseType)
+        xhr.responseType = details.responseType;
+    if (details.timeout)
+        xhr.timeout = details.timeout;
+    if (details.overrideMimeType)
+        xhr.overrideMimeType(details.overrideMimeType);
+    for (const [k, v] of Object.entries(details.headers || {}))
+        xhr.setRequestHeader(k, v);
+    xhr.send(details.data);
+    return { abort: () => xhr.abort() };
+}
+
+// The promise-based Greasemonkey 4 API.
+const GM = {
+    info: GM_info,
+    setValue: async (k, v) => GM_setValue(k, v),
+    getValue: async (k, d) => GM_getValue(k, d),
+    deleteValue: async k => GM_deleteValue(k),
+    listValues: async () => GM_listValues(),
+    addStyle: async css => GM_addStyle(css),
+    addElement: (...a) => GM_addElement(...a),
+    openInTab: async url => GM_openInTab(url),
+    setClipboard: async text => GM_setClipboard(text),
+    notification: async (d, t) => GM_notification(d, t),
+    registerMenuCommand: async () => 0,
+    getResourceUrl: async () => null,
+    xmlHttpRequest: details => new Promise((resolve, reject) => {
+        GM_xmlhttpRequest(Object.assign({}, details, {
+            onload: r => { details.onload && details.onload(r); resolve(r); },
+            onerror: r => { details.onerror && details.onerror(r); reject(r); },
+        }));
+    }),
+};
+
+// The scripts already run in the page's own world.
+const unsafeWindow = window;
+
+)JS";
+
+constexpr auto kWrapperTail = R"JS(
+})();
+)JS";
+
+QJsonArray toJson(const QStringList &list)
+{
+    return QJsonArray::fromStringList(list);
+}
+
+QString jsString(const QString &s)
+{
+    // A JSON string is a valid JS string literal. Wrap it in an array to get
+    // QJsonDocument to serialize it, then strip the brackets.
+    const QByteArray json = QJsonDocument(QJsonArray{s}).toJson(QJsonDocument::Compact);
+    return QString::fromUtf8(json.mid(1, json.size() - 2));
+}
+
+QString wrap(const Greasemonkey::Metadata &meta, const QString &requiredCode, const QString &source)
+{
+    QJsonObject script{
+        {QStringLiteral("name"), meta.name},
+        {QStringLiteral("namespace"), meta.nameSpace},
+        {QStringLiteral("description"), meta.description},
+        {QStringLiteral("version"), meta.version},
+        {QStringLiteral("includes"), toJson(meta.includes)},
+        {QStringLiteral("excludes"), toJson(meta.excludes)},
+        {QStringLiteral("matches"), toJson(meta.matches)},
+        {QStringLiteral("run-at"), meta.runAt},
+    };
+    QJsonObject info{
+        {QStringLiteral("script"), script},
+        {QStringLiteral("scriptMetaStr"), meta.block},
+        {QStringLiteral("scriptHandler"), QStringLiteral("att")},
+        {QStringLiteral("version"), QCoreApplication::applicationVersion()},
+    };
+
+    QString head = QString::fromUtf8(kWrapperHead);
+    head.replace(QLatin1String("@INFO@"),
+                 QString::fromUtf8(QJsonDocument(info).toJson(QJsonDocument::Compact)));
+    head.replace(QLatin1String("@KEY@"), jsString(meta.nameSpace + QLatin1Char('/') + meta.name));
+
+    // Newlines between the parts: a script may end in a // comment without one,
+    // which would otherwise swallow the closing "})();".
+    return head + requiredCode + QLatin1Char('\n') + source + QString::fromUtf8(kWrapperTail);
+}
+
+QWebEngineScript::InjectionPoint injectionPoint(const QString &runAt)
+{
+    if (runAt == QLatin1String("document-start"))
+        return QWebEngineScript::DocumentCreation;
+    if (runAt == QLatin1String("document-idle"))
+        return QWebEngineScript::Deferred;
+    return QWebEngineScript::DocumentReady; // Greasemonkey's default.
+}
+
+} // namespace
+
+Greasemonkey::Metadata Greasemonkey::parseMetadata(const QString &source)
+{
+    Metadata meta;
+    meta.runAt = QStringLiteral("document-end");
+
+    static const QRegularExpression line(QStringLiteral(R"(^\s*//\s*@(\S+)(?:[ \t]+(.*?))?\s*$)"));
+
+    bool inBlock = false;
+    QStringList block;
+    for (const QStringView raw : QStringView(source).split(QLatin1Char('\n'))) {
+        const QString text = raw.trimmed().toString();
+        if (!inBlock) {
+            if (text.startsWith(QLatin1String("// ==UserScript==")))
+                inBlock = true;
+            continue;
+        }
+        if (text.startsWith(QLatin1String("// ==/UserScript==")))
+            break;
+        block << text;
+
+        const QRegularExpressionMatch m = line.match(text);
+        if (!m.hasMatch())
+            continue;
+        const QString key = m.captured(1);
+        const QString value = m.captured(2);
+        if (key == QLatin1String("name"))
+            meta.name = value;
+        else if (key == QLatin1String("namespace"))
+            meta.nameSpace = value;
+        else if (key == QLatin1String("description"))
+            meta.description = value;
+        else if (key == QLatin1String("version"))
+            meta.version = value;
+        else if (key == QLatin1String("run-at"))
+            meta.runAt = value;
+        else if (key == QLatin1String("noframes"))
+            meta.noFrames = true;
+        else if (key == QLatin1String("include"))
+            meta.includes << value;
+        else if (key == QLatin1String("exclude"))
+            meta.excludes << value;
+        else if (key == QLatin1String("match"))
