commit: 7811803f05a811dab3e9220060a9539844cf8e74 Author: Andreas Sturmlechner <andreas.sturmlechner <AT> gmail <DOT> com> AuthorDate: Sun Sep 11 16:07:57 2016 +0000 Commit: Michael Palimaka <kensington <AT> gentoo <DOT> org> CommitDate: Wed Sep 14 14:13:12 2016 +0000 URL: https://gitweb.gentoo.org/proj/kde.git/commit/?id=7811803f
kde-frameworks/kfilemetadata: Fix segfault w/ ODF and EPUB files Package-Manager: portage-2.3.0 ...ilemetadata-5.26.0-epubextractor-segfault.patch | 149 +++++++++++++++++++++ ...filemetadata-5.26.0-odfextractor-segfault.patch | 66 +++++++++ .../kfilemetadata/kfilemetadata-5.26.0-r1.ebuild | 55 ++++++++ 3 files changed, 270 insertions(+) diff --git a/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-epubextractor-segfault.patch b/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-epubextractor-segfault.patch new file mode 100644 index 0000000..b738d1a --- /dev/null +++ b/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-epubextractor-segfault.patch @@ -0,0 +1,149 @@ +From: Christoph Cullmann <[email protected]> +Date: Sun, 11 Sep 2016 17:14:51 +0000 +Subject: Improve epub extractor, less segfaults +X-Git-Url: http://quickgit.kde.org/?p=kfilemetadata.git&a=commitdiff&h=47f6e57b2fa3768feb4f1f4a2cd3ce46660d90f2 +--- +Improve epub extractor, less segfaults + +Improve epub extractor: + +1) check for more nullpointers (e.g. data can be null for some fields, iterators, ...) +2) actually close the epub file again at all +3) iterator seems to handle clink as stated in docs, fix double free + +e.g. see bug 361727 +could be the double freed clink in the last iterator + +BUG: 361727 +REVIEW: 128888 +--- + + +--- a/src/extractors/epubextractor.cpp ++++ b/src/extractors/epubextractor.cpp +@@ -1,5 +1,6 @@ + /* + Copyright (C) 2013 Vishesh Handa <[email protected]> ++ Copyright (C) 2016 Christoph Cullmann <[email protected]> + + This library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public +@@ -46,11 +47,14 @@ + QString fetchMetadata(struct epub* e, const epub_metadata& type) + { + int size = 0; +- + unsigned char** data = epub_get_metadata(e, type, &size); + if (data) { + QStringList strList; + for (int i = 0; i < size; i++) { ++ // skip nullptr entries, can happen for broken xml files ++ if (!data[i]) ++ continue; ++ + strList << QString::fromUtf8((char*)data[i]); + free(data[i]); + } +@@ -65,7 +69,8 @@ + + void EPubExtractor::extract(ExtractionResult* result) + { +- struct epub* ePubDoc = epub_open(result->inputUrl().toUtf8().constData(), 1); ++ // open epub, return on exit, file will be closed again at end of function ++ auto ePubDoc = epub_open(result->inputUrl().toUtf8().constData(), 1); + if (!ePubDoc) { + qWarning() << "Invalid document"; + return; +@@ -138,49 +143,49 @@ + // + // Plain Text + // +- if (!