libtorrent manual

Author: Arvid Norberg, arvid@rasterbar.com

Table of contents

introduction

libtorrent is a C++ library that aims to be a good alternative to all the other bittorrent implementations around. It is a library and not a full featured client, although it comes with a working example client.

The main goals of libtorrent are:

  • to be cpu efficient
  • to be memory efficient
  • to be very easy to use

libtorrent is still being developed, however it is stable. It is an ongoing project (including this documentation). The current state includes the following features:

  • multitracker extension support (as specified by John Hoffman)
  • serves multiple torrents on a single port and a single thread
  • supports http proxies and proxy authentication
  • gzipped tracker-responses
  • HTTP seeding, as specified by Michael Burford of GetRight.
  • piece picking on block-level like in Azureus (as opposed to piece-level). This means it can download parts of the same piece from different peers. It will also prefer to download whole pieces from single peers if the download speed is high enough from that particular peer.
  • queues torrents for file check, instead of checking all of them in parallel.
  • uses separate threads for checking files and for main downloader, with a fool-proof thread-safe library interface. (i.e. There's no way for the user to cause a deadlock). (see threads)
  • can limit the upload and download bandwidth usage and the maximum number of unchoked peers
  • piece-wise, unordered, incremental file allocation
  • implements fair trade. User settable trade-ratio, must at least be 1:1, but one can choose to trade 1 for 2 or any other ratio that isn't unfair to the other party.
  • fast resume support, a way to get rid of the costly piece check at the start of a resumed torrent. Saves the storage state, piece_picker state as well as all local peers in a separate fast-resume file.
  • supports an extension protocol. See extensions.
  • supports files > 2 gigabytes.
  • supports the no_peer_id=1 extension that will ease the load off trackers.
  • supports the udp-tracker protocol by Olaf van der Spek.
  • possibility to limit the number of connections.
  • delays have messages if there's no other outgoing traffic to the peer, and doesn't send have messages to peers that already has the piece. This saves bandwidth.
  • does not have any requirements on the piece order in a torrent that it resumes. This means it can resume a torrent downloaded by any client.
  • adjusts the length of the request queue depending on download rate.
  • supports the compact=1 tracker parameter.
  • selective downloading. The ability to select which parts of a torrent you want to download.
  • ip filter

libtorrent is portable at least among Windows, MacOS X and other UNIX-systems. It uses Boost.Thread, Boost.Filesystem, Boost.Date_time and various other boost libraries as well as zlib (shipped) and asio (shipped). At least version 1.33.1 of boost is required.

libtorrent has been successfully compiled and tested on:

  • Windows 2000 vc7.1
  • Linux x86 GCC 3.0.4, GCC 3.2.3, GCC 3.4.2
  • MacOS X (darwin), (Apple's) GCC 3.3, (Apple's) GCC 4.0
  • SunOS 5.8 GCC 3.1
  • Cygwin GCC 3.3.3

Fails on:

  • GCC 2.95.4
  • msvc6 sp5

libtorrent is released under the BSD-license.

downloading and building

To acquire the latest version of libtorrent, you'll have to grab it from CVS. You'll find instructions on how to do this here (see Anonymous CVS access).

The build systems supported "out of the box" in libtorrent are boost-build v2 (BBv2) and autotools (for unix-like systems). If you still can't build after following these instructions, you can usually get help in the #libtorrent IRC channel on irc.freenode.net.

building with BBv2

The primary reason to use boost-build is that it will automatically build the dependent boost libraries with the correct compiler settings, in order to ensure that the build targets are link compatible (see boost guidelines for some details on this issue).

Since BBv2 will build the boost libraries for you, you need the full boost source package. Having boost installed via some package system is usually not enough (and even if it is enough, the necessary environment variables are usually not set by the package installer).

Step 1: Download boost

You'll find boost here.

Extract the archive to some directory where you want it. For the sake of this guide, let's assume you extract the package to c:\boost_1_33_0 (I'm using a windows path in this example since if you're on linux/unix you're more likely to use the autotools). You'll need at least version 1.32 of the boost library in order to build libtorrent.

If you use 1.32, you need to download BBv2 separately, so for now, let's assume you will use version 1.33.

Step 2: Setup BBv2

First you need to build bjam. You do this by opening a terminal (In windows, run cmd). Change directory to c:\boost_1_33_0\tools\build\jam_src. Then run the script called build.bat or build.sh on a unix system. This will build bjam and place it in a directory starting with bin. and then have the name of your platform. Copy the bjam.exe (or bjam on a unix system) to a place that's in you shell's PATH. On linux systems a place commonly used may be /usr/local/bin or on windows c:\windows (you can also add directories to the search paths by modifying the environment variable called PATH).

Now you have bjam installed. bjam can be considered an interpreter that the boost-build system is implemented on. So boost-build uses bjam. So, to complete the installation you need to make two more things. You need to set the environment variable BOOST_BUILD_PATH. This is the path that tells bjam where it can find boost-build, your configuration file and all the toolsets (descriptions used by boost-build to know how to use different compilers on different platforms). Assuming the boost install path above, set it to c:\boost_1_33_0\tools\build\v2.

The last thing to do to complete the setup of BBv2 is to modify your user-config.jam file. It is located in c:\boost_1_33\tools\build\v2. Depending on your platform and which compiler you're using, you should add a line for each compiler and compiler version you have installed on your system that you want to be able to use with BBv2. For example, if you're using Microsoft Visual Studio 7.1 (2003), just add a line:

using msvc : 7.1 ;

If you use GCC, add the line:

using gcc ;

If you have more than one version of GCC installed, you can add the commandline used to invoke g++ after the version number, like this:

using gcc : 3.3 : g++-3.3 ;
using gcc : 4.0 : g++-4.0 ;

Another toolset worth mentioning is the darwin toolset (For MacOS X). From Tiger (10.4) MacOS X comes with both GCC 3.3 and GCC 4.0. Then you can use the following toolsets:

using darwin : 3.3 : g++-3.3 ;
using darwin : 4.0 : g++-4.0 ;

Note that the spaces around the semi-colons and colons are important!

Step 3: Building libtorrent

When building libtorrent, the Jamfile expects the environment variable BOOST_ROOT to be set to the boost installation directory. It uses this to find the boost libraries it depends on, so they can be built and their headers files found. So, set this to c:\boost_1_33_0.

Then the only thing left is simply to invoke bjam. If you want to specify a specific toolset to use (compiler) you can just add that to the commandline. For example:

bjam msvc-7.1 link=static
bjam gcc-3.3 link=static
bjam darwin-4.0 link=static

To build different versions you can also just add the name of the build variant. Some default build variants in BBv2 are release, debug, profile.

You can build libtorrent as a dll too, by typing link=shared, or link=static to build a static library. link=shared is the default.

If you want to explicitly say how to link against the runtime library, you can set the runtime-link feature on the commandline, either to shared or static. Most operating systems will only allow linking shared against the runtime, but on windows you can do both. Example:

bjam msvc-7.1 link=static runtime-link=static

Warning

If you link statically to the runtime library, you cannot build libtorrent as a shared library (DLL), since you will get separate heaps in the library and in the client application. It will result in crashes.

The build targets are put in a directory called bin, and under it they are sorted in directories depending on the toolset and build variant used.

To build the examples, just change directory to the examples directory and invoke bjam from there. To build and run the tests, go to the test directory and run bjam.

Note that if you're building on windows using the msvc toolset, you cannot run it from a cygwin terminal, you'll have to run it from a cmd terminal. The same goes for cygwin, if you're building with gcc in cygwin you'll have to run it from a cygwin terminal. Also, make sure the paths are correct in the different environments. In cygwin, the paths (BOOST_BUILD_PATH and BOOST_ROOT) should be in the typical unix-format (e.g. /cygdrive/c/boost_1_33_0). In the windows environment, they should have the typical windows format (c:/boost_1_33_0).

The Jamfile will define NDEBUG when it's building a release build. There are two other build variants available in the Jamfile. debug_log and release_log, these two variants inherits from the debug and release variants respectively, but adds extra logging (TORRENT_VERBOSE_LOGGING). For more build configuration flags see Build configurations.

The Jamfile has the following build variants:

  • release - release version without any logging
  • release_log - release version with standard logging
  • release_vlog - release version with verbose logging (all peer connections are logged)
  • debug - debug version without any logging
  • debug_log - debug version with standard logging
  • debug_vlog - debug version with verbose logging

When building the example client on windows, you need to build with link=static otherwise you may get unresolved external symbols for some boost.program-options symbols.

building with autotools

First of all, you need to install automake and autoconf. Many unix/linux systems comes with these preinstalled.

Step 1: Running configure

In your shell, change directory to the libtorrent directory and run ./configure. This will look for libraries and C++ features that libtorrent is dependent on. If something is missing or can't be found it will print an error telling you what failed.

The most likely problem you may encounter is that the configure script won't find the boost libraries. Make sure you have boost installed on your system. The easiest way to install boost is usually to use the preferred package system on your platform. Usually libraries and headers are installed in standard directories where the compiler will find them, but sometimes that may not be the case. For example when installing boost on darwin using darwinports (the package system based on BSD ports) all libraries are installed to /opt/local/lib and headers are installed to /opt/local/include. By default the compiler will not look in these directories. You have to set the enviornment variables LDFLAGS and CXXFLAGS in order to make the compiler find those libs. In this example you'd set them like this:

export LDFLAGS=-L/opt/local/lib
export CXXFLAGS=-I/opt/local/include

It was observed on FreeBSD (release 6.0) that one needs to add '-lpthread' to LDFLAGS, as Boost::Thread detection will fail without it, even if Boost::Thread is installed.

If you need to set these variables, it may be a good idea to add those lines to your ~/.profile or ~/.tcshrc depending on your shell.

You know that the boost libraries were found if you see the following output from the configure script:

checking whether the Boost::DateTime library is available... yes
checking for main in -lboost_date_time... yes
checking whether the Boost::Filesystem library is available... yes
checking for main in -lboost_filesystem... yes
checking whether the Boost::Thread library is available... yes
checking for main in -lboost_thread... yes

Another possible source of problems may be if the path to your libtorrent directory contains spaces. Make sure you either rename the directories with spaces in their names to remove the spaces or move the libtorrent directory.

Creating a debug build

To tell configure to build a debug version (with debug info, asserts and invariant checks enabled), you have to run the configure script with the following option:

./configure --enable-debug=yes

Creating a release build

To tell the configure to build a release version (without debug info, asserts and invariant checks), you have to run the configure script with the following option:

./configure --enable-debug=no

The above option make use of -DNDEBUG, which is used throughout libtorrent.

