#include "dir_hierarchy.h" #include "filelist_io.h" #define DEBUG_DIRECTORY_STORAGE 1 /******************************************************************************************************************/ /* Internal File Hierarchy Storage */ /******************************************************************************************************************/ template typename std::set::iterator erase_from_set(typename std::set& s,const typename std::set::iterator& it) { typename std::set::iterator tmp(it); ++tmp; s.erase(it) ; return tmp; } // This class handles the file hierarchy // A Mutex is used to ensure total coherence at this level. So only abstracted operations are allowed, // so that the hierarchy stays completely coherent between calls. InternalFileHierarchyStorage::InternalFileHierarchyStorage() : mRoot(0) { DirEntry *de = new DirEntry("") ; de->row=0; de->parent_index=0; de->dir_modtime=0; mNodes.push_back(de) ; } int InternalFileHierarchyStorage::parentRow(DirectoryStorage::EntryIndex e) { if(!checkIndex(e,FileStorageNode::TYPE_DIR | FileStorageNode::TYPE_FILE) || e==0) return -1 ; return mNodes[mNodes[e]->parent_index]->row; } // high level modification routines bool InternalFileHierarchyStorage::isIndexValid(DirectoryStorage::EntryIndex e) const { return e < mNodes.size() && mNodes[e] != NULL ; } bool InternalFileHierarchyStorage::stampDirectory(const DirectoryStorage::EntryIndex& indx) { if(!checkIndex(indx,FileStorageNode::TYPE_DIR)) return false; static_cast(mNodes[indx])->dir_modtime = time(NULL) ; return true; } bool InternalFileHierarchyStorage::updateSubDirectoryList(const DirectoryStorage::EntryIndex& indx,const std::map& subdirs) { if(!checkIndex(indx,FileStorageNode::TYPE_DIR)) return false; DirEntry& d(*static_cast(mNodes[indx])) ; std::map should_create(subdirs); for(uint32_t i=0;i(mNodes[d.subdirs[i]])->dir_name) == subdirs.end()) { std::cerr << "[directory storage] Removing subdirectory " << static_cast(mNodes[d.subdirs[i]])->dir_name << " with index " << d.subdirs[i] << std::endl; if( !removeDirectory(d.subdirs[i])) i++ ; } else { std::cerr << "[directory storage] Keeping existing subdirectory " << static_cast(mNodes[d.subdirs[i]])->dir_name << " with index " << d.subdirs[i] << std::endl; should_create.erase(static_cast(mNodes[d.subdirs[i]])->dir_name) ; ++i; } for(std::map::const_iterator it(should_create.begin());it!=should_create.end();++it) { std::cerr << "[directory storage] adding new subdirectory " << it->first << " at index " << mNodes.size() << std::endl; DirEntry *de = new DirEntry(it->first) ; de->row = mNodes.size(); de->parent_index = indx; de->dir_modtime = it->second; d.subdirs.push_back(mNodes.size()) ; mNodes.push_back(de) ; } return true; } bool InternalFileHierarchyStorage::removeDirectory(DirectoryStorage::EntryIndex indx) // no reference here! Very important. Otherwise, the messign we do inside can change the value of indx!! { // check that it's a directory if(!checkIndex(indx,FileStorageNode::TYPE_DIR)) return false; if(indx == 0) return nodeAccessError("checkIndex(): Cannot remove top level directory") ; #ifdef DEBUG_DIRECTORY_STORAGE std::cerr << "(--) Removing directory " << indx << std::endl; print(); #endif // remove from parent DirEntry& d(*static_cast(mNodes[indx])) ; DirEntry& parent_dir(*static_cast(mNodes[d.parent_index])); for(uint32_t i=0;i= mNodes.size() || mNodes[indx] == NULL) return nodeAccessError("checkIndex(): Node does not exist") ; if(! (mNodes[indx]->type() & type)) return nodeAccessError("checkIndex(): Node is of wrong type") ; return true; } bool