metaforce/hecl/include/HECL/Database.hpp

#ifndef HECLDATABASE_HPP
#define HECLDATABASE_HPP

#include <iterator>
#include <string>
#include <functional>
#include <vector>
#include <map>
#include <list>
#include <unordered_map>
#include <unordered_set>
#include <memory>
#include <atomic>
#include <fstream>
#include <stdint.h>
#include <assert.h>

#include <angelscript.h>
#include <Athena/IStreamReader.hpp>
#include <LogVisor/LogVisor.hpp>

#include "HECL.hpp"

namespace HECL
{
namespace Database
{
class Project;

extern AngelScript::asIScriptEngine* asENGINE;
void InitASEngine();

template <class ASCLASS>
class ASType
{
    static void Constructor(ASCLASS* self)
    {
        new(self) ASCLASS();
    }
    static void Destructor(ASCLASS* self)
    {
        self->~ASCLASS();
    }
    const char* m_name;
    int m_typeid;
public:
    ASType(const char* namesp, const char* name) : m_name(name)
    {
        InitASEngine();
        assert(asENGINE->SetDefaultNamespace(namesp) >= 0);
        assert((m_typeid = asENGINE->RegisterObjectType(name, sizeof(ASCLASS), AngelScript::asOBJ_VALUE)) >= 0);
        assert(asENGINE->RegisterObjectBehaviour(name, AngelScript::asBEHAVE_CONSTRUCT, "void f()",
                                                 AngelScript::asFUNCTION(Constructor),
                                                 AngelScript::asCALL_CDECL_OBJLAST) >= 0);
        assert(asENGINE->RegisterObjectBehaviour(name, AngelScript::asBEHAVE_DESTRUCT, "void f()",
                                                 AngelScript::asFUNCTION(Destructor),
                                                 AngelScript::asCALL_CDECL_OBJLAST) >= 0);
    }
    inline const char* getName() const {return m_name;}
    inline int getTypeID() const {return m_typeid;}
};

template <class ASELEMCLASS>
class ASListType
{
    struct ASListInst
    {
        std::vector<ASELEMCLASS*> m_items;
        ASListInst(void* list)
        {
            AngelScript::asUINT count = *(AngelScript::asUINT*)list;
            ASELEMCLASS* items = (ASELEMCLASS*)((char*)list + 4);
            m_items.reserve(count);
            for (AngelScript::asUINT i=0 ; i<count ; ++i)
                m_items.push_back(&items[i]);
        }
    };
    static void ListConstructor(void* list, ASListInst* self)
    {
        new(self) ASListInst(list);
    }
    static void ListDestructor(ASListInst* self)
    {
        self->~ASListInst();
    }
    const char* m_name;
    const char* m_elemName;
    int m_typeid;
public:
    ASListType(const char* namesp, const char* name, const char* elemName) : m_name(name), m_elemName(elemName)
    {
        InitASEngine();
        assert(asENGINE->SetDefaultNamespace(namesp) >= 0);
        assert((m_typeid = asENGINE->RegisterObjectType(name, sizeof(ASListInst), AngelScript::asOBJ_VALUE)) >= 0);
        assert(asENGINE->RegisterObjectBehaviour(name, AngelScript::asBEHAVE_LIST_CONSTRUCT,
                                                 ("void f(int &in) {repeat " + std::string(elemName) + "}").c_str(),
                                                 AngelScript::asFUNCTION(ListConstructor),
                                                 AngelScript::asCALL_CDECL_OBJLAST) >= 0);
        assert(asENGINE->RegisterObjectBehaviour(name, AngelScript::asBEHAVE_DESTRUCT, "void f()",
                                                 AngelScript::asFUNCTION(ListDestructor),
                                                 AngelScript::asCALL_CDECL_OBJLAST) >= 0);
    }
    inline const char* getName() const {return m_name;}
    inline const char* getElemName() const {return m_elemName;}
    inline int getTypeID() const {return m_typeid;}
    inline std::vector<ASELEMCLASS*>& vectorCast(void* addr) const
    {return static_cast<ASListInst*>(addr)->m_items;}
    inline const std::vector<ASELEMCLASS*>& vectorCast(const void* addr) const
    {return static_cast<const ASListInst*>(addr)->m_items;}
};

struct ASStringType : ASType<std::string>
{ASStringType();};
extern ASStringType asSTRINGTYPE;