Step 2: Building libtorrent

Once the configure script is run successfully, you just type make and libtorrent, the examples and the tests will be built.

When libtorrent is built it may be a good idea to run the tests, you do this by running make check.

If you want to build a release version (without debug info, asserts and invariant checks), you have to rerun the configure script and rebuild, like this:

./configure --disable-debug
make clean
make

generating the build system

No build system is present if libtorrent is checked out from CVS - it needs to be generated first.

Execute the following commands to generate the build system:

  • autoheader
  • aclocal -I m4
  • libtoolize --copy --force
  • automake --add-missing --copy --gnu
  • autoconf

After generating the build system, run configure and build libtorrent. This was described earlier.

building with other build systems

If you're making your own project file, note that there are two versions of the file abstraction. There's one file_win.cpp which relies on windows file API that supports files larger than 2 Gigabytes. This does not work in vc6 for some reason, possibly because it may require windows NT and above. The other file, file.cpp is the default implementation that simply relies on the standard low level io routines (read(), write(), open() etc.), this implementation doesn't do anything special to support unicode filenames, so if your target is Windows 2000 and up, you may want to use file_win.cpp which supports unicode filenames.

If you're building in MS Visual Studio, you may have to set the compiler options "force conformance in for loop scope", "treat wchar_t as built-in type" and "Enable Run-Time Type Info" to Yes.

build configurations

By default libtorrent is built In debug mode, and will have pretty expensive invariant checks and asserts built into it. If you want to disable such checks (you want to do that in a release build) you can see the table below for which defines you can use to control the build.

macro description
NDEBUG If you define this macro, all asserts, invariant checks and general debug code will be removed. This option takes precedence over other debug settings.
TORRENT_LOGGING This macro will enable logging of the session events, such as tracker announces and incoming connections (as well as blocked connections).
TORRENT_VERBOSE_LOGGING If you define this macro, every peer connection will log its traffic to a log file as well as the session log.
TORRENT_STORAGE_DEBUG This will enable extra expensive invariant checks in the storage, including logging of piece sorting.
UNICODE If building on windows this will make sure the UTF-8 strings in pathnames are converted into UTF-16 before they are passed to the file operations.
LITTLE_ENDIAN This will use the little endian version of the sha-1 code. If defined on a big-endian system the sha-1 hashes will be incorrect and fail. If it is not defined and __BIG_ENDIAN__ isn't defined either (it is defined by Apple's GCC) both little-endian and big-endian versions will be built and the correct code will be chosen at run-time.
TORRENT_LINKING_SHARED If this is defined when including the libtorrent headers, the classes and functions will be tagged with __declspec(dllimport) on msvc and default visibility on GCC 4 and later. Set this in your project if you're linking against libtorrent as a shared library. (This is set by the Jamfile when link=shared is set).
TORRENT_BUILDING_SHARED If this is defined, the functions and classes in libtorrent are marked with __declspec(dllexport) on msvc, or with default visibility on GCC 4 and later. This should be defined when building libtorrent as a shared library. (This is set by the Jamfile when link=shared is set).

If you experience that libtorrent uses unreasonable amounts of cpu, it will definately help to define NDEBUG, since it will remove the invariant checks within the library.

overview

The interface of libtorrent consists of a few classes. The main class is the session, it contains the main loop that serves all torrents.

The basic usage is as follows:

Each class and function is described in this manual.

session

The session class has the following synopsis:

class session: public boost::noncopyable
{

        session(const fingerprint& print
                = libtorrent::fingerprint("LT", 0, 1, 0, 0));

        session(
                const fingerprint& print
                , std::pair<int, int> listen_port_range
                , const char* listen_interface = 0);

        torrent_handle add_torrent(
                torrent_info const& ti
                , boost::filesystem::path const& save_path
                , entry const& resume_data = entry()
                , bool compact_mode = true
                , int block_size = 16 * 1024);

        torrent_handle add_torrent(
                char const* tracker_url
                , sha1_hash const& info_hash
                , boost::filesystem::path const& save_path
                , entry const& resume_data = entry()
                , bool compact_mode = true
                , int block_size = 16 * 1024);

        void remove_torrent(torrent_handle const& h);

        void disable_extensions();
        void enable_extension(peer_connection::extension_index);

        void set_http_settings(const http_settings& settings);

        void set_upload_rate_limit(int bytes_per_second);
        void set_download_rate_limit(int bytes_per_second);
        void set_max_uploads(int limit);
        void set_max_connections(int limit);
        void set_max_half_open_connections(int limit);

        void set_ip_filter(ip_filter const& f);

        session_status status() const;

        bool is_listening() const;
        unsigned short listen_port() const;
        bool listen_on(
                std::pair<int, int> const& port_range
                , char const* interface = 0);


        std::auto_ptr<alert> pop_alert();
        void set_severity_level(alert::severity_t s);
};

Once it's created, the session object will spawn the main thread that will do all the work. The main thread will be idle as long it doesn't have any torrents to participate in.

session()

session(fingerprint const& print = libtorrent::fingerprint("LT", 0, 1, 0, 0));
session(fingerprint const& print
        , std::pair<int, int> listen_port_range
        , char const* listen_interface = 0);

If the fingerprint in the first overload is ommited, the client will get a default fingerprint stating the version of libtorrent. The fingerprint is a short string that will be used in the peer-id to identify the client and the client's version. For more details see the fingerprint class. The constructor that only takes a fingerprint will not open a listen port for the session, to get it running you'll have to call session::listen_on(). The other constructor, that takes a port range and an interface as well as the fingerprint will automatically try to listen on a port on the given interface. For more information about the parameters, see listen_on() function.

~session()

The destructor of session will notify all trackers that our torrents have been shut down. If some trackers are down, they will time out. All this before the destructor of session returns. So, it's adviced that any kind of interface (such as windows) are closed before destructing the sessoin object. Because it can take a few second for it to finish. The timeout can be set with set_http_settings().

add_torrent()

torrent_handle add_torrent(
        torrent_info const& ti
        , boost::filesystem::path const& save_path
        , entry const& resume_data = entry()
        , bool compact_mode = true
        , int block_size = 16 * 1024);

torrent_handle add_torrent(
        char const* tracker_url
        , sha1_hash const& info_hash
        , boost::filesystem::path const& save_path
        , entry const& resume_data = entry()
        , bool compact_mode = true
        , int block_size = 16 * 1024);

You add torrents through the add_torrent() function where you give an object representing the information found in the torrent file and the path where you want to save the files. The save_path will be prepended to the directory structure in the torrent-file.

If the torrent you are trying to add already exists in the session (is either queued for checking, being checked or downloading) add_torrent() will throw duplicate_torrent which derives from std::exception.

The optional parameter, resume_data can be given if up to date fast-resume data is available. The fast-resume data can be acquired from a running torrent by calling torrent_handle::write_resume_data(). See fast resume.

The compact_mode paramater refers to the layout of the storage for this torrent. If set to true (default), the storage will grow as more pieces are downloaded, and pieces are rearranged to finally be in their correct places once the entire torrent has been downloaded. If it is false, the entire storage is allocated before download begins. I.e. the files contained in the torrent are filled with zeroes, and each downloaded piece is put in its final place directly when downloaded. For more info, see storage allocation.

block_size sets the preferred request size, i.e. the number of bytes to request from a peer at a time. This block size must be a divisor of the piece size, and since the piece size is an even power of 2, so must the block size be. If the block size given here turns out to be greater than the piece size, it will simply be clamped to the piece size.

The torrent_handle returned by add_torrent() can be used to retrieve information about the torrent's progress, its peers etc. It is also used to abort a torrent.

The second overload that takes a tracker url and an info-hash instead of metadata (torrent_info) can be used with torrents where (at least some) peers support the metadata extension. For the overload to be available, libtorrent must be built with extensions enabled (TORRENT_ENABLE_EXTENSIONS defined).

remove_torrent()

void remove_torrent(torrent_handle const& h);

remove_torrent() will close all peer connections associated with the torrent and tell the tracker that we've stopped participating in the swarm.

disable_extensions() enable_extension()

void disable_extensions();
void enable_extension(peer_connection::extension_index);

disable_extensions() will disable all extensions available in libtorrent. enable_extension() will enable a single extension. The available extensions are enumerated in the peer_connection class. These are the available extensions:

enum extension_index
{
        extended_chat_message,
        extended_metadata_message,
        extended_peer_exchange_message,
        extended_listen_port_message,
        num_supported_extensions
};

peer_exchange is not implemented yet

By default, all extensions are enabled. For more information about the extensions, see the extensions section.

set_upload_rate_limit() set_download_rate_limit()

void set_upload_rate_limit(int bytes_per_second);
void set_download_rate_limit(int bytes_per_second);

set_upload_rate_limit() set the maximum number of bytes allowed to be sent to peers per second. This bandwidth is distributed among all the peers. If you don't want to limit upload rate, you can set this to -1 (the default). set_download_rate_limit() works the same way but for download rate instead of upload rate.

set_max_uploads() set_max_connections()

void set_max_uploads(int limit);
void set_max_connections(int limit);

These functions will set a global limit on the number of unchoked peers (uploads) and the number of connections opened. The number of connections is set to a hard minimum of at least two connections per torrent, so if you set a too low connections limit, and open too many torrents, the limit will not be met. The number of uploads is at least one per torrent.

set_max_half_open_connections()

void set_max_half_open_connections(int limit);

Sets the maximum number of half-open connections libtorrent will have when connecting to peers. A half-open connection is one where connect() has been called, but the connection still hasn't been established (nor filed). Windows XP Service Pack 2 sets a defaul, system wide, limit of the number of half-open connections to 10. So, this limit can be used to work nicer together with other network applications on that system. The default is to have no limit, and passing -1 as the limit, means to have no limit. When limiting the number of simultaneous connection attempts, peers will be put in a queue waiting for their turn to get connected.

set_ip_filter()

void set_ip_filter(ip_filter const& filter);

Sets a filter that will be used to reject and accept incoming as well as outgoing connections based on their originating ip address. The default filter will allow connections to any ip address. To build a set of rules for which addresses are accepted and not, see ip_filter.

status()

session_status status() const;

status() returns session wide-statistics and status. The session_status struct has the following members:

struct session_status
{
        bool has_incoming_connections;

        float upload_rate;
        float download_rate;

        float payload_upload_rate;
        float payload_download_rate;

        size_type total_download;
        size_type total_upload;

        size_type total_payload_download;
        size_type total_payload_upload;

        int num_peers;
};

has_incoming_connections is false as long as no incoming connections have been established on the listening socket. Every time you change the listen port, this will be reset to false.

upload_rate, download_rate, payload_download_rate and payload_upload_rate are the total download and upload rates accumulated from all torrents. The payload versions is the payload download only.

total_download and total_upload are the total number of bytes downloaded and uploaded to and from all torrents. total_payload_download and total_payload_upload are the same thing but where only the payload is considered.

num_peers is the total number of peer connections this session have.

is_listening() listen_port() listen_on()

bool is_listening() const;
unsigned short listen_port() const;
bool listen_on(
        std::pair<int, int> const& port_range
        , char const* interface = 0);

is_listening() will tell you wether or not the session has successfully opened a listening port. If it hasn't, this function will return false, and then you can use listen_on() to make another try.

listen_port() returns the port we ended up listening on. Since you just pass a port-range to the constructor and to listen_on(), to know which port it ended up using, you have to ask the session using this function.

listen_on() will change the listen port and/or the listen interface. If the session is already listening on a port, this socket will be closed and a new socket will be opened with these new settings. The port range is the ports it will try to listen on, if the first port fails, it will continue trying the next port within the range and so on. The interface parameter can be left as 0, in that case the os will decide which interface to listen on, otherwise it should be the ip-address of the interface you want the listener socket bound to. listen_on() returns true if it managed to open the socket, and false if it failed. If it fails, it will also generate an appropriate alert (listen_failed_alert).