(result->inputFlags() & ExtractionResult::ExtractPlainText)) { +- return; ++ if (result->inputFlags() & ExtractionResult::ExtractPlainText) { ++ if (auto iter = epub_get_iterator(ePubDoc, EITERATOR_SPINE, 0)) { ++ do { ++ char* curr = epub_it_get_curr(iter); ++ if (!curr) ++ continue; ++ ++ QString html = QString::fromUtf8(curr); ++ html.remove(QRegularExpression(QStringLiteral("<[^>]*>"))); ++ result->append(html); ++ } while (epub_it_get_next(iter)); ++ ++ epub_free_iterator(iter); ++ } ++ ++ auto tit = epub_get_titerator(ePubDoc, TITERATOR_NAVMAP, 0); ++ if (!tit) { ++ tit = epub_get_titerator(ePubDoc, TITERATOR_GUIDE, 0); ++ } ++ if (tit) { ++ if (epub_tit_curr_valid(tit)) { ++ do { ++ // get link, iterator handles freeing of it ++ char* clink = epub_tit_get_curr_link(tit); ++ ++ // epub_get_data returns -1 on failure ++ char* data = nullptr; ++ const int size = epub_get_data(ePubDoc, clink, &data); ++ if (size >= 0 && data) { ++ QString html = QString::fromUtf8(data, size); ++ // strip html tags ++ html.remove(QRegularExpression(QStringLiteral("<[^>]*>"))); ++ ++ result->append(html); ++ free(data); ++ } ++ } while (epub_tit_next(tit)); ++ } ++ epub_free_titerator(tit); ++ } + } + +- struct eiterator* iter = epub_get_iterator(ePubDoc, EITERATOR_SPINE, 0); +- do { +- char* curr = epub_it_get_curr(iter); +- if (!curr) +- continue; +- QString html = QString::fromUtf8(curr); +- html.remove(QRegularExpression(QStringLiteral("<[^>]*>"))); +- +- result->append(html); +- } while (epub_it_get_next(iter)); +- +- epub_free_iterator(iter); +- +- struct titerator* tit; +- +- tit = epub_get_titerator(ePubDoc, TITERATOR_NAVMAP, 0); +- if (!tit) { +- tit = epub_get_titerator(ePubDoc, TITERATOR_GUIDE, 0); +- } +- +- if (epub_tit_curr_valid(tit)) { +- do { +- char* clink = epub_tit_get_curr_link(tit); +- +- char* data; +- int size = epub_get_data(ePubDoc, clink, &data); +- free(clink); +- +- // epub_get_data returns -1 on failure +- if (size > 0 && data) { +- QString html = QString::fromUtf8(data, size); +- // strip html tags +- html.remove(QRegularExpression(QStringLiteral("<[^>]*>"))); +- +- result->append(html); +- free(data); +- } +- } while (epub_tit_next(tit)); +- } +- epub_free_titerator(tit); ++ // close epub file again ++ epub_close(ePubDoc); + } + + diff --git a/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-odfextractor-segfault.patch b/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-odfextractor-segfault.patch new file mode 100644 index 0000000..9f3029b --- /dev/null +++ b/kde-frameworks/kfilemetadata/files/kfilemetadata-5.26.0-odfextractor-segfault.patch @@ -0,0 +1,66 @@ +From: Christoph Cullmann <[email protected]> +Date: Sun, 11 Sep 2016 13:07:47 +0000 +Subject: Make odf indexer more error prove, check if the files are there (and are files at all) (meta.xml + content.xml) +X-Git-Url: http://quickgit.kde.org/?p=kfilemetadata.git&a=commitdiff&h=40730d75397aefb92145f86fc6abc9b303c56cfe +--- +Make odf indexer more error prove, check if the files are there (and are files at all) (meta.xml + content.xml) + +REVIEW: 128886 +BUG 364748 + +=> if you download this odt's to indexed directories your baloo will die on each index, be careful +--- + + +--- a/src/extractors/odfextractor.cpp ++++ b/src/extractors/odfextractor.cpp +@@ -2,6 +2,7 @@ + <one line to give the library's name and an idea of what it does.