InternalFileHierarchyStorage::updateSubFilesList(const DirectoryStorage::EntryIndex& indx,const std::map& subfiles,std::map& new_files) { if(!checkIndex(indx,FileStorageNode::TYPE_DIR)) return false; DirEntry& d(*static_cast(mNodes[indx])) ; new_files = subfiles ; // remove from new_files the ones that already exist and have a modf time that is not older. for(uint32_t i=0;i(mNodes[d.subfiles[i]])) ; std::map::const_iterator it = subfiles.find(f.file_name) ; if(it == subfiles.end()) // file does not exist anymore => delete { std::cerr << "[directory storage] removing non existing file " << f.file_name << " at index " << d.subfiles[i] << std::endl; delete mNodes[d.subfiles[i]] ; mNodes[d.subfiles[i]] = NULL ; d.subfiles[i] = d.subfiles[d.subfiles.size()-1] ; d.subfiles.pop_back(); continue; } if(it->second.modtime != f.file_modtime || it->second.size != f.file_size) // file is newer and/or has different size { f.file_hash.clear(); // hash needs recomputing f.file_modtime = it->second.modtime; f.file_size = it->second.size; } new_files.erase(f.file_name) ; ++i; } for(std::map::const_iterator it(new_files.begin());it!=new_files.end();++it) { std::cerr << "[directory storage] adding new file " << it->first << " at index " << mNodes.size() << std::endl; d.subfiles.push_back(mNodes.size()) ; mNodes.push_back(new FileEntry(it->first,it->second.size,it->second.modtime)); mNodes.back()->row = mNodes.size()-1; mNodes.back()->parent_index = indx; } return true; } bool InternalFileHierarchyStorage::updateHash(const DirectoryStorage::EntryIndex& file_index,const RsFileHash& hash) { if(!checkIndex(file_index,FileStorageNode::TYPE_FILE)) { std::cerr << "[directory storage] (EE) cannot update file at index " << file_index << ". Not a valid index, or not a file." << std::endl; return false; } std::cerr << "[directory storage] updating hash at index " << file_index << ", hash=" << hash << std::endl; RsFileHash& old_hash (static_cast(mNodes[file_index])->file_hash) ; old_hash = hash ; return true; } bool InternalFileHierarchyStorage::updateFile(const DirectoryStorage::EntryIndex& file_index,const RsFileHash& hash, const std::string& fname,uint64_t size, const time_t modf_time) { if(!checkIndex(file_index,FileStorageNode::TYPE_FILE)) { std::cerr << "[directory storage] (EE) cannot update file at index " << file_index << ". Not a valid index, or not a file." << std::endl; return false; } FileEntry& fe(*static_cast(mNodes[file_index])) ; std::cerr << "[directory storage] updating file entry at index " << file_index << ", name=" << fe.file_name << " size=" << fe.file_size << ", hash=" << fe.file_hash << std::endl; fe.file_hash = hash; fe.file_size = size; fe.file_modtime = modf_time; fe.file_name = fname; return true; } bool InternalFileHierarchyStorage::updateDirEntry(const DirectoryStorage::EntryIndex& indx,const std::string& dir_name,time_t most_recent_time,time_t dir_modtime,const std::vector& subdirs_array,const std::vector& subfiles_array) { if(!checkIndex(indx,FileStorageNode::TYPE_DIR)) { std::cerr << "[directory storage] (EE) cannot update dir at index " << indx << ". Not a valid index, or not an existing dir." << std::endl; return false; } DirEntry& d(*static_cast(mNodes[indx])) ; d.dir_most_recent_time = most_recent_time; d.dir_modtime = dir_modtime; d.dir_update_time = time(NULL); d.dir_name = dir_name; d.subfiles = subfiles_array ; d.subdirs = subdirs_array ; // check that all subdirs already exist. If not, create. for(uint32_t i=0;i= mNodes.size() ) mNodes.resize(subdirs_array[i]+1,NULL); FileStorageNode *& node(mNodes[subdirs_array[i]]); if(node != NULL && node->type() != FileStorageNode::TYPE_DIR) { delete node ; node = NULL ; } if(node == NULL) node = new DirEntry(""); ((DirEntry*&)node)->dir_parent_path = d.dir_parent_path + "/" + dir_name ; node->row = i ; node->parent_index = indx ; } for(uint32_t i=0;i= mNodes.size() ) mNodes.resize(subfiles_array[i]+1,NULL); FileStorageNode *& node(mNodes[subfiles_array[i]]); if(node != NULL && node->type() != FileStorageNode::TYPE_FILE) { delete node ; node = NULL ; } if(node == NULL) node = new FileEntry("",0,0); node->row = subdirs_array.size()+i ; node->parent_index = indx ; } return true; } bool InternalFileHierarchyStorage::getDirUpdateTS(const DirectoryStorage::EntryIndex& index,time_t& recurs_max_modf_TS,time_t& local_update_TS) { if(!checkIndex(index,FileStorageNode::TYPE_DIR)) { std::cerr << "[directory storage] (EE) cannot update TS for index " << index << ". Not a valid index or not a directory." << std::endl; return false; } DirEntry& d(*static_cast(mNodes[index])) ; recurs_max_modf_TS = d.dir_most_recent_time ; local_update_TS = d.dir_update_time ; return true; } bool InternalFileHierarchyStorage::setDirUpdateTS(const DirectoryStorage::EntryIndex& index,time_t& recurs_max_modf_TS,time_t& local_update_TS) { if(!checkIndex(index,FileStorageNode::TYPE_DIR)) { std::cerr << "[directory storage] (EE) cannot update TS for index " << index << ". Not a valid index or not a directory." << std::endl; return false; } DirEntry& d(*static_cast(mNodes[index])) ; d.dir_most_recent_time = recurs_max_modf_TS ; d.dir_update_time = local_update_TS ; return true; } // Do a complete recursive sweep over sub-directories and files, and update the lst modf TS. This could be also performed by a cleanup method. time_t InternalFileHierarchyStorage::recursUpdateLastModfTime(const DirectoryStorage::EntryIndex& dir_index) { DirEntry& d(*static_cast(mNodes[dir_index])) ; time_t largest_modf_time = d.dir_modtime ; for(uint32_t i=0;i(mNodes[d.subfiles[i]])->file_modtime) ; for(uint32_t i=0;i(mNodes[indx]) ; } const InternalFileHierarchyStorage::FileEntry *InternalFileHierarchyStorage::getFileEntry(DirectoryStorage::EntryIndex indx) const { if(!checkIndex(indx,FileStorageNode::TYPE_FILE)) return NULL ; return static_cast(mNodes[indx]) ; } uint32_t InternalFileHierarchyStorage::getType(DirectoryStorage::EntryIndex indx) const { if(checkIndex(indx,FileStorageNode::TYPE_FILE | FileStorageNode::TYPE_DIR)) return mNodes[indx]->type() ; else return FileStorageNode::TYPE_UNKNOWN; } DirectoryStorage::EntryIndex InternalFileHierarchyStorage::getSubFileIndex(DirectoryStorage::EntryIndex parent_index,uint32_t file_tab_index) { if(!checkIndex(parent_index,FileStorageNode::TYPE_DIR)) return DirectoryStorage::NO_INDEX; if(static_cast(mNodes[parent_index])->subfiles.size() <= file_tab_index) return DirectoryStorage::NO_INDEX; return static_cast(mNodes[parent_index])->subfiles[file_tab_index]; } DirectoryStorage::EntryIndex InternalFileHierarchyStorage::getSubDirIndex(DirectoryStorage::EntryIndex parent_index,uint32_t dir_tab_index) { if(!checkIndex(parent_index,FileStorageNode::TYPE_DIR)) return DirectoryStorage::NO_INDEX; if(static_cast(mNodes[parent_index])->subdirs.size() <= dir_tab_index) return DirectoryStorage::NO_INDEX; return static_cast(mNodes[parent_index])->subdirs[dir_tab_index]; } bool InternalFileHierarchyStorage::searchHash(const RsFileHash& hash,std::list& results) { std::map::const_iterator it = mFileHashes.find(hash); if( it != mFileHashes.end() ) { results.clear(); results.push_back(it->second) ; return true ; } else return false; } bool InternalFileHierarchyStorage::check(std::string& error_string) const // checks consistency of storage. { // recurs go through all entries, check that all std::vector hits(mNodes.size(),0) ; // count hits of children. Should be 1 for all in the end. Otherwise there's an error. hits[0] = 1 ; // because 0 is never the child of anyone for(uint32_t i=0;itype() == FileStorageNode::TYPE_DIR) { // stamp the kids const DirEntry& de = *static_cast(mNodes[i]) ; for(uint32_t j=0;j= mNodes.size()) { error_string = "Node child out of tab!" ; return false ; } if(hits[de.subdirs[j]] != 0) { error_string = "Double hit on a single node" ; return false; } hits[de.subdirs[j]] = 1; } for(uint32_t j=0;j= mNodes.size()) { error_string = "Node child out of tab!" ; return false ; } if(hits[de.subfiles[j]] != 0) { error_string = "Double hit on a single node" ; return false; } hits[de.subfiles[j]] = 1; } } for(uint32_t i=0;itype() == FileStorageNode::TYPE_DIR) { std::cerr << " Node " << i << ": type=" << mNodes[i]->type() << std::endl; ++ndirs; } else if(mNodes[i]->type() == FileStorageNode::TYPE_FILE) { std::cerr << " Node " << i << ": type=" << mNodes[i]->type() << std::endl; ++nfiles; } else { ++nunknown; std::cerr << "(EE) Error: unknown type node found!" << std::endl; } std::cerr << "Total nodes: " << mNodes.size() << " (" << nfiles << " files, " << ndirs << " dirs, " << nempty << " empty slots)" << std::endl; recursPrint(0,DirectoryStorage::EntryIndex(0)); } void InternalFileHierarchyStorage::recursPrint(int depth,DirectoryStorage::EntryIndex node) const { std::string indent(2*depth,' '); if(mNodes[node] == NULL) { std::cerr << "EMPTY NODE !!" << std::endl; return ; } DirEntry& d(*static_cast(mNodes[node])); std::cerr << indent << "dir:" << d.dir_name << ", modf time: " << d.dir_modtime << ", recurs_last_modf_time: " << d.dir_most_recent_time << ", parent: " << d.parent_index << ", row: " << d.row << ", subdirs: " ; for(uint32_t i=0;i(mNodes[d.subfiles[i]])); std::cerr << indent << " hash:" << f.file_hash << " ts:" << (uint64_t)f.file_modtime << " " << f.file_size << " " << f.file_name << ", parent: " << f.parent_index << ", row: " << f.row << std::endl; } } bool InternalFileHierarchyStorage::nodeAccessError(const std::string& s) { std::cerr << "(EE) InternalDirectoryStructure: ERROR: " << s << std::endl; return false ; } // Removes the given subdirectory from the parent node and all its pendign subdirs. Files are kept, and will go during the cleaning // phase. That allows to keep file information when moving them around. bool InternalFileHierarchyStorage::recursRemoveDirectory(DirectoryStorage::EntryIndex dir) { DirEntry& d(*static_cast(mNodes[dir])) ; for(uint32_t i=0;itype() == FileStorageNode::TYPE_FILE) { const FileEntry& fe(*static_cast(mNodes[i])) ; unsigned char *file_section_data = NULL ; uint32_t file_section_offset = 0 ; uint32_t file_section_size = 0; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_PARENT_INDEX ,(uint32_t)fe.parent_index)) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_ROW ,(uint32_t)fe.row )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_ENTRY_INDEX ,(uint32_t)i )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_FILE_NAME ,fe.file_name )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_FILE_SIZE ,fe.file_size )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_FILE_SHA1_HASH,fe.file_hash )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(file_section_data,file_section_size,file_section_offset,FILE_LIST_IO_TAG_MODIF_TS ,(uint32_t)fe.file_modtime)) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(buffer,buffer_size,buffer_offset,FILE_LIST_IO_TAG_LOCAL_FILE_ENTRY,file_section_data,file_section_offset)) throw