The interface parameter can also be a hostname that will resolve to the device you want to listen on.

pop_alert() set_severity_level()

std::auto_ptr<alert> pop_alert();
void set_severity_level(alert::severity_t s);

pop_alert() is used to ask the session if any errors or events has occured. With set_severity_level() you can filter how serious the event has to be for you to receive it through pop_alert(). For information, see alerts.

entry

The entry class represents one node in a bencoded hierarchy. It works as a variant type, it can be either a list, a dictionary (std::map), an integer or a string. This is its synopsis:

class entry
{
public:

        typedef std::map<std::string, entry> dictionary_type;
        typedef std::string string_type;
        typedef std::list<entry> list_type;
        typedef size_type integer_type;

        enum data_type
        {
                int_t,
                string_t,
                list_t,
                dictionary_t,
                undefined_t
        };

        data_type type() const;

        entry(dictionary_type const&);
        entry(string_type const&);
        entry(list_type const&);
        entry(integer_type const&);

        entry();
        entry(data_type t);
        entry(entry const& e);
        ~entry();

        void operator=(entry const& e);
        void operator=(dictionary_type const&);
        void operator=(string_type const&);
        void operator=(list_type const&);
        void operator=(integer_type const&);

        integer_type& integer();
        integer_type const& integer() const;
        string_type& string();
        string_type const& string() const;
        list_type& list();
        list_type const& list() const;
        dictionary_type& dict();
        dictionary_type const& dict() const;

        // these functions requires that the entry
        // is a dictionary, otherwise they will throw
        entry& operator[](char const* key);
        entry& operator[](std::string const& key);
        entry const& operator[](char const* key) const;
        entry const& operator[](std::string const& key) const;
        entry* find_key(char const* key);
        entry const* find_key(char const* key) const;

        void print(std::ostream& os, int indent = 0) const;
};

TODO: finish documentation of entry.

integer() string() list() dict() type()

integer_type& integer();
integer_type const& integer() const;
string_type& string();
string_type const& string() const;
list_type& list();
list_type const& list() const;
dictionary_type& dict();
dictionary_type const& dict() const;

The integer(), string(), list() and dict() functions are accessors that return the respecive type. If the entry object isn't of the type you request, the accessor will throw type_error (which derives from std::runtime_error). You can ask an entry for its type through the type() function.

The print() function is there for debug purposes only.

If you want to create an entry you give it the type you want it to have in its constructor, and then use one of the non-const accessors to get a reference which you then can assign the value you want it to have.

The typical code to get info from a torrent file will then look like this:

entry torrent_file;
// ...

// throws if this is not a dictionary
entry::dictionary_type const& dict = torrent_file.dict();
entry::dictionary_type::const_iterator i;
i = dict.find("announce");
if (i != dict.end())
{
        std::string tracker_url = i->second.string();
        std::cout << tracker_url << "\n";
}

The following code is equivalent, but a little bit shorter:

entry torrent_file;
// ...

// throws if this is not a dictionary
if (entry* i = torrent_file.find_key("announce"))
{
        std::string tracker_url = i->string();
        std::cout << tracker_url << "\n";
}

To make it easier to extract information from a torrent file, the class torrent_info exists.

operator[]

entry& operator[](char const* key);
entry& operator[](std::string const& key);
entry const& operator[](char const* key) const;
entry const& operator[](std::string const& key) const;

All of these functions requires the entry to be a dictionary, if it isn't they will throw libtorrent::type_error.

The non-const versions of the operator[] will return a reference to either the existing element at the given key or, if there is no element with the given key, a reference to a newly inserted element at that key.

The const version of operator[] will only return a reference to an existing element at the given key. If the key is not found, it will throw libtorrent::type_error.

find_key()

entry* find_key(char const* key);
entry const* find_key(char const* key) const;

These functions requires the entry to be a dictionary, if it isn't they will throw libtorrent::type_error.

They will look for an element at the given key in the dictionary, if the element cannot be found, they will return 0. If an element with the given key is found, the return a pointer to it.

torrent_info

The torrent_info has the following synopsis:

class torrent_info
{
public:

        torrent_info();
        torrent_info(sha1_hash const& info_hash);
        torrent_info(entry const& torrent_file);

        entry create_torrent() const;
        void set_comment(char const* str);
        void set_piece_size(int size);
        void set_creator(char const* str);
        void set_hash(int index, sha1_hash const& h);
        void add_tracker(std::string const& url, int tier = 0);
        void add_file(boost::filesystem::path file, size_type size);
        void add_url_seed(std::string const& url);

        typedef std::vector<file_entry>::const_iterator file_iterator;
        typedef std::vector<file_entry>::const_reverse_iterator
                reverse_file_iterator;

        file_iterator begin_files() const;
        file_iterator end_files() const;
        reverse_file_iterator rbegin_files() const;
        reverse_file_iterator rend_files() const;

        int num_files() const;
        file_entry const& file_at(int index) const;

        std::vector<file_slice> map_block(int piece, size_type offset
                , int size) const;
        peer_request map_file(int file_index, size_type file_offset
                , int size) const;

        std::vector<announce_entry> const& trackers() const;

        std::vector<std::string> const& url_seeds() const;

        size_type total_size() const;
        size_type piece_length() const;
        int num_pieces() const;
        sha1_hash const& info_hash() const;
        std::stirng const& name() const;
        std::string const& comment() const;
        std::string const& creator() const;

        boost::optional<boost::posix_time::ptime>
        creation_date() const;


        void print(std::ostream& os) const;

        size_type piece_size(unsigned int index) const;
        sha1_hash const& hash_for_piece(unsigned int index) const;
};

torrent_info()

torrent_info();
torrent_info(sha1_hash const& info_hash);
torrent_info(entry const& torrent_file);

The default constructor of torrent_info is used when creating torrent files. It will initialize the object to an empty torrent, containing no files. The info hash will be set to 0 when this constructor is used. To use the empty torrent_info object, add files and piece hashes, announce URLs and optionally a creator tag and comment. To do this you use the members set_comment(), set_piece_size(), set_creator(), set_hash() etc.

The contructor that takes an info-hash is identical to the default constructor with the exception that it will initialize the info-hash to the given value. This is used internally when downloading torrents without the metadata. The metadata will be created by libtorrent as soon as it has been downloaded from the swarm.

The last constructor is the one that is used in most cases. It will create a torrent_info object from the information found in the given torrent_file. The entry represents a tree node in an bencoded file. To load an ordinary .torrent file into an entry, use bdecode(), see bdecode() bencode().

set_comment() set_piece_size() set_creator() set_hash() add_tracker() add_file()

void set_comment(char const* str);
void set_piece_size(int size);
void set_creator(char const* str);
void set_hash(int index, sha1_hash const& h);
void add_tracker(std::string const& url, int tier = 0);
void add_file(boost::filesystem::path file, size_type size);

These files are used when creating a torrent file. set_comment() will simply set the comment that belongs to this torrent. The comment can be retrieved with the comment() member. The string should be UTF-8 encoded.

set_piece_size() will set the size of each piece in this torrent. The piece size must be an even multiple of 2. i.e. usually something like 256 kiB, 512 kiB, 1024 kiB etc. The size is given in number of bytes.

set_creator() is an optional attribute that can be used to identify your application that was used to create the torrent file. The string should be UTF-8 encoded.

set_hash() writes the hash for the piece with the given piece-index. You have to call this function for every piece in the torrent. Usually the hasher is used to calculate the sha1-hash for a piece.

add_tracker() adds a tracker to the announce-list. The tier determines the order in which the trackers are to be tried. For more iformation see trackers().

add_file() adds a file to the torrent. The order in which you add files will determine the order in which they are placed in the torrent file. You have to add at least one file to the torrent. The path you give has to be a relative path from the root directory of the torrent. The size is given in bytes.

When you have added all the files and hashes to your torrent, you can generate an entry which then can be encoded as a .torrent file. You do this by calling create_torrent().

For a complete example of how to create a torrent from a file structure, see make_torrent.

create_torrent()

entry create_torrent();

Returns an entry representing the bencoded tree of data that makes up a .torrent file. You can save this data as a torrent file with bencode() (see bdecode() bencode()), for a complete example, see make_torrent.

This function is not const because it will also set the info-hash of the torrent_info object.

begin_files() end_files() rbegin_files() rend_files()

file_iterator begin_files() const;
file_iterator end_files() const;
reverse_file_iterator rbegin_files() const;
reverse_file_iterator rend_files() const;

This class will need some explanation. First of all, to get a list of all files in the torrent, you can use begin_files(), end_files(), rbegin_files() and rend_files(). These will give you standard vector iterators with the type file_entry.

The path is the full (relative) path of each file. i.e. if it is a multi-file torrent, all the files starts with a directory with the same name as torrent_info::name(). The filenames are encoded with UTF-8.

struct file_entry
{
        boost::filesystem::path path;
        size_type size;
};

num_files() file_at()

int num_files() const;
file_entry const& file_at(int index) const;

If you need index-access to files you can use the num_files() and file_at() to access files using indices.

map_block()

std::vector<file_slice> map_block(int piece, size_type offset
        , int size) const;

This function will map a piece index, a byte offset within that piece and a size (in bytes) into the corresponding files with offsets where that data for that piece is supposed to be stored.

