fs: add zip mount support. hash: fix not checking open result for file. fs: fix stdio not checking nullptr access.

sphaira/CMakeLists.txt

@@ -93,8 +93,7 @@ add_executable(sphaira
     source/utils/devoptab_save.cpp
     source/utils/devoptab_nsp.cpp
     source/utils/devoptab_xci.cpp
-    # todo:
-    # source/utils/devoptab_zip.cpp
+    source/utils/devoptab_zip.cpp
 
     source/usb/base.cpp
     source/usb/usbds.cpp

sphaira/include/yati/source/file.hpp

@@ -10,6 +10,7 @@ namespace sphaira::yati::source {
 struct File final : Base {
     File(fs::Fs* fs, const fs::FsPath& path);
     Result Read(void* buf, s64 off, s64 size, u64* bytes_read) override;
+    Result GetSize(s64* out);
 
 private:
     fs::Fs* m_fs{};

sphaira/source/fatfs.cpp

@@ -192,7 +192,7 @@ Result ReadStorage(FsStorage* storage, std::span<LruBufferedData> lru_cache, voi
 
         // if the dst is big enough, read data in place.
         if (read_size > alloc_size) {
-            if (R_SUCCEEDED(fsStorageRead(storage, file_off, dst, read_size))) {
+            R_TRY(fsStorageRead(storage, file_off, dst, read_size));
             const auto bytes_read = read_size;
             read_size -= bytes_read;
             file_off += bytes_read;
@@ -204,8 +204,8 @@ Result ReadStorage(FsStorage* storage, std::span<LruBufferedData> lru_cache, voi
             m_buffered->off = file_off - max_advance;
             m_buffered->size = max_advance;
             std::memcpy(m_buffered->data, dst - max_advance, max_advance);
-            }
-        } else if (R_SUCCEEDED(fsStorageRead(storage, file_off, m_buffered->data, alloc_size))) {
+        } else {
+            R_TRY(fsStorageRead(storage, file_off, m_buffered->data, alloc_size));
 			const auto bytes_read = alloc_size;
 			const auto max_advance = std::min<u64>(read_size, bytes_read);
             std::memcpy(dst, m_buffered->data, max_advance);

sphaira/source/fs.cpp

@@ -498,6 +498,8 @@ Result File::Read( s64 off, void* buf, u64 read_size, u32 option, u64* bytes_rea
     if (m_fs->IsNative()) {
         R_TRY(fsFileRead(&m_native, off, buf, read_size, option, bytes_read));
     } else {
+        R_UNLESS(m_stdio, Result_FsUnknownStdioError);
+
         if (m_stdio_off != off) {
             m_stdio_off = off;
             std::fseek(m_stdio, off, SEEK_SET);
@@ -524,6 +526,8 @@ Result File::Write(s64 off, const void* buf, u64 write_size, u32 option) {
     if (m_fs->IsNative()) {
         R_TRY(fsFileWrite(&m_native, off, buf, write_size, option));
     } else {
+        R_UNLESS(m_stdio, Result_FsUnknownStdioError);
+
         if (m_stdio_off != off) {
             log_write("[FS] diff seek\n");
             m_stdio_off = off;
@@ -546,6 +550,7 @@ Result File::SetSize(s64 sz) {
     if (m_fs->IsNative()) {
         R_TRY(fsFileSetSize(&m_native, sz));
     } else {
+        R_UNLESS(m_stdio, Result_FsUnknownStdioError);
         const auto fd = fileno(m_stdio);
         R_UNLESS(fd > 0, Result_FsUnknownStdioError);
         R_UNLESS(!ftruncate(fd, sz), Result_FsUnknownStdioError);
@@ -560,6 +565,8 @@ Result File::GetSize(s64* out) {
     if (m_fs->IsNative()) {
         R_TRY(fsFileGetSize(&m_native, out));
     } else {
+        R_UNLESS(m_stdio, Result_FsUnknownStdioError);
+
         struct stat st;
         R_UNLESS(!fstat(fileno(m_stdio), &st), Result_FsUnknownStdioError);
         *out = st.st_size;

sphaira/source/hasher.cpp

@@ -18,10 +18,12 @@ struct FileSource final : BaseSource {
     }
 
     Result Size(s64* out) override {
+        R_TRY(m_open_result);
         return m_file.GetSize(out);
     }
 
