Workaround for the issue #411. TagLib already supports large files on the systems that long is 64-bit. (e.g. GCC on 64-bit Linux, MacOSX etc.) However, MP4::Atom can't handle 64-bit atoms even on such systems.

This is just a workaround, so I think that we should change MP4::Atom::offset and length to long long or offset_t in taglib2 branch.

Using this code (from @jacobvosmaer) and these Ogg Vorbis files (from @nguillaumin):

#include <iostream>
#include <taglib/fileref.h>

int main(int argc, char **argv) {
  if (argc != 2) {
    std::cout << "usage: " << argv[0] << " file.ogg" << std::endl;
    exit(1);
  }

  const char *filename = argv[1];

  TagLib::FileRef file(filename);

  std::cerr << "saving " << filename << " with taglib." << std::endl;

  if (!file.save()) {
    std::cerr << "file save failed" << std::endl;
    exit(1);
  }
}

The saved Ogg Vorbis files can't be decoded any program I've tried.

I created a framework from the v1.9.1 source code with the following cmake:

cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_FRAMEWORK=ON -DCMAKE_C_COMPILER=/usr/bin/gcc -DCMAKE_CXX_COMPILER=/usr/bin/c++ -DCMAKE_OSX_SYSROOT=/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.11.sdk/ -DCMAKE_OSX_DEPLOYMENT_TARGET=10.10 -DCMAKE_OSX_ARCHITECTURES="x86_64"

I then drag-and-dropped the tag.framework file (folder) from /Library/Frameworks/ into my Xcode 7.0 project, ticked the "Copy files if needed" checkbox and checked that the file was indeed present in my project folder. I then added the framework to the "Embedded Binaries", which automatically also adds it to "Linked Frameworks and Libraries". I checked that the framework was present in the Build Phases under the "Link binary with libraries" and "Embed frameworks" sections.

I can link to the header files with no error from Xcode (i.e. #include <tag/mpegfile.h>) but when I try to compile, I get an error: Command /usr/bin/codesign failed with exit code 1

The compiler gives the following additional info: /Users/valentin/Library/Developer/Xcode/DerivedData/Squeed-gsytetlenuelapbdgypechfnxbht/Build/Products/Debug/Squeed.app/Contents/Frameworks/tag.framework/Versions/A: No such file or directory

I went and looked in the framework and, indeed, there is no "A" folder in the Versions, but a "1.14.0" folder.. Why is the compiler trying to look at a version that's not there?

Thanks !

This patch adds support for mod, s3m, it and xm files.

These files actually only have a title tag, but the names of the samples are commonly (always) abused as a multiline comment so I parse them into the comment tag. I also provide a trackerName tag which is only really available for xm. For the other formats I derive this tag from the file format. For s3m it is always "ScreamTracker III", for it it is "Impulse Tracker". For mod it is "ProTracker", "StarTrekker", "Atari Oktalyzer", "TakeTracker" or "NoiseTracker", depending on the modId.

I support writing of the title and comment tags and for xm files also the trackerName tag. But note that all these fields are limited in size (the limit depends on the file format). For it files the comment tag does not just consist of the instrument and sample names but also of the special message tag it supports. This would actually be the right place to store a comment, but no one seem to do that. Instead the instrument and sample names are abused as with the other formats. Because I do not completely rewrite the files but only overwrite the tags in place the length of characters I can put in this tag is limited by the already existing length. If no message tag (or a zero length message tag) exists, then I append the message at the end of the file. In this case the message can be the maximum of 8000 characters (including the termination NUL character) that is supported by Impulse Tracker 1.04+.

I also parse some other properties of these files, but the standard bitrate and sampleRate properties are not supported, because they simply do not apply for this file formats. Also I do not support the length property, because the only way to find out how long the songs are is to dummy-interpret them (which common players actually do, but I don't think replicating that functionality here is a good idea).

Also note that the readProperties parameter is ignored, because the tags and properties are so intertwined that I have to read both anyway. Should I remove the readProperties parameter from the constructors?

Noticed a change in Taglib 1.10 from 1.9 in the way ID3v1 tags are returned and wondering if this is deliberate or a side-effect, and either way is it desireable? I have not been able to test 1.11 yet to see what happens there.

In 1.9 accessing ID3v1 tags using to8Bit returned a standard nul terminated string the size of the actual text as happens for ID3v2, but in 1.10 this string is always padded with nul characters to a fixed size. This means that using size() or empty() on the returned string is meaningless. Generally the trailing nul padding is not too much an issue, but it makes it harder to detect an empty tag. The wiki says that this should return nul, but for ID3v1 it is returning a string of nul characters instead.