The file slice struct looks like this:

struct file_slice
{
        int file_index;
        size_type offset;
        size_type size;
};

The file_index refers to the index of the file (in the torrent_info). To get the path and filename, use file_at() and give the file_index as argument. The offset is the byte offset in the file where the range starts, and size is the number of bytes this range is. The size + offset will never be greater than the file size.

map_file()

peer_request map_file(int file_index, size_type file_offset
        , int size) const;

This function will map a range in a specific file into a range in the torrent. The file_offset parameter is the offset in the file, given in bytes, where 0 is the start of the file. The peer_request structure looks like this:

struct peer_request
{
        int piece;
        int start;
        int length;
        bool operator==(peer_request const& r) const;
};

piece is the index of the piece in which the range starts. start is the offset within that piece where the range starts. length is the size of the range, in bytes.

The input range is assumed to be valid within the torrent. file_offset + size is not allowed to be greater than the file size. file_index must refer to a valid file, i.e. it cannot be >= num_files().

url_seeds()

std::vector<std::string> const& url_seeds() const;
void add_url_seed(std::string const& url);

If there are any url-seeds in this torrent, url_seeds() will return a vector of those urls. If you're creating a torrent file, add_url_seed() adds one url to the list of url-seeds. Currently, the only transport protocol supported for the url is http.

See HTTP seeding for more information.

print()

void print(std::ostream& os) const;

The print() function is there for debug purposes only. It will print the info from the torrent file to the given outstream.

trackers()

std::vector<announce_entry> const& trackers() const;

The trackers() function will return a sorted vector of announce_entry. Each announce entry contains a string, which is the tracker url, and a tier index. The tier index is the high-level priority. No matter which trackers that works or not, the ones with lower tier will always be tried before the one with higher tier number.

struct announce_entry
{
        announce_entry(std::string const& url);
        std::string url;
        int tier;
};

total_size() piece_length() piece_size() num_pieces()

size_type total_size() const;
size_type piece_length() const;
size_type piece_size(unsigned int index) const;
int num_pieces() const;

total_size(), piece_length() and num_pieces() returns the total number of bytes the torrent-file represents (all the files in it), the number of byte for each piece and the total number of pieces, respectively. The difference between piece_size() and piece_length() is that piece_size() takes the piece index as argument and gives you the exact size of that piece. It will always be the same as piece_length() except in the case of the last piece, which may be smaller.

hash_for_piece() info_hash()

size_type piece_size(unsigned int index) const;
sha1_hash const& hash_for_piece(unsigned int index) const;

hash_for_piece() takes a piece-index and returns the 20-bytes sha1-hash for that piece and info_hash() returns the 20-bytes sha1-hash for the info-section of the torrent file. For more information on the sha1_hash, see the big_number class. info_hash() will only return a valid hash if the torrent_info was read from a .torrent file or if an entry was created from it (through create_torrent).

name() comment() creation_date() creator()

std::string const& name() const;
std::string const& comment() const;
boost::optional<boost::posix_time::ptime> creation_date() const;

name() returns the name of the torrent.

comment() returns the comment associated with the torrent. If there's no comment, it will return an empty string. creation_date() returns a boost::posix_time::ptime object, representing the time when this torrent file was created. If there's no timestamp in the torrent file, this will return a date of january 1:st 1970.

Both the name and the comment is UTF-8 encoded strings.

creator() returns the creator string in the torrent. If there is no creator string it will return an empty string.

torrent_handle

You will usually have to store your torrent handles somewhere, since it's the object through which you retrieve infromation about the torrent and aborts the torrent. Its declaration looks like this:

struct torrent_handle
{
        torrent_handle();

        torrent_status status();
        void get_download_queue(std::vector<partial_piece_info>& queue) const;
        void get_peer_info(std::vector<peer_info>& v) const;
        torrent_info const& get_torrent_info() const;
        bool is_valid() const;

        entry write_resume_data() const;
        std::vector<char> const& metadata() const;
        void force_reannounce() const;
        void connect_peer(asio::ipv4::tcp::endpoint const& adr) const;

        void set_tracker_login(std::string const& username
                , std::string const& password) const;

        std::vector<announce_entry> const& trackers() const;
        void replace_trackers(std::vector<announce_entry> const&);

        void add_url_seed(std::string const& url);

        void set_ratio(float ratio) const;
        void set_max_uploads(int max_uploads) const;
        void set_max_connections(int max_connections) const;
        void set_upload_limit(int limit) const;
        void set_download_limit(int limit) const;

        void set_peer_upload_limit(asio::ipv4::tcp::endpoint ip, int limit) const;
        void set_peer_download_limit(asio::ipv4::tcp::endpoint ip, int limit) const;

        void use_interface(char const* net_interface) const;

        void pause() const;
        void resume() const;
        bool is_paused() const;
        bool is_seed() const;

        void filter_piece(int index, bool filter) const;
        void filter_pieces(std::vector<bool> const& bitmask) const;
        bool is_piece_filtered(int index) const;
        std::vector<bool> filtered_pieces() const;

        void filter_files(std::vector<bool> const& files) const;

        bool has_metadata() const;

        boost::filsystem::path save_path() const;
        bool move_storage(boost::filesystem::path const& save_path) const;

        sha1_hash info_hash() const;

        bool operator==(torrent_handle const&) const;
        bool operator!=(torrent_handle const&) const;
        bool operator<(torrent_handle const&) const;
};

The default constructor will initialize the handle to an invalid state. Which means you cannot perform any operation on it, unless you first assign it a valid handle. If you try to perform any operation on an uninitialized handle, it will throw invalid_handle.

TODO: document filter_piece(), filter_pieces(), is_piece_filtered(), filtered_pieces() and filter_files()

save_path()

boost::filsystem::path save_path() const;

save_path() returns the path that was given to add_torrent() when this torrent was started.

move_storage()

bool move_storage(boost::filsystem::path const& save_path) const;

Moves the file(s) that this torrent are currently seeding from or downloading to. This operation will only have the desired effect if the given save_path is located on the same drive as the original save path. If the move operation fails, this function returns false, otherwise true. Post condition for successful operation is: save_path() == save_path.

force_reannounce()

void force_reannounce() const;

force_reannounce() will force this torrent to do another tracker request, to receive new peers. If the torrent is invalid, queued or in checking mode, this functions will throw invalid_handle.

connect_peer()

void connect_peer(asio::ipv4::tcp::endpoint const& adr) const;

connect_peer() is a way to manually connect to peers that one believe is a part of the torrent. If the peer does not respond, or is not a member of this torrent, it will simply be disconnected. No harm can be done by using this other than an unnecessary connection attempt is made. If the torrent is uninitialized or in queued or checking mode, this will throw invalid_handle.

set_ratio()

void set_ratio(float ratio) const;

set_ratio() sets the desired download / upload ratio. If set to 0, it is considered being infinite. i.e. the client will always upload as much as it can, no matter how much it gets back in return. With this setting it will work much like the standard clients.

Besides 0, the ratio can be set to any number greater than or equal to 1. It means how much to attempt to upload in return for each download. e.g. if set to 2, the client will try to upload 2 bytes for every byte received. The default setting for this is 0, which will make it work as a standard client.

set_upload_limit() set_download_limit()

void set_upload_limit(int limit) const;
void set_download_limit(int limit) const;

set_upload_limit will limit the upload bandwidth used by this particular torrent to the limit you set. It is given as the number of bytes per second the torrent is allowed to upload. set_download_limit works the same way but for download bandwidth instead of upload bandwidth. Note that setting a higher limit on a torrent then the global limit (session::set_upload_rate_limit) will not override the global rate limit. The torrent can never upload more than the global rate limit.

set_peer_upload_limit() set_peer_download_limit()

void set_peer_upload_limit(asio::ipv4::tcp::endpoint ip, int limit) const;
void set_peer_download_limit(asio::ipv4::tcp::endpoint ip, int limit) const;

Works like set_upload_limit and set_download_limit respectively, but controls individual peer instead of the whole torrent.

pause() resume() is_paused()

void pause() const;
void resume() const;
bool is_paused() const;

pause(), and resume() will disconnect all peers and reconnect all peers respectively. When a torrent is paused, it will however remember all share ratios to all peers and remember all potential (not connected) peers. You can use is_paused() to determine if a torrent is currently paused. Torrents may be paused automatically if there is a file error (eg. disk full) or something similar. See file_error_alert.

is_seed()

bool is_seed() const;

Returns true if the torrent is in seed mode (i.e. if it has finished downloading).

has_metadata()

bool has_metadata() const;

Returns true if this torrent has metadata (either it was started from a .torrent file or the metadata has been downloaded). The only scenario where this can return false is when the torrent was started torrent-less (i.e. with just an info-hash and tracker ip). Note that if the torrent doesn't have metadata, the member get_torrent_info() will throw.

set_tracker_login()

void set_tracker_login(std::string const& username
        , std::string const& password) const;

set_tracker_login() sets a username and password that will be sent along in the HTTP-request of the tracker announce. Set this if the tracker requires authorization.

trackers() replace_trackers()

std::vector<announce_entry> const& trackers() const;
void replace_trackers(std::vector<announce_entry> const&) const;

trackers() will return the list of trackers for this torrent. The announce entry contains both a string url which specifu the announce url for the tracker as well as an int tier, which is specifies the order in which this tracker is tried. If you want libtorrent to use another list of trackers for this torrent, you can use replace_trackers() which takes a list of the same form as the one returned from trackers() and will replace it. If you want an immediate effect, you have to call force_reannounce().

add_url_seed()

void add_url_seed(std::string const& url);

add_url_seed() adds another url to the torrent's list of url seeds. If the given url already exists in that list, the call has no effect. The torrent will connect to the server and try to download pieces from it, unless it's paused, queued, checking or seeding.

See HTTP seeding for more information.

use_interface()

void use_interface(char const* net_interface) const;

use_interface() sets the network interface this torrent will use when it opens outgoing connections. By default, it uses the same interface as the session uses to listen on. The parameter can be a string containing an ip-address or a hostname.

info_hash()

sha1_hash info_hash() const;

info_hash() returns the info-hash for the torrent.

set_max_uploads() set_max_connections()

void set_max_uploads(int max_uploads) const;
void set_max_connections(int max_connections) const;

set_max_uploads() sets the maximum number of peers that's unchoked at the same time on this torrent. If you set this to -1, there will be no limit.

set_max_connections() sets the maximum number of connection this torrent will open. If all connections are used up, incoming connections may be refused or poor connections may be closed. This must be at least 2. The default is unlimited number of connections. If -1 is given to the function, it means unlimited.

write_resume_data()

entry write_resume_data() const;

write_resume_data() generates fast-resume data and returns it as an entry. This entry is suitable for being bencoded. For more information about how fast-resume works, see fast resume.