     Result Read(void* buf, s64 off, s64 size, u64* bytes_read) override {
+        R_TRY(m_open_result);
         const auto rc = m_file.Read(off, buf, size, 0, bytes_read);
         if (m_fs->IsNative() && m_is_file_based_emummc) {
             svcSleepThread(2e+6); // 2ms

sphaira/source/ui/menus/filebrowser.cpp

@@ -690,6 +690,8 @@ void FsView::OnClick() {
             MountNspFs();
         } else if (IsExtension(entry.GetExtension(), XCI_EXTENSIONS)) {
             MountXciFs();
+        } else if (IsExtension(entry.GetExtension(), "zip")) {
+            MountZipFs();
         } else if (IsExtension(entry.GetExtension(), INSTALL_EXTENSIONS)) {
             InstallFiles();
         } else if (IsSd()) {
@@ -1991,7 +1993,17 @@ void FsView::MountXciFs() {
 }
 
 void FsView::MountZipFs() {
-    //todo:
+    fs::FsPath mount;
+    const auto rc = devoptab::MountZip(GetFs(), GetNewPathCurrent(), mount);
+    App::PushErrorBox(rc, "Failed to mount zip."_i18n);
+
+    if (R_SUCCEEDED(rc)) {
+        auto fs = std::make_shared<FsStdioWrapper>(mount, [mount](){
+            devoptab::UmountZip(mount);
+        });
+
+        MountFsHelper(fs, GetEntry().GetName());
+    }
 }
 
 Base::Base(u32 flags, u32 options)