If 1.10 behaviour is to be kept, then it would be good to highlight this change in the release notes.

I'm using TagLib to get tags from audio files for my application that I'm making using wxWidgets. I have set it up to fetch multiple tags like artist, title, album and all, some files load fine, but some when loaded crashes my application and gives error saying,

./build.sh: line 12:  5891 Segmentation fault      ./SampleBrowser

I tried running with GDB,

Thread 1 "SampleBrowser" received signal SIGSEGV, Segmentation fault.
Browser::AddSamples (this=0x555555a73e00, file=...) at ../src/Browser.cpp:248
248         std::string Artist = File.tag()->artist().to8Bit(true);
(gdb)

Here is the function that is responsible for fetching tags,

void Browser::AddSamples(wxString file)
{
    TagLib::FileRef f (static_cast<const char*>(file), true, TagLib::AudioProperties::ReadStyle::Average);

    if (!f.isNull() && f.tag() && f.audioProperties())
    {
        TagLib::Tag* tag = f.tag();
        TagLib::AudioProperties* properties = f.audioProperties();

        TagLib::String Artist = tag->artist();
        TagLib::String Album = tag->album();
        TagLib::String Genre = tag->genre();
        TagLib::String Title = tag->title();
        TagLib::String Comment = tag->comment();

        int seconds = properties->length() % 60;
        int minutes = (properties->length() - seconds) / 60;

        int Bitrate = properties->bitrate();
        int Channels = properties->channels();
        int Length = properties->lengthInMilliseconds();
        int LengthSec = properties->lengthInSeconds();
        int SampleRate = properties->sampleRate();

        wxString Path = file;
        wxString Filename = file.AfterLast('/').BeforeLast('.');

        Data.clear();
        Data.push_back(false);
        Data.push_back(Filename);
        Data.push_back(wxString::FromUTF8(Artist.toCString(true)));
        Data.push_back(wxString::Format("%d", Channels));
        Data.push_back(wxString::Format("%d", Length));
        Data.push_back(wxString::Format("%d", SampleRate));
        Data.push_back(wxString::Format("%d", Bitrate));
        Data.push_back(wxString::FromUTF8(Comment.toCString(true)));

        SampleListView->AppendItem(Data);

        db.InsertSample(0, Filename.ToStdString(), Artist.to8Bit(true),
                        Channels, Length, SampleRate, Bitrate,
                        Comment.to8Bit(true), Path.ToStdString());
    }
    else
    {
        for (int i = 0; i < f.isNull(); i++)
        {
            const wxString& msg = wxString::Format("Error! Cannot open file %s", f.file()->name());
            wxMessageDialog *msgDialog = new wxMessageDialog(NULL, msg, "Error", wxOK | wxICON_ERROR);
            msgDialog->ShowModal();
        }
    }
}

I have this AddSamples(), that I call inside 2 other functions, as I have different ways of adding files to my application, via drag and drop and by file browser inside the application, both of which uses this AddSamples() to get tags and insert them in the wxDataViewListCtrl and SQLite3 database. file here is the parameter passed in the function, which is of wxString type. I am able to successfully add files to the application using AddSamples(), but say there are 5 files in a directory, 3 will add fine, but other 2 crashes the application and causes segmentation fault.

As you can see, there are 30 files in the directory, Kick 001 and Kick 006 fails to add to the application. I tried opening those 2 files in a audio player, it plays fine, file is not corrupt. Also say user drags and drops 1000+ files at once, a lot of files are going to fail to be added this way.

taglib_set_strings_unicode will make taglib output in only one encoding(unicode or Latin1), if the original id3 information is like GBK charset, then the output encoding is incorrect. how to make the c api output the original characters?

This adds the management of tags for the DSD audio files types: DSF and DSDIFF

• DSF: backport from the work done by sbooth on the taglib2 branch, ID3v2 tags as per the DSF standard
• DSDIFF: manages 'DIIN' chunk (Edited Master Information chunk, for title and artist) as per the DSDIFF 1.5 standard, and adds management of an 'ID3 ' (ID3v2) chunk as found in popular softwares understanding DSDIFF that store it at either the root level, or for some as a child of the 'PROP' chunk

This patch enables TagLib to return the length of audio in milliseconds in addition to seconds.

As suggested in #503, it adds an abstract function like this to AudioProperties and its derived classes.