There are three cases where this function will just return an empty entry:

  1. The torrent handle is invalid.
  2. The torrent is checking (or is queued for checking) its storage, it will obviously not be ready to write resume data.
  3. The torrent hasn't received valid metadata and was started without metadata (see libtorrent's metadata from peers extension)

Note that by the time this function returns, the resume data may already be invalid if the torrent is still downloading! The recommended practice is to first pause the torrent, then generate the fast resume data, and then close it down.

metadata()

std::vector<char> const& metadata() const;

metadata() will return a reference to a buffer containing the exact info part of the .torrent file. This buffer will be valid as long as the torrent is still running. When hashed, it will produce the same hash as the info-hash.

status()

torrent_status status() const;

status() will return a structure with information about the status of this torrent. If the torrent_handle is invalid, it will throw invalid_handle exception. See torrent_status.

get_download_queue()

void get_download_queue(std::vector<partial_piece_info>& queue) const;

get_download_queue() takes a non-const reference to a vector which it will fill with information about pieces that are partially downloaded or not downloaded at all but partially requested. The entry in the vector (partial_piece_info) looks like this:

struct partial_piece_info
{
        enum { max_blocks_per_piece };
        int piece_index;
        int blocks_in_piece;
        std::bitset<max_blocks_per_piece> requested_blocks;
        std::bitset<max_blocks_per_piece> finished_blocks;
        address peer[max_blocks_per_piece];
        int num_downloads[max_blocks_per_piece];
};

piece_index is the index of the piece in question. blocks_in_piece is the number of blocks in this particular piece. This number will be the same for most pieces, but the last piece may have fewer blocks than the standard pieces.

requested_blocks is a bitset with one bit per block in the piece. If a bit is set, it means that that block has been requested, but not necessarily fully downloaded yet. To know from whom the block has been requested, have a look in the peer array. The bit-index in the requested_blocks and finished_blocks correspons to the array-index into peers and num_downloads. The array of peers is contains the address of the peer the piece was requested from. If a piece hasn't been requested (the bit in requested_blocks is not set) the peer array entry will be undefined.

The finished_blocks is a bitset where each bit says if the block is fully downloaded or not. And the num_downloads array says how many times that block has been downloaded. When a piece fails a hash verification, single blocks may be redownloaded to see if the hash teast may pass then.

get_peer_info()

void get_peer_info(std::vector<peer_info>&) const;

get_peer_info() takes a reference to a vector that will be cleared and filled with one entry for each peer connected to this torrent, given the handle is valid. If the torrent_handle is invalid, it will throw invalid_handle exception. Each entry in the vector contains information about that particular peer. See peer_info.

get_torrent_info()

torrent_info const& get_torrent_info() const;

Returns a const reference to the torrent_info object associated with this torrent. This reference is valid as long as the torrent_handle is valid, no longer. If the torrent_handle is invalid or if it doesn't have any metadata, invalid_handle exception will be thrown. The torrent may be in a state without metadata only if it was started without a .torrent file, i.e. by using the libtorrent extension of just supplying a tracker and info-hash.

is_valid()

bool is_valid() const;

Returns true if this handle refers to a valid torrent and false if it hasn't been initialized or if the torrent it refers to has been aborted. Note that a handle may become invalid after it has been added to the session. Usually this is because the storage for the torrent is somehow invalid or if the filenames are not allowed (and hence cannot be opened/created) on your filesystem. If such an error occurs, a file_error_alert is generated and all handles that refers to that torrent will become invalid.

TODO: document storage

torrent_status

It contains the following fields:

struct torrent_status
{
        enum state_t
        {
                queued_for_checking,
                checking_files,
                connecting_to_tracker,
                downloading,
                finished,
                seeding,
                allocating
        };

        state_t state;
        bool paused;
        float progress;
        boost::posix_time::time_duration next_announce;
        boost::posix_time::time_duration announce_interval;

        std::string current_tracker;

        size_type total_download;
        size_type total_upload;

        size_type total_payload_download;
        size_type total_payload_upload;

        size_type total_failed_bytes;
        size_type total_redundant_bytes;

        float download_rate;
        float upload_rate;

        float download_payload_rate;
        float upload_payload_rate;

        int num_peers;

        int num_complete;
        int num_incomplete;

        const std::vector<bool>* pieces;
        int num_pieces;

        size_type total_done;
        size_type total_wanted_done;
        size_type total_wanted;

        int num_seeds;
        float distributed_copies;

        int block_size;
};

progress is a value in the range [0, 1], that represents the progress of the torrent's current task. It may be checking files or downloading. The torrent's current task is in the state member, it will be one of the following:

queued_for_checking The torrent is in the queue for being checked. But there currently is another torrent that are being checked. This torrent will wait for its turn.
checking_files The torrent has not started its download yet, and is currently checking existing files.
connecting_to_tracker The torrent has sent a request to the tracker and is currently waiting for a response
downloading The torrent is being downloaded. This is the state most torrents will be in most of the time. The progress meter will tell how much of the files that has been downloaded.
finished In this state the torrent has finished downloading but still doesn't have the entire torrent. i.e. some pieces are filtered and won't get downloaded.
seeding In this state the torrent has finished downloading and is a pure seeder.
allocating If the torrent was started in full allocation mode, this indicates that the (disk) storage for the torrent is allocated.

When downloading, the progress is total_wanted_done / total_wanted.

paused is set to true if the torrent is paused and false otherwise.

next_announce is the time until the torrent will announce itself to the tracker. And announce_interval is the time the tracker want us to wait until we announce ourself again the next time.

current_tracker is the URL of the last working tracker. If no tracker request has been successful yet, it's set to an empty string.

total_download and total_upload is the number of bytes downloaded and uploaded to all peers, accumulated, this session only.

total_payload_download and total_payload_upload counts the amount of bytes send and received this session, but only the actual oayload data (i.e the interesting data), these counters ignore any protocol overhead.

total_failed_bytes is the number of bytes that has been downloaded and that has failed the piece hash test. In other words, this is just how much crap that has been downloaded.

total_redundant_bytes is the number of bytes that has been downloaded even though that data already was downloaded. The reason for this is that in some situations the same data can be downloaded by mistake. When libtorrent sends requests to a peer, and the peer doesn't send a response within a certain timeout, libtorrent will re-request that block. Another situation when libtorrent will re-request blocks is when the requests it sends out are not replyed in FIFO-order (it will re-request blocks that are skipped by an out of order block). This is supposed to be as low as possible.

pieces is the bitmask that represents which pieces we have (set to true) and the pieces we don't have. It's a pointer and may be set to 0 if the torrent isn't downloading or seeding.

num_pieces is the number of pieces that has been downloaded. It is equivalent to: std::accumulate(pieces->begin(), pieces->end()). So you don't have to count yourself. This can be used to see if anything has updated since last time if you want to keep a graph of the pieces up to date.

download_rate and upload_rate are the total rates for all peers for this torrent. These will usually have better precision than summing the rates from all peers. The rates are given as the number of bytes per second. The download_payload_rate and upload_payload_rate respectively is the total transfer rate of payload only, not counting protocol chatter. This might be slightly smaller than the other rates, but if projected over a long time (e.g. when calculating ETA:s) the difference may be noticable.

num_peers is the number of peers this torrent currently is connected to. Peer connections that are in the half-open state (is attempting to connect) or are queued for later connection attempt do not count. Although they are visible in the peer list when you call get_peer_info().

num_complete and num_incomplete are set to -1 if the tracker did not send any scrape data in its announce reply. This data is optional and may not be available from all trackers. If these are not -1, they are the total number of peers that are seeding (complete) and the total number of peers that are still downloading (incomplete) this torrent.

total_done is the total number of bytes of the file(s) that we have. All this does not necessarily has to be downloaded during this session (that's total_download_payload).

total_wanted_done is the number of bytes we have downloadd, only counting the pieces that we actually want to download. i.e. excluding any pieces that we have but are filtered as not wanted.

total_wanted is the total number of bytes we want to download. This is also excluding pieces that have been filtered.

num_seeds is the number of peers that are seeding that this client is currently connected to.

distributed_copies is the number of distributed copies of the torrent. Note that one copy may be spread out among many peers. The integer part tells how many copies there are currently of the rarest piece(s) among the peers this client is connected to. The fractional part tells the share of pieces that have more copies than the rarest piece(s). For example: 2.5 would mean that the rarest pieces have only 2 copies among the peers this torrent is connected to, and that 50% of all the pieces have more than two copies.

block_size is the size of a block, in bytes. A block is a sub piece, it is the number of bytes that each piece request asks for and the number of bytes that each bit in the partial_piece_info's bitset represents (see get_download_queue()). This is typically 16 kB, but it may be larger if the pieces are larger.

peer_info

It contains the following fields:

struct peer_info
{
        enum
        {
                interesting = 0x1,
                choked = 0x2,
                remote_interested = 0x4,
                remote_choked = 0x8,
                supports_extensions = 0x10,
                local_connection = 0x20,
                handshake = 0x40,
                connecting = 0x80,
                queued = 0x100
        };
        unsigned int flags;
        asio::ipv4::tcp::endpoint ip;
        float up_speed;
        float down_speed;
        float payload_up_speed;
        float payload_down_speed;
        size_type total_download;
        size_type total_upload;
        peer_id pid;
        std::vector<bool> pieces;
        bool seed;
        int upload_limit;
        int download_limit;

        size_type load_balancing;

        int download_queue_length;
        int upload_queue_length;

        int downloading_piece_index;
        int downloading_block_index;
        int downloading_progress;
        int downloading_total;

        std::string client;

        enum
        {
                standard_bittorrent = 0,
                web_seed = 1
        };
        int connection_type;
};

The flags attribute tells you in which state the peer is. It is set to any combination of the enums above. The following table describes each flag:

interesting we are interested in pieces from this peer.
choked we have choked this peer.
remote_interested the peer is interested in us
remote_choked the peer has choked us.
support_extensions means that this peer supports the extension protocol.
local_connection The connection was initiated by us, the peer has a listen port open, and that port is the same as in the address of this peer. If this flag is not set, this peer connection was opened by this peer connecting to us.
handshake The connection is opened, and waiting for the handshake. Until the handshake is done, the peer cannot be identified.
connecting The connection is in a half-open state (i.e. it is being connected).
queued The connection is currently queued for a connection attempt. This may happen if there is a limit set on the number of half-open TCP connections.