/**
 * @brief AngelScript Module Wrapper designed for HECL usage
 *
 * Behaves similarly to unique_ptr. Move construction/assignment only
 * Implicit conversion is supplied for seamless usage as a module reference
 */
class ASUniqueModule
{
    AngelScript::asIScriptModule* m_mod;
    ASUniqueModule() {m_mod = nullptr;}
    ASUniqueModule(AngelScript::asIScriptModule* mod) : m_mod(mod) {}
public:
    ~ASUniqueModule() {if (m_mod) m_mod->Discard();}
    ASUniqueModule(ASUniqueModule&& other) = default;
    ASUniqueModule(const ASUniqueModule& other) = delete;
    ASUniqueModule& operator=(ASUniqueModule&& other) = default;
    ASUniqueModule& operator=(const ASUniqueModule& other) = delete;
    inline operator AngelScript::asIScriptModule&() {return *m_mod;}
    inline operator bool() {return m_mod != nullptr;}
    static ASUniqueModule CreateFromCode(const char* module, const char* code)
    {
        AngelScript::asIScriptModule* mod = asENGINE->GetModule(module, AngelScript::asGM_ALWAYS_CREATE);
        assert(mod);
        assert(mod->AddScriptSection(module, code) >= 0);
        if (mod->Build() >= 0)
            return ASUniqueModule(mod);
        else
            return ASUniqueModule();
    }
    static ASUniqueModule CreateFromPath(const ProjectPath& path)
    {
        std::string asStr;
        std::ifstream(path.getAbsolutePath()) >> asStr;
        return CreateFromCode(path.getRelativePathUTF8().c_str(), asStr.c_str());
    }
};

extern LogVisor::LogModule LogModule;

/**
 * @brief Nodegraph class for gathering dependency-resolved objects for packaging
 */
class PackageDepsgraph
{
public:
    struct Node
    {
        enum
        {
            NODE_DATA,
            NODE_GROUP
        } type;
        ProjectPath path;
        ProjectPath cookedPath;
        class ObjectBase* projectObj;
        Node* sub;
        Node* next;
    };
private:
    friend class Project;
    std::vector<Node> m_nodes;
public:
    const Node* getRootNode() const {return &m_nodes[0];}
};

/**
 * @brief Subclassed by dataspec entries to manage per-game aspects of the data pipeline
 *
 * The DataSpec class manages interfaces for unpackaging, cooking, and packaging
 * of data for interacting with a specific system/game-engine.
 */
class IDataSpec
{
public:
    virtual ~IDataSpec() {}

    /**
     * @brief Extract Pass Info
     *
     * An extract pass iterates through a source package or image and
     * reverses the cooking process by emitting editable resources
     */
    struct ExtractPassInfo
    {
        SystemString srcpath;
        std::vector<SystemString> extractArgs;
        bool force;
    };

    /**
     * @brief Extract Report Representation
     *
     * Constructed by canExtract() to advise the user of the content about
     * to be extracted
     */
    struct ExtractReport
    {
        SystemString name;
        SystemString desc;
        std::vector<ExtractReport> childOpts;
    };

    typedef std::function<void(const HECL::SystemChar*, int, float)> FExtractProgress;

    virtual bool canExtract(Project&, const ExtractPassInfo& info, std::vector<ExtractReport>& reps)
    {(void)info;LogModule.report(LogVisor::Error, "not implemented");return false;}
    virtual void doExtract(Project&, const ExtractPassInfo& info, FExtractProgress progress)
    {(void)info;(void)progress;}