+            meta.matches << value;
+        else if (key == QLatin1String("require"))
+            meta.requireUrls << value;
+    }
+    meta.block = block.join(QLatin1Char('\n'));
+    return meta;
+}
+
+Greasemonkey::Greasemonkey(QWebEngineProfile *profile, QObject *parent)
+    : QObject(parent)
+    , m_profile(profile)
+    , m_network(new QNetworkAccessManager(this))
+    , m_dir(QStandardPaths::writableLocation(QStandardPaths::AppDataLocation)
+            + QStringLiteral("/greasemonkey"))
+{
+    m_dir.mkpath(QStringLiteral("requires"));
+}
+
+QString Greasemonkey::requirePath(const QString &url) const
+{
+    const QByteArray hash = QCryptographicHash::hash(url.toUtf8(), QCryptographicHash::Sha1).toHex();
+    return m_dir.filePath(QStringLiteral("requires/") + QString::fromLatin1(hash) + QStringLiteral(".js"));
+}
+
+void Greasemonkey::reload(bool announce)
+{
+    // A reload that starts while an older one is still downloading wins.
+    const int generation = ++m_generation;
+
+    QList<Script> scripts;
+    QStringList missing;
+    const QStringList files = m_dir.entryList({QStringLiteral("*.js")}, QDir::Files, QDir::Name);
+    for (const QString &fileName : files) {
+        QFile file(m_dir.filePath(fileName));
+        if (!file.open(QIODevice::ReadOnly)) {
+            emit message(QStringLiteral("Greasemonkey: cannot read %1").arg(fileName), true);
+            continue;
+        }
+        Script script{fileName, QString::fromUtf8(file.readAll()), {}};
+        script.meta = parseMetadata(script.source);
+        if (script.meta.name.isEmpty())
+            script.meta.name = fileName;
+        for (const QString &url : std::as_const(script.meta.requireUrls)) {
+            if (!QFile::exists(requirePath(url)) && !missing.contains(url))
+                missing << url;
+        }
+        scripts << script;
+    }
+
+    if (missing.isEmpty()) {
+        install(scripts, {}, announce);
+        return;
+    }
+
+    emit message(QStringLiteral("Greasemonkey: downloading %1 @require file(s)...").arg(missing.size()),
+                 false);
+    auto remaining = std::make_shared<int>(int(missing.size()));
+    auto failed = std::make_shared<QStringList>();
+    for (const QString &url : std::as_const(missing)) {
+        QNetworkRequest request{QUrl(url)};
+        request.setAttribute(QNetworkRequest::RedirectPolicyAttribute,
+                             QNetworkRequest::NoLessSafeRedirectPolicy);
+        QNetworkReply *reply = m_network->get(request);
+        connect(reply, &QNetworkReply::finished, this,
+                [this, reply, url, remaining, failed, scripts, generation, announce] {
+                    reply->deleteLater();
+                    QSaveFile file(requirePath(url));
+                    if (reply->error() != QNetworkReply::NoError || !file.open(QIODevice::WriteOnly)
+                        || file.write(reply->readAll()) < 0 || !file.commit())
+                        *failed << url;
+                    if (--*remaining == 0 && generation == m_generation)
+                        install(scripts, *failed, announce);
+                });
+    }
+}
+
+void Greasemonkey::install(const QList<Script> &scripts, const QStringList &failedRequires,
+                           bool announce)
+{
+    QWebEngineScriptCollection *collection = m_profile->scripts();
+    const QList<QWebEngineScript> all = collection->toList();
+    for (const QWebEngineScript &old : all) {
+        if (old.name().startsWith(QLatin1String(kScriptNamePrefix)))
+            collection->remove(old);
+    }
+
+    int installed = 0;
+    QStringList skipped;
+    for (const Script &script : scripts) {
+        QString requiredCode;
+        bool complete = true;
+        for (const QString &url : script.meta.requireUrls) {
+            QFile file(requirePath(url));
+            if (failedRequires.contains(url) || !file.open(QIODevice::ReadOnly)) {
+                complete = false;
+                break;
+            }
+            requiredCode += QString::fromUtf8(file.readAll()) + QLatin1Char('\n');
+        }
+        // Running a script without the library it needs would just throw.
+        if (!complete) {
+            skipped << script.meta.name;
+            continue;
+        }
+
+        QWebEngineScript web;
+        web.setName(QString::fromLatin1(kScriptNamePrefix) + script.fileName);
+        web.setWorldId(QWebEngineScript::MainWorld);
+        // setSourceCode first: QtWebEngine parses the ==UserScript== block here,
+        // keeping @include/@exclude/@match for Chromium to check per page, and
+        // would overwrite an injection point set earlier.
+        web.setSourceCode(wrap(script.meta, requiredCode, script.source));
+        web.setInjectionPoint(injectionPoint(script.meta.runAt));
+        web.setRunsOnSubFrames(!script.meta.noFrames);
+        collection->insert(web);
+        ++installed;
+    }
+
+    if (!skipped.isEmpty())
+        emit message(QStringLiteral("Greasemonkey: skipped %1 (an @require failed to download)")
+                         .arg(skipped.join(QLatin1String(", "))),
+                     true);
+    else if (announce)
+        emit message(QStringLiteral("Greasemonkey: loaded %1 script(s); reload pages to apply").arg(installed),
+                     false);
+}