The ip field is the IP-address to this peer. The type is an asio endpoint. For more info, see the asio documentation.

up_speed and down_speed contains the current upload and download speed we have to and from this peer (including any protocol messages). The transfer rates of payload data only are found in payload_up_speed and payload_down_speed. These figures are updated aproximately once every second.

total_download and total_upload are the total number of bytes downloaded from and uploaded to this peer. These numbers do not include the protocol chatter, but only the payload data.

pid is the peer's id as used in the bit torrent protocol. This id can be used to extract 'fingerprints' from the peer. Sometimes it can tell you which client the peer is using. See identify_client()_

pieces is a vector of booleans that has as many entries as there are pieces in the torrent. Each boolean tells you if the peer has that piece (if it's set to true) or if the peer miss that piece (set to false).

seed is true if this peer is a seed.

upload_limit is the number of bytes per second we are allowed to send to this peer every second. It may be -1 if there's no limit. The upload limits of all peers should sum up to the upload limit set by session::set_upload_limit.

download_limit is the number of bytes per second this peer is allowed to receive. -1 means it's unlimited.

load_balancing is a measurment of the balancing of free download (that we get) and free upload that we give. Every peer gets a certain amount of free upload, but this member says how much extra free upload this peer has got. If it is a negative number it means that this was a peer from which we have got this amount of free download.

download_queue_length is the number of piece-requests we have sent to this peer that hasn't been answered with a piece yet.

upload_queue_length is the number of piece-requests we have received from this peer that we haven't answered with a piece yet.

You can know which piece, and which part of that piece, that is currently being downloaded from a specific peer by looking at the next four members. downloading_piece_index is the index of the piece that is currently being downloaded. This may be set to -1 if there's currently no piece downloading from this peer. If it is >= 0, the other three members are valid. downloading_block_index is the index of the block (or sub-piece) that is being downloaded. downloading_progress is the number of bytes of this block we have received from the peer, and downloading_total is the total number of bytes in this block.

client is a string describing the software at the other end of the connection. In some cases this information is not available, then it will contain a string that may give away something about which software is running in the other end. In the case of a web seed, the server type and version will be a part of this string.

connection_type can currently be one of standard_bittorrent or web_seed. These are currently the only implemented protocols.

http_settings

You have some control over tracker requests through the http_settings object. You create it and fill it with your settings and then use session::set_http_settings() to apply them. You have control over proxy and authorization settings and also the user-agent that will be sent to the tracker. The user-agent is a good way to identify your client.

struct http_settings
{
        http_settings();
        std::string proxy_ip;
        int proxy_port;
        std::string proxy_login;
        std::string proxy_password;
        std::string user_agent;
        int tracker_completion_timeout;
        int tracker_receive_timeout;
        int tracker_maximum_response_length;
};

proxy_ip may be a hostname or ip to a http proxy to use. If this is an empty string, no http proxy will be used.

proxy_port is the port on which the http proxy listens. If proxy_ip is empty, this will be ignored.

proxy_login should be the login username for the http proxy, if this empty, the http proxy will be tried to be used without authentication.

proxy_password the password string for the http proxy.

user_agent this is the client identification to the tracker. It will be followed by the string "(libtorrent)" to identify that this library is being used. This should be set to your client's name and version number. This name will not only be used when making HTTP requests, but also when sending extended headers to peers that support that extension.

tracker_completion_timeout is the number of seconds the tracker connection will wait from when it sent the request until it considers the tracker to have timed-out. Default value is 60 seconds.

tracker_receive_timeout is the number of seconds to wait to receive any data from the tracker. If no data is received for this number of seconds, the tracker will be considered as having timed out. If a tracker is down, this is the kind of timeout that will occur. The default value is 20 seconds.

tracker_maximum_response_length is the maximum number of bytes in a tracker response. If a response size passes this number it will be rejected and the connection will be closed. On gzipped responses this size is measured on the uncompressed data. So, if you get 20 bytes of gzip response that'll expand to 2 megs, it will be interrupted before the entire response has been uncompressed (given your limit is lower than 2 megs). Default limit is 1 megabyte.

ip_filter

The ip_filter class is a set of rules that uniquely categorizes all ip addresses as allowed or disallowed. The default constructor creates a single rule that allowes all addresses (0.0.0.0 - 255.255.255.255). The address type here is asio::ipv4::address. It can also be accessed as libtorrent::address.

class ip_filter
{
public:
        enum access_flags { blocked = 1 };

        ip_filter();
        void add_rule(address first, address last, int flags);
        int access(address const& addr) const;

        struct ip_range
        {
                address first;
                address last;
                int flags;
        };

        std::vector<ip_range> export_filter() const;
};

ip_filter()

ip_filter()

Creates a default filter that doesn't filter any address.

postcondition: access(x) == 0 for every x

add_rule()

void add_rule(address first, address last, int flags);

Adds a rule to the filter. first and last defines a range of ip addresses that will be marked with the given flags. The flags can currenly be 0, which means allowed, or ip_filter::blocked, which means disallowed.

postcondition: access(x) == flags for every x in the range [first, last]

This means that in a case of overlapping ranges, the last one applied takes precedence.

access()

int access(address const& addr) const;

Returns the access permissions for the given address (addr). The permission can currently be 0 or ip_filter::blocked. The complexity of this operation is O(log n), where n is the minimum number of non-overlapping ranges to describe the current filter.

export_filter()

std::vector<ip_range> export_filter() const;

This function will return the current state of the filter in the minimum number of ranges possible. They are sorted from ranges in low addresses to high addresses. Each entry in the returned vector is a range with the access control specified in its flags field.

big_number

Both the peer_id and sha1_hash types are typedefs of the class big_number. It represents 20 bytes of data. Its synopsis follows:

class big_number
{
public:
        bool operator==(const big_number& n) const;
        bool operator!=(const big_number& n) const;
        bool operator<(const big_number& n) const;

        const unsigned char* begin() const;
        const unsigned char* end() const;

        unsigned char* begin();
        unsigned char* end();
};

The iterators gives you access to individual bytes.

hasher

This class creates sha1-hashes. Its declaration looks like this:

class hasher
{
public:
        hasher();
        hasher(char const* data, unsigned int len);

        void update(char const* data, unsigned int len);
        sha1_hash final();
        void reset();
};

You use it by first instantiating it, then call update() to feed it with data. i.e. you don't have to keep the entire buffer of which you want to create the hash in memory. You can feed the hasher parts of it at a time. When You have fed the hasher with all the data, you call final() and it will return the sha1-hash of the data.

The constructor that takes a char const* and an integer will construct the sha1 context and feed it the data passed in.

If you want to reuse the hasher object once you have created a hash, you have to call reset() to reinitialize it.

The sha1-algorithm used was implemented by Steve Reid and released as public domain. For more info, see src/sha1.cpp.

fingerprint

The fingerprint class represents information about a client and its version. It is used to encode this information into the client's peer id.

This is the class declaration:

struct fingerprint
{
        fingerprint(const char* id_string, int major, int minor
                , int revision, int tag);

        std::string to_string() const;

        char name[2];
        char major_version;
        char minor_version;
        char revision_version;
        char tag_version;

};

The constructor takes a char const* that should point to a string constant containing exactly two characters. These are the characters that should be unique for your client. Make sure not to clash with anybody else. Here are some taken id's:

id chars client
'AZ' Azureus
'LT' libtorrent (default)
'BX' BittorrentX
'MT' Moonlight Torrent
'TS' Torrent Storm
'SS' Swarm Scope
'XT' Xan Torrent

There's currently an informal directory of client id's here.

The major, minor, revision and tag parameters are used to identify the version of your client. All these numbers must be within the range [0, 9].

to_string() will generate the actual string put in the peer-id, and return it.

free functions

identify_client()

std::string identify_client(peer_id const& id);

This function is declared in the header <libtorrent/identify_client.hpp>. It can can be used to extract a string describing a client version from its peer-id. It will recognize most clients that have this kind of identification in the peer-id.

bdecode() bencode()

template<class InIt> entry bdecode(InIt start, InIt end);
template<class OutIt> void bencode(OutIt out, const entry& e);

These functions will encode data to bencoded or decode bencoded data.

The entry class is the internal representation of the bencoded data and it can be used to retreive information, an entry can also be build by the program and given to bencode() to encode it into the OutIt iterator.

The OutIt and InIt are iterators (InputIterator and OutputIterator respectively). They are templates and are usually instantiated as ostream_iterator, back_insert_iterator or istream_iterator. These functions will assume that the iterator refers to a character (char). So, if you want to encode entry e into a buffer in memory, you can do it like this:

std::vector<char> buffer;
bencode(std::back_inserter(buf), e);

If you want to decode a torrent file from a buffer in memory, you can do it like this:

std::vector<char> buffer;
// ...
entry e = bdecode(buf.begin(), buf.end());

Or, if you have a raw char buffer:

const char* buf;
// ...
entry e = bdecode(buf, buf + data_size);

Now we just need to know how to retrieve information from the entry.

If bdecode() encounters invalid encoded data in the range given to it it will throw invalid_encoding.

alerts

The pop_alert() function on session is the interface for retrieving alerts, warnings, messages and errors from libtorrent. If there hasn't occured any errors (matching your severity level) pop_alert() will return a zero pointer. If there has been some error, it will return a pointer to an alert object describing it. You can then use the alert object and query it for information about the error or message. To retrieve any alerts, you have to select a severity level using session::set_severity_level(). It defaults to alert::none, which means that you don't get any messages at all, ever. You have the following levels to select among:

none No alert will ever have this severity level, which effectively filters all messages.
fatal Fatal errors will have this severity level. Examples can be disk full or something else that will make it impossible to continue normal execution.
critical Signals errors that requires user interaction or messages that almost never should be ignored. For example, a chat message received from another peer is announced as severity critical.
warning Messages with the warning severity can be a tracker that times out or responds with invalid data. It will be retried automatically, and the possible next tracker in a multitracker sequence will be tried. It does not require any user interaction.
info Events that can be considered normal, but still deserves an event. This could be a piece hash that fails.
debug This will include alot of debug events that can be used both for debugging libtorrent but also when debugging other clients that are connected to libtorrent. It will report strange behaviors among the connected peers.

When setting a severity level, you will receive messages of that severity and all messages that are more sever. If you set alert::none (the default) you will not recieve any events at all.

When you set a severuty level other than none, you have the responsibility to call pop_alert() from time to time. If you don't do that, the alert queue will just grow.

When you get an alert, you can use typeid() or dynamic_cast<> to get more detailed information on exactly which type it is. i.e. what kind of error it is. You can also use a dispatcher mechanism that's available in libtorrent.

All alert types are defined in the <libtorrent/alert_types.hpp> header file.

The alert class is the base class that specific messages are derived from. This is its synopsis:

class alert
{
public:

        enum severity_t { debug, info, warning, critital, fatal, none };

        alert(severity_t severity, const std::string& msg);
        virtual ~alert();

        std::string const& msg() const;
        severity_t severity() const;

        virtual std::auto_ptr<alert> clone() const = 0;
};

This means that all alerts have at least a string describing it. They also have a severity level that can be used to sort them or present them to the user in different ways.

