Re-implement processParsedCif() to reduce memory use by 10%. Further reductions are neccessary

cif/helpers.go

@@ -1,12 +1,10 @@
 package cif
 
 import (
-	"encoding/json"
 	"errors"
 	"os"
 	"time"
 
-	"git.fjla.uk/owlboard/go-types/pkg/upstreamApi"
 	"git.fjla.uk/owlboard/timetable-mgr/helpers"
 	"git.fjla.uk/owlboard/timetable-mgr/log"
 	"go.uber.org/zap"
@@ -108,34 +106,6 @@ func parseDaysRun(daysBinary *string) []string {
 	return result
 }
 
-func debugWriteFile(header *upstreamApi.JsonTimetableV1, schedule *[]upstreamApi.JsonScheduleV1) {
-	if helpers.Runtime == "debug" {
-		log.Msg.Debug("Writing CIF Header and Schedule elements to file")
-		filepath := "./cif_debug_data/"
-		filename := time.Now().In(londonTimezone).Format("2006-01-02_15:04:05_ParsedCIF")
-
-		data, err := json.MarshalIndent(map[string]interface{}{
-			"header":   header,
-			"schedule": schedule,
-		}, "", "  ")
-		if err != nil {
-			log.Msg.Error("Error marshalling data", zap.Error(err))
-			return
-		}
-
-		err = os.MkdirAll(filepath, 0777)
-		if err != nil {
-			log.Msg.Error("Error creating directory", zap.Error(err))
-			return
-		}
-
-		writeErr := os.WriteFile(filepath+filename+".json", data, 0777)
-		if writeErr != nil {
-			log.Msg.Error("Error writing debug file to disk", zap.Error(writeErr))
-		}
-	}
-}
-
 func debugWriteDownload(input *[]byte) {
 	if helpers.Runtime == "debug" {
 		log.Msg.Debug("Writing CIF Download to file")
@@ -148,9 +118,25 @@ func debugWriteDownload(input *[]byte) {
 			return
 		}
 
-		err = os.WriteFile(filepath+filename+".jsonl", *input, 0777)
+		file, err := os.Create(filepath + filename + ".jsonl")
 		if err != nil {
-			log.Msg.Error("Error writing debug file to disk", zap.Error(err))
+			log.Msg.Error("Error creating file", zap.Error(err))
+			return
+		}
+		defer file.Close()
+
+		// Write data to file in smaller chunks
+		bufferSize := 4096 // Adjust the buffer size as needed
+		for i := 0; i < len(*input); i += bufferSize {
+			end := i + bufferSize
+			if end > len(*input) {
+				end = len(*input)
+			}
+			_, err := file.Write((*input)[i:end])
+			if err != nil {
+				log.Msg.Error("Error writing data to file", zap.Error(err))
+				return
+			}
 		}
 	}
 }

cif/parse.go

@@ -3,6 +3,7 @@ package cif
 import (
 	"bytes"
 	"encoding/json"
+	"errors"
 
 	"git.fjla.uk/owlboard/go-types/pkg/upstreamApi"
 	"git.fjla.uk/owlboard/timetable-mgr/log"
@@ -10,55 +11,55 @@ import (
 )
 
 // Unmarshalls data into the correct types for processing
-func parseCifData(data []byte) (*parsedData, error) {
-	// Split the data into lines
-	lines := bytes.Split(data, []byte("\n"))
+func parseCifData(data *[]byte) (*parsedData, error) {
+	log.Msg.Debug("Starting CIF Data parsing")
+	if data == nil {
+		return nil, errors.New("unable to parse nil pointer")
+	}
 
-	// Initialise variable for the parsed data
+	// Initialise data structures
 	var parsed parsedData
 	parsed.assoc = make([]upstreamApi.JsonAssociationV1, 0)
 	parsed.sched = make([]upstreamApi.JsonScheduleV1, 0)
 
-	for _, line := range lines {
+	// Create JSON Decoder
+	decoder := json.NewDecoder(bytes.NewReader(*data))
 
