Compute checksums of segments and store them in the local database.

[cumulus.git] / scandir.cc

diff --git a/scandir.cc b/scandir.cc
index 6e672c6..82160a1 100644 (file)
--- a/scandir.cc
+++ b/scandir.cc
@@ -135,6 +135,7 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path,
 {
     int64_t size = 0;
     list<string> object_list;
+    const char *status = NULL;          /* Status indicator printed out */
 
     /* Look up this file in the old stat cache, if we can.  If the stat
      * information indicates that the file has not changed, do not bother
@@ -152,6 +153,7 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path,
             const ObjectReference &ref = *i;
             if (!db->IsAvailable(ref)) {
                 cached = false;
+                status = "repack";
                 break;
             }
         }
@@ -173,8 +175,6 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path,
     /* If the file is new or changed, we must read in the contents a block at a
      * time. */
     if (!cached) {
-        printf("    [new]\n");
-
         SHA1Checksum hash;
         while (true) {
             ssize_t bytes = file_read(fd, block_buf, LBS_BLOCK_SIZE);
@@ -211,10 +211,14 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path,
                  * Additionally, keep track of the age of the data by looking
                  * up the age of the block which was expired and using that
                  * instead of the current time. */
-                if (db->IsOldObject(block_csum, bytes, &block_age))
+                if (db->IsOldObject(block_csum, bytes, &block_age)) {
                     o->set_group("compacted");
-                else
+                    if (status == NULL)
+                        status = "partial";
+                } else {
                     o->set_group("data");
+                    status = "new";
+                }
 
                 o->set_data(block_buf, bytes);
                 o->write(tss);
@@ -227,11 +231,17 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path,
             segment_list.insert(ref.get_segment());
             db->UseObject(ref);
             size += bytes;
+
+            if (status == NULL)
+                status = "old";
         }
 
         file_info["checksum"] = hash.checksum_str();
     }
 
+    if (status != NULL)
+        printf("    [%s]\n", status);
+
     statcache->Save(path, &stat_buf, file_info["checksum"], object_list);
 
     /* For files that only need to be broken apart into a few objects, store
@@ -295,6 +305,13 @@ void dump_inode(const string& path,         // Path within snapshot
         file_info["group"] += " (" + uri_encode(grp->gr_name) + ")";
     }
 
+    if (stat_buf.st_nlink > 1 && (stat_buf.st_mode & S_IFMT) != S_IFDIR) {
+        file_info["links"] = encode_int(stat_buf.st_nlink);
+        file_info["inode"] = encode_int(major(stat_buf.st_dev))
+            + "/" + encode_int(minor(stat_buf.st_dev))
+            + "/" + encode_int(stat_buf.st_ino);
+    }
+
     char inode_type;
 
     switch (stat_buf.st_mode & S_IFMT) {
@@ -334,7 +351,6 @@ void dump_inode(const string& path,         // Path within snapshot
 
         file_size = dumpfile(fd, file_info, path, stat_buf);
         file_info["size"] = encode_int(file_size);
-        close(fd);
 
         if (file_size < 0)
             return;             // error occurred; do not dump file
@@ -562,7 +578,6 @@ void usage(const char *program)
 
 int main(int argc, char *argv[])
 {
-    string backup_source = ".";
     string backup_dest = "";
     string localdb_dir = "";
     string backup_scheme = "";
@@ -617,15 +632,14 @@ int main(int argc, char *argv[])
         }
     }
 
-    searches.push_back(".");
     if (optind == argc) {
-        add_include(".");
-    } else {
-        for (int i = optind; i < argc; i++)
-            add_include(argv[i]);
+        usage(argv[0]);
+        return 1;
     }
 
-    backup_source = argv[optind];
+    searches.push_back(".");
+    for (int i = optind; i < argc; i++)
+        add_include(argv[i]);
 
     if (backup_dest == "") {
         fprintf(stderr,
@@ -661,7 +675,6 @@ int main(int argc, char *argv[])
             printf("    %s\n", i->c_str());
     }
 
-    tss = new TarSegmentStore(backup_dest);
     block_buf = new char[LBS_BLOCK_SIZE];
 
     /* Store the time when the backup started, so it can be included in the
@@ -681,6 +694,8 @@ int main(int argc, char *argv[])
     db->Open(database_path.c_str(), desc_buf,
              backup_scheme.size() ? backup_scheme.c_str() : NULL);
 
+    tss = new TarSegmentStore(backup_dest, db);
+
     /* Initialize the stat cache, for skipping over unchanged files. */
     statcache = new StatCache;
     statcache->Open(localdb_dir.c_str(), desc_buf,
@@ -701,8 +716,6 @@ int main(int argc, char *argv[])
     string backup_root = root->get_ref().to_string();
     delete root;
 
-    db->Close();
-
     statcache->Close();
     delete statcache;
 
@@ -710,6 +723,8 @@ int main(int argc, char *argv[])
     tss->dump_stats();
     delete tss;
 
+    db->Close();
+
     /* Write a backup descriptor file, which says which segments are needed and
      * where to start to restore this snapshot.  The filename is based on the
      * current time. */