transmission/libtransmission/makemeta.cc

// This file Copyright © 2010-2022 Mnemosyne LLC.
// It may be used under GPLv2 (SPDX: GPL-2.0), GPLv3 (SPDX: GPL-3.0),
// or any future license endorsed by Mnemosyne LLC.
// License text can be found in the licenses/ folder.

#include <algorithm>
#include <cerrno>
#include <cstdint>
#include <cstdlib> /* qsort */
#include <cstring> /* strcmp, strlen */
#include <mutex>
#include <string_view>

#include <event2/util.h> /* evutil_ascii_strcasecmp() */

#include "transmission.h"

#include "crypto-utils.h"
#include "error.h"
#include "file.h"
#include "log.h"
#include "makemeta.h"
#include "platform.h" /* threads, locks */
#include "session.h"
#include "tr-assert.h"
#include "utils.h" /* buildpath */
#include "variant.h"
#include "version.h"
#include "web-utils.h"

using namespace std::literals;

/****
*****
****/

struct FileList
{
    uint64_t size;
    char* filename;
    struct FileList* next;
};

static struct FileList* getFiles(char const* dir, char const* base, struct FileList* list)
{
    if (dir == nullptr || base == nullptr)
    {
        return nullptr;
    }

    auto buf = tr_strvPath(dir, base);
    tr_sys_path_native_separators(std::data(buf));

    tr_sys_path_info info;
    if (tr_error* error = nullptr; !tr_sys_path_get_info(buf.c_str(), 0, &info, &error))
    {
        tr_logAddError(_("Torrent Creator is skipping file \"%s\": %s"), buf.c_str(), error->message);
        tr_error_free(error);
        return list;
    }

    if (tr_sys_dir_t odir = info.type == TR_SYS_PATH_IS_DIRECTORY ? tr_sys_dir_open(buf.c_str(), nullptr) : TR_BAD_SYS_DIR;
        odir != TR_BAD_SYS_DIR)
    {
        char const* name = nullptr;
        while ((name = tr_sys_dir_read_name(odir, nullptr)) != nullptr)
        {
            if (name[0] != '.') /* skip dotfiles */
            {
                list = getFiles(buf.c_str(), name, list);
            }
        }

        tr_sys_dir_close(odir, nullptr);
    }
    else if (info.type == TR_SYS_PATH_IS_FILE && info.size > 0)
    {
        auto* const node = tr_new0(FileList, 1);
        node->size = info.size;
        node->filename = tr_strvDup(buf);
        node->next = list;
        list = node;
    }

    return list;
}

static uint32_t bestPieceSize(uint64_t totalSize)
{
    uint32_t const KiB = 1024;
    uint32_t const MiB = 1048576;
    uint32_t const GiB = 1073741824;

    if (totalSize >= 2 * GiB)
    {
        return 2 * MiB;
    }

    if (totalSize >= 1 * GiB)
    {
        return 1 * MiB;
    }

    if (totalSize >= 512 * MiB)
    {
        return 512 * KiB;
    }

    if (totalSize >= 350 * MiB)
    {
        return 256 * KiB;
    }

    if (totalSize >= 150 * MiB)
    {
        return 128 * KiB;
    }

    if (totalSize >= 50 * MiB)
    {
        return 64 * KiB;
    }

    return 32 * KiB; /* less than 50 meg */
}

tr_metainfo_builder* tr_metaInfoBuilderCreate(char const* topFileArg)
{
    char* const real_top = tr_sys_path_resolve(topFileArg, nullptr);

    if (real_top == nullptr)
    {
        /* TODO: Better error reporting */
        return nullptr;
    }

    auto* const ret = tr_new0(tr_metainfo_builder, 1);

    ret->top = real_top;

    {
        tr_sys_path_info info;
        ret->isFolder = tr_sys_path_get_info(ret->top, 0, &info, nullptr) && info.type == TR_SYS_PATH_IS_DIRECTORY;
    }