-		// Skip empty lines to avoid logging errors when there is no error
-		if len(bytes.TrimSpace(line)) == 0 {
-			continue
-		}
-
-		// Map each line for processing
+	// Iterate over JSON Objects using stream decoder
+	for decoder.More() {
 		var obj map[string]json.RawMessage
-		if err := json.Unmarshal(line, &obj); err != nil {
-			log.Msg.Error("Error decoding line", zap.String("line", string(line)), zap.Error(err))
-			continue
+		if err := decoder.Decode(&obj); err != nil {
+			log.Msg.Error("Error decoding JSON String")
+			return nil, err
 		}
 
-		// Loop through the mapped data and unmarshal to the correct type
+		// Handle parsed data
 		for key, value := range obj {
 			switch key {
 			case "JsonTimetableV1":
 				var timetable upstreamApi.JsonTimetableV1
 				if err := json.Unmarshal(value, &timetable); err != nil {
-					log.Msg.Error("Unable to parse JSON Timetable", zap.Error(err), zap.String("line", string(value)))
+					log.Msg.Error("Error decoding JSONTimetableV1 object", zap.Error(err))
 					continue
 				}
 				parsed.header = timetable
 			case "JsonAssociationV1":
 				var association upstreamApi.JsonAssociationV1
 				if err := json.Unmarshal(value, &association); err != nil {
-					log.Msg.Error("Error decoding JSON Association", zap.Error(err))
+					log.Msg.Error("Error decoding JSONAssociationV1 object", zap.Error(err))
 					continue
 				}
 				parsed.assoc = append(parsed.assoc, association)
 			case "JsonScheduleV1":
 				var schedule upstreamApi.JsonScheduleV1
 				if err := json.Unmarshal(value, &schedule); err != nil {
-					log.Msg.Error("Error decoding JSON Schedule", zap.Error(err))
+					log.Msg.Error("Error decoding JSONScheduleV1 object", zap.Error(err))
 					continue
 				}
 				parsed.sched = append(parsed.sched, schedule)
 			}
 		}
 	}
+	log.Msg.Debug("CIF Parsing completed")
 	return &parsed, nil
 }

cif/process.go

@@ -12,40 +12,61 @@ import (
 
 // Processes parsed CIF data and applies the data to the database
 func processParsedCif(data *parsedData) error {
-	createTasks := make([]*upstreamApi.JsonScheduleV1, 0)
-	deleteTasks := make([]*upstreamApi.JsonScheduleV1, 0)
+	log.Msg.Debug("Starting CIF Processing")
 
-	for _, item := range data.sched {
-		switch item.TransactionType {
-		case "Delete":
-			deleteItem := item // Create new variable to ensure repetition of pointers
-			deleteTasks = append(deleteTasks, &deleteItem)
-		case "Create":
-			createItem := item // Create new variable to ensure repetition of pointers
-			createTasks = append(createTasks, &createItem)
-		default:
-			log.Msg.Error("Unknown transaction type in CIF Schedule", zap.String("TransactionType", item.TransactionType))
+	// Batch size for processing
+	batchSize := 750 // Needs optimising for better RAM use.  1000 ~ 5.7GB, 500 ~ 5.5GB, 750 ~ 5.2GB
+
+	// Process deletions in batches
+	for i := 0; i < len(data.sched); i += batchSize {
+		end := i + batchSize
+		if end > len(data.sched) {
+			end = len(data.sched)
+		}
+		deleteBatch := make([]*upstreamApi.JsonScheduleV1, 0)
+		for _, item := range data.sched[i:end] {
+			if item.TransactionType == "Delete" {
+				deleteItem := item
+				deleteBatch = append(deleteBatch, &deleteItem)
+			}
+		}
+		if len(deleteBatch) > 0 {
+			err := doDeletions(deleteBatch)
+			if err != nil {
+				log.Msg.Error("Error deleting CIF Entries", zap.Error(err))
+				return err
+			}
 		}
 	}
 
