librats 1.0.2 → 2.1.4

package/native-src/src/librats/bittorrent/torrent_info.cpp

@@ -0,0 +1,314 @@
+#include "librats/bittorrent/torrent_info.h"
+
+#include "librats/bittorrent/bencode.h"
+#include "librats/bittorrent/magnet_uri.h"
+#include "librats/crypto/sha1.h"
+#include "librats/util/fs.h"
+
+#include <cctype>
+#include <cstddef>
+#include <cstring>
+
+namespace librats::bittorrent {
+
+namespace {
+
+// ----------------------------------------------------------------------------
+// Minimal bencode byte-span scanner.
+//
+// We must hash the *original* bytes of the `info` value (re-encoding the parsed
+// structure could reorder dict keys and change the hash). The eager BencodeValue
+// decoder discards positions, so this scanner walks the raw buffer to locate the
+// info value's byte range without building a tree.
+// ----------------------------------------------------------------------------
+
+/// Read a bencoded byte string at @p pos, advancing @p pos past it.
+bool read_bstring(const std::uint8_t* d, std::size_t n, std::size_t& pos, std::string& out) {
+    std::size_t colon = pos;
+    while (colon < n && d[colon] != ':') {
+        if (d[colon] < '0' || d[colon] > '9') return false;
+        ++colon;
+    }
+    if (colon >= n || colon == pos) return false;
+    std::uint64_t len = 0;
+    for (std::size_t i = pos; i < colon; ++i) {
+        len = len * 10 + std::uint64_t(d[i] - '0');
+        if (len > n) return false;  // can't be longer than the whole buffer
+    }
+    const std::size_t start = colon + 1;
+    if (start + len > n) return false;
+    out.assign(reinterpret_cast<const char*>(d + start), len);
+    pos = start + len;
+    return true;
+}
+
+/// Advance @p pos past exactly one bencoded value (int / string / list / dict).
+bool skip_value(const std::uint8_t* d, std::size_t n, std::size_t& pos) {
+    if (pos >= n) return false;
+    const char c = char(d[pos]);
+    if (c == 'i') {
+        std::size_t e = pos + 1;
+        while (e < n && d[e] != 'e') ++e;
+        if (e >= n) return false;
+        pos = e + 1;
+        return true;
+    }
+    if (c == 'l' || c == 'd') {
+        ++pos;
+        while (pos < n && d[pos] != 'e') {
+            if (!skip_value(d, n, pos)) return false;
+        }
+        if (pos >= n) return false;
+        ++pos;  // consume 'e'
+        return true;
+    }
+    if (c >= '0' && c <= '9') {
+        std::string tmp;
+        return read_bstring(d, n, pos, tmp);
+    }
+    return false;
+}
+
+/// Locate the byte range [start, end) of the top-level dict's `info` value.
+std::optional<std::pair<std::size_t, std::size_t>> find_info_span(const Bytes& buf) {
+    const std::uint8_t* d = buf.data();
+    const std::size_t   n = buf.size();
+    if (n == 0 || d[0] != 'd') return std::nullopt;
+
+    std::size_t pos = 1;
+    while (pos < n && d[pos] != 'e') {
+        std::string key;
+        if (!read_bstring(d, n, pos, key)) return std::nullopt;
+        const std::size_t value_start = pos;
+        if (!skip_value(d, n, pos)) return std::nullopt;
+        if (key == "info") return std::make_pair(value_start, pos);
+    }
+    return std::nullopt;
+}
+
+// ---- safe bencode accessors (the BencodeValue accessors throw on mismatch) ----
+
+const librats::BencodeValue* find(const librats::BencodeValue& dict, const char* key) {
+    return dict.find(key);
+}
+
+const std::string* find_string(const librats::BencodeValue& dict, const char* key) {
+    const auto* v = find(dict, key);
+    return (v && v->is_string()) ? &v->as_string() : nullptr;
+}
+
+std::optional<std::int64_t> find_int(const librats::BencodeValue& dict, const char* key) {
+    const auto* v = find(dict, key);
+    if (v && v->is_integer()) return v->as_integer();
+    return std::nullopt;
+}
+
+/// A path component is safe if it is non-empty, not "." / ".." and free of
+/// separators / NULs — guarding the disk layer against traversal.