The specific alerts, that all derives from alert, are:

listen_failed_alert

This alert is generated when none of the ports, given in the port range, to session can be opened for listening. This alert is generated as severity level fatal.

struct listen_failed_alert: alert
{
        listen_failed_alert(const std::string& msg);
        virtual std::auto_ptr<alert> clone() const;
};

file_error_alert

If the storage fails to read or write files that it needs access to, this alert is generated and the torrent is paused. It is generated as severity level fatal.

struct file_error_alert: alert
{
        file_error_alert(
                const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

tracker_announce_alert

This alert is generated each time a tracker announce is sent (or attempted to be sent). It is generated at severity level info.

struct tracker_announce_alert: alert
{
        tracker_announce_alert(
                const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

tracker_alert

This alert is generated on tracker time outs, premature disconnects, invalid response or a HTTP response other than "200 OK". From the alert you can get the handle to the torrent the tracker belongs to. This alert is generated as severity level warning.

The times_in_row member says how many times in a row this tracker has failed. status_code is the code returned from the HTTP server. 401 means the tracker needs authentication, 404 means not found etc. If the tracker timed out, the code will be set to 0.

struct tracker_alert: alert
{
        tracker_alert(const torrent_handle& h, int times, int status
                , const std::string& msg);
        virtual std::auto_ptr<alert> clone() const;

        torrent_handle handle;
        int times_in_row;
        int status_code;
};

tracker_reply_alert

This alert is only for informational purpose. It is generated when a tracker announce succeeds. It is generated with severity level info.

struct tracker_reply_alert: alert
{
        tracker_reply_alert(const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

tracker_warning_alert

This alert is triggered if the tracker reply contains a warning field. Usually this means that the tracker announce was successful, but the tracker has a message to the client. The message string in the alert will contain the warning message from the tracker. It is generated with severity level warning.

struct tracker_warning_alert: alert
{
        tracker_warning_alert(torrent_handle const& h
                , std::string const& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

url_seed_alert

This alert is generated when a HTTP seed name lookup fails. This alert is generated as severity level warning.

It contains url to the HTTP seed that failed along with an error message.

struct url_seed_alert: alert
{
        url_seed_alert(std::string const& h, const std::string& msg);
        virtual std::auto_ptr<alert> clone() const;

        std::string url;
};

hash_failed_alert

This alert is generated when a finished piece fails its hash check. You can get the handle to the torrent which got the failed piece and the index of the piece itself from the alert. This alert is generated as severity level info.

struct hash_failed_alert: alert
{
        hash_failed_alert(
                const torrent_handle& h
                , int index
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
        int piece_index;
};

peer_ban_alert

This alert is generated when a peer is banned because it has sent too many corrupt pieces to us. It is generated at severity level info. The handle member is a torrent_handle to the torrent that this peer was a member of.

struct peer_ban_alert: alert
{
        peer_ban_alert(
                asio::ipv4::tcp::endpoint const& pip
                , torrent_handle h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        asio::ipv4::tcp::endpoint ip;
        torrent_handle handle;
};

peer_error_alert

This alert is generated when a peer sends invalid data over the peer-peer protocol. The peer will be disconnected, but you get its ip address from the alert, to identify it. This alert is generated as severity level debug.

struct peer_error_alert: alert
{
        peer_error_alert(
                asio::ipv4::tcp::endpoint const& pip
                , peer_id const& pid
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        asio::ipv4::tcp::endpoint ip;
        peer_id id;
};

invalid_request_alert

This is a debug alert that is generated by an incoming invalid piece request. The handle is a handle to the torrent the peer is a member of. ìp is the address of the peer and the request is the actual incoming request from the peer. The alert is generated as severity level debug.

struct invalid_request_alert: alert
{
        invalid_request_alert(
                peer_request const& r
                , torrent_handle const& h
                , asio::ipv4::tcp::endpoint const& send
                , peer_id const& pid
                , std::string const& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
        asio::ipv4::tcp::endpoint ip;
        peer_request request;
        peer_id id;
};


struct peer_request
{
        int piece;
        int start;
        int length;
        bool operator==(peer_request const& r) const;
};

The peer_request contains the values the client sent in its request message. piece is the index of the piece it want data from, start is the offset within the piece where the data should be read, and length is the amount of data it wants.

torrent_finished_alert

This alert is generated when a torrent switches from being a downloader to a seed. It will only be generated once per torrent. It contains a torrent_handle to the torrent in question. This alert is generated as severity level info.

struct torrent_finished_alert: alert
{
        torrent_finished_alert(
                const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

metadata_failed_alert

This alert is generated when the metadata has been completely received and the info-hash failed to match it. i.e. the metadata that was received was corrupt. libtorrent will automatically retry to fetch it in this case. This is only relevant when running a torrent-less download, with the metadata extension provided by libtorrent. It is generated at severity level info.

struct metadata_received_alert: alert
{
        metadata_received_alert(
                const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

metadata_received_alert

This alert is generated when the metadata has been completely received and the torrent can start downloading. It is not generated on torrents that are started with metadata, but only those that needs to download it from peers (when utilizing the libtorrent extension). It is generated at severity level info.

struct metadata_received_alert: alert
{
        metadata_received_alert(
                const torrent_handle& h
                , const std::string& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

fastresume_rejected_alert

This alert is generated when a fastresume file has been passed to add_torrent but the files on disk did not match the fastresume file. The string explaints the reason why the resume file was rejected. It is generated at severity level warning.

struct fastresume_rejected_alert: alert
{
        fastresume_rejected_alert(torrent_handle const& h
                , std::string const& msg);

        virtual std::auto_ptr<alert> clone() const;
        torrent_handle handle;
};

dispatcher

TODO: describe the dispatcher mechanism

exceptions

There are a number of exceptions that can be thrown from different places in libtorrent, here's a complete list with description.

invalid_handle

This exception is thrown when querying information from a torrent_handle that hasn't been initialized or that has become invalid.

struct invalid_handle: std::exception
{
        const char* what() const throw();
};

duplicate_torrent

This is thrown by add_torrent() if the torrent already has been added to the session.

struct duplicate_torrent: std::exception
{
        const char* what() const throw();
};

invalid_encoding

This is thrown by bdecode() if the input data is not a valid bencoding.

struct invalid_encoding: std::exception
{
        const char* what() const throw();
};

type_error

This is thrown from the accessors of entry if the data type of the entry doesn't match the type you want to extract from it.

struct type_error: std::runtime_error
{
        type_error(const char* error);
};

invalid_torrent_file

This exception is thrown from the constructor of torrent_info if the given bencoded information doesn't meet the requirements on what information has to be present in a torrent file.

struct invalid_torrent_file: std::exception
{
        const char* what() const throw();
};

examples

Except for the example programs in this manual, there's also a bigger example of a (little bit) more complete client, client_test. There are separate instructions for how to use it here if you'd like to try it.

dump_torrent

This is an example of a program that will take a torrent-file as a parameter and print information about it to std out:

#include <iostream>
#include <fstream>
#include <iterator>
#include <iomanip>

#include "libtorrent/entry.hpp"
#include "libtorrent/bencode.hpp"
#include "libtorrent/torrent_info.hpp"


int main(int argc, char* argv[])
{
        using namespace libtorrent;

        if (argc != 2)
        {
                std::cerr << "usage: dump_torrent torrent-file\n";
                return 1;
        }

        try
        {
                std::ifstream in(argv[1], std::ios_base::binary);
                in.unsetf(std::ios_base::skipws);
                entry e = bdecode(std::istream_iterator<char>(in), std::istream_iterator<char>());

                std::cout << "\n\n----- raw info -----\n\n";
                e.print(std::cout);

                torrent_info t(e);

                // print info about torrent
                std::cout << "\n\n----- torrent file info -----\n\n";
                std::cout << "trackers:\n";
                for (std::vector<announce_entry>::const_iterator i = t.trackers().begin();
                        i != t.trackers().end(); ++i)
                {
                        std::cout << i->tier << ": " << i->url << "\n";
                }

                std::cout << "number of pieces: " << t.num_pieces() << "\n";
                std::cout << "piece length: " << t.piece_length() << "\n";
                std::cout << "info hash: " << t.info_hash() << "\n";
                std::cout << "comment: " << t.comment() << "\n";
                std::cout << "created by: " << t.creator() << "\n";
                std::cout << "files:\n";
                for (torrent_info::file_iterator i = t.begin_files();
                        i != t.end_files(); ++i)
                {
                        std::cout << "  " << std::setw(11) << i->size
                                << " " << i->path.string() << "\n";
                }

        }
        catch (std::exception& e)
        {
                std::cout << e.what() << "\n";
        }

        return 0;
}

simple client

This is a simple client. It doesn't have much output to keep it simple:

#include <iostream>
#include <fstream>
#include <iterator>
#include <exception>

#include <boost/format.hpp>
#include <boost/date_time/posix_time/posix_time.hpp>

#include "libtorrent/entry.hpp"
#include "libtorrent/bencode.hpp"
#include "libtorrent/session.hpp"
#include "libtorrent/http_settings.hpp"

int main(int argc, char* argv[])
{
        using namespace libtorrent;

        if (argc != 2)
        {
                std::cerr << "usage: ./simple_cient torrent-file\n"
                        "to stop the client, press return.\n";
                return 1;
        }

        try
        {
                session s;
                s.listen_on(std::make_pair(6881, 6889));

                std::ifstream in(argv[1], std::ios_base::binary);
                in.unsetf(std::ios_base::skipws);
                entry e = bdecode(std::istream_iterator<char>(in)
                        , std::istream_iterator<char>());
                s.add_torrent(torrent_info(e), "");

                // wait for the user to end
                char a;
                std::cin.unsetf(std::ios_base::skipws);
                std::cin >> a;
        }
        catch (std::exception& e)
        {
                std::cout << e.what() << "\n";
        }
        return 0;
}

make_torrent

Shows how to create a torrent from a directory tree:

#include <iostream>
#include <fstream>
#include <iterator>
#include <iomanip>

#include "libtorrent/entry.hpp"
#include "libtorrent/bencode.hpp"
#include "libtorrent/torrent_info.hpp"
#include "libtorrent/file.hpp"
#include "libtorrent/storage.hpp"
#include "libtorrent/hasher.hpp"