-	err := doDeletions(deleteTasks)
-	if err != nil {
-		log.Msg.Error("Error deleting CIF Entries", zap.Error(err))
-		return err
-	}
-	err = doCreations(createTasks)
-	if err != nil {
-		log.Msg.Error("Error creating CIF Entries", zap.Error(err))
-		return err
+	// Process creations in batches
+	for i := 0; i < len(data.sched); i += batchSize {
+		end := i + batchSize
+		if end > len(data.sched) {
+			end = len(data.sched)
+		}
+		createBatch := make([]*upstreamApi.JsonScheduleV1, 0)
+		for _, item := range data.sched[i:end] {
+			if item.TransactionType == "Create" {
+				createItem := item
+				createBatch = append(createBatch, &createItem)
+			}
+		}
+		if len(createBatch) > 0 {
+			err := doCreations(createBatch)
+			if err != nil {
+				log.Msg.Error("Error creating CIF Entries", zap.Error(err))
+				return err
+			}
+		}
 	}
 
+	log.Msg.Debug("CIF Processing complete")
 	return nil
 }
 
 // Create delete query types and pass to the function which batches the deletions
 func doDeletions(deletions []*upstreamApi.JsonScheduleV1) error {
-	log.Msg.Info("Preparing CIF update Delete tasks", zap.Int("Delete task count", len(deletions)))
-
 	deleteQueries := make([]database.DeleteQuery, 0)
 	for _, item := range deletions {
 		query := database.DeleteQuery{
@@ -68,8 +89,6 @@ func doDeletions(deletions []*upstreamApi.JsonScheduleV1) error {
 
 // Convert to the correct struct for the database and pass to the function which batches insertions
 func doCreations(creations []*upstreamApi.JsonScheduleV1) error {
-	log.Msg.Info("Preparing CIF update Create tasks", zap.Int("Create task count", len(creations)))
-
 	createDocuments := make([]database.Service, 0)
 	for _, item := range creations {
 		document, err := ConvertServiceType(item, false)

cif/update.go

@@ -27,7 +27,7 @@ func runCifFullDownload(cfg *helpers.Configuration) error {
 
 	// If debug mode is on, call debugWriteDownload
 	if helpers.Runtime == "debug" {
-		debugWriteDownload(&data)
+		debugWriteDownload(data)
 	}
 
 	// Parse CIF file
@@ -37,14 +37,12 @@ func runCifFullDownload(cfg *helpers.Configuration) error {
 		return err
 	}
 
+	// Make `data` a nil pointer as it is no longer required
+	data = nil
+
 	// Drop timetable collection
 	dbAccess.DropCollection(dbAccess.TimetableCollection) // I should edit this to prevent removal of VSTP entries in the database.
 
-	// If debug mode is on, call debugWriteFile
-	if helpers.Runtime == "debug" {
-		debugWriteFile(&parsed.header, &parsed.sched)
-	}
-
 	// Process CIF file
 	err = processParsedCif(parsed)
 	if err != nil {
@@ -77,7 +75,7 @@ func runCifUpdateDownload(cfg *helpers.Configuration, metadata *dbAccess.CifMeta
 
 		// If debug mode is on, call debugWriteDownload
 		if helpers.Runtime == "debug" {
-			debugWriteDownload(&data)
+			debugWriteDownload(data)
 		}
 
 		// Parse CIF file
@@ -87,11 +85,10 @@ func runCifUpdateDownload(cfg *helpers.Configuration, metadata *dbAccess.CifMeta
 			return err
 		}
 
-		// If debug mode is on, call debugWriteFile
-		if helpers.Runtime == "debug" {
-			debugWriteFile(&parsed.header, &parsed.sched)
-		}
+		// Make `data` a nil pointer as it is no longer required
+		data = nil
 
+		log.Msg.Debug("Starting metadata checks")
 		// Check CIF Sequence
 		// Skip if LastSequence is >= to this sequence
 		if metadata.LastSequence >= parsed.header.Metadata.Sequence {
@@ -105,6 +102,7 @@ func runCifUpdateDownload(cfg *helpers.Configuration, metadata *dbAccess.CifMeta
 			log.Msg.Error("CIF sequence not as expected", zap.Error(err), zap.Int64("LastSequence", metadata.LastSequence), zap.Int64("New Sequence", parsed.header.Metadata.Sequence))
 			return err
 		}
+		log.Msg.Debug("Metadata checks complete")
 