    /*!
     * Returns the length of the file in milliseconds.
     */
    virtual int lengthInMilliseconds() const = 0;

This also enables to calculate the actual average bitrate of Vorbis/Speex files and removes some useless data members from AudioProperties related classes.

Save tags in latest available version realy not good idea. For example windows media player and windows explorer can't read ID3v2.4 tags.

Fixed realy strange behavior.

TagLib::MPEG::FileRef fref(filename); // .mp3 file without any tags TagLib::MPEG::File* file = fref.file(); TagLib::ID3v2::Tag *tag = file->ID3v2Tag(); ASSERT(tag); tag->header()->setMajorVersion(3); // !!! // set other tag fields fref.save();

Got ID3v2.4 tags in file.

PS. Also added constructor MP4::Tag::Tag()

Hello, I have taglib 1.10 on Arch linux and due to this library the complete Kodi is constantly crashing.

Important lines from gdb stacktrace:

[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
Core was generated by `/usr/lib/kodi/kodi.bin --standalone'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x0000000000c842c2 in CTagLoaderTagLib::ParseID3v2Tag(TagLib::ID3v2::Tag*, MUSIC_INFO::EmbeddedArt*, MUSIC_INFO::CMusicInfoTag&) ()
[Current thread is 1 (Thread 0x7eff91f64700 (LWP 24064))]

Full crashlog is here: http://pastebin.com/iiQ8eSae

Folks from Kodi user forums suggested to downgrade to the taglib 1.9.1 which is not crashing. Here: http://forum.kodi.tv/showthread.php?tid=254289&pid=2200524

Will be this problem fixed in the near future? Do you need details or the given MP3 file (on which taglib is segfaulting)?

Thanks very much

Hello. I'm trying to use taglib on AWS Lambda and almost everything is ok, text data is being updated correctly, but picture adding doesn't work. Does the taglib need some extra dependencies?

Installed version is 1.12 https://www.linuxfromscratch.org/blfs/view/svn/multimedia/taglib.html

I can't really pin point what taglib is doing, but whenever I try to read any of the attached files memory gets corrupted and the application eventually SEGFAULTs. The only thing I manage to see with valgrind was an ilegal read on the string comparator, but I couldn't find out why. Here's valgrind output

These are the test files that generate the error: test.tar.gz

This PR changes the ID3v2-FrameFactory so it is possible to subclass it for implementing own frames. Additionally it splits up the createFrame() method to make its usage in subclasses a bit more easy.

As this is a binary-incompatible change I do not expect you to take this PR in consideration until next major release.

OpenMPT, a popular Windows module tracker has support for storing artist information in FastTracker II (.xm) and Impulse Tracker (.it) modules through extension chunks. It would be awesome if TagLib could extract this metadata.

File format reference: https://wiki.openmpt.org/Development:_OpenMPT_Format_Extensions

Example files: openmpt-artist-examples.tar.gz

I'm using taglib for a media player Is very fast to get mp3 tags and embedded pixmaps, but destructor is very slow: in Ryzen 3900X /linux: parse time arround 1-25ms per mp3 destructor is arround 100ms !!!

watcher->setFuture(QtConcurrent::run([filename](){
QStringList ret;
        QElapsedTimer timer;
        timer.start();
        qDebug() << "-- TAG (CHECK) --"<<filename ;
        {
            TagLib::FileRef f(filename.toStdString().c_str(),true,TagLib::AudioProperties::Fast);
            if(!f.isNull() && f.tag()) {
                TagLib::Tag *tag = f.tag();
                ret.push_back(QString::fromUtf8(tag->artist().toCString(true)));
                ret.push_back(QString::fromUtf8((tag->title().toCString(true))));
                ret.push_back(QString::fromUtf8((tag->album().toCString(true))));
            }
             if(!f.isNull() && f.audioProperties()) {
                TagLib::AudioProperties *properties = f.audioProperties();
                ret.push_back(QString::number(properties->length()*1000));
            }
            qDebug() << "----> parse interval"<<timer.elapsed()<<"ms";
            timer.start();
        **}  //TagLib::FileRef  DESTRUCTIOR**
        qDebug() << "----> destructor interval "<<timer.elapsed()<<"ms";
        ...

----> parse interval 15 ms ----> parse interval 21 ms ----> parse interval 24 ms ----> parse interval 25 ms ----> parse interval 35 ms ----> parse interval 34 ms ----> parse interval 40 ms ----> destructor interval 105 ms ----> destructor interval 109 ms ----> destructor interval 106 ms ----> destructor interval 107 ms ----> destructor interval 102 ms ----> destructor interval 106 ms

any idea?