X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=scandir.cc;h=30b8d4572aa6eab67a9796018f8ba8c59310e6a4;hb=6ac0cb32a08c81ddb87cd3a1373e8e64e8757534;hp=16161870cfa0a3b796025f19f67215f4749326a3;hpb=d60ac6ffaac11851042a5978ac10f52293a50b50;p=cumulus.git diff --git a/scandir.cc b/scandir.cc index 1616187..30b8d45 100644 --- a/scandir.cc +++ b/scandir.cc @@ -15,11 +15,12 @@ #include #include #include +#include #include +#include #include "format.h" #include "store.h" -#include "tarstore.h" #include "sha1.h" using std::list; @@ -30,15 +31,45 @@ using std::ostream; static TarSegmentStore *tss = NULL; /* Buffer for holding a single block of data read from a file. */ -static const int LBS_BLOCK_SIZE = 1024 * 1024; +static const size_t LBS_BLOCK_SIZE = 1024 * 1024; static char *block_buf; -void scandir(const string& path, std::ostream& metadata); +static const size_t LBS_METADATA_BLOCK_SIZE = 65536; -/* Converts time to microseconds since the epoch. */ -int64_t encode_time(time_t time) +/* Contents of the root object. This will contain a set of indirect links to + * the metadata objects. */ +std::ostringstream metadata_root; + +/* Buffer for building up metadata. */ +std::ostringstream metadata; + +/* Keep track of all segments which are needed to reconstruct the snapshot. */ +std::set segment_list; + +void scandir(const string& path); + +/* Ensure contents of metadata are flushed to an object. */ +void metadata_flush() { - return (int64_t)time * 1000000; + string m = metadata.str(); + if (m.size() == 0) + return; + + /* Write current metadata information to a new object. */ + LbsObject *meta = new LbsObject; + meta->set_group("root"); + meta->set_data(m.data(), m.size()); + meta->write(tss); + meta->checksum(); + + /* Write a reference to this block in the root. */ + ObjectReference ref = meta->get_ref(); + metadata_root << "@" << ref.to_string() << "\n"; + segment_list.insert(ref.get_segment()); + + delete meta; + + metadata.str(""); } /* Read data from a file descriptor and return the amount of data read. A @@ -68,15 +99,15 @@ size_t file_read(int fd, char *buf, size_t maxlen) /* Read the contents of a file (specified by an open file descriptor) and copy * the data to the store. */ -void dumpfile(int fd, dictionary &file_info, ostream &metadata) +void dumpfile(int fd, dictionary &file_info) { struct stat stat_buf; fstat(fd, &stat_buf); int64_t size = 0; - list segment_list; + list object_list; if ((stat_buf.st_mode & S_IFMT) != S_IFREG) { - printf("file is no longer a regular file!\n"); + fprintf(stderr, "file is no longer a regular file!\n"); return; } @@ -93,8 +124,13 @@ void dumpfile(int fd, dictionary &file_info, ostream &metadata) hash.process(block_buf, bytes); // tarstore processing - string blockid = tss->write_object(block_buf, bytes, "data"); - segment_list.push_back(blockid); + LbsObject *o = new LbsObject; + o->set_group("data"); + o->set_data(block_buf, bytes); + o->write(tss); + object_list.push_back(o->get_name()); + segment_list.insert(o->get_ref().get_segment()); + delete o; size += bytes; } @@ -104,28 +140,33 @@ void dumpfile(int fd, dictionary &file_info, ostream &metadata) /* For files that only need to be broken apart into a few objects, store * the list of objects directly. For larger files, store the data * out-of-line and provide a pointer to the indrect object. */ - if (segment_list.size() < 8) { + if (object_list.size() < 8) { string blocklist = ""; - for (list::iterator i = segment_list.begin(); - i != segment_list.end(); ++i) { - if (i != segment_list.begin()) + for (list::iterator i = object_list.begin(); + i != object_list.end(); ++i) { + if (i != object_list.begin()) blocklist += " "; blocklist += *i; } file_info["data"] = blocklist; } else { string blocklist = ""; - for (list::iterator i = segment_list.begin(); - i != segment_list.end(); ++i) { + for (list::iterator i = object_list.begin(); + i != object_list.end(); ++i) { blocklist += *i + "\n"; } - string indirect = tss->write_object(blocklist.data(), blocklist.size(), - "indirect"); - file_info["data"] = "@" + indirect; + + LbsObject *i = new LbsObject; + i->set_group("indirect"); + i->set_data(blocklist.data(), blocklist.size()); + i->write(tss); + file_info["data"] = "@" + i->get_name(); + segment_list.insert(i->get_ref().get_segment()); + delete i; } } -void scanfile(const string& path, ostream &metadata) +void scanfile(const string& path) { int fd; long flags; @@ -176,16 +217,14 @@ void scanfile(const string& path, ostream &metadata) buf = new char[stat_buf.st_size + 2]; len = readlink(path.c_str(), buf, stat_buf.st_size + 1); if (len < 0) { - printf("error reading symlink: %m\n"); + fprintf(stderr, "error reading symlink: %m\n"); } else if (len <= stat_buf.st_size) { buf[len] = '\0'; - printf(" contents=%s\n", buf); + file_info["contents"] = uri_encode(buf); } else if (len > stat_buf.st_size) { - printf("error reading symlink: name truncated\n"); + fprintf(stderr, "error reading symlink: name truncated\n"); } - file_info["contents"] = uri_encode(buf); - delete[] buf; break; case S_IFREG: @@ -210,7 +249,7 @@ void scanfile(const string& path, ostream &metadata) fcntl(fd, F_SETFL, flags & ~O_NONBLOCK); file_info["size"] = encode_int(stat_buf.st_size); - dumpfile(fd, file_info, metadata); + dumpfile(fd, file_info); close(fd); break; @@ -225,23 +264,26 @@ void scanfile(const string& path, ostream &metadata) } file_info["type"] = string(1, inode_type); - metadata << "type: " << inode_type << "\n"; dict_output(metadata, file_info); metadata << "\n"; + // Break apart metadata listing if it becomes too large. + if (metadata.str().size() > LBS_METADATA_BLOCK_SIZE) + metadata_flush(); + // If we hit a directory, now that we've written the directory itself, // recursively scan the directory. if (recurse) - scandir(path, metadata); + scandir(path); } -void scandir(const string& path, ostream &metadata) +void scandir(const string& path) { DIR *dir = opendir(path.c_str()); if (dir == NULL) { - printf("Error: %m\n"); + fprintf(stderr, "Error: %m\n"); return; } @@ -259,7 +301,7 @@ void scandir(const string& path, ostream &metadata) for (vector::iterator i = contents.begin(); i != contents.end(); ++i) { const string& filename = *i; - scanfile(path + "/" + filename, metadata); + scanfile(path + "/" + filename); } closedir(dir); @@ -269,18 +311,41 @@ int main(int argc, char *argv[]) { block_buf = new char[LBS_BLOCK_SIZE]; - tss = new TarSegmentStore("."); + string backup_dest = "."; + + if (argc > 1) + backup_dest = argv[1]; - std::ostringstream metadata; + tss = new TarSegmentStore(backup_dest); + + string desc_filename = backup_dest + "/snapshot.lbs"; + std::ofstream descriptor(desc_filename.c_str()); try { - scanfile(".", metadata); + scanfile("."); } catch (IOException e) { fprintf(stderr, "IOException: %s\n", e.getError().c_str()); } - const string md = metadata.str(); - string root = tss->write_object(md.data(), md.size(), "root"); + metadata_flush(); + const string md = metadata_root.str(); + + LbsObject *root = new LbsObject; + root->set_group("root"); + root->set_data(md.data(), md.size()); + root->write(tss); + root->checksum(); + + segment_list.insert(root->get_ref().get_segment()); + descriptor << "root: " << root->get_ref().to_string() << "\n\n"; + + delete root; + + descriptor << "segments:\n"; + for (std::set::iterator i = segment_list.begin(); + i != segment_list.end(); ++i) { + descriptor << " " << *i << "\n"; + } tss->sync(); delete tss;