 		// Do further sequence checks - parsed.header.Metadata.Sequence MUST = metadata.LastSequence + 1
 
@@ -123,7 +121,7 @@ func runCifUpdateDownload(cfg *helpers.Configuration, metadata *dbAccess.CifMeta
 }
 
 // Wraps nrod.NrodDownload() into a function which can handle downloading data for a given day
-func fetchUpdate(t time.Time, cfg *helpers.Configuration) ([]byte, error) {
+func fetchUpdate(t time.Time, cfg *helpers.Configuration) (*[]byte, error) {
 	url, err := getUpdateUrl("daily")
 	if err != nil {
 		return nil, err

corpus/update.go

@@ -16,7 +16,7 @@ func RunCorpusUpdate(cfg *helpers.Configuration) error {
 		return err
 	}
 
-	unsortedCorpusData, err := parseCorpusData(&resp)
+	unsortedCorpusData, err := parseCorpusData(resp)
 	if err != nil {
 		log.Msg.Error("Error parsing Corpus data", zap.Error(err))
 		return err

dbAccess/cif.go

@@ -91,16 +91,12 @@ func DeleteCifEntries(deletions []database.DeleteQuery) error {
 		bulkDeletions = append(bulkDeletions, mongo.NewDeleteManyModel().SetFilter(filter))
 	}
 
-	log.Msg.Info("Running `Delete` tasks from CIF Update", zap.Int("Required deletions", len(deletions)))
-
-	result, err := collection.BulkWrite(context.Background(), bulkDeletions)
+	_, err := collection.BulkWrite(context.Background(), bulkDeletions)
 	if err != nil {
 		log.Msg.Error("Error deleting documents", zap.Error(err))
 		return err
 	}
 
-	log.Msg.Info("Deleted CIF Documents", zap.Int64("Deletion count", result.DeletedCount))
-
 	return nil
 }
 
@@ -123,14 +119,11 @@ func CreateCifEntries(schedules []database.Service) error {
 
 	bulkWriteOptions := options.BulkWrite().SetOrdered(false)
 
-	log.Msg.Info("Running `Create` tasks from CIF Update", zap.Int("Documents to create", len(schedules)))
-	result, err := collection.BulkWrite(context.Background(), models, bulkWriteOptions)
+	_, err := collection.BulkWrite(context.Background(), models, bulkWriteOptions)
 	if err != nil {
 		log.Msg.Error("Error inserting documents", zap.Error(err))
 		return err
 	}
 
-	log.Msg.Info("Inserted CIF Documents", zap.Int64("Insertion count", result.InsertedCount))
-
 	return nil
 }

nrod/download.go

@@ -13,7 +13,7 @@ import (
 )
 
 // Downloads NROD Data over HTTP from the given URL, extracted data is returned
-func NrodDownload(url string, cfg *helpers.Configuration) ([]byte, error) {
+func NrodDownload(url string, cfg *helpers.Configuration) (*[]byte, error) {
 	log.Msg.Debug("Fetching NROD data", zap.String("Request URL", url))
 	client := http.Client{
 		Timeout: time.Second * 10,
@@ -49,7 +49,7 @@ func NrodDownload(url string, cfg *helpers.Configuration) ([]byte, error) {
 }
 
 // Extracts GZIP Data from an HTTP Response and returns the decompresses data as a byte array
-func nrodExtract(resp http.Response) ([]byte, error) {
+func nrodExtract(resp http.Response) (*[]byte, error) {
 	log.Msg.Debug("Extracting HTTP Response Data")
 	gzReader, err := gzip.NewReader(resp.Body)
 	if err != nil {
@@ -59,7 +59,7 @@ func nrodExtract(resp http.Response) ([]byte, error) {
 			log.Msg.Error("Unable to read response body")
 			return nil, err
 		}
-		return data, nil
+		return &data, nil
 	}
 
 	defer gzReader.Close()
@@ -70,5 +70,5 @@ func nrodExtract(resp http.Response) ([]byte, error) {
 		log.Msg.Error("Failed to read GZIPped data", zap.Error(err))
 	}
 
-	return extractedData, nil
+	return &extractedData, nil
 }