sphaira/source/utils/devoptab_zip.cpp

@@ -0,0 +1,766 @@
+
+#include "utils/devoptab.hpp"
+#include "defines.hpp"
+#include "log.hpp"
+
+#include "yati/source/file.hpp"
+
+#include <cstring>
+#include <cstdio>
+#include <cerrno>
+#include <algorithm>
+#include <sys/iosupport.h>
+#include <zlib.h>
+
+namespace sphaira::devoptab {
+namespace {
+
+struct BufferedFileData {
+    BufferedFileData(std::unique_ptr<yati::source::Base>&& _source, u64 _size)
+    : source{std::forward<decltype(_source)>(_source)}
+    , capacity{_size} {
+
+    }
+
+    Result Read(void *_buffer, u64 file_off, u64 read_size);
+
+private:
+    std::unique_ptr<yati::source::Base> source;
+    const u64 capacity;
+
+    u64 m_off{};
+    u64 m_size{};
+    u8 m_data[1024 * 64]{};
+};
+
+Result BufferedFileData::Read(void *_buffer, u64 file_off, u64 read_size) {
+    auto dst = static_cast<u8*>(_buffer);
+    size_t amount = 0;
+
+    R_UNLESS(file_off < capacity, FsError_UnsupportedOperateRangeForFileStorage);
+    read_size = std::min(read_size, capacity - file_off);
+
+    if (m_size) {
+        // check if we can read this data into the beginning of dst.
+        if (file_off < m_off + m_size && file_off >= m_off) {
+            const auto off = file_off - m_off;
+            const auto size = std::min<s64>(read_size, m_size - off);
+            if (size) {
+                std::memcpy(dst, m_data + off, size);
+
+                read_size -= size;
+                file_off += size;
+                amount += size;
+                dst += size;
+            }
+        }
+    }
+
+    if (read_size) {
+        const auto alloc_size = sizeof(m_data);
+        m_off = 0;
+        m_size = 0;
+        u64 bytes_read;
+
+        // if the dst is big enough, read data in place.
+        if (read_size > alloc_size) {
+            R_TRY(source->Read(dst, file_off, read_size, &bytes_read));
+
+            read_size -= bytes_read;
+            file_off += bytes_read;
+            amount += bytes_read;
+            dst += bytes_read;
+
+            // save the last chunk of data to the m_buffered io.
+            const auto max_advance = std::min<u64>(amount, alloc_size);
+            m_off = file_off - max_advance;
+            m_size = max_advance;
+            std::memcpy(m_data, dst - max_advance, max_advance);
+        } else {
+            R_TRY(source->Read(m_data, file_off, alloc_size, &bytes_read));
+			const auto bytes_read = alloc_size;
+			const auto max_advance = std::min<u64>(read_size, bytes_read);
+            std::memcpy(dst, m_data, max_advance);
+
+            m_off = file_off;
+            m_size = bytes_read;
+
+            read_size -= max_advance;
+            file_off += max_advance;
+            amount += max_advance;
+            dst += max_advance;
+        }
+    }
+
+    R_SUCCEED();
+}
+
+#define LOCAL_HEADER_SIG 0x4034B50
+#define FILE_HEADER_SIG 0x2014B50
+#define DATA_DESCRIPTOR_SIG 0x8074B50
+#define END_RECORD_SIG 0x6054B50
+
+enum mmz_Flag {
+    mmz_Flag_Encrypted = 1 << 0,
+    mmz_Flag_DataDescriptor = 1 << 3,
+    mmz_Flag_StrongEncrypted = 1 << 6,
+};
+
+enum mmz_Compression {
+    mmz_Compression_None = 0,
+    mmz_Compression_Deflate = 8,
+};
+
+// 30 bytes (0x1E)
+#pragma pack(push,1)
+typedef struct mmz_LocalHeader {
+    uint32_t sig;
+    uint16_t version;
+    uint16_t flags;
+    uint16_t compression;
+    uint16_t modtime;
+    uint16_t moddate;
+    uint32_t crc32;
+    uint32_t compressed_size;
+    uint32_t uncompressed_size;
+    uint16_t filename_len;
+    uint16_t extrafield_len;
+} mmz_LocalHeader;
+#pragma pack(pop)
+
+#pragma pack(push,1)
+typedef struct mmz_DataDescriptor {
+    uint32_t sig;
+    uint32_t crc32;
+    uint32_t compressed_size;
+    uint32_t uncompressed_size;
+} mmz_DataDescriptor;
+#pragma pack(pop)
+
+// 46 bytes (0x2E)
+#pragma pack(push,1)