> + Copyright (C) 2013 Vishesh Handa <[email protected]> + Copyright (C) 2012 Jörg Ehrichs <[email protected]> ++ Copyright (C) 2016 Christoph Cullmann <[email protected]> + + This library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public +@@ -59,19 +60,18 @@ + return; + } + +- const QStringList entries = directory->entries(); +- if (!entries.contains(QStringLiteral("meta.xml"))) { ++ // we need a meta xml file in the archive! ++ const auto metaXml = directory->entry(QStringLiteral("meta.xml")); ++ if (!metaXml || !metaXml->isFile()) { + qWarning() << "Invalid document structure (meta.xml is missing)"; + return; + } + + QDomDocument metaData(QStringLiteral("metaData")); +- const KArchiveFile* file = static_cast<const KArchiveFile*>(directory->entry(QStringLiteral("meta.xml"))); +- metaData.setContent(file->data()); ++ metaData.setContent(static_cast<const KArchiveFile*>(metaXml)->data()); + + // parse metadata ... + QDomElement docElem = metaData.documentElement(); +- + QDomNode n = docElem.firstChild().firstChild(); // <office:document-meta> ... <office:meta> ... content + while (!n.isNull()) { + QDomElement e = n.toElement(); +@@ -129,9 +129,14 @@ + return; + } + +- const KArchiveFile* contentsFile = static_cast<const KArchiveFile*>(directory->entry(QStringLiteral("content.xml"))); +- QXmlStreamReader xml(contentsFile->createDevice()); ++ // for content indexing, we need content xml file ++ const auto contentXml = directory->entry(QStringLiteral("content.xml")); ++ if (!contentXml || !contentXml->isFile()) { ++ qWarning() << "Invalid document structure (content.xml is missing)"; ++ return; ++ } + ++ QXmlStreamReader xml(static_cast<const KArchiveFile*>(contentXml)->createDevice()); + while (!xml.atEnd()) { + xml.readNext(); + if (xml.isCharacters()) { + diff --git a/kde-frameworks/kfilemetadata/kfilemetadata-5.26.0-r1.ebuild b/kde-frameworks/kfilemetadata/kfilemetadata-5.26.0-r1.ebuild new file mode 100644 index 0000000..ecf7e65 --- /dev/null +++ b/kde-frameworks/kfilemetadata/kfilemetadata-5.26.0-r1.ebuild @@ -0,0 +1,55 @@ +# Copyright 1999-2016 Gentoo Foundation +# Distributed under the terms of the GNU General Public License v2 +# $Id$ + +EAPI=6 + +inherit eutils kde5 + +DESCRIPTION="Library for extracting file metadata" +KEYWORDS="~amd64 ~arm ~x86" +IUSE="epub exif ffmpeg libav pdf taglib" + +RDEPEND=" + $(add_frameworks_dep karchive) + $(add_frameworks_dep ki18n) + $(add_qt_dep qtxml) + epub? ( app-text/ebook-tools ) + exif? ( media-gfx/exiv2:= ) + ffmpeg? ( + libav? ( media-video/libav:= ) + !libav? ( media-video/ffmpeg:0= ) + ) + pdf? ( app-text/poppler[qt5] ) + taglib? ( media-libs/taglib ) +" +DEPEND="${RDEPEND} + kernel_linux? ( sys-apps/attr ) +" + +PATCHES=( + "${FILESDIR}/${P}-odfextractor-segfault.patch" + "${FILESDIR}/${P}-epubextractor-segfault.patch" +) + +src_configure() { + local mycmakeargs=( + $(cmake-utils_use_find_package epub EPub) + $(cmake-utils_use_find_package exif Exiv2) + $(cmake-utils_use_find_package ffmpeg FFmpeg) + $(cmake-utils_use_find_package pdf PopplerQt5) + $(cmake-utils_use_find_package taglib Taglib) + ) + + kde5_src_configure +} + +pkg_postinst() { + kde5_pkg_postinst + + if ! has_version app-text/catdoc || ! has_version dev-libs/libxls; then + elog "To get additional features, optional runtime dependencies may be installed:" + optfeature "indexing of Microsoft Word or Powerpoint files" app-text/catdoc + optfeature "indexing of Microsoft Excel files" dev-libs/libxls + fi +}
