Diffstat (limited to 'main.go')
| -rw-r--r-- | main.go | 174 |
1 files changed, 147 insertions, 27 deletions
@@ -2,10 +2,15 @@ package main import ( "bytes" + "encoding/csv" + "encoding/json" "flag" + "hash/crc32" + "io" "log/slog" "os" "os/signal" + "strconv" "strings" "sync" "syscall" @@ -16,8 +21,10 @@ import ( ) var ( - configPath = flag.String("c", "config.json", "Path to the configuration file") - sigTermChan = make(chan os.Signal, 1) + configPath = flag.String("c", "config.json", "Path to the configuration file") + sigTermChan = make(chan os.Signal, 1) + cacheMapMutex = &sync.RWMutex{} + cacheMap = make(map[string]map[uint32]struct{}) ) func main() { @@ -48,8 +55,10 @@ func main() { } converters := make([]converter.Converter, 0, len(cfg.Converters)) - var converterTypes []string + converterTypes := make([]string, 0, len(cfg.Converters)) + converterHashes := make([]uint32, 0, len(cfg.Converters)) for _, converterCfg := range cfg.Converters { + converterBytes, _ := json.Marshal(converterCfg) conv, err := converter.NewConverterMap[converterCfg.Type](&converterCfg) if err != nil { slog.Error("fail to initialize converter", slog.String("error", err.Error())) @@ -57,6 +66,7 @@ func main() { } converters = append(converters, conv) converterTypes = append(converterTypes, converterCfg.Type) + converterHashes = append(converterHashes, crc32.ChecksumIEEE(converterBytes)) } generalLogger := slog.With(slog.String("input_storage", cfg.Input.Storage.Type)) @@ -84,52 +94,134 @@ func main() { default: } + if cfg.Input.CacheProcessed { + cacheFile, err := os.OpenFile(cfg.Input.CacheProcessedCsvPath, os.O_CREATE|os.O_RDONLY, 0644) + if err != nil { + generalLogger.Error("fail to initialize cache file", slog.String("cache_path", cfg.Input.CacheProcessedCsvPath), slog.String("error", err.Error())) + os.Exit(1) + } + defer cacheFile.Close() + + csvReader := csv.NewReader(cacheFile) + for { + rec, err := csvReader.Read() + if err == io.EOF { + break + } + if err != nil { + generalLogger.Error("fail to initialize cache while reading file", slog.String("cache_path", cfg.Input.CacheProcessedCsvPath), slog.String("error", err.Error())) + os.Exit(1) + } + if len(rec) != 2 { + generalLogger.Error("fail to initialize cache while reading file", slog.Int("record_length", len(rec)), slog.String("error", "incorrect format: expected '{image-name},{semicolon-separated-processor-hashes}'")) + os.Exit(1) + } + hashes := strings.Split(rec[1], ";") + cacheMap[rec[0]] = make(map[uint32]struct{}) + for _, hash := range hashes { + hashUint, err := strconv.ParseUint(hash, 10, 32) + if err != nil { + generalLogger.Error("fail to initialize cache while reading file", slog.Int("record_length", len(rec)), slog.String("error", err.Error())) + } + cacheMap[rec[0]][uint32(hashUint)] = struct{}{} + } + } + } + + select { + case <-sigTermChan: + generalLogger.Info("exiting due to termination signal") + os.Exit(130) + default: + } + processSemaphore := make(chan struct{}, cfg.MaxProcessThreads) queueSemaphore := make(chan struct{}, cfg.MaxPreProcessThreads) var wg sync.WaitGroup wg.Add(fileCount) + processTerminating := false + for i, file := range files { queueSemaphore <- struct{}{} - go func(index int, inputName string) { - fileLogger := generalLogger.With(slog.String("input_path", inputName), slog.Int("file_index", index)) - threadSigTermChannel := make(chan os.Signal, 1) - signal.Notify(threadSigTermChannel, os.Interrupt, syscall.SIGTERM) + select { + case <-sigTermChan: + generalLogger.Info("exiting due to termination signal") + processTerminating = true + default: + } + go func(index int, inputName string, earlyTerminate bool) { defer func() { <-queueSemaphore; wg.Done() }() - select { - case <-threadSigTermChannel: - fileLogger.Info("exiting due to termination signal") + fileLogger := generalLogger.With(slog.String("input_path", inputName), slog.Int("file_index", index)) + + if earlyTerminate { + fileLogger.Info("skip processing file (process is terminating)") return - default: } - inputMetadata, err := inputClient.ReadMetadata(inputName) - if err != nil { - fileLogger.Warn("fail to read metadata of (supposedly