    /* build a list of files containing top file and,
       if it's a directory, all of its children */
    FileList* files = nullptr;
    {
        char* dir = tr_sys_path_dirname(ret->top, nullptr);
        char* base = tr_sys_path_basename(ret->top, nullptr);
        files = getFiles(dir, base, nullptr);
        tr_free(base);
        tr_free(dir);
    }

    for (struct FileList* walk = files; walk != nullptr; walk = walk->next)
    {
        ++ret->fileCount;
    }

    ret->files = tr_new0(tr_metainfo_builder_file, ret->fileCount);

    int i = 0;
    while (files != nullptr)
    {
        struct FileList* const tmp = files;
        files = files->next;

        tr_metainfo_builder_file* const file = &ret->files[i++];
        file->filename = tmp->filename;
        file->size = tmp->size;

        ret->totalSize += tmp->size;

        tr_free(tmp);
    }

    std::sort(
        ret->files,
        ret->files + ret->fileCount,
        [](auto const& a, auto const& b) { return evutil_ascii_strcasecmp(a.filename, b.filename) < 0; });

    tr_metaInfoBuilderSetPieceSize(ret, bestPieceSize(ret->totalSize));

    return ret;
}

static bool isValidPieceSize(uint32_t n)
{
    bool const isPowerOfTwo = n != 0 && (n & (n - 1)) == 0;

    return isPowerOfTwo;
}

bool tr_metaInfoBuilderSetPieceSize(tr_metainfo_builder* b, uint32_t bytes)
{
    if (!isValidPieceSize(bytes))
    {
        tr_logAddError(
            _("Failed to set piece size to %s, leaving it at %s"),
            tr_formatter_mem_B(bytes).c_str(),
            tr_formatter_mem_B(b->pieceSize).c_str());
        return false;
    }

    b->pieceSize = bytes;
    b->pieceCount = (int)(b->totalSize / b->pieceSize);

    if (b->totalSize % b->pieceSize != 0)
    {
        ++b->pieceCount;
    }

    return true;
}

void tr_metaInfoBuilderFree(tr_metainfo_builder* builder)
{
    if (builder != nullptr)
    {
        for (uint32_t i = 0; i < builder->fileCount; ++i)
        {
            tr_free(builder->files[i].filename);
        }

        tr_free(builder->files);
        tr_free(builder->top);
        tr_free(builder->comment);
        tr_free(builder->source);

        for (int i = 0; i < builder->trackerCount; ++i)
        {
            tr_free(builder->trackers[i].announce);
        }

        tr_free(builder->trackers);
        tr_free(builder->outputFile);
        tr_free(builder);
    }
}

/****
*****
****/

static std::vector<std::byte> getHashInfo(tr_metainfo_builder* b)
{
    auto ret = std::vector<std::byte>(std::size(tr_sha1_digest_t{}) * b->pieceCount);

    if (b->totalSize == 0)
    {
        return {};
    }

    b->pieceIndex = 0;
    uint64_t totalRemain = b->totalSize;
    uint32_t fileIndex = 0;
    auto* walk = std::data(ret);
    auto buf = std::vector<char>(b->pieceSize);
    uint64_t off = 0;
    tr_error* error = nullptr;

    tr_sys_file_t fd = tr_sys_file_open(b->files[fileIndex].filename, TR_SYS_FILE_READ | TR_SYS_FILE_SEQUENTIAL, 0, &error);
    if (fd == TR_BAD_SYS_FILE)
    {
        b->my_errno = error->code;
        tr_strlcpy(b->errfile, b->files[fileIndex].filename, sizeof(b->errfile));
        b->result = TR_MAKEMETA_IO_READ;
        tr_error_free(error);
        return {};
    }

    while (totalRemain != 0)
    {
        TR_ASSERT(b->pieceIndex < b->pieceCount);

        auto* bufptr = std::data(buf);
        uint32_t const thisPieceSize = std::min(uint64_t{ b->pieceSize }, totalRemain);
        uint64_t leftInPiece = thisPieceSize;

        while (leftInPiece != 0)
        {
            uint64_t const n_this_pass = std::min(b->files[fileIndex].size - off, leftInPiece);
            uint64_t n_read = 0;
            (void)tr_sys_file_read(fd, bufptr, n_this_pass, &n_read, nullptr);
            bufptr += n_read;
            off += n_read;
            leftInPiece -= n_read;

            if (off == b->files[fileIndex].size)
            {
                off = 0;
                tr_sys_file_close(fd, nullptr);
                fd = TR_BAD_SYS_FILE;

                if (++fileIndex < b->fileCount)
                {
                    fd = tr_sys_file_open(b->files[fileIndex].filename, TR_SYS_FILE_READ | TR_SYS_FILE_SEQUENTIAL, 0, &error);

                    if (fd == TR_BAD_SYS_FILE)
                    {
                        b->my_errno = error->code;
                        tr_strlcpy(b->errfile, b->files[fileIndex].filename, sizeof(b->errfile));
                        b->result = TR_MAKEMETA_IO_READ;
                        tr_error_free(error);
                        return {};
                    }
                }
            }
        }