+typedef struct mmz_FileHeader {
+    uint32_t sig;
+    uint16_t version;
+    uint16_t version_needed;
+    uint16_t flags;
+    uint16_t compression;
+    uint16_t modtime;
+    uint16_t moddate;
+    uint32_t crc32;
+    uint32_t compressed_size;
+    uint32_t uncompressed_size;
+    uint16_t filename_len;
+    uint16_t extrafield_len;
+    uint16_t filecomment_len;
+    uint16_t disk_start; // wat
+    uint16_t internal_attr; // wat
+    uint32_t external_attr; // wat
+    uint32_t local_hdr_off;
+} mmz_FileHeader;
+#pragma pack(pop)
+
+#pragma pack(push,1)
+typedef struct mmz_EndRecord {
+    uint32_t sig;
+    uint16_t disk_number;
+    uint16_t disk_wcd;
+    uint16_t disk_entries;
+    uint16_t total_entries;
+    uint32_t central_directory_size;
+    uint32_t file_hdr_off;
+    uint16_t comment_len;
+} mmz_EndRecord;
+#pragma pack(pop)
+
+struct FileEntry {
+    std::string path;
+    u16 flags;
+    u16 compression_type;
+    u16 modtime;
+    u16 moddate;
+    u32 compressed_size; // may be zero.
+    u32 uncompressed_size; // may be zero.
+    u32 local_file_header_off;
+};
+
+struct DirectoryEntry {
+    std::string path;
+    std::vector<DirectoryEntry> dir_child;
+    std::vector<FileEntry> file_child;
+};
+
+using FileTableEntries = std::vector<FileEntry>;
+
+struct Device {
+    std::unique_ptr<BufferedFileData> source;
+    DirectoryEntry root;
+};
+
+struct Zfile {
+    z_stream z; // zlib stream.
+    Bytef* buffer; // buffer that compressed data is read into.
+    size_t buffer_size; // size of the above buffer.
+    size_t compressed_off; // offset of the compressed file.
+};
+
+struct File {
+    Device* device;
+    const FileEntry* entry;
+    Zfile zfile; // only used if the file is compressed.
+    size_t data_off; // offset of the file data.
+    size_t off;
+};
+
+struct Dir {
+    Device* device;
+    const DirectoryEntry* entry;
+    u32 index;
+};
+
+const FileEntry* find_file_entry(const DirectoryEntry& dir, std::string_view path) {
+    if (path.starts_with(dir.path)) {
+        // todo: check if / comes after file name in order to only check dirs.
+        for (const auto& e : dir.file_child) {
+            if (e.path == path) {
+                return &e;
+            }
+        }
+
+        for (const auto& e : dir.dir_child) {
+            if (auto entry = find_file_entry(e, path)) {
+                return entry;
+            }
+        }
+    }
+
+    return nullptr;
+}
+
+const DirectoryEntry* find_dir_entry(const DirectoryEntry& dir, std::string_view path) {
+    if (dir.path == path) {
+        return &dir;
+    }
+
+    if (path.starts_with(dir.path)) {
+        for (const auto& e : dir.dir_child) {
+            if (auto entry = find_dir_entry(e, path)) {
+                return entry;
+            }
+        }
+    }
+
+    return nullptr;
+}
+
+void set_stat_file(const FileEntry* entry, struct stat *st) {
+    std::memset(st, 0, sizeof(*st));
+
+    st->st_mode = S_IFREG | S_IRUSR | S_IRGRP | S_IROTH;
+    st->st_size = entry->uncompressed_size;
+    st->st_nlink = 1;
+
+    struct tm tm{};
+    tm.tm_sec = (entry->modtime & 0x1F) << 1;
+    tm.tm_min = (entry->modtime >> 5) & 0x3F;
+    tm.tm_hour = (entry->modtime >> 11);
+    tm.tm_mday = (entry->moddate & 0x1F);
+    tm.tm_mon = ((entry->moddate >> 5) & 0xF) - 1;
+    tm.tm_year = (entry->moddate >> 9) + 80;
+
+    st->st_atime = mktime(&tm);
+    st->st_mtime = st->st_atime;
+    st->st_ctime = st->st_atime;
+}
+
+int set_errno(struct _reent *r, int err) {
+    r->_errno = err;
+    return -1;
+}
+
+int devoptab_open(struct _reent *r, void *fileStruct, const char *_path, int flags, int mode) {
+    auto device = (Device*)r->deviceData;
+    auto file = static_cast<File*>(fileStruct);
+    std::memset(file, 0, sizeof(*file));
+
+    char path[FS_MAX_PATH]{};
+    if (!fix_path(_path, path)) {
