| 1 | /* |
|---|---|
| 2 | SPDX-FileCopyrightText: 2012 Vishesh Handa <me@vhanda.in> |
| 3 | SPDX-FileCopyrightText: 2016 Varun Joshi <varunj.1011@gmail.com> |
| 4 | |
| 5 | SPDX-License-Identifier: LGPL-2.1-or-later |
| 6 | */ |
| 7 | |
| 8 | #include "extractorcollection.h" |
| 9 | #include "extractor_p.h" |
| 10 | #include "extractorplugin.h" |
| 11 | #include "externalextractor.h" |
| 12 | #include "kfilemetadata_debug.h" |
| 13 | #include "config-kfilemetadata.h" |
| 14 | |
| 15 | #include <KPluginMetaData> |
| 16 | #include <QCoreApplication> |
| 17 | #include <QDir> |
| 18 | #include <QMimeDatabase> |
| 19 | #include <vector> |
| 20 | |
| 21 | using namespace KFileMetaData; |
| 22 | |
| 23 | class KFileMetaData::ExtractorCollectionPrivate |
| 24 | { |
| 25 | public: |
| 26 | QMultiHash<QString, Extractor*> m_mimeExtractors; |
| 27 | |
| 28 | std::vector<Extractor> m_allExtractors; |
| 29 | |
| 30 | void findExtractors(); |
| 31 | QList<Extractor*> getExtractors(const QString& mimetype); |
| 32 | }; |
| 33 | |
| 34 | ExtractorCollection::ExtractorCollection() |
| 35 | : d(new ExtractorCollectionPrivate) |
| 36 | { |
| 37 | d->findExtractors(); |
| 38 | } |
| 39 | |
| 40 | ExtractorCollection::~ExtractorCollection() = default; |
| 41 | |
| 42 | |
| 43 | QList<Extractor*> ExtractorCollection::allExtractors() |
| 44 | { |
| 45 | QList<Extractor*> plugins; |
| 46 | for (auto& ex : d->m_allExtractors) { |
| 47 | if (ex.d->initPlugin()) { |
| 48 | plugins.push_back(t: &ex); |
| 49 | } |
| 50 | } |
| 51 | return plugins; |
| 52 | } |
| 53 | |
| 54 | void ExtractorCollectionPrivate::findExtractors() |
| 55 | { |
| 56 | const QList<KPluginMetaData> kfilemetadataPlugins = |
| 57 | KPluginMetaData::findPlugins(QStringLiteral("kf6/kfilemetadata"), filter: {}, options: KPluginMetaData::AllowEmptyMetaData); |
| 58 | for (const KPluginMetaData &plugin : kfilemetadataPlugins) { |
| 59 | Extractor extractor; |
| 60 | extractor.d->m_pluginPath = plugin.fileName(); |
| 61 | extractor.setAutoDeletePlugin(Extractor::DoNotDeletePlugin); |
| 62 | |
| 63 | if (!plugin.rawData().isEmpty()) { |
| 64 | const auto metadataMap = plugin.rawData().toVariantMap(); |
| 65 | if (!metadataMap.contains(key: QLatin1String("Name"))) { |
| 66 | qCWarning(KFILEMETADATA_LOG) << "Found plugin with invalid metadata (missing Name):"<< extractor.d->m_pluginPath; |
| 67 | } else if (!metadataMap.contains(key: QLatin1String("MimeTypes"))) { |
| 68 | qCWarning(KFILEMETADATA_LOG) << "Found plugin with invalid metadata (missing MimeTypes):"<< extractor.d->m_pluginPath; |
| 69 | } else { |
| 70 | qCDebug(KFILEMETADATA_LOG) << "Found plugin with metadata:"<< extractor.d->m_pluginPath; |
| 71 | extractor.setMetaData(metadataMap); |
| 72 | m_allExtractors.push_back(x: std::move(extractor)); |
| 73 | } |
| 74 | } else { |
| 75 | qCWarning(KFILEMETADATA_LOG) << "Found plugin without metadata (deprecated):"<< extractor.d->m_pluginPath; |
| 76 | if (extractor.d->initPlugin() && !extractor.mimetypes().isEmpty()) { |
| 77 | m_allExtractors.push_back(x: std::move(extractor)); |
| 78 | } |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | QStringList externalPlugins; |
| 83 | const QDir externalPluginDir(QStringLiteral(LIBEXEC_INSTALL_DIR "/kfilemetadata/externalextractors")); |
| 84 | qCDebug(KFILEMETADATA_LOG) << "Searching for external extractors:"<< externalPluginDir.path(); |
| 85 | // For external plugins, we look into the directories |
| 86 | const QStringList externalPluginEntryList = externalPluginDir.entryList(filters: QDir::Dirs | QDir::NoDotAndDotDot); |
| 87 | for (const QString &externalPlugin : externalPluginEntryList) { |
| 88 | if (externalPlugins.contains(str: externalPlugin)) { |