+bool safe_component(const std::string& c) {
+    if (c.empty() || c == "." || c == "..") return false;
+    return c.find('/') == std::string::npos && c.find('\\') == std::string::npos
+        && c.find('\0') == std::string::npos;
+}
+
+} // namespace
+
+bool TorrentInfo::parse_info_dict(const librats::BencodeValue& info, TorrentParseError* err) {
+    auto fail = [&](const char* m) { if (err) err->message = m; return false; };
+
+    const std::string* name = find_string(info, "name");
+    const auto piece_length  = find_int(info, "piece length");
+    const std::string* pieces = find_string(info, "pieces");
+    if (!name || !piece_length || !pieces) return fail("info dict missing name/piece length/pieces");
+    if (*piece_length <= 0 || *piece_length > (1 << 30)) return fail("invalid piece length");
+    if (pieces->size() % kInfoHashSize != 0) return fail("pieces string not a multiple of 20");
+    if (!safe_component(*name)) return fail("unsafe torrent name");
+
+    FileStorage files;
+    files.set_piece_length(std::uint32_t(*piece_length));
+    files.set_name(*name);
+
+    if (const auto* file_list = find(info, "files"); file_list && file_list->is_list()) {
+        // Multi-file: each entry has a length and a path component list.
+        for (const auto& entry : file_list->as_list()) {
+            const auto length = find_int(entry, "length");
+            const auto* path  = find(entry, "path");
+            if (!length || *length < 0 || !path || !path->is_list()) return fail("bad file entry");
+            std::string rel = *name;
+            for (const auto& comp : path->as_list()) {
+                if (!comp.is_string() || !safe_component(comp.as_string())) return fail("unsafe file path");
+                rel += '/';
+                rel += comp.as_string();
+            }
+            if (!files.add_file(std::move(rel), *length)) return fail("file size overflow");
+        }
+    } else {
+        // Single-file: the info dict itself carries the length.
+        const auto length = find_int(info, "length");
+        if (!length || *length < 0) return fail("single-file torrent missing length");
+        if (!files.add_file(*name, *length)) return fail("file size overflow");
+    }
+
+    // The piece-hash count must match the file layout exactly.
+    if (pieces->size() / kInfoHashSize != files.num_pieces()) return fail("piece count mismatch");
+
+    files_        = std::move(files);
+    name_         = *name;
+    piece_hashes_.assign(pieces->begin(), pieces->end());
+    is_private_   = find_int(info, "private").value_or(0) == 1;
+    has_metadata_ = true;
+    return true;
+}
+
+std::optional<TorrentInfo> TorrentInfo::from_info_dict(const Bytes& info_dict_bytes,
+                                                       const InfoHash& expected,
+                                                       TorrentParseError* err) {
+    try {
+        librats::BencodeValue info = librats::BencodeDecoder::decode(info_dict_bytes);
+        TorrentInfo ti;
+        if (!ti.parse_info_dict(info, err)) return std::nullopt;
+        ti.info_dict_bytes_ = info_dict_bytes;
+        ti.info_hash_       = SHA1::hash_raw(info_dict_bytes.data(), info_dict_bytes.size());
+        if (!is_all_zero(expected) && ti.info_hash_ != expected) {
+            if (err) err->message = "info-hash mismatch";
+            return std::nullopt;
+        }
+        return ti;
+    } catch (const std::exception& e) {
+        if (err) err->message = e.what();
+        return std::nullopt;
+    }
+}
+
+std::optional<TorrentInfo> TorrentInfo::from_bytes(const Bytes& data, TorrentParseError* err) {
+    try {
+        librats::BencodeValue root = librats::BencodeDecoder::decode(data);
+        if (!root.is_dict()) { if (err) err->message = "torrent is not a dict"; return std::nullopt; }
+
+        const auto span = find_info_span(data);
+        const auto* info = find(root, "info");
+        if (!span || !info) { if (err) err->message = "missing info dict"; return std::nullopt; }
+
+        TorrentInfo ti;
+        if (!ti.parse_info_dict(*info, err)) return std::nullopt;
+
+        ti.info_dict_bytes_.assign(data.begin() + std::ptrdiff_t(span->first),
+                                   data.begin() + std::ptrdiff_t(span->second));
+        ti.info_hash_ = SHA1::hash_raw(ti.info_dict_bytes_.data(), ti.info_dict_bytes_.size());
+
+        // Optional top-level fields.
