| 45 | } |
| 46 | |
| 47 | Status BeginNextFile(SourceType** ret) { |
| 48 | ++files_cursor_; |
| 49 | |
| 50 | if (files_cursor_ >= source_.size()) { |
| 51 | return error::OutOfRange("All files completed"); |
| 52 | } |
| 53 | |
| 54 | current_ = &source_[files_cursor_]; |
| 55 | |
| 56 | FileSystem* fs = NULL; |
| 57 | Status s = env_->GetFileSystem(current_->path, &fs); |
| 58 | LOG_RETURN_IF_NOT_OK(s) |
| 59 | |
| 60 | if (! SingleThreadMode(current_->path)) { |
| 61 | uint64_t file_size = 0; |
| 62 | s = fs->GetRecordCount(current_->path, &file_size); |
| 63 | LOG_RETURN_IF_NOT_OK(s) |
| 64 | |
| 65 | int32_t slice_id = 0; |
| 66 | int32_t slice_count = 1; |
| 67 | if (IsDistributeShared(current_->path)) { |
| 68 | slice_id = env_->GetServerId() * thread_num_ + thread_id_; |
| 69 | slice_count = env_->GetServerCount() * thread_num_; |
| 70 | } else if (current_->local_shared) { |
| 71 | slice_id = thread_id_; |
| 72 | slice_count = thread_num_; |
| 73 | } |
| 74 | |
| 75 | DataSlicer slicer(slice_id, slice_count, file_size); |
| 76 | offset_ = slicer.LocalStart(); |
| 77 | end_ = offset_ + slicer.LocalSize(); |
| 78 | LOG(INFO) << "file_size:" << file_size |
| 79 | << "thread id:" << thread_id_ |
| 80 | << ", thread num:" << thread_num_ |
| 81 | << ", offset:" << offset_ |
| 82 | << ", end:" << end_; |
| 83 | s = fs->NewStructuredAccessFile(current_->path, offset_, end_, &reader_); |
| 84 | } else { |
| 85 | s = fs->NewStructuredAccessFile(current_->path, 0, 0, &reader_); |
| 86 | } |
| 87 | RETURN_IF_NOT_OK(s) |
| 88 | |
| 89 | // set schema for hdfsFS |
| 90 | std::vector<DataType> types; |
| 91 | types.push_back(kInt64); |
| 92 | if (std::is_same<typename std::decay<SourceType>::type, EdgeSource>::value){ |
| 93 | types.push_back(kInt64); |
| 94 | } |
| 95 | if (current_->IsWeighted()) { |
| 96 | types.push_back(kFloat); |
| 97 | } |
| 98 | if (current_->IsLabeled()) { |
| 99 | types.push_back(kInt32); |
| 100 | } |
| 101 | if (current_->IsAttributed()) { |
| 102 | types.push_back(kString); |
| 103 | } |
| 104 | reader_->SetSchema(types); |
nothing calls this directly
no test coverage detected