        TR_ASSERT(bufptr - std::data(buf) == (int)thisPieceSize);
        TR_ASSERT(leftInPiece == 0);
        auto const digest = tr_sha1(buf);
        if (!digest)
        {
            b->my_errno = EIO;
            tr_snprintf(b->errfile, sizeof(b->errfile), "error hashing piece %" PRIu32, b->pieceIndex);
            b->result = TR_MAKEMETA_IO_READ;
            return {};
        }

        walk = std::copy(std::begin(*digest), std::end(*digest), walk);

        if (b->abortFlag)
        {
            b->result = TR_MAKEMETA_CANCELLED;
            break;
        }

        totalRemain -= thisPieceSize;
        ++b->pieceIndex;
    }

    TR_ASSERT(b->abortFlag || size_t(walk - std::data(ret)) == std::size(ret));
    TR_ASSERT(b->abortFlag || !totalRemain);

    if (fd != TR_BAD_SYS_FILE)
    {
        tr_sys_file_close(fd, nullptr);
    }

    return ret;
}

static void getFileInfo(
    char const* topFile,
    tr_metainfo_builder_file const* file,
    tr_variant* uninitialized_length,
    tr_variant* uninitialized_path)
{
    /* get the file size */
    tr_variantInitInt(uninitialized_length, file->size);

    /* how much of file->filename to walk past */
    size_t offset = strlen(topFile);
    if (offset > 0 && topFile[offset - 1] != TR_PATH_DELIMITER)
    {
        ++offset; /* +1 for the path delimiter */
    }

    /* build the path list */
    tr_variantInitList(uninitialized_path, 0);

    auto filename = std::string_view{ file->filename };
    if (std::size(filename) > offset)
    {
        filename.remove_prefix(offset);
        auto token = std::string_view{};
        while (tr_strvSep(&filename, &token, TR_PATH_DELIMITER))
        {
            tr_variantListAddStr(uninitialized_path, token);
        }
    }
}

static void makeInfoDict(tr_variant* dict, tr_metainfo_builder* builder)
{
    tr_variantDictReserve(dict, 5);

    if (builder->isFolder) /* root node is a directory */
    {
        tr_variant* list = tr_variantDictAddList(dict, TR_KEY_files, builder->fileCount);

        for (uint32_t i = 0; i < builder->fileCount; ++i)
        {
            tr_variant* d = tr_variantListAddDict(list, 2);
            tr_variant* length = tr_variantDictAdd(d, TR_KEY_length);
            tr_variant* pathVal = tr_variantDictAdd(d, TR_KEY_path);
            getFileInfo(builder->top, &builder->files[i], length, pathVal);
        }
    }
    else
    {
        tr_variantDictAddInt(dict, TR_KEY_length, builder->files[0].size);
    }

    if (auto* const base = tr_sys_path_basename(builder->top, nullptr); base != nullptr)
    {
        tr_variantDictAddStr(dict, TR_KEY_name, base);
        tr_free(base);
    }

    tr_variantDictAddInt(dict, TR_KEY_piece_length, builder->pieceSize);

    if (auto const piece_hashes = getHashInfo(builder); !std::empty(piece_hashes))
    {
        tr_variantDictAddRaw(dict, TR_KEY_pieces, std::data(piece_hashes), std::size(piece_hashes));
    }

    tr_variantDictAddInt(dict, TR_KEY_private, builder->isPrivate ? 1 : 0);
    if (builder->source != nullptr)
    {
        tr_variantDictAddStr(dict, TR_KEY_source, builder->source);
    }
}

static void tr_realMakeMetaInfo(tr_metainfo_builder* builder)
{
    tr_variant top;

    /* allow an empty set, but if URLs *are* listed, verify them. #814, #971 */
    for (int i = 0; i < builder->trackerCount && builder->result == TR_MAKEMETA_OK; ++i)
    {
        if (!tr_urlIsValidTracker(builder->trackers[i].announce))
        {
            tr_strlcpy(builder->errfile, builder->trackers[i].announce, sizeof(builder->errfile));
            builder->result = TR_MAKEMETA_URL;
        }
    }

    tr_variantInitDict(&top, 6);

    if (builder->fileCount == 0 || builder->totalSize == 0 || builder->pieceSize == 0 || builder->pieceCount == 0)
    {
        builder->errfile[0] = '\0';
        builder->my_errno = ENOENT;
        builder->result = TR_MAKEMETA_IO_READ;
        builder->isDone = true;
    }