    /**
     * @brief Cook Task Info
     *
     * A cook task takes a single tracked path and generates the
     * corresponding cooked version
     */
    struct CookTaskInfo
    {
        ProjectPath path;
        ProjectPath cookedPath;
    };
    virtual bool canCook(const Project&, const CookTaskInfo& info,
                         SystemString& reasonNo)
    {(void)info;reasonNo=_S("not implemented");return false;}
    virtual void doCook(const Project&, const CookTaskInfo& info)
    {(void)info;}

    /**
     * @brief Package Pass Info
     *
     * A package pass performs last-minute queries of source resources,
     * gathers dependencies and packages cooked data together in the
     * most efficient form for the dataspec
     */
    struct PackagePassInfo
    {
        const PackageDepsgraph& depsgraph;
        ProjectPath subpath;
        ProjectPath outpath;
    };
    virtual bool canPackage(const Project&, const PackagePassInfo& info,
                            SystemString& reasonNo)
    {(void)info;reasonNo=_S("not implemented");return false;}
    virtual void gatherDependencies(const Project&, const PackagePassInfo& info,
                                    std::unordered_set<ProjectPath>& implicitsOut)
    {(void)info;(void)implicitsOut;}
    virtual void doPackage(const Project&, const PackagePassInfo& info)
    {(void)info;}
};

/**
 * @brief Pre-emptive indication of what the constructed DataSpec is used for
 */
enum DataSpecTool
{
    TOOL_EXTRACT,
    TOOL_COOK,
    TOOL_PACKAGE
};

extern std::vector<const struct DataSpecEntry*> DATA_SPEC_REGISTRY;

/**
 * @brief IDataSpec registry entry
 *
 * Auto-registers with data spec registry
 */
struct DataSpecEntry
{
    const SystemChar* m_name;
    const SystemChar* m_desc;
    std::function<IDataSpec*(DataSpecTool)> m_factory;

    DataSpecEntry(const SystemChar* name, const SystemChar* desc,
                  std::function<IDataSpec*(DataSpecTool)>&& factory)
    : m_name(std::move(name)), m_desc(std::move(desc)), m_factory(std::move(factory))
    {
        DATA_SPEC_REGISTRY.push_back(this);
    }
};

/**
 * @brief Base object to subclass for integrating with key project operations
 *
 * All project objects are provided with IDataObject pointers to their database
 * entries. Subclasses register themselves with a type registry so instances
 * are automatically constructed when performing operations like cooking and packaging.
 *
 * DO NOT CONSTRUCT THIS OR SUBCLASSES DIRECTLY!!
 */
class ObjectBase
{
    friend class Project;
    SystemString m_path;
protected:

    /**
     * @brief Byte-order of target system
     */
    enum DataEndianness
    {
        DE_NONE,
        DE_BIG, /**< Big-endian (PowerPC) */
        DE_LITTLE /**< Little-endian (Intel) */
    };

    /**
     * @brief Data-formats of target system
     */
    enum DataPlatform
    {
        DP_NONE,
        DP_GENERIC, /**< Scanline textures and 3-way shader bundle (GLSL, HLSL, SPIR-V) */
        DP_REVOLUTION, /**< Tiled textures and GX register buffers */
        DP_CAFE /**< Swizzled textures and R700 shader objects */
    };

    typedef std::function<void(const void* data, size_t len)> FDataAppender;

    /**
     * @brief Optional private method implemented by subclasses to cook objects
     * @param dataAppender subclass calls this function zero or more times to provide cooked-data linearly
     * @param endianness byte-order to target
     * @param platform data-formats to target
     * @return true if cook succeeded
     *
     * This method is called during IProject::cookPath().
     * Part of the cooking process may include embedding database-refs to dependencies.
     * This method should store the 64-bit value provided by IDataObject::id() when doing this.
     */
    virtual bool cookObject(FDataAppender dataAppender,
                            DataEndianness endianness, DataPlatform platform)
    {(void)dataAppender;(void)endianness;(void)platform;return true;}

    typedef std::function<void(ObjectBase*)> FDepAdder;