| 89 | qCDebug(KFILEMETADATA_LOG) << "Skipping duplicate - "<< externalPluginDir.path() << ":"<< externalPlugin; |
| 90 | continue; |
| 91 | } |
| 92 | |
| 93 | qCDebug(KFILEMETADATA_LOG) << "Adding plugin - "<< externalPluginDir.path() << ":"<< externalPlugin; |
| 94 | externalPlugins << externalPlugin; |
| 95 | |
| 96 | Extractor extractor; |
| 97 | auto pluginPath = externalPluginDir.absoluteFilePath(fileName: externalPlugin); |
| 98 | ExternalExtractor *plugin = new ExternalExtractor(pluginPath); |
| 99 | if (plugin && !plugin->mimetypes().isEmpty()) { |
| 100 | extractor.setExtractorPlugin(plugin); |
| 101 | extractor.setAutoDeletePlugin(Extractor::AutoDeletePlugin); |
| 102 | m_allExtractors.push_back(x: std::move(extractor)); |
| 103 | } |
| 104 | } |
| 105 | externalPlugins.clear(); |
| 106 | |
| 107 | for (Extractor& extractor : m_allExtractors) { |
| 108 | auto pluginProperties = extractor.extractorProperties(); |
| 109 | if (!pluginProperties.isEmpty()) { |
| 110 | auto mimetypeProperties = pluginProperties[QLatin1String("MimeTypes")]; |
| 111 | if (mimetypeProperties.metaType() != QMetaType(QMetaType::QVariantMap)) { |
| 112 | qCWarning(KFILEMETADATA_LOG) << "Invalid MimeTypes property in"<< pluginProperties; |
| 113 | } else { |
| 114 | const auto mimetypes = mimetypeProperties.toMap(); |
| 115 | for (const auto [mimetype, v] : mimetypes.asKeyValueRange()) { |
| 116 | m_mimeExtractors.insert(key: mimetype, value: &extractor); |
| 117 | } |
| 118 | } |
| 119 | } else if (extractor.d->m_plugin) { |
| 120 | const auto mimetypes = extractor.mimetypes(); |
| 121 | for (const QString &mimetype : mimetypes) { |
| 122 | m_mimeExtractors.insert(key: mimetype, value: &extractor); |
| 123 | } |
| 124 | } |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | QList<Extractor*> ExtractorCollectionPrivate::getExtractors(const QString& mimetype) |
| 129 | { |
| 130 | QList<Extractor*> extractors = m_mimeExtractors.values(key: mimetype); |
| 131 | |
| 132 | if (extractors.isEmpty()) { |
| 133 | qCDebug(KFILEMETADATA_LOG) << "No extractor for"<< mimetype; |
| 134 | return extractors; |
| 135 | } |
| 136 | |
| 137 | qCDebug(KFILEMETADATA_LOG) << "Fetching extractors for"<< mimetype; |
| 138 | Extractor* failed = nullptr; |
| 139 | for (auto ex : extractors) { |
| 140 | if (!ex->d->initPlugin()) { |
| 141 | failed = ex; |
| 142 | break; |
| 143 | } |
| 144 | } |
| 145 | |
| 146 | if (!failed) { |
| 147 | return extractors; |
| 148 | } |
| 149 | |
| 150 | auto it = m_mimeExtractors.begin(); |
| 151 | while (it != m_mimeExtractors.end()) { |
| 152 | if (it.value() == failed) { |
| 153 | it = m_mimeExtractors.erase(it); |
| 154 | } else { |
| 155 | ++it; |
| 156 | } |
| 157 | } |
| 158 | return getExtractors(mimetype); |
| 159 | } |
| 160 | |
| 161 | QList<Extractor*> ExtractorCollection::fetchExtractors(const QString& mimetype) const |
| 162 | { |
| 163 | QList<Extractor*> plugins = d->getExtractors(mimetype); |
| 164 | if (!plugins.isEmpty()) { |
| 165 | return plugins; |
| 166 | } |
| 167 | |
| 168 | // try to find the best matching more generic extractor by mimetype inheritance |
| 169 | QMimeDatabase db; |
| 170 | auto type = db.mimeTypeForName(nameOrAlias: mimetype); |
| 171 | const QStringList ancestors = type.allAncestors(); |
| 172 | |
| 173 | for (const auto &ancestor : ancestors) { |
| 174 | if (ancestor == QLatin1String("application/octet-stream")) { |
| 175 | continue; |
| 176 | } |
| 177 | QList<Extractor*> plugins = d->getExtractors(mimetype: ancestor); |
| 178 | if (!plugins.isEmpty()) { |
| 179 | qCDebug(KFILEMETADATA_LOG) << "Using inherited mimetype"<< ancestor << "for"<< mimetype; |
| 180 | return plugins; |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | return plugins; |
| 185 | } |
| 186 |
