From 15f7cef45e82e54a039e5486f08134c4f9a05471 Mon Sep 17 00:00:00 2001
From: Michael Vrable <mvrable@cs.ucsd.edu>
Date: Wed, 7 Feb 2007 13:45:30 -0800
Subject: [PATCH] Tag objects with a 4-byte type code.

---
 scandir.cc |  8 +++++---
 store.cc   | 30 +++++++++++++++++-------------
 store.h    | 13 +++++++++----
 3 files changed, 31 insertions(+), 20 deletions(-)

diff --git a/scandir.cc b/scandir.cc
index c2a093f..e253527 100644
--- a/scandir.cc
+++ b/scandir.cc
@@ -82,7 +82,8 @@ void dumpfile(int fd, dictionary &file_info)
     struct uuid segment_uuid;
     int object_id;
     OutputStream *index_data = index_segment->new_object(&segment_uuid,
-                                                         &object_id);
+                                                         &object_id,
+                                                         "DREF");
 
     SHA1Checksum hash;
     while (true) {
@@ -95,7 +96,8 @@ void dumpfile(int fd, dictionary &file_info)
 
         hash.process(block_buf, bytes);
         OutputStream *block = data_segment->new_object(&block_segment_uuid,
-                                                       &block_object_id);
+                                                       &block_object_id,
+                                                       "DATA");
         block->write(block_buf, bytes);
         index_data->write_uuid(block_segment_uuid);
         index_data->write_u32(block_object_id);
@@ -250,7 +252,7 @@ int main(int argc, char *argv[])
 
     segment_store = new SegmentStore(".");
     SegmentWriter *sw = segment_store->new_segment();
-    info_dump = sw->new_object(NULL);
+    info_dump = sw->new_object(NULL, "ROOT");
 
     index_segment = new SegmentPartitioner(segment_store);
     data_segment = new SegmentPartitioner(segment_store);
diff --git a/store.cc b/store.cc
index e7c373c..0a833ab 100644
--- a/store.cc
+++ b/store.cc
@@ -223,8 +223,9 @@ SegmentWriter::~SegmentWriter()
 
     for (object_table::const_iterator i = objects.begin();
          i != objects.end(); ++i) {
-        out->write_s64(i->first);
-        out->write_s64(i->second);
+        out->write_s64(i->offset);
+        out->write_s64(i->size);
+        out->write(i->type, sizeof(i->type));
     }
 
     static const char signature2[] = "LBSEND";
@@ -243,18 +244,21 @@ SegmentWriter::~SegmentWriter()
     delete raw_out;
 }
 
-OutputStream *SegmentWriter::new_object(int *id)
+OutputStream *SegmentWriter::new_object(int *id, const char *type)
 {
     if (object_stream)
         finish_object();
 
-    object_start_offset = out->get_pos();
-    object_stream = new WrapperOutputStream(*out);
-
-    if (id != NULL) {
+    if (id != NULL)
         *id = objects.size();
-    }
 
+    struct index_info info;
+    info.offset = out->get_pos();
+    info.size = -1;             // Will be filled in when object is finished
+    strncpy(info.type, type, sizeof(info.type));
+    objects.push_back(info);
+
+    object_stream = new WrapperOutputStream(*out);
     return object_stream;
 }
 
@@ -262,9 +266,8 @@ void SegmentWriter::finish_object()
 {
     assert(object_stream != NULL);
 
-    // store (start, length) information for locating this object
-    objects.push_back(std::make_pair(object_start_offset,
-                                     object_stream->get_pos()));
+    // Fill in object size, which could not be stored at start
+    objects.back().size = object_stream->get_pos();
 
     delete object_stream;
     object_stream = NULL;
@@ -321,7 +324,8 @@ SegmentPartitioner::~SegmentPartitioner()
         delete segment;
 }
 
-OutputStream *SegmentPartitioner::new_object(struct uuid *uuid, int *id)
+OutputStream *SegmentPartitioner::new_object(struct uuid *uuid, int *id,
+                                             const char *type)
 {
     if (segment != NULL && segment->get_size() > target_size) {
         delete segment;
@@ -334,5 +338,5 @@ OutputStream *SegmentPartitioner::new_object(struct uuid *uuid, int *id)
     if (uuid != NULL)
         *uuid = segment->get_uuid();
 
-    return segment->new_object(id);
+    return segment->new_object(id, type);
 }
diff --git a/store.h b/store.h
index e1244f0..5d877c6 100644
--- a/store.h
+++ b/store.h
@@ -162,7 +162,7 @@ public:
     struct uuid get_uuid() const { return id; }
 
     // Start writing out a new object to this segment.
-    OutputStream *new_object(int *id);
+    OutputStream *new_object(int *id, const char *type);
     void finish_object();
 
     // Determine size of segment data written out so far.
@@ -173,13 +173,18 @@ public:
     static std::string format_uuid(const struct uuid u);
 
 private:
-    typedef std::vector<std::pair<int64_t, int64_t> > object_table;
+    struct index_info {
+        int64_t offset;         // File offset at which object starts
+        int64_t size;           // Size of object in bytes
+        char type[4];           // Object type code
+    };
+
+    typedef std::vector<struct index_info> object_table;
 
     ChecksumOutputStream *out;  // Output stream with checksumming enabled
     OutputStream *raw_out;      // Raw output stream, without checksumming
     struct uuid id;
 
-    int64_t object_start_offset;
     OutputStream *object_stream;
 
     object_table objects;
@@ -209,7 +214,7 @@ public:
     explicit SegmentPartitioner(SegmentStore *s);
     ~SegmentPartitioner();
 
-    OutputStream *new_object(struct uuid *uuid, int *id);
+    OutputStream *new_object(struct uuid *uuid, int *id, const char *type);
 
 private:
     size_t target_size;
-- 
2.20.1