#include <boost/filesystem/operations.hpp>
#include <boost/filesystem/path.hpp>
#include <boost/filesystem/fstream.hpp>

using namespace boost::filesystem;
using namespace libtorrent;

void add_files(torrent_info& t, path const& p, path const& l)
{
        path f(p / l);
        if (is_directory(f))
        {
                for (directory_iterator i(f), end; i != end; ++i)
                        add_files(t, p, l / i->leaf());
        }
        else
        {
                std::cerr << "adding \"" << l.string() << "\"\n";
                file fi(f, file::in);
                fi.seek(0, file::end);
                libtorrent::size_type size = fi.tell();
                t.add_file(l, size);
        }
}

int main(int argc, char* argv[])
{
        using namespace libtorrent;
        using namespace boost::filesystem;

        if (argc != 4)
        {
                std::cerr << "usage: make_torrent <output torrent-file> "
                        "<announce url> <file or directory to create torrent from>\n";
                return 1;
        }

        boost::filesystem::path::default_name_check(native);

        try
        {
                torrent_info t;
                path full_path = initial_path() / path(argv[3]);
                ofstream out(initial_path() / path(argv[1]), std::ios_base::binary);

                int piece_size = 256 * 1024;
                char const* creator_str = "libtorrent";

                add_files(t, full_path.branch_path(), full_path.leaf());
                t.set_piece_size(piece_size);

                storage st(t, full_path.branch_path());
                t.add_tracker(argv[2]);

                // calculate the hash for all pieces
                int num = t.num_pieces();
                std::vector<char> buf(piece_size);
                for (int i = 0; i < num; ++i)
                {
                        st.read(&buf[0], i, 0, t.piece_size(i));
                        hasher h(&buf[0], t.piece_size(i));
                        t.set_hash(i, h.final());
                        std::cerr << (i+1) << "/" << num << "\r";
                }

                t.set_creator(creator_str);

                // create the torrent and print it to out
                entry e = t.create_torrent();
                libtorrent::bencode(std::ostream_iterator<char>(out), e);
        }
        catch (std::exception& e)
        {
                std::cerr << e.what() << "\n";
        }

        return 0;
}

fast resume

The fast resume mechanism is a way to remember which pieces are downloaded and where they are put between sessions. You can generate fast resume data by calling torrent_handle::write_resume_data() on torrent_handle. You can then save this data to disk and use it when resuming the torrent. libtorrent will not check the piece hashes then, and rely on the information given in the fast-resume data. The fast-resume data also contains information about which blocks, in the unfinished pieces, were downloaded, so it will not have to start from scratch on the partially downloaded pieces.

To use the fast-resume data you simply give it to add_torrent(), and it will skip the time consuming checks. It may have to do the checking anyway, if the fast-resume data is corrupt or doesn't fit the storage for that torrent, then it will not trust the fast-resume data and just do the checking.

file format

The file format is a bencoded dictionary containing the following fields:

file-format string: "libtorrent resume file"
file-version integer: 1
info-hash string, the info hash of the torrent this data is saved for.
blocks per piece integer, the number of blocks per piece. Must be: piece_size / (16 * 1024). Clamped to be within the range [1, 256]. It is the number of blocks per (normal sized) piece. Usually each block is 16 * 1024 bytes in size. But if piece size is greater than 4 megabytes, the block size will increase.
slots

list of integers. The list mappes slots to piece indices. It tells which piece is on which slot. If piece index is -2 it means it is free, that there's no piece there. If it is -1, means the slot isn't allocated on disk yet. The pieces have to meet the following requirement:

If there's a slot at the position of the piece index, the piece must be located in that slot.

peers

list of dictionaries. Each dictionary has the following layout:

ip string, the ip address of the peer.
port integer, the listen port of the peer

These are the local peers we were connected to when this fast-resume data was saved.

unfinished

list of dictionaries. Each dictionary represents an piece, and has the following layout:

piece integer, the index of the piece this entry refers to.
bitmask string, a binary bitmask representing the blocks that have been downloaded in this piece.
adler32 The adler32 checksum of the data in the blocks specified by bitmask.
file sizes list where each entry corresponds to a file in the file list in the metadata. Each entry has a list of two values, the first value is the size of the file in bytes, the second is the timestamp when the last time someone wrote to it. This information is used to compare with the files on disk. All the files must match exactly this information in order to consider the resume data as current. Otherwise a full re-check is issued.

threads

libtorrent starts 3 threads.

  • The first thread is the main thread that will sit idle in a select() call most of the time. This thread runs the main loop that will send and receive data on all connections.
  • The second thread is a hash-check thread. Whenever a torrent is added it will first be passed to this thread for checking the files that may already have been downloaded. If there is any resume data this thread will make sure it is valid and matches the files. Once the torrent has been checked, it is passed on to the main thread that will start it. The hash-check thread has a queue of torrents, it will only check one torrent at a time.
  • The third thread is spawned the first time a tracker is contacted. It is used for doing calls to gethostbyname(). Since this call is blocking (and may block for several seconds if the dns server is down or slow) it is necessary to run this in its own thread to avoid stalling the main thread.

storage allocation

There are two modes in which storage (files on disk) are allocated in libtorrent.

  • The traditional full allocation mode, where the entire files are filled up with zeroes before anything is downloaded.
  • And the compact allocation mode, where only files are allocated for actual pieces that have been downloaded. This is the default allocation mode in libtorrent.

The allocation mode is selected when a torrent is started. It is passed as a boolean argument to session::add_torrent() (see add_torrent()). These two modes have different drawbacks and benefits.

full allocation

When a torrent is started in full allocation mode, the checker thread (see threads) will make sure that the entire storage is allocated, and fill any gaps with zeroes. It will of course still check for existing pieces and fast resume data. The main drawbacks of this mode are:

  • It will take longer to start the torrent, since it will need to fill the files with zeroes. This delay is linearly dependent on the size of the download.
  • The download will occupy unnecessary disk space between download sessions.
  • Disk caches usually perform extremely poorly with random access to large files and may slow down a download considerably.

The benefit of thise mode are:

  • Downloaded pieces are written directly to their final place in the files and the total number of disk operations will be fewer and may also play nicer to filesystems' file allocation, and reduce fragmentation.
  • No risk of a download failing because of a full disk during download.

compact allocation

The compact allocation will only allocate as much storage as it needs to keep the pieces downloaded so far. This means that pieces will be moved around to be placed at their final position in the files while downloading (to make sure the completed download has all its pieces in the correct place). So, the main drawbacks are:

  • More disk operations while downloading since pieces are moved around.
  • Potentially more fragmentation in the filesystem.

The benefits though, are:

  • No startup delay, since the files doesn't need allocating.
  • The download will not use unnecessary disk space.
  • Disk caches perform much better than in full allocation and raises the download speed limit imposed by the disk.

The algorithm that is used when allocating pieces and slots isn't very complicated. For the interested, a description follows.

storing a piece:

  1. let A be a newly downloaded piece, with index n.
  2. let s be the number of slots allocated in the file we're downloading to. (the number of pieces it has room for).
  3. if n >= s then allocate a new slot and put the piece there.
  4. if n < s then allocate a new slot, move the data at slot n to the new slot and put A in slot n.

allocating a new slot:

  1. if there's an unassigned slot (a slot that doesn't contain any piece), return that slot index.
  2. append the new slot at the end of the file (or find an unused slot).
  3. let i be the index of newly allocated slot
  4. if we have downloaded piece index i already (to slot j) then
    1. move the data at slot j to slot i.
    2. return slot index j as the newly allocated free slot.
  5. return i as the newly allocated slot.

extensions

These extensions all operates within the extension protocol. The name of the extension is the name used in the extension-list packets, and the payload is the data in the extended message (not counting the length-prefix, message-id nor extension-id).

Note that since this protocol relies on one of the reserved bits in the handshake, it may be incompatible with future versions of the mainline bittorrent client.

These are the extensions that are currently implemented.

chat messages

Extension name: "chat"

The payload in the packet is a bencoded dictionary with any combination of the following entries:

"msg" This is a string that contains a message that should be displayed to the user.
"ctrl" This is a control string that can tell a client that it is ignored (to make the user aware of that) and it can also tell a client that it is no longer ignored. These notifications are encoded as the strings: "ignored" and "not ignored". Any unrecognized strings should be ignored.

metadata from peers

Extension name: "metadata"

The point with this extension is that you don't have to distribute the metadata (.torrent-file) separately. The metadata can be distributed through the bittorrent swarm. The only thing you need to download such a torrent is the tracker url and the info-hash of the torrent.

It works by assuming that the initial seeder has the metadata and that the metadata will propagate through the network as more peers join.

There are three kinds of messages in the metadata extension. These packets are put as payload to the extension message. The three packets are:

  • request metadata
  • metadata
  • don't have metadata

request metadata:

size name description
uint8_t msg_type Determines the kind of message this is 0 means 'request metadata'
uint8_t start The start of the metadata block that is requested. It is given in 256:ths of the total size of the metadata, since the requesting client don't know the size of the metadata.
uint8_t size The size of the metadata block that is requested. This is also given in 256:ths of the total size of the metadata. The size is given as size-1. That means that if this field is set 0, the request wants one 256:th of the metadata.

metadata:

size name description
uint8_t msg_type 1 means 'metadata'
int32_t total_size The total size of the metadata, given in number of bytes.
int32_t offset The offset of where the metadata block in this message belongs in the final metadata. This is given in bytes.
uint8_t[] metadata The actual metadata block. The size of this part is given implicit by the length prefix in the bittorrent protocol packet.

Don't have metadata:

size name description
uint8_t msg_type 2 means 'I don't have metadata'. This message is sent as a reply to a metadata request if the the client doesn't have any metadata.

HTTP seeding

The HTTP seed extension implements this specification.

The libtorrent implementation assumes that, if the URL ends with a slash ('/'), the filename should be appended to it in order to request pieces from that file. The way this works is that if the torrent is a single-file torrent, only that filename is appended. If the torrent is a multi-file torrent, the torrent's name '/' the file name is appended. This is the same directory structure that libtorrent will download torrents into.

filename checks

Boost.Filesystem will by default check all its paths to make sure they conform to filename requirements on many platforms. If you don't want this check, you can set it to either only check for native filesystem requirements or turn it off alltogether. You can use:

boost::filesystem::path::default_name_check(boost::filesystem::native);

for example. For more information, see the Boost.Filesystem docs.

acknowledgements

Written by Arvid Norberg. Copyright © 2003-2005

Contributions by Magnus Jonsson, Daniel Wallin and Cory Nelson

Big thanks to Michael Wojciechowski and Peter Koeleman for making the autotools scripts.

Thanks to Reimond Retz for bugfixes, suggestions and testing

Thanks to University of Umeå for providing development and test hardware.

Project is hosted by sourceforge.

sf_logo