+        if (const auto* a = find_string(root, "announce")) ti.announce_ = *a;
+        if (const auto* c = find_string(root, "comment")) ti.comment_ = *c;
+        if (const auto* cb = find_string(root, "created by")) ti.created_by_ = *cb;
+        if (auto cd = find_int(root, "creation date")) ti.creation_date_ = *cd;
+
+        if (const auto* al = find(root, "announce-list"); al && al->is_list()) {
+            for (const auto& tier : al->as_list()) {
+                if (!tier.is_list()) continue;
+                TrackerTier t;
+                for (const auto& url : tier.as_list())
+                    if (url.is_string()) t.push_back(url.as_string());
+                if (!t.empty()) ti.announce_list_.push_back(std::move(t));
+            }
+        }
+
+        if (const auto* ul = find(root, "url-list")) {  // web seeds (BEP 19)
+            if (ul->is_string()) ti.web_seeds_.push_back(ul->as_string());
+            else if (ul->is_list())
+                for (const auto& u : ul->as_list())
+                    if (u.is_string()) ti.web_seeds_.push_back(u.as_string());
+        }
+
+        if (const auto* nodes = find(root, "nodes"); nodes && nodes->is_list()) {
+            for (const auto& node : nodes->as_list()) {
+                if (!node.is_list() || node.as_list().size() != 2) continue;
+                const auto& pair = node.as_list();
+                if (pair[0].is_string() && pair[1].is_integer())
+                    ti.dht_nodes_.push_back(DhtNode{pair[0].as_string(),
+                                                    std::uint16_t(pair[1].as_integer())});
+            }
+        }
+
+        return ti;
+    } catch (const std::exception& e) {
+        if (err) err->message = e.what();
+        return std::nullopt;
+    }
+}
+
+std::optional<TorrentInfo> TorrentInfo::from_file(const std::string& path, TorrentParseError* err) {
+    std::size_t size = 0;
+    void* buf = read_file_binary(path.c_str(), &size);
+    if (!buf) { if (err) err->message = "cannot read file"; return std::nullopt; }
+    Bytes data(static_cast<std::uint8_t*>(buf), static_cast<std::uint8_t*>(buf) + size);
+    free_file_buffer(buf);
+    return from_bytes(data, err);
+}
+
+std::optional<TorrentInfo> TorrentInfo::from_magnet(const std::string& uri, TorrentParseError* err) {
+    auto magnet = MagnetUri::parse(uri);
+    if (!magnet) { if (err) err->message = "invalid magnet uri"; return std::nullopt; }
+
+    TorrentInfo ti;
+    ti.info_hash_    = magnet->info_hash;
+    ti.name_         = magnet->display_name;
+    ti.web_seeds_    = magnet->web_seeds;
+    ti.has_metadata_ = false;  // file list & piece hashes still need fetching
+    for (const auto& tr : magnet->trackers) ti.announce_list_.push_back({tr});
+    if (!ti.announce_list_.empty() && !ti.announce_list_.front().empty())
+        ti.announce_ = ti.announce_list_.front().front();
+    return ti;
+}
+
+bool TorrentInfo::set_metadata(const Bytes& info_dict_bytes) {
+    auto completed = from_info_dict(info_dict_bytes, info_hash_, nullptr);
+    if (!completed) return false;
+    // Preserve discovery fields learned from the magnet; adopt the parsed payload.