    /**
     * @brief Optional private method implemented by CProjectObject subclasses to resolve dependencies
     * @param depAdder subclass calls this function zero or more times to register each dependency
     *
     * This method is called during IProject::packagePath().
     * Dependencies registered via this method will eventually have this method called on themselves
     * as well. This is a non-recursive operation, no need for subclasses to implement recursion-control.
     */
    virtual void gatherDeps(FDepAdder depAdder)
    {(void)depAdder;}

    /**
     * @brief Get a packagable FourCC representation of the object's type
     * @return FourCC of the type
     */
    virtual FourCC getType() const
    {return FourCC("NULL");}

public:
    ObjectBase(const SystemString& path)
    : m_path(path) {}

    inline const SystemString& getPath() const {return m_path;}

};


/**
 * @brief Main project interface
 *
 * Projects are intermediate working directories used for staging
 * resources in their ideal editor-formats. This interface exposes all
 * primary operations to perform on a given project.
 */
class Project
{
public:
    struct ProjectDataSpec
    {
        const DataSpecEntry& spec;
        ProjectPath cookedPath;
        bool active;
    };
private:
    ProjectRootPath m_rootPath;
    ProjectPath m_dotPath;
    ProjectPath m_cookedRoot;
    std::vector<ProjectDataSpec> m_compiledSpecs;
public:
    Project(const HECL::ProjectRootPath& rootPath);

    /**
     * @brief Configuration file handle
     *
     * Holds a path to a line-delimited textual configuration file;
     * opening a locked handle for read/write transactions
     */
    class ConfigFile
    {
        SystemString m_filepath;
        std::list<std::string> m_lines;
        FILE* m_lockedFile = NULL;
    public:
        ConfigFile(const Project& project, const SystemString& name,
                   const SystemString& subdir=_S("/.hecl/"));
        std::list<std::string>& lockAndRead();
        void addLine(const std::string& line);
        void removeLine(const std::string& refLine);
        bool checkForLine(const std::string& refLine);
        void unlockAndDiscard();
        bool unlockAndCommit();
    };
    ConfigFile m_specs;
    ConfigFile m_paths;
    ConfigFile m_groups;

    /**
     * @brief Internal packagePath() exception
     *
     * Due to the recursive nature of packagePath(), there are potential
     * pitfalls like infinite-recursion. HECL throws this whenever there
     * are uncooked dependencies or if the maximum dependency-recursion
     * limit is exceeded.
     */
    class PackageException : public std::runtime_error {};

    /**
     * @brief A rough description of how 'expensive' a given cook operation is
     *
     * This is used to provide pretty colors during the cook operation
     */
    enum Cost
    {
        C_NONE,
        C_LIGHT,
        C_MEDIUM,
        C_HEAVY
    };

    /**
     * @brief Get the path of the project's root-directory
     * @return project root path
     *
     * Self explanatory
     */
    inline const ProjectRootPath& getProjectRootPath() const {return m_rootPath;}

    /**
     * @brief Get the path of project's cooked directory for a specific DataSpec
     * @param DataSpec to retrieve path for
     * @return project cooked path
     *
     * The cooked path matches the directory layout of the working directory,
     * except data is
     */
    inline const ProjectPath& getProjectCookedPath(const DataSpecEntry& spec) const
    {
        for (const ProjectDataSpec& sp : m_compiledSpecs)
            if (&sp.spec == &spec)
                return sp.cookedPath;
        LogModule.report(LogVisor::FatalError, "Unable to find spec '%s'", spec.m_name);
        return m_cookedRoot;
    }

