Streaming data into the CIF Parse function reduces memory down to 4.1GB - also looking to stream the parsed data directly into the database if feasable.

nrod/download.go

@@ -39,7 +39,7 @@ func NrodDownload(url string, cfg *helpers.Configuration) (*[]byte, error) {
 	}
 
 	// Yes, I know `readedData` is not proper English.  But readData reads more like a verb action.
-	readedData, err := nrodExtract(*resp)
+	readedData, err := nrodExtract(resp)
 	if err != nil {
 		log.Msg.Error("Unable to read response data")
 		return nil, err
@@ -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 {

nrod/streams.go

@@ -0,0 +1,60 @@
+package nrod
+
+import (
+	"compress/gzip"
+	"fmt"
+	"io"
+	"net/http"
+	"time"
+
+	"git.fjla.uk/owlboard/timetable-mgr/helpers"
+	"git.fjla.uk/owlboard/timetable-mgr/log"
+	"go.uber.org/zap"
+)
+
+// Downloads NROD Data and extracts if GZIP, returns a io.Reader
+func NrodStream(url string, cfg *helpers.Configuration) (io.ReadCloser, error) {
+	log.Msg.Debug("Fetching NROD data stream", zap.String("Request URL", url))
+
+	client := http.Client{
+		Timeout: time.Second * 300,
+	}
+
+	req, err := http.NewRequest("GET", url, nil)
+	if err != nil {
+		log.Msg.Error("Error creating HTTP Request", zap.Error(err))
+		return nil, err
+	}
+
+	req.Header.Add("Authorization", "Basic "+helpers.BasicAuth(cfg.NrodUser, cfg.NrodPass))
+
+	resp, err := client.Do(req)
+	if err != nil {
+		log.Msg.Error("Error carrying out HTTP Request", zap.Error(err), zap.Int("STATUS", resp.StatusCode))
+		return nil, err
+	}
+
+	if resp.StatusCode != http.StatusOK {
+		err := fmt.Errorf("unexpected status code: %d", resp.StatusCode)
+		log.Msg.Error("Non-successful status code", zap.Error(err))
+		return nil, err
+	}
+
+	// Run the data through the extractor function and return io.ReadCloser, error from
+	// directly
+	return NrodStreamExtract(resp)
+}
+
+func NrodStreamExtract(resp *http.Response) (io.ReadCloser, error) {
+	log.Msg.Debug("Extracting NROD Download")
+
+	log.Msg.Debug("Content Type", zap.String("Content-Encoding", resp.Header.Get("Content-Encoding")))
+
+	gzReader, err := gzip.NewReader(resp.Body)
+	if err != nil {
+		log.Msg.Warn("Unable to create GZIP Reader, data probably not gzipped")
+		return resp.Body, err
+	}
+
+	return gzReader, nil
+}