existing) input file", slog.String("error", err.Error())) - return + var inputMetadata *input.MetadataStruct + + id := inputClient.ID(file) + cacheMapMutex.Lock() + if _, ok := cacheMap[id]; !ok { + cacheMap[id] = make(map[uint32]struct{}) } + cacheMapMutex.Unlock() convertersToLaunch := []int{} for j, conv := range converters { + if cfg.Input.CacheProcessed { + cacheMapMutex.RLock() + if _, ok := cacheMap[id][converterHashes[j]]; ok { + cacheMapMutex.RUnlock() + fileLogger.Info("skip already processed file (based on cache file containing it and processor)", + slog.String("file_id", id), + slog.Uint64("conv_hash", uint64(converterHashes[j])), + slog.Int("conv_index", j)) + continue + } + cacheMapMutex.RUnlock() + } outputName := conv.DeductOutputPath(inputName) originalInputHash := "" convLogger := fileLogger.With(slog.String("output_path", outputName), slog.Int("conv_index", j)) - if !cfg.ForceRewrite && !conv.IsMissing(outputName) { - outputMetadata, err := conv.ReadMetadata(outputName) + if inputMetadata == nil { + inputMetadata, err = inputClient.ReadMetadata(inputName) if err != nil { - convLogger.Warn("fail to read metadata of (supposedly existing) output file", slog.String("error", err.Error())) - continue + fileLogger.Warn("fail to read metadata of (supposedly existing) input file", slog.String("error", err.Error())) + return } - originalInputHash = outputMetadata.HashOriginal - if inputMetadata.Hash == originalInputHash { - convLogger.Info("skip already processed file (based on equal hash)", slog.String("input_hash", inputMetadata.Hash)) + } + + switch cfg.Converters[j].Output.RewriteOn { + case "Never": + if !conv.IsMissing(outputName) { + convLogger.Info("skip already existing file") continue } + case "UnequalHashInCache": + if !conv.IsMissing(outputName) { + outputMetadata, err := conv.ReadMetadata(outputName) + if err != nil { + convLogger.Warn("fail to read metadata of (supposedly existing) output file", slog.String("error", err.Error())) + continue + } + originalInputHash = outputMetadata.HashOriginal + if inputMetadata.Hash == originalInputHash { + convLogger.Info("skip already processed file (based on equal hash)", slog.String("input_hash", inputMetadata.Hash)) + cacheMapMutex.Lock() + cacheMap[id][converterHashes[j]] = struct{}{} + cacheMapMutex.Unlock() + continue + } + } + case "Always": } + convertersToLaunch = append(convertersToLaunch, j) } @@ -147,10 +239,14 @@ func main() { return } - fileContent := make([]byte, inputMetadata.Size) - if _, err = reader.Read(fileContent); err != nil { + fileContent, err := io.ReadAll(reader) + if err != nil { fileLogger.Warn("fail to read content of input file", slog.String("error", err.Error())) return + } else if int64(len(fileContent)) != inputMetadata.Size { + fileLogger.Warn("the downloaded file seems to be missing some of its content", + slog.Int("actual_size_bytes", len(fileContent)), + slog.Int64("expected_size_bytes", inputMetadata.Size)) } reader.Close() @@ -165,8 +261,11 @@ func main() { } convLogger.Info("successfully processed file", slog.String("input_hash", inputMetadata.Hash)) + cacheMapMutex.Lock() + cacheMap[id][converterHashes[convIndex]] = struct{}{} + cacheMapMutex.Unlock() } - }(i, file) + }(i, file, processTerminating) } wg.Wait() @@ -176,7 +275,28 @@ func main() { generalLogger.Info("exiting due to termination signal") os.Exit(130) default: - generalLogger.Info("all files processed successfully, exiting") - os.Exit(0) } + + generalLogger.Info("all files processed successfully") + + if cfg.Input.CacheProcessed { + generalLogger.Info("writing cache file") + cacheFile, err := os.OpenFile(cfg.Input.CacheProcessedCsvPath, os.O_TRUNC|os.O_CREATE|os.O_WRONLY, 0644) + if err != nil { + generalLogger.Error("error writing cache into file") + } else { + for id, hashes := range cacheMap { + var hashStrings = make([]string, 0, len(hashes)) + for hash := range hashes { + hashStrings = append(hashStrings, strconv.FormatUint(uint64(hash), 10)) + } + cacheFile.Write([]byte(id)) + cacheFile.Write([]byte{','}) + cacheFile.WriteString(strings.Join(hashStrings, ";")) + cacheFile.Write([]byte{'\n'}) + } + } + } + + os.Exit(0) } |