X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=store.h;h=6fec7634fe51fc3113612f75cf22aaf246a55336;hb=3cfc1643082e60fea72539033b7adb094d236b0a;hp=3725a9f3cd99ff5018e7d234eb58f78660dd8ce5;hpb=a81fd6242d4a4076b7c79213de8cc7e13a311e56;p=cumulus.git diff --git a/store.h b/store.h index 3725a9f..6fec763 100644 --- a/store.h +++ b/store.h @@ -2,19 +2,27 @@ * Copyright (C) 2006 Michael Vrable * * Backup data is stored in a collection of objects, which are grouped together - * into segments for storage purposes. This file provides interfaces for - * reading and writing objects and segments. */ + * into segments for storage purposes. This implementation of the object store + * is built on top of libtar, and represents segments as TAR files and objects + * as files within them. */ #ifndef _LBS_STORE_H #define _LBS_STORE_H #include +#include -#include +#include #include +#include #include +#include #include +#include "sha1.h" + +class LbsObject; + /* In memory datatype to represent key/value pairs of information, such as file * metadata. Currently implemented as map. */ typedef std::map dictionary; @@ -32,75 +40,113 @@ public: std::string getError() const { return error; } }; -/* OutputStream is an abstract interface for writing data without seeking. - * Output could be to a file, to an object within a segment, or even to a - * memory buffer to help serialize data. */ -class OutputStream { +/* A simple wrapper around a single TAR file to represent a segment. Objects + * may only be written out all at once, since the tar header must be written + * first; incremental writing is not supported. */ +class Tarfile { public: - OutputStream(); - virtual ~OutputStream() { } - - // Write the given data buffer - void write(const void *data, size_t len); - - // Return the total number of bytes written so far - int64_t get_pos() const { return bytes_written; } - - // Convenience functions for writing other data types. Values are always - // written out in little-endian order. - void write_u8(uint8_t val); - void write_u16(uint16_t val); - void write_u32(uint32_t val); - void write_u64(uint64_t val); - - void write_s32(int32_t val) { write_u32((uint32_t)val); } - void write_s64(int64_t val) { write_u64((uint64_t)val); } + Tarfile(const std::string &path, const std::string &segment); + ~Tarfile(); - void write_varint(uint64_t val); + void write_object(int id, const char *data, size_t len); - void write_string(const std::string &s); - void write_dictionary(const dictionary &d); + // Return an estimate of the size of the file. + size_t size_estimate() { return size; } -protected: - // Function which actually causes a write: must be overridden by - // implementation. - virtual void write_internal(const void *data, size_t len) = 0; + void internal_write_object(const std::string &path, + const char *data, size_t len); private: - int64_t bytes_written; + size_t size; + std::string segment_name; + std::ostringstream checksums; + TAR *t; }; -/* An OutputStream implementation which writes data to memory and returns the - * result as a string. */ -class StringOutputStream : public OutputStream { +class TarSegmentStore { public: - StringOutputStream(); - std::string contents() const { return buf.str(); } + // New segments will be stored in the given directory. + TarSegmentStore(const std::string &path) { this->path = path; } + ~TarSegmentStore() { sync(); } -protected: - virtual void write_internal(const void *data, size_t len); + // Writes an object to segment in the store, and returns the name + // (segment/object) to refer to it. The optional parameter group can be + // used to control object placement; objects with different group + // parameters are kept in separate segments. + std::string write_object(const char *data, size_t len, + const std::string &group = "", + const std::list &refs = norefs); + + // Ensure all segments have been fully written. + void sync(); private: - std::stringstream buf; + struct segment_info { + Tarfile *file; + std::string name; // UUID + std::set refs; // Other segments this one refers to + int count; // Objects written to this segment + }; + + std::string path; + std::map segments; + + // An empty list which can be used as an argument to write_object to + // indicate that this object depends on no others. + static std::list norefs; + + // Ensure that all segments in the given group have been fully written. + void close_segment(const std::string &group); + + // Parse an object reference string and return just the segment name + // portion. + std::string object_reference_to_segment(const std::string &object); }; -/* An OutputStream implementation which writes data via the C stdio layer. */ -class FileOutputStream : public OutputStream { +/* An in-memory representation of an object, which can be incrementally built + * before it is written out to a segment. */ +class LbsObject { public: - explicit FileOutputStream(FILE *file); - virtual ~FileOutputStream(); - -protected: - virtual void write_internal(const void *data, size_t len); + LbsObject(); + ~LbsObject(); + + // If an object is placed in a group, it will be written out to segments + // only containing other objects in the same group. A group name is simply + // a string. + //std::string get_group() const { return group; } + void set_group(const std::string &g) { group = g; } + + // Data in an object must be written all at once, and cannot be generated + // incrementally. Data can be an arbitrary block of binary data of any + // size. The pointer to the data need only remain valid until write() is + // called. + //const char *get_data() const { return data; } + //size_t get_data_len() const { return data_len; } + void set_data(const char *d, size_t len) { data = d; data_len = len; } + + // Write an object to a segment, thus making it permanent. This function + // can be called at most once. + void write(TarSegmentStore *store); + + // An object is assigned a permanent name once it has been written to a + // segment. Until that time, its name cannot be determined. + std::string get_name() const { return name; } + + // Logically, one object may reference other objects (such as a metadata + // listing referncing actual file data blocks). Such references should be + // noted explicitly. It may eventually be used to build up a tree of + // checksums for later verifying integrity. + void add_reference(const LbsObject *o); private: - FILE *f; -}; + std::string group; + const char *data; + size_t data_len; -/* Simple wrappers that encode integers using a StringOutputStream and return - * the encoded result. */ -std::string encode_u16(uint16_t val); -std::string encode_u32(uint32_t val); -std::string encode_u64(uint64_t val); + bool written; + std::string name; + + std::set refs; +}; #endif // _LBS_STORE_H