    if (builder->result == TR_MAKEMETA_OK && builder->trackerCount != 0)
    {
        int prevTier = -1;
        tr_variant* tier = nullptr;

        if (builder->trackerCount > 1)
        {
            tr_variant* annList = tr_variantDictAddList(&top, TR_KEY_announce_list, 0);

            for (int i = 0; i < builder->trackerCount; ++i)
            {
                if (prevTier != builder->trackers[i].tier)
                {
                    prevTier = builder->trackers[i].tier;
                    tier = tr_variantListAddList(annList, 0);
                }

                tr_variantListAddStr(tier, builder->trackers[i].announce);
            }
        }

        tr_variantDictAddStr(&top, TR_KEY_announce, builder->trackers[0].announce);
    }

    if (builder->result == TR_MAKEMETA_OK && !builder->abortFlag)
    {
        if (!tr_str_is_empty(builder->comment))
        {
            tr_variantDictAddStr(&top, TR_KEY_comment, builder->comment);
        }

        tr_variantDictAddStrView(&top, TR_KEY_created_by, TR_NAME "/" LONG_VERSION_STRING);
        tr_variantDictAddInt(&top, TR_KEY_creation_date, time(nullptr));
        tr_variantDictAddStrView(&top, TR_KEY_encoding, "UTF-8");
        makeInfoDict(tr_variantDictAddDict(&top, TR_KEY_info, 666), builder);
    }

    /* save the file */
    if ((builder->result == TR_MAKEMETA_OK) && (!builder->abortFlag) &&
        (tr_variantToFile(&top, TR_VARIANT_FMT_BENC, builder->outputFile) != 0))
    {
        builder->my_errno = errno;
        tr_strlcpy(builder->errfile, builder->outputFile, sizeof(builder->errfile));
        builder->result = TR_MAKEMETA_IO_WRITE;
    }

    /* cleanup */
    tr_variantFree(&top);

    if (builder->abortFlag)
    {
        builder->result = TR_MAKEMETA_CANCELLED;
    }

    builder->isDone = true;
}

/***
****
****  A threaded builder queue
****
***/

static tr_metainfo_builder* queue = nullptr;

static tr_thread* workerThread = nullptr;

static std::recursive_mutex queue_mutex_;

static void makeMetaWorkerFunc(void* /*user_data*/)
{
    for (;;)
    {
        tr_metainfo_builder* builder = nullptr;

        /* find the next builder to process */
        queue_mutex_.lock();

        if (queue != nullptr)
        {
            builder = queue;
            queue = queue->nextBuilder;
        }

        queue_mutex_.unlock();

        /* if no builders, this worker thread is done */
        if (builder == nullptr)
        {
            break;
        }

        tr_realMakeMetaInfo(builder);
    }

    workerThread = nullptr;
}

void tr_makeMetaInfo(
    tr_metainfo_builder* builder,
    char const* outputFile,
    tr_tracker_info const* trackers,
    int trackerCount,
    char const* comment,
    bool isPrivate,
    char const* source)
{
    /* free any variables from a previous run */
    for (int i = 0; i < builder->trackerCount; ++i)
    {
        tr_free(builder->trackers[i].announce);
    }

    tr_free(builder->trackers);
    tr_free(builder->comment);
    tr_free(builder->source);
    tr_free(builder->outputFile);

    /* initialize the builder variables */
    builder->abortFlag = false;
    builder->result = TR_MAKEMETA_OK;
    builder->isDone = false;
    builder->pieceIndex = 0;
    builder->trackerCount = trackerCount;
    builder->trackers = tr_new0(tr_tracker_info, builder->trackerCount);

    for (int i = 0; i < builder->trackerCount; ++i)
    {
        builder->trackers[i].tier = trackers[i].tier;
        builder->trackers[i].announce = tr_strdup(trackers[i].announce);
    }

    builder->comment = tr_strdup(comment);
    builder->isPrivate = isPrivate;
    builder->source = tr_strdup(source);

    if (!tr_str_is_empty(outputFile))
    {
        builder->outputFile = tr_strdup(outputFile);
    }
    else
    {
        builder->outputFile = tr_strvDup(tr_strvJoin(builder->top, ".torrent"sv));
    }

    /* enqueue the builder */
    auto const lock = std::lock_guard(queue_mutex_);

    builder->nextBuilder = queue;
    queue = builder;

    if (workerThread == nullptr)
    {
        workerThread = tr_threadNew(makeMetaWorkerFunc, nullptr);
    }
}