+        return set_errno(r, ENOENT);
+    }
+
+    const auto entry = find_file_entry(device->root, path);
+    if (!entry) {
+        return set_errno(r, ENOENT);
+    }
+
+    if ((entry->flags & mmz_Flag_Encrypted) || (entry->flags & mmz_Flag_StrongEncrypted)) {
+        log_write("[ZIP] encrypted zip not supported\n");
+        return set_errno(r, ENOENT);
+    }
+
+    if (entry->compression_type != mmz_Compression_None && entry->compression_type != mmz_Compression_Deflate) {
+        log_write("[ZIP] unsuported compression type: %u\n", entry->compression_type);
+        return set_errno(r, ENOENT);
+    }
+
+    mmz_LocalHeader local_hdr{};
+    auto offset = entry->local_file_header_off;
+    if (R_FAILED(device->source->Read(&local_hdr, offset, sizeof(local_hdr)))) {
+        return set_errno(r, ENOENT);
+    }
+
+    if (local_hdr.sig != LOCAL_HEADER_SIG) {
+        return set_errno(r, ENOENT);
+    }
+
+    offset += sizeof(local_hdr) + local_hdr.filename_len + local_hdr.extrafield_len;
+
+    // todo: does a decs take prio over file header?
+    if (local_hdr.flags & mmz_Flag_DataDescriptor) {
+        mmz_DataDescriptor data_desc{};
+        if (R_FAILED(device->source->Read(&data_desc, offset, sizeof(data_desc)))) {
+            return set_errno(r, ENOENT);
+        }
+
+        if (data_desc.sig != DATA_DESCRIPTOR_SIG) {
+            return set_errno(r, ENOENT);
+        }
+
+        offset += sizeof(data_desc);
+    }
+
+    if (entry->compression_type == mmz_Compression_Deflate) {
+        auto& zfile = file->zfile;
+        zfile.buffer_size = 1024 * 64;
+        zfile.buffer = (Bytef*)calloc(1, zfile.buffer_size);
+        if (!zfile.buffer) {
+            return set_errno(r, ENOENT);
+        }
+
+        // skip zlib header.
+        if (Z_OK != inflateInit2(&zfile.z, -MAX_WBITS)) {
+            free(zfile.buffer);
+            zfile.buffer = nullptr;
+            return set_errno(r, ENOENT);
+        }
+    }
+
+    file->device = device;
+    file->entry = entry;
+    file->data_off = offset;
+    return r->_errno = 0;
+}
+
+int devoptab_close(struct _reent *r, void *fd) {
+    auto file = static_cast<File*>(fd);
+
+    if (file->entry->compression_type == mmz_Compression_Deflate) {
+        inflateEnd(&file->zfile.z);
+
+        if (file->zfile.buffer) {
+            free(file->zfile.buffer);
+        }
+    }
+
+    std::memset(file, 0, sizeof(*file));
+    return r->_errno = 0;
+}
+
+ssize_t devoptab_read(struct _reent *r, void *fd, char *ptr, size_t len) {
+    auto file = static_cast<File*>(fd);
+    len = std::min(len, file->entry->uncompressed_size - file->off);
+
+    if (!len) {
+        return 0;
+    }
+
+    if (file->entry->compression_type == mmz_Compression_None) {
+        if (R_FAILED(file->device->source->Read(ptr, file->data_off + file->off, len))) {
+            return set_errno(r, ENOENT);
+        }
+    } else if (file->entry->compression_type == mmz_Compression_Deflate) {
+        auto& zfile = file->zfile;
+        zfile.z.next_out = (Bytef*)ptr;
+        zfile.z.avail_out = len;
+
+        // run until we have inflated enough data.
+        while (zfile.z.avail_out) {
+            // check if we need to fetch more data.
+            if (!zfile.z.next_in || !zfile.z.avail_in) {
+                const auto clen = std::min(zfile.buffer_size, file->entry->compressed_size - zfile.compressed_off);
+                if (R_FAILED(file->device->source->Read(zfile.buffer, file->data_off + zfile.compressed_off, clen))) {
+                    return set_errno(r, ENOENT);
+                }
+
+                zfile.compressed_off += clen;
+                zfile.z.next_in = zfile.buffer;
+                zfile.z.avail_in = clen;
+            }
+
+            const auto rc = inflate(&zfile.z, Z_SYNC_FLUSH);
+            if (Z_OK != rc) {