+    files_           = std::move(completed->files_);
+    piece_hashes_    = std::move(completed->piece_hashes_);
+    info_dict_bytes_ = std::move(completed->info_dict_bytes_);
+    is_private_      = completed->is_private_;
+    if (name_.empty()) name_ = completed->name_;
+    has_metadata_    = true;
+    return true;
+}
+
+std::array<std::uint8_t, 20> TorrentInfo::piece_hash(std::uint32_t index) const {
+    std::array<std::uint8_t, 20> out{};
+    const std::size_t off = std::size_t(index) * kInfoHashSize;
+    if (off + kInfoHashSize <= piece_hashes_.size())
+        std::memcpy(out.data(), piece_hashes_.data() + off, kInfoHashSize);
+    return out;
+}
+
+std::vector<std::string> TorrentInfo::all_trackers() const {
+    std::vector<std::string> all;
+    if (!announce_.empty()) all.push_back(announce_);
+    for (const auto& tier : announce_list_)
+        for (const auto& url : tier)
+            if (url != announce_) all.push_back(url);
+    return all;
+}
+
+std::string TorrentInfo::to_magnet_uri(bool include_trackers) const {
+    std::string uri = "magnet:?xt=urn:btih:" + to_hex(info_hash_);
+    if (!name_.empty()) {
+        uri += "&dn=";
+        // Percent-encode anything that isn't an unreserved URI character.
+        for (unsigned char c : name_) {
+            if (std::isalnum(c) || c == '-' || c == '_' || c == '.' || c == '~') {
+                uri += char(c);
+            } else {
+                static const char hex[] = "0123456789ABCDEF";
+                uri += '%';
+                uri += hex[c >> 4];
+                uri += hex[c & 0x0F];
+            }
+        }
+    }
+    if (include_trackers)
+        for (const auto& tr : all_trackers()) uri += "&tr=" + tr;
+    return uri;
+}
+
+} // namespace librats::bittorrent

package/native-src/src/librats/bittorrent/torrent_info.h

@@ -0,0 +1,118 @@
+#pragma once
+
+/**
+ * @file torrent_info.h
+ * @brief Parsed torrent metadata: file layout, piece hashes, trackers, info-hash.
+ *
+ * A TorrentInfo is built from a .torrent file, a bare info dictionary (received
+ * via ut_metadata) or a magnet link. From a magnet it is "valid but without
+ * metadata" — it knows only the info-hash until set_metadata() completes it.
+ *
+ * The info-hash is the SHA-1 of the *exact* bytes of the bencoded `info`
+ * dictionary. We therefore keep those bytes verbatim (info_dict_bytes_) rather
+ * than re-encoding the parsed structure, so re-serving them over BEP 9 and
+ * re-hashing always reproduce the original hash.
+ */
+
+#include "librats/util/rats_export.h"
+#include "librats/bittorrent/file_storage.h"
+#include "librats/bittorrent/types.h"
+#include "librats/core/bytes.h"
+
+#include <array>
+#include <cstdint>
+#include <optional>
+#include <string>
+#include <vector>
+
+namespace librats { class BencodeValue; }
+
+namespace librats::bittorrent {
+
+/// Diagnostic for a parse failure (optional out-param of the factories).
+struct TorrentParseError {
+    std::string message;
+};
+
+class RATS_API TorrentInfo {
+public:
+    using TrackerTier = std::vector<std::string>;
+    struct DhtNode { std::string host; std::uint16_t port = 0; };
+
+    TorrentInfo() = default;
+
+    // ---- Factories ----
+    static std::optional<TorrentInfo> from_bytes(const Bytes& data, TorrentParseError* err = nullptr);
+    static std::optional<TorrentInfo> from_file(const std::string& path, TorrentParseError* err = nullptr);
+    static std::optional<TorrentInfo> from_magnet(const std::string& uri, TorrentParseError* err = nullptr);
+    /// Build directly from a bencoded info dict. If @p expected is non-zero, the
+    /// computed info-hash must match it.