std::runtime_error("Write error") ; free(file_section_data) ; } else if(mNodes[i] != NULL && mNodes[i]->type() == FileStorageNode::TYPE_DIR) { const DirEntry& de(*static_cast(mNodes[i])) ; unsigned char *dir_section_data = NULL ; uint32_t dir_section_offset = 0 ; uint32_t dir_section_size = 0; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_PARENT_INDEX ,(uint32_t)de.parent_index )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_ROW ,(uint32_t)de.row )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_ENTRY_INDEX ,(uint32_t)i )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_FILE_NAME ,de.dir_name )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_FILE_SIZE ,de.dir_parent_path )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_MODIF_TS ,(uint32_t)de.dir_modtime )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_UPDATE_TS ,(uint32_t)de.dir_update_time )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_RECURS_MODIF_TS,(uint32_t)de.dir_most_recent_time )) throw std::runtime_error("Write error") ; if(!FileListIO::writeField(dir_section_data,dir_section_size,dir_section_offset,FILE_LIST_IO_TAG_RAW_NUMBER,(uint32_t)de.subdirs.size())) throw std::runtime_error("Write error") ; for(uint32_t j=0;j= mNodes.size()) mNodes.resize(node_index+1,NULL) ; FileEntry *fe = new FileEntry(file_name,file_size,file_modtime,file_hash); fe->parent_index = parent_index ; fe->row = row ; mNodes[node_index] = fe ; } else if(FileListIO::readField(buffer,buffer_size,buffer_offset,FILE_LIST_IO_TAG_LOCAL_DIR_ENTRY,node_section_data,node_section_size)) { uint32_t node_index ; std::string dir_name ; std::string dir_parent_path ; uint32_t dir_modtime ; uint32_t dir_update_time ; uint32_t dir_most_recent_time ; uint32_t row ; uint32_t parent_index ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_PARENT_INDEX ,parent_index )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_ROW ,row )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_ENTRY_INDEX ,node_index )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_FILE_NAME ,dir_name )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_FILE_SIZE ,dir_parent_path )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_MODIF_TS ,dir_modtime )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_UPDATE_TS ,dir_update_time )) throw std::runtime_error("Read error") ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_RECURS_MODIF_TS,dir_most_recent_time )) throw std::runtime_error("Read error") ; if(node_index >= mNodes.size()) mNodes.resize(node_index+1,NULL) ; DirEntry *de = new DirEntry(dir_name) ; de->dir_name = dir_name ; de->dir_parent_path = dir_parent_path ; de->dir_modtime = dir_modtime ; de->dir_update_time = dir_update_time ; de->dir_most_recent_time = dir_most_recent_time ; de->parent_index = parent_index ; de->row = row ; uint32_t n_subdirs = 0 ; uint32_t n_subfiles = 0 ; if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_RAW_NUMBER,n_subdirs)) throw std::runtime_error("Read error") ; for(uint32_t j=0;jsubdirs.push_back(di) ; } if(!FileListIO::readField(node_section_data,node_section_size,node_section_offset,FILE_LIST_IO_TAG_RAW_NUMBER,n_subfiles)) throw std::runtime_error("Read error") ; for(uint32_t j=0;jsubfiles.push_back(fi) ; } mNodes[node_index] = de ; } else throw std::runtime_error("Unknown node section.") ; free(node_section_data) ; } free(buffer) ; return true ; } catch(std::exception& e) { std::cerr << "Error while reading: " << e.what() << std::endl; if(buffer != NULL) free(buffer) ; return false; } }