+                if (Z_STREAM_END == rc) {
+                    len -= zfile.z.avail_out;
+                } else {
+                    log_write("[ZLIB] failed to inflate: %d %s\n", rc, zfile.z.msg);
+                    return set_errno(r, ENOENT);
+                }
+            }
+        }
+    }
+
+    file->off += len;
+    return len;
+}
+
+off_t devoptab_seek(struct _reent *r, void *fd, off_t pos, int dir) {
+    auto file = static_cast<File*>(fd);
+
+    // seek like normal.
+    if (file->entry->compression_type == mmz_Compression_None) {
+        if (dir == SEEK_CUR) {
+            pos += file->off;
+        } else if (dir == SEEK_END) {
+            pos = file->entry->uncompressed_size;
+        }
+    } else if (file->entry->compression_type == mmz_Compression_Deflate) {
+        // limited seek options.
+        // todo: support seeking to end and then back to orig position.
+        if (dir == SEEK_SET) {
+            if (pos == 0) {
+                inflateReset(&file->zfile.z);
+            } else if (pos != file->off) {
+                // seeking to the end is fine as it may be used to calc size.
+                if (pos != file->entry->uncompressed_size) {
+                    // random access seek is not supported.
+                    pos = file->off;
+                }
+            }
+        } else if (dir == SEEK_CUR) {
+            // random access seek is not supported.
+            pos = file->off;
+        } else if (dir == SEEK_END) {
+            pos = file->entry->uncompressed_size;
+        }
+    }
+
+    r->_errno = 0;
+    return file->off = std::clamp<u64>(pos, 0, file->entry->uncompressed_size);
+}
+
+int devoptab_fstat(struct _reent *r, void *fd, struct stat *st) {
+    auto file = static_cast<File*>(fd);
+
+    std::memset(st, 0, sizeof(*st));
+    set_stat_file(file->entry, st);
+    return r->_errno = 0;
+}
+
+DIR_ITER* devoptab_diropen(struct _reent *r, DIR_ITER *dirState, const char *_path) {
+    auto device = (Device*)r->deviceData;
+    auto dir = static_cast<Dir*>(dirState->dirStruct);
+    std::memset(dir, 0, sizeof(*dir));
+
+    char path[FS_MAX_PATH];
+    if (!fix_path(_path, path)) {
+        set_errno(r, ENOENT);
+        return NULL;
+    }
+
+    const auto entry = find_dir_entry(device->root, path);
+    if (!entry) {
+        set_errno(r, ENOENT);
+        return NULL;
+    }
+
+    dir->device = device;
+    dir->entry = entry;
+    r->_errno = 0;
+    return dirState;
+}
+
+int devoptab_dirreset(struct _reent *r, DIR_ITER *dirState) {
+    auto dir = static_cast<Dir*>(dirState->dirStruct);
+
+    dir->index = 0;
+
+    return r->_errno = 0;
+}
+
+int devoptab_dirnext(struct _reent *r, DIR_ITER *dirState, char *filename, struct stat *filestat) {
+    auto dir = static_cast<Dir*>(dirState->dirStruct);
+    std::memset(filestat, 0, sizeof(*filestat));
+
+    u32 index = dir->index;
+    if (index >= dir->entry->dir_child.size()) {
+        index -= dir->entry->dir_child.size();
+        if (index >= dir->entry->file_child.size()) {
+            return set_errno(r, ENOENT);
+        } else {
+            const auto& entry = dir->entry->file_child[index];
+            const auto rel_path = entry.path.substr(entry.path.find_last_of('/') + 1);
+
+            set_stat_file(&entry, filestat);
+            std::strcpy(filename, rel_path.c_str());
+        }
+
+    } else {
+        const auto& entry = dir->entry->dir_child[index];
+        const auto rel_path = entry.path.substr(entry.path.find_last_of('/') + 1);
+
+        filestat->st_mode = S_IFDIR | S_IRUSR | S_IRGRP | S_IROTH;
+        std::strcpy(filename, rel_path.c_str());
+    }
+
+    dir->index++;
+    return r->_errno = 0;
+}
+
+int devoptab_dirclose(struct _reent *r, DIR_ITER *dirState) {
+    auto dir = static_cast<Dir*>(dirState->dirStruct);
+    std::memset(dir, 0, sizeof(*dir));
+