    /**
     * @brief Add given file(s) to the database
     * @param path file or pattern within project
     * @return true on success
     *
     * This method blocks while object hashing takes place
     */
    bool addPaths(const std::vector<ProjectPath>& paths);

    /**
     * @brief Remove a given file or file-pattern from the database
     * @param paths file(s) or pattern(s) within project
     * @param recursive traverse into matched subdirectories
     * @return true on success
     *
     * This method will not delete actual working files from the project
     * directory. It will delete associated cooked objects though.
     */
    bool removePaths(const std::vector<ProjectPath>& paths, bool recursive=false);

    /**
     * @brief Register a working sub-directory as a Dependency Group
     * @param path directory to register as Dependency Group
     * @return true on success
     *
     * Dependency Groups are used at runtime to stage burst load-transactions.
     * They may only be added to directories and will automatically claim
     * subdirectories as well.
     *
     * Cooked objects in dependency groups will be packaged contiguously
     * and automatically duplicated if shared with other dependency groups.
     * This contiguous storage makes for optimal loading from slow block-devices
     * like optical drives.
     */
    bool addGroup(const ProjectPath& path);

    /**
     * @brief Unregister a working sub-directory as a dependency group
     * @param path directory to unregister as Dependency Group
     * @return true on success
     */
    bool removeGroup(const ProjectPath& path);

    /**
     * @brief Re-reads the data store holding user's spec preferences
     *
     * Call periodically in a long-term use of the HECL::Database::Project class.
     * Install filesystem event-hooks if possible.
     */
    void rescanDataSpecs();

    /**
     * @brief Return map populated with dataspecs targetable by this project interface
     * @return Platform map with name-string keys and enable-status values
     */
    inline const std::vector<ProjectDataSpec>& getDataSpecs() {return m_compiledSpecs;}

    /**
     * @brief Enable persistent user preference for particular spec string(s)
     * @param specs String(s) representing unique spec(s) from listDataSpecs
     * @return true on success
     */
    bool enableDataSpecs(const std::vector<SystemString>& specs);

    /**
     * @brief Disable persistent user preference for particular spec string(s)
     * @param specs String(s) representing unique spec(s) from listDataSpecs
     * @return true on success
     */
    bool disableDataSpecs(const std::vector<SystemString>& specs);

    /**
     * @brief Begin cook process for specified directory
     * @param path directory of intermediates to cook
     * @param feedbackCb a callback to run reporting cook-progress
     * @param recursive traverse subdirectories to cook as well
     * @return true on success
     *
     * Object cooking is generally an expensive process for large projects.
     * This method blocks execution during the procedure, with periodic
     * feedback delivered via feedbackCb.
     */
    bool cookPath(const ProjectPath& path,
                  std::function<void(SystemString&, Cost, unsigned)> feedbackCb,
                  bool recursive=false);

    /**
     * @brief Interrupts a cook in progress (call from SIGINT handler)
     *
     * Database corruption is bad! sqlite is pretty robust at avoiding data corruption,
     * but HECL spreads its data objects through the filesystem; this ensures that
     * those objects are cleanly finalized or discarded before stopping.
     *
     * Note that this method returns immediately; the resumed cookPath()
     * call will return as quickly as possible.
     */
    void interruptCook();

    /**
     * @brief Delete cooked objects for directory
     * @param path directory of intermediates to clean
     * @param recursive traverse subdirectories to clean as well
     * @return true on success
     *
     * Developers understand how useful 'clean' is. While ideally not required,
     * it's useful for verifying that a rebuild from ground-up is doable.
     */
    bool cleanPath(const ProjectPath& path, bool recursive=false);

    /**
     * @brief Constructs a full depsgraph of the project-subpath provided
     * @param path Subpath of project to root depsgraph at
     * @return Populated depsgraph ready to traverse
     */
    PackageDepsgraph buildPackageDepsgraph(const ProjectPath& path);

};

}
}

#endif // HECLDATABASE_HPP