[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-commits
Subject:    [kfilemetadata] /: Make odf indexer more error prove, check if the files are there (and are files at
From:       Christoph Cullmann <cullmann () kde ! org>
Date:       2016-09-11 13:08:56
Message-ID: E1bj4VE-0000wI-Ql () code ! kde ! org
[Download RAW message or body]

Git commit 40730d75397aefb92145f86fc6abc9b303c56cfe by Christoph Cullmann.
Committed on 11/09/2016 at 13:07.
Pushed by cullmann into branch 'master'.

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

M  +19   -3    autotests/odfextractortest.cpp
M  +3    -0    autotests/odfextractortest.h
A  +-    --    autotests/samplefiles/test_missing_content.odt
A  +-    --    autotests/samplefiles/test_missing_meta.odt
M  +12   -7    src/extractors/odfextractor.cpp

http://commits.kde.org/kfilemetadata/40730d75397aefb92145f86fc6abc9b303c56cfe

diff --git a/autotests/odfextractortest.cpp b/autotests/odfextractortest.cpp
index bea3640..1b68d3b 100644
--- a/autotests/odfextractortest.cpp
+++ b/autotests/odfextractortest.cpp
@@ -1,6 +1,7 @@
 /*
  * <one line to give the library's name and an idea of what it does.>
  * Copyright (C) 2014  Vishesh Handa <me@vhanda.in>
+ * Copyright (C) 2016  Christoph Cullmann <cullmann@kde.org>
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
@@ -37,7 +38,7 @@ QString OdfExtractorTest::testFilePath(const QString& fileName) \
const  
 void OdfExtractorTest::testText()
 {
-    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor(this));
+    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor());
 
     SimpleExtractionResult result(testFilePath(QStringLiteral("test.odt")), \
QStringLiteral("application/vnd.oasis.opendocument.text"));  \
plugin->extract(&result); @@ -62,7 +63,7 @@ void OdfExtractorTest::testText()
 
 void OdfExtractorTest::testTextMetaDataOnly()
 {
-    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor(this));
+    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor());
 
     SimpleExtractionResult result(testFilePath(QStringLiteral("test.odt")), \
QStringLiteral("application/vnd.oasis.opendocument.text"), \
ExtractionResult::ExtractMetaData);  plugin->extract(&result);
@@ -74,7 +75,7 @@ void OdfExtractorTest::testTextMetaDataOnly()
 
 void OdfExtractorTest::testPresentation()
 {
-    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor(this));
+    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor());
 
     SimpleExtractionResult result(testFilePath(QStringLiteral("test.odp")), \
QStringLiteral("application/vnd.oasis.opendocument.presentation"));  \
plugin->extract(&result); @@ -90,5 +91,20 @@ void \
OdfExtractorTest::testPresentation()  QCOMPARE(result.text(), \
QStringLiteral("KFileMetaData Pres "));  }
 
+void OdfExtractorTest::testTextMissingMetaNoCrash()
+{
+    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor());
+
+    SimpleExtractionResult \
result(testFilePath(QStringLiteral("test_missing_meta.odt")), \
QStringLiteral("application/vnd.oasis.opendocument.text")); +    \
plugin->extract(&result); +}
+
+void OdfExtractorTest::testTextMissingContentNoCrash()
+{
+    QScopedPointer<ExtractorPlugin> plugin(new OdfExtractor());
+
+    SimpleExtractionResult \
result(testFilePath(QStringLiteral("test_missing_content.odt")), \
QStringLiteral("application/vnd.oasis.opendocument.text")); +    \
plugin->extract(&result); +}
 
 QTEST_MAIN(OdfExtractorTest)
diff --git a/autotests/odfextractortest.h b/autotests/odfextractortest.h
index 3e6a1c7..89642ce 100644
--- a/autotests/odfextractortest.h
+++ b/autotests/odfextractortest.h
@@ -1,6 +1,7 @@
 /*
  * <one line to give the library's name and an idea of what it does.>
  * Copyright (C) 2014  Vishesh Handa <me@vhanda.in>
+ * Copyright (C) 2016  Christoph Cullmann <cullmann@kde.org>
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
@@ -35,6 +36,8 @@ private Q_SLOTS:
 
     void testPresentation();
 
+    void testTextMissingMetaNoCrash();
+    void testTextMissingContentNoCrash();
 };
 
 #endif // ODFEXTRACTORTEST_H
diff --git a/autotests/samplefiles/test_missing_content.odt \
b/autotests/samplefiles/test_missing_content.odt new file mode 100644
index 0000000..7f7ac4b
Binary files /dev/null and b/autotests/samplefiles/test_missing_content.odt differ
diff --git a/autotests/samplefiles/test_missing_meta.odt \
b/autotests/samplefiles/test_missing_meta.odt new file mode 100644
index 0000000..8549639
Binary files /dev/null and b/autotests/samplefiles/test_missing_meta.odt differ
diff --git a/src/extractors/odfextractor.cpp b/src/extractors/odfextractor.cpp
index 948f3e0..3d224dc 100644
--- 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 <me@vhanda.in>
     Copyright (C) 2012  Jörg Ehrichs <joerg.ehrichs@gmx.de>
+    Copyright (C) 2016  Christoph Cullmann <cullmann@kde.org>
 
     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 @@ void OdfExtractor::extract(ExtractionResult* result)
         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 @@ void OdfExtractor::extract(ExtractionResult* result)
         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()) {


[prev in list] [next in list] [prev in thread] [next in thread]