+    return r->_errno = 0;
+}
+
+int devoptab_lstat(struct _reent *r, const char *_path, struct stat *st) {
+    auto device = (Device*)r->deviceData;
+
+    if (!device) {
+        return set_errno(r, ENOENT);
+    }
+
+    char path[FS_MAX_PATH];
+    if (!fix_path(_path, path)) {
+        return set_errno(r, ENOENT);
+    }
+
+    std::memset(st, 0, sizeof(*st));
+    st->st_nlink = 1;
+
+    if (find_dir_entry(device->root, path)) {
+        st->st_mode = S_IFDIR | S_IRUSR | S_IRGRP | S_IROTH;
+    } else if (auto entry = find_file_entry(device->root, path)) {
+        set_stat_file(entry, st);
+    } else {
+        log_write("[ZIP] didn't find in lstat\n");
+        return set_errno(r, ENOENT);
+    }
+
+    return r->_errno = 0;
+}
+
+constexpr devoptab_t DEVOPTAB = {
+    .structSize   = sizeof(File),
+    .open_r       = devoptab_open,
+    .close_r      = devoptab_close,
+    .read_r       = devoptab_read,
+    .seek_r       = devoptab_seek,
+    .fstat_r      = devoptab_fstat,
+    .stat_r       = devoptab_lstat,
+    .dirStateSize = sizeof(Dir),
+    .diropen_r    = devoptab_diropen,
+    .dirreset_r   = devoptab_dirreset,
+    .dirnext_r    = devoptab_dirnext,
+    .dirclose_r   = devoptab_dirclose,
+    .lstat_r      = devoptab_lstat,
+};
+
+auto BuildPath(const std::string& path) -> std::string {
+    if (path.starts_with('/')) {
+        return path;
+    }
+    return "/" + path;
+}
+
+void Parse(const FileTableEntries& entries, u32& index, DirectoryEntry& out) {
+    for (; index < entries.size(); index++) {
+        const auto path = BuildPath(entries[index].path);
+
+        // check if this path belongs to this dir.
+        if (!path.starts_with(out.path)) {
+            return;
+        }
+
+        if (path.ends_with('/')) {
+            auto& new_entry = out.dir_child.emplace_back();
+            new_entry.path = path.substr(0, path.length() - 1);
+
+            u32 new_index = index + 1;
+            Parse(entries, new_index, new_entry);
+            index = new_index - 1;
+        } else {
+            // check if this file actually belongs to this folder.
+            const auto idx = path.find_first_of('/', out.path.length() + 1);
+            const auto sub = path.substr(0, idx);
+
+            if (idx != path.npos && out.path != sub) {
+                auto& new_entry = out.dir_child.emplace_back();
+                new_entry.path = sub;
+                Parse(entries, index, new_entry);
+            } else {
+                auto& new_entry = out.file_child.emplace_back(entries[index]);
+                new_entry.path = path;
+            }
+        }
+    }
+}
+
+void Parse(const FileTableEntries& entries, DirectoryEntry& out) {
+    u32 index = 0;
+    out.path = "/"; // add root folder.
+    Parse(entries, index, out);
+}
+
+Result find_central_dir_offset(BufferedFileData* source, s64 size, mmz_EndRecord* record) {
+    // check if the record is at the end (no extra header).
+    auto offset = size - sizeof(*record);
+    R_TRY(source->Read(record, offset, sizeof(*record)));
+
+    if (record->sig == END_RECORD_SIG) {
+        R_SUCCEED();
+    }
+
+    // failed, find the sig by reading the last 64k and loop across it.
+    const auto rsize = std::min<u64>(UINT16_MAX, size);
+    offset = size - rsize;
+    std::vector<u8> data(rsize);
+    R_TRY(source->Read(data.data(), offset, data.size()));
+
+    // check in reverse order as it's more likely at the end.
+    for (s64 i = data.size() - sizeof(*record); i >= 0; i--) {
+        u32 sig;
+        std::memcpy(&sig, data.data() + i, sizeof(sig));
+        if (sig == END_RECORD_SIG) {
+            std::memcpy(record, data.data() + i, sizeof(*record));
+            R_SUCCEED();
+        }
+    }
+
+    R_THROW(0x1);
+}
+
+Result ParseZip(BufferedFileData* source, s64 size, FileTableEntries& out) {
+    mmz_EndRecord end_rec;
