Support for spreading objects across segments.

[cumulus.git] / store.h

diff --git a/store.h b/store.h
index 0cd2c4b..7629673 100644 (file)
--- a/store.h
+++ b/store.h
@@ -16,6 +16,8 @@
 #include <sstream>
 #include <vector>
 
+#include "sha1.h"
+
 /* In memory datatype to represent key/value pairs of information, such as file
  * metadata.  Currently implemented as map<string, string>. */
 typedef std::map<std::string, std::string> dictionary;
@@ -112,6 +114,26 @@ private:
     OutputStream &real;
 };
 
+/* Like WrapperOutputStream, but additionally computes a checksum of data as it
+ * is written. */
+class ChecksumOutputStream : public OutputStream {
+public:
+    explicit ChecksumOutputStream(OutputStream &o);
+    virtual ~ChecksumOutputStream() { }
+
+    /* Once a checksum is computed, no further data should be written to the
+     * stream. */
+    const uint8_t *finish_and_checksum();
+    size_t checksum_size() const { return csum.checksum_size(); }
+
+protected:
+    virtual void write_internal(const void *data, size_t len);
+
+private:
+    OutputStream &real;
+    SHA1Checksum csum;
+};
+
 /* Simple wrappers that encode integers using a StringOutputStream and return
  * the encoded result. */
 std::string encode_u16(uint16_t val);
@@ -139,6 +161,9 @@ public:
     OutputStream *new_object();
     void finish_object();
 
+    // Determine size of segment data written out so far.
+    size_t get_size() const { return raw_out->get_pos(); }
+
     // Utility functions for generating and formatting UUIDs for display.
     static struct uuid generate_uuid();
     static std::string format_uuid(const struct uuid u);
@@ -146,7 +171,8 @@ public:
 private:
     typedef std::vector<std::pair<int64_t, int64_t> > object_table;
 
-    OutputStream *out;
+    ChecksumOutputStream *out;  // Output stream with checksumming enabled
+    OutputStream *raw_out;      // Raw output stream, without checksumming
     struct uuid id;
 
     int64_t object_start_offset;
@@ -170,4 +196,23 @@ private:
     std::string directory;
 };
 
+/* A SegmentPartitioner helps to divide objects up among a collection of
+ * segments to meet a rough size limit per segment.  Like a SegmentWriter, only
+ * one object should be written at a time; however, multiple
+ * SegmentPartitioners can be created using the same base SegmentStore. */
+class SegmentPartitioner {
+public:
+    explicit SegmentPartitioner(SegmentStore *s);
+    ~SegmentPartitioner();
+
+    OutputStream *new_object();
+
+private:
+    size_t target_size;
+
+    SegmentStore *store;
+    SegmentWriter *segment;
+    OutputStream *object;
+};
+
 #endif // _LBS_STORE_H