+    static std::optional<TorrentInfo> from_info_dict(const Bytes& info_dict_bytes,
+                                                     const InfoHash& expected,
+                                                     TorrentParseError* err = nullptr);
+
+    /// Complete a magnet-built TorrentInfo with the info dict fetched from peers.
+    /// Returns false (leaving the object unchanged) if the bytes don't hash to
+    /// the expected info-hash or fail to parse.
+    bool set_metadata(const Bytes& info_dict_bytes);
+
+    // ---- Core properties ----
+    bool is_valid()     const noexcept { return !is_all_zero(info_hash_); }
+    bool has_metadata() const noexcept { return has_metadata_; }
+
+    const InfoHash&   info_hash()     const noexcept { return info_hash_; }
+    std::string       info_hash_hex() const           { return to_hex(info_hash_); }
+    const std::string& name()         const noexcept { return name_; }
+    const std::string& comment()      const noexcept { return comment_; }
+    const std::string& created_by()   const noexcept { return created_by_; }
+    std::int64_t       creation_date()const noexcept { return creation_date_; }
+    bool               is_private()   const noexcept { return is_private_; }
+
+    // ---- File & piece info ----
+    const FileStorage& files()        const noexcept { return files_; }
+    std::int64_t       total_size()   const noexcept { return files_.total_size(); }
+    std::size_t        num_files()    const noexcept { return files_.num_files(); }
+    std::uint32_t      piece_length() const noexcept { return files_.piece_length(); }
+    std::uint32_t      num_pieces()   const noexcept { return files_.num_pieces(); }
+    std::uint32_t      piece_size(std::uint32_t i) const noexcept { return files_.piece_size(i); }
+
+    /// 20-byte SHA-1 of piece @p index (all-zero if out of range).
+    std::array<std::uint8_t, 20> piece_hash(std::uint32_t index) const;
+    const Bytes& piece_hashes() const noexcept { return piece_hashes_; }
+
+    // ---- Trackers / seeds / nodes ----
+    const std::string&              announce()      const noexcept { return announce_; }
+    const std::vector<TrackerTier>& announce_list() const noexcept { return announce_list_; }
+    std::vector<std::string>        all_trackers()  const;
+    const std::vector<std::string>& web_seeds()     const noexcept { return web_seeds_; }
+    const std::vector<DhtNode>&     dht_nodes()      const noexcept { return dht_nodes_; }
+
+    // ---- Raw info dict (for BEP 9) ----
+    const Bytes& info_dict_bytes() const noexcept { return info_dict_bytes_; }
+    std::size_t  metadata_size()   const noexcept { return info_dict_bytes_.size(); }
+
+    /// Build a magnet URI for this torrent.
+    std::string to_magnet_uri(bool include_trackers = true) const;
+
+private:
+    /// Parse name/piece-length/pieces/files/private out of a bencoded info dict
+    /// into files_ and piece_hashes_. Does not touch info_hash_/info_dict_bytes_.
+    bool parse_info_dict(const librats::BencodeValue& info, TorrentParseError* err);
+
+    InfoHash                 info_hash_{};
+    std::string              name_;
+    std::string              comment_;
+    std::string              created_by_;
+    std::int64_t             creation_date_ = 0;
+    bool                     is_private_   = false;
+    bool                     has_metadata_ = false;
+    FileStorage              files_;
+    Bytes                    piece_hashes_;
+    std::string              announce_;
+    std::vector<TrackerTier> announce_list_;
+    std::vector<std::string> web_seeds_;
+    std::vector<DhtNode>     dht_nodes_;
+    Bytes                    info_dict_bytes_;
+};
+
+} // namespace librats::bittorrent