+    R_TRY(find_central_dir_offset(source, size, &end_rec));
+
+    out.reserve(end_rec.total_entries);
+    auto file_header_off = end_rec.file_hdr_off;
+
+    for (u16 i = 0; i < end_rec.total_entries; i++) {
+        // read the file header.
+        mmz_FileHeader file_hdr{};
+        R_TRY(source->Read(&file_hdr, file_header_off, sizeof(file_hdr)));
+
+        if (file_hdr.sig != FILE_HEADER_SIG) {
+            log_write("[ZIP] invalid file record\n");
+            R_THROW(0x1);
+        }
+
+        // save all the data hat we care about.
+        auto& new_entry = out.emplace_back();
+        new_entry.flags = file_hdr.flags;
+        new_entry.compression_type = file_hdr.compression;
+        new_entry.modtime = file_hdr.modtime;
+        new_entry.moddate = file_hdr.moddate;
+        new_entry.compressed_size = file_hdr.compressed_size;
+        new_entry.uncompressed_size = file_hdr.uncompressed_size;
+        new_entry.local_file_header_off = file_hdr.local_hdr_off;
+
+        // read the file name.
+        const auto filename_off = file_header_off + sizeof(file_hdr);
+        new_entry.path.resize(file_hdr.filename_len);
+        R_TRY(source->Read(new_entry.path.data(), filename_off, new_entry.path.size()));
+
+        // advance the offset.
+        file_header_off += sizeof(file_hdr) + file_hdr.filename_len + file_hdr.extrafield_len + file_hdr.filecomment_len;
+    }
+
+    R_SUCCEED();
+}
+
+struct Entry {
+    Device device;
+    devoptab_t devoptab;
+    fs::FsPath path;
+    fs::FsPath mount;
+    char name[32];
+    s32 ref_count;
+};
+
+Mutex g_mutex;
+std::vector<Entry> g_entries;
+u32 g_mount_idx;
+
+} // namespace
+
+Result MountZip(fs::Fs* fs, const fs::FsPath& path, fs::FsPath& out_path) {
+    SCOPED_MUTEX(&g_mutex);
+
+    // check if we already have the save mounted.
+    for (auto& e : g_entries) {
+        if (e.path == path) {
+            e.ref_count++;
+            out_path = e.mount;
+            R_SUCCEED();
+        }
+    }
+
+    // create new entry.
+    auto& entry = g_entries.emplace_back();
+    auto source = std::make_unique<yati::source::File>(fs, path);
+
+    s64 size;
+    R_TRY(source->GetSize(&size));
+    log_write("[ZIP] got size\n");
+
+    auto buffered = std::make_unique<BufferedFileData>(std::move(source), size);
+
+    FileTableEntries table_entries;
+    R_TRY(ParseZip(buffered.get(), size, table_entries));
+    log_write("[ZIP] parsed zip\n");
+
+    DirectoryEntry root;
+    Parse(table_entries, root);
+
+    entry.path = path;
+    entry.devoptab = DEVOPTAB;
+    entry.devoptab.name = entry.name;
+    entry.devoptab.deviceData = &entry.device;
+    entry.device.source = std::move(buffered);
+    entry.device.root = root;
+    std::snprintf(entry.name, sizeof(entry.name), "zip_%u", g_mount_idx);
+    std::snprintf(entry.mount, sizeof(entry.mount), "zip_%u:/", g_mount_idx);
+
+    R_UNLESS(AddDevice(&entry.devoptab) >= 0, 0x1);
+    log_write("[ZIP] DEVICE SUCCESS %s %s\n", path.s, entry.name);
+
+    out_path = entry.mount;
+    entry.ref_count++;
+    g_mount_idx++;
+
+    R_SUCCEED();
+}
+
+void UmountZip(const fs::FsPath& mount) {
+    SCOPED_MUTEX(&g_mutex);
+
+    auto itr = std::ranges::find_if(g_entries, [&mount](auto& e){
+        return mount == e.mount;
+    });
+
+    if (itr == g_entries.end()) {
+        return;
+    }
+
+    if (itr->ref_count) {
+        itr->ref_count--;
+    }
+
+    if (!itr->ref_count) {
+        RemoveDevice(mount);
+        g_entries.erase(itr);
+    }
+}
+
+} // namespace sphaira::devoptab

sphaira/source/yati/source/file.cpp

@@ -11,4 +11,8 @@ Result File::Read(void* buf, s64 off, s64 size, u64* bytes_read) {
     return m_file.Read(off, buf, size, 0, bytes_read);
 }
 
+Result File::GetSize(s64* out) {
+    return m_file.GetSize(out);
+}
+
 } // namespace sphaira::yati::source