src/greasemonkey.h

@@ -0,0 +1,68 @@
+#pragma once
+
+#include <QDir>
+#include <QList>
+#include <QObject>
+#include <QString>
+#include <QStringList>
+
+class QNetworkAccessManager;
+class QWebEngineProfile;
+
+// Greasemonkey userscripts, the way qutebrowser does them: every *.js file in
+// <AppDataLocation>/greasemonkey is injected into the shared profile, so it runs
+// in every window.
+//
+// URL matching (@include, @exclude, @match) is left to QtWebEngine, which
+// parses the ==UserScript== block itself and checks it inside Chromium. att
+// parses the block too, for @run-at, @noframes and @require, and for GM_info.
+// Scripts run in the page's main world, wrapped in a function that provides
+// the common GM_* / GM.* API (see kWrapper in greasemonkey.cpp).
+class Greasemonkey : public QObject
+{
+    Q_OBJECT
+
+public:
+    struct Metadata {
+        QString name;
+        QString nameSpace;
+        QString description;
+        QString version;
+        QString runAt; // document-start, document-end (default) or document-idle
+        bool noFrames = false;
+        QStringList includes;
+        QStringList excludes;
+        QStringList matches;
+        QStringList requireUrls;
+        QString block; // the raw ==UserScript== block, for GM_info.scriptMetaStr
+    };
+
+    static Metadata parseMetadata(const QString &source);
+
+    Greasemonkey(QWebEngineProfile *profile, QObject *parent = nullptr);
+
+    QString directory() const { return m_dir.path(); }
+
+    // Re-reads the scripts. @require'd files are downloaded once and cached
+    // under greasemonkey/requires; scripts are installed when they are in.
+    // Errors are always reported; success only if announce is set.
+    void reload(bool announce = true);
+
+signals:
+    void message(const QString &text, bool isError);
+
+private:
+    struct Script {
+        QString fileName;
+        QString source;
+        Metadata meta;
+    };
+
+    QString requirePath(const QString &url) const;
+    void install(const QList<Script> &scripts, const QStringList &failedRequires, bool announce);
+
+    QWebEngineProfile *m_profile;
+    QNetworkAccessManager *m_network;
+    QDir m_dir;
+    int m_generation = 0;
+};

src/main.cpp

@@ -1,6 +1,7 @@
 #include "adblock.h"
 #include "browserwindow.h"
 #include "gpuworkaround.h"
+#include "greasemonkey.h"
 #include "history.h"
 #include "urlutil.h"
 
@@ -85,12 +86,15 @@ int main(int argc, char *argv[])
 
     // Shared by every window; after QApplication, which QtSql needs.
     History history;
-    const Session session{profile, adBlocker, &history};
+    auto *greasemonkey = new Greasemonkey(profile, &app);
+    const Session session{profile, adBlocker, greasemonkey, &history};
 
     // Windows delete themselves on close; the app quits with the last one.
     (new BrowserWindow(session, startUrl))->show();
-    // After the window exists, so its status bar shows the update messages.
+    // After the window exists, so its status bar shows their messages. The
+    // scripts are in place before the event loop starts the first page load.
     adBlocker->start();
+    greasemonkey->reload